Professional Documents
Culture Documents
Infrared Optical Materials and Their Antireflection Coatings
Infrared Optical Materials and Their Antireflection Coatings
j,
J A Savage
Royal Signals and Radar Establishment, Malvern
Savage, J .A.
Infrared optical materials and their antireflection
coatings.
1. Infrared spectroscopy 2. Optical instruments
I. Title
535.8'42 QC457
ISBN 0-85274-790-X
Preface ix
1 Introduction 1
References 233
Index 257
I:
Preface
Up until the 1940s the branch of science called optics dealt mainly with the
ultraviolet and visible parts of the spectrum. The need for optical materials
was mostly fulfilled by bulk glass and a few halide crystals. Since that time
a new generation of infrared optical systems has been researched and is now
being marketed. The developments in infrared lasers and detectors have
enabled many laser systems, thermal systems and glass fibre communi-
cations systems to be produced. The production of these systems has in turn
created a need for infrared optical materials transmitting electromagnetic
radiation essentially from the visible out to 14 f'm, thus spanning the near
middle and part of the far infrared spectrum. Thus materials such as silicon,
germanium, gallium arsenide, calcium aluminate glass, chalcogenide glass,
zinc sulphide, zinc selenide, the alkali and alkaline earth halides and silicate
glass fibres have all been researched and in many cases developed and pro-
duced as infrared optical materials during the last 25 years. Most bulk
infrared optical materials possess a high refractive index ( > 2) and there-
fore must be antireflection coated. Hence a parallel coatings development
has taken place, yielding high efficiency and ultra-durable coatings to a
technical standard not thought possible only a few years ago.
Workers have tended to concentrate on a particular area such as fibres,
bulk optical materials or antireflection coatings, but many of the problems
encountered in the research, development and production of these materials
are common, so that these different areas can be thought of as a continuum
within one field. Most of the data in this field is scattered in the literature
and it is timely that this is gathered together in a single volume, containing
a comprehensive reference list to allow the reader to delve more deeply into
any particular part of the subject as he or she wishes. The treatment of the
subject concentrates more on the optical and general physical properties
and the material science aspects of the materials, rather than the solid state
physics of them.
It is hoped that this volume, besides providing a useful summary of the
state of the art, will enable users to bec.ome more familiar with these new
infrared optical materials and encourage researchers and producers to
X Preface
continue to gather data on existing materials and find new ones to fill the
very obvious gaps in knowledge which exist at present.
I am indebted w Professor W T Welford for encouraging me to under-
take this work and for reading the typescript, to many of my friends and
colleagues for direct and indirect help given and finally w my wife Anna
and dog Grendel for their patience during the preparation of this volume.
J A Savage
Malvern
September 1984
1
Introduction
·The ability to synthesise materials and use them to perform various func-
tions has been a major factor in human technical progress. For instance the
cutting of wood and other vegetation, the breaking of ground and the reap-
ing of sown crops were difficult until the invention of metal tools. The
knowledge and use of simple metal synthesis technology allowed such tools
to be made, and initiated a great leap forward in man's quality of life.
At a much later stage in human progress, the understanding and use of more
advanced iron and steel synthesis technology has affected our lives in many
obvious ways to the point where iron and steel products are now taken for
granted.
We may now be said to be living in a silicon age which may become
known as an information technology age. The personal computer and glass
fibre cable transmission medium allow many new functions to be performed
and many existing ones to be performed more efficiently. Undoubtedly this
silicon age will have a major impact on our lives. By now most of us are
aware that the microchip has made this possible and that the glass fibre
cable using optical carrier waves is able to carry the large quantity of infor-
mation that microchip systems are able to generate. Thus there is a carrier
system available well suited to the task of information exchange on a vast
and hitherto uncontemplated scale. Perhaps even now we are beginning to
take this technology for granted. But how many of us are aware of the
importance of the physics, chemistry, metallurgy, glass and plastics tech-
nologies, commonly called materials science, involved in the production of
the microchip and the glass fibre? The manufacture of these items, at the
heart of current information technology, rests on the ability to synthesise
essentially two materials to standards of purity and perfection previously
not achieved in our human progress. These materials are the element silicon
and the compound silica (SiOz).
In practice the silicon is doped to control the electrical activity, and
dielectrics together with metals are used in creating a microchip. The silica
glass fibre composition is adjusted to allow it to perform its waveguiding
function by the addition of 10-20"7o of GeOz or PzOs. The exacting stand-
ards required for the synthesis of Si and Si02 to the quality required for
2 Introduction
information technology applications are not generally well known. Yet this
materials technology is at the beginning of the device and component
technology and the subsequent system or consumer product. Countries
which recognise the importance of materials technology and are able to
exploit it and the devices which spring from it to the full are at the leading
edge of technological progress. The struggle, for that is what it is, to achieve
the perfection and purity required for device and component use in some
cases can require a similar effort to that more familiar to scientists in
general of creating a whole system or consumer product. But that struggle
offers an opportunity for invention, creativity, involvement and personal
satisfaction in what the author has found to be one of the most fascinating
fields of human endeavour. The urge to communicate the fascination and
primary importance of materials science and technology to others is the
motivation for this work on the subject of infrared optical materials and
their antireflection coatings.
The manufacture of laser systems e.g. COz (Ream 1982, Hinman and
Cannizzo 1983) and thermal systems e.g. thermal imagers (Kuhl 1983) has
created the need for bulk 3-5 p.m and 8-12 p.m infrared optical materials
and coatings. Similarly the need for a low loss high information carrying
capacity medium has stimulated the development of the low Joss optical
fibre. In the initial region of the infrared spectrum from 0.75 p.m to 14 p.m
the absorptions of the minor atmospheric constituents, water vapour and
carbon dioxide, result in three main 'windows' in the atmosphere (Kruse et
a/1962); one from 0.75-2.5 p.m (near infrared), another from 3.0-5.0 p.m
(middle infrared) and a third from 7.5-14 p.m (far infrared) as shown in
figure 1.1. Image intensifiers operate in the near infrared and are able to use
~
80
"-
~
~ 60
~
·;;
~
~
40
.=" 20
0 2 4 6 8
Wavelen:;~th (~m)
'I t t t t t t
o, H,O co, co, 0, H,O co,
Absorbing molecule
Figure 1.1 Atmospheric transmittance over a 6000 ft sea level path.
Introduction 3
existing optical glasses (Kuhl 1983) and do not require specialist new
infrared optical materials. From the black body spectral emittance curves
shown in figure 1.2 it is clear that to detect relatively hot objects (engine ex-
hausts) the 3-5 p.m window is most suitable, and to detect objects at room
temperature (human body) the 7.5-14 p.m (8-12 p.m hereafter) window is
most suitable. Thus at present there is a major interest in thermal systems
which are concerned with wavelengths up to about 12 p.m. In order to
process this thermal radiation optically, windows, lenses and telescopes are
required. Generally the requirements for infrared transmitting materials are
set primarily by the atmospheric transmission and secondarily by the opera-
tional wavelength range of the sources and detectors and by the power
handling requirements of particular systems. On average, components up
to 150 mm in diameter and up to 20 mm in thickness are required but there
are occasional needs for much larger ones. In a useful historical review of
infrared photography and thermography O'Neill (1983) discusses the
industrial, medical and scientific uses of thermal systems. In the case of
glass fibres for optical communications the operational wavelengths are in
the near infrared, 0.8-1.55 p.m, a region close to the optical loss minima
in the glasses and at particular wavelengths where semiconducting LED or
laser sources are available. Fibres of the order of 50/125 p.m core/core
plus cladding diameters in lengths of at least 1 km and of low loss
(0.5-3 dB km _, depending upon application) are required for optical
communications applications. Tebo (1983) discusses the uses of infrared
fibres in medicine, optical transfer optics and long distance communications
links.
105 r - - - - - - - - - - - - ,
4
' 10
3.
7b 103
e
3
c
102
~ 10
~
§
~ 10-1
~ 10-2
;:
Jr 1o-3
1o'~~~~~~~_L~~~u
Wavelergth l~ml
The drive to market laser, thermal and fibre optic systems requmng
optical components with reproducible properties and reliable performances
is establishing the commercial production of infrared optical materials and
in some cases new and improved physical measurement techniques for their
characterisation. In recent years the technology has been advancing rapidly
in the whole field of optical materials, from the ultraviolet to the infrared,
as discussed in a review by Musikant (1983). New products are emerging or
have emerged in the field of infrared optical materials: spinel (MgA\z04),
ALON (5AlN .9Ah0 3) and stabilised Zr02 for 3-5 !Lm use as discussed
together with existing materials in Chapter 3; germanium, scaled up GaAs,
chalcogenide glass (Ge- [As or Sb ]-Se-Te), ZnS, ZnSe, CaLa2S4, dia-
mond and scaled up alkali halides (KCl, NaCI) for 8-12 /Lm and beyond
discussed in Chapters 4 and 5; silicate and fluoride optical glass fibres for
the near and mid infrared; other halide (Cl, Br, I) and chalcogenide glasses
together with crystalline materials (AgCI, AgBr, KRS5) being researched
mainly for 8-12 JLm fibre applications as discussed in Chapter 7; a-carbon
coating and other amorphous insulators discussed in Chapter 9. Laser
damage at wavelengths of 1.06 and 10.6 !Lm is discussed for some materials
in Chapter 6. The development of synthesising and processing techniques
such as distillation, vacuum melting, hot isostatic pressing, vapour growth
and melt growth including reactive atmosphere processing are discussed in
relation to specific materials as appropriate in Chapters 3, 4, 5, 7 and 9. An
indication of some of the specialist characterisation techniques such as
those for refractive index, MTF, absorption, fracture stress and fracture
toughness, thermal shock and rain erosion, necessary for property measure-
ment and quality assessment of infrared optical materials is given in
Chapter 8. A diagramatic comparison of transmittance range, refractive
index, thermal expansion coefficient, hardness, Young's modulus, rupture
modulus and major extrinsic absorption wavelength amongst many infrared
optical materials is given in Chapter 2.
2
Loss Mechanisms in Infrared Optical
Materials
1 o'
1 o'
1 o'
'CE 1 o' ~
~
~
-"' c
~ .2
c
·~ 10 E-
0
c
.2
~
:E
~
~
"€l ~
8 3
~
0
c I ~
c
~
~
0
;g. I ~
~
~
"'e- 0c
~
~
0
\ 0
« 1 o- 1 ~
~
I ~
I 5
11f'
\
\
1 o·'
5 4 3
10 10 10 1 o' 10
Wavelength (cm-1)
Figure 2.1 The transmittance of AszS, glass showing the uv and
vibrational absorption edges and the window region (Strom et ali974).
i;
lattice absorptions as illustrated in figure 2.1. These lattice absorptions
result from vibrational modes of the atoms in these materials. In ionic·
crystals vibrations of large amplitude will occur when incident radiation is
of the same frequency as the resonant frequency of the atomic units and this
is termed the reststrahl frequency. These atomic units must possess a perma-
nent dipole moment which can be activated by the oscillating electric field
of the incident radiation. Non-polar solids would be expected to be infrared
'' '
'
,.'''
8 Loss mechanisms in infrared optical materials
inactive but usually exhibit induced dipole effects. For example, a non-ionic
solid can have an effective charge and thus a dipole moment if the atoms
are not identical. Induced dipole moments are also possible in elemental
materials such as diamond and silicon. These homo-polar materials do not
possess a permanent dipole moment but an infrared inactive vibrational
mode may induce charges on the atoms and a second mode may
simultaneously cause a vibration of these charged atoms. These second-
order effects are likely to be of low magnitude but nevertheless they are
likely to absorb infrared radiation in solids of useful thickness and hence
limit the transmittance. The fundamental absorption frequency can be
calculated for a linear polar diatomic molecule consisting of two point
masses m1 and m2. The frequency of vibration, V, of the simple harmonic
motion of the two masses along a line joining them is given by (Dekker 1960)
materials such as the alkali halides (Sparks and Sham 1973), the alkaline
earth fluorides (Lipson et a/ 1976), and in semiconductors (Deutch 1975).
Highly purified samples of these materials exhibit an absorption coefficient,
(3, in this multiphonon region which reduces e~ponentially (Bendow 1975).
This exponential tail can be represented by
(2.4)
where A and-y are material dependent parameters and w is the frequency.
The temperature dependence of equation (2.4) has been shown to vary as
some power of the temperature at high temperature and becomes
temperature independent at low temperatures (McGill 1975). Detailed
discussions of phonons in solids are given by Mitra and Gielisse (1965) and
by Wang (1966) and multiphonon processes are discussed by Mitra and
Bendow (1975). For present purposes, if the lattice vibrations or phonons
in a periodic crystal lattice are treated as harmonic oscillators coupled to
their nearest neighbour oscillations, then both transverse and longitudinal
modes of oscillation with different velocities transmit the energy across the
crystal. Where there is more than one atom per unit cell the adjacent atoms
can oscillate in and out of pnase with one another and these oscillations are
termed acoustic and optical phonons respectively. The phonons can be
classified into four groups called longitudinal optical (LO ), transverse
optical (TO), longitudinal acoustic (LA) and transverse acoustic (TA).
Several of these phonons add together to yield an individual Gaussian
contribution to the overall absorption tail. It is the summation of all of
these individual contributions resulting from each particular crystal struc-
ture and unit cell which yields the exponential absorption between the
transparent region and the fundamental lattice absorption frequencies.
Thus the fundamental absorption processes which limit the transparency
range of insulators and semiconductors are due to electronic transitions
across the band gap at short wavelengths or lattice vibrations at longer
wavelengths. The absorption coefficient on the long wavelength side of the
band gap and on the short wavelength side of the lattice absorption exhibits
an exponential dependence on frequency.
Additionally, in semiconductors free electron absorption is important in
the region of transparency besides the exponential tails of the absorption
edges. The effect of free carriers on the optical properties becomes impor-
tant at wavelengths larger than the intrinsic electronic absorption edge and
involves only the energy band containing the carriers and is described as an
intraband effect (Willardson and Beer 1967). There are in addition interband
effects which involve another energy band and these transitions give rise to
absorption bands at specific wavelengths. The absorption coefficient, f3c,
dependent on free carriers can be calculated using
N'A2e'
f3c = -=..::..:....::.,...,.
2 3 (2.5)
p:1rnm* c
10 Loss mechanisms in infrared optical materials
Intrinsic scatter and absorption are responsible for limiting the optical loss I:
in an insulator. These effects are best illustrated in the case of a glass fibre I
optic since this is available in path lengths of several tens of kilometres, thus
making the problem of obtaining accurate loss measurements at low levels
relatively easy in comparison with other infrared optical materials which are
only available in thicknesses of a few centimetres. Pinnow eta/ (1973) have
identified the fundamental optical scattering and absorption mechanisms
which limit the light transmission in silica fibre optic waveguide. Scattering
loss in glasses is known to be caused by microscopic variations in the local
dielectric constant associated with the random structure of these materials.
The magnitude of the scattering from this effect can be calculated using
classical electromagnetic theory and thermodynamics. In addition, the
random structure of glass is determined by the fictive temperature (the
temperature at which a glass would come into thermodynamic equilibrium
if heated). By incorporating this fictive temperature concept into the
classical theory, Pinnow eta/ (1968) were able to account for light scattering
in fused silica. The situation is more complex in multi-component glasses
where there is an additional mechanism arising from the statistically ran-
dom distribution of polarisable components which produces further local
variations in the dielectric constant. Pinnow has also developed a quan-
titative model to account for this effect. The model yields a total scattering
loss for Si0 2 glass in agreement with experimentally determined values of
the order of 2 dB km - 1 at 0.8 I'm. It is known that the random structure
in glasses gives rise to varying local electric fields on a microscopic scale.
Recent theoretical work by Dow and Redfield (1971, 1972) and by Tauc
(1975) provides evidence that such local microfields cause intrinsic absorp-
tion in chemically pure materials in what is normally the transparent region
below the fundamental interband absorption edge. The mechanism is due
to local field induced broadening of the exciton levels which are created in
optical absorption energies close to but below the interband edge. This
should produce a tail in the uv interband edge varying exponentially as
previously described. Experiments on silica fibre confirm an exponential . '
Halides
In general these materials possess high band gaps and are transparent in the
uv, visible and the IR. They possess low to medium refractive indices,
1.5-2.5 (antireflection coatings are not absolutely necessary), low absorp-
tion coefficients but relatively poor mechanical, thermal and chemical
durability prcperties.
Intrinsic loss 13
Oxides
These possess medium band gaps of the order of 3-6 eV, medium refractive
indices, 1.5-2.0 (antireflection coatings are not absolutely necessary),
moderately low absorption coefficients and good thermal and mechanical
properties.
frequency varies as the reciprocal of the square root of the reduced mass
and as the square root of the force constant. The force constant decreases
with increasing interatomic distance and with increasing cation coordina-
tion number. This is because the charge must be shared between more ions
in structures with high coordination numbers. Thus with these few simple
criteria it is possible to shortlist new materials for investigation of their
infrared transmittance. Glasses and cubic materials with heavy ions, large
interatomic distances and high coordination numbers are considered.
1---------- -_-_-_-_-_-_-s--Uip~~;laS,-
GaAs
r-- --- - - Sulphide glass- - -
- - - - - - Selenideglo.ss I
Selenide glass
- - - - - ZnSI8-12~ml I
ZnS
- - - - - - - ZnS ICiearlran) ! ZnSe
- - - - - - - ZnSe I CdTe
------CdTe
Cha~o9,enides Chalcogenides
0.1 0.2 0.40.61D 2 4 6810 20 406080
Wavelength l~ml 0 2 3 4 5
Refractive index at 4.0 ~m
Figure 2.2 The relative transmittance Figure 2.3 The relative refractive
ranges of some halide, oxide, semicon- indices at 4.0 I'm for some halide,
ductor and chalcogenide materials. oxide, semiconductor and chalcogenide
materials.
near and mid infrared. If far infrared transmittance and beyond is required,
then a halide material would be suitable but the mechanical properties leave
much to be desired. A useful compromise in this case might be to choose
one of the chalcogenide or semiconductor materials.
j I
oi'----'1o"----~2'*0'----___;3~0----"4~'-o---"'so,_----"'{60 o 2 4 6 8 10 12 14
~--Mgf, MgF2
f-----CaF, 1--- CaF2
f - - - - - - B a F2 f-- Ba F2 i I
0 10 20 30 40 50
Thermal expansion coefficient
Figure 2.4 The relative thermal expan- Figure 2.5 The relative hardness of
sion coefficients of some halide, oxide, some halide, oxide, semiconductor and
semiconductor and chalcogenide chalcogenide materials.
materials.
the physics and chemistry of the.material and its method of synthesis, they
will be discussed in relation to each material in the following chapters. In
this way a true appreciation of the problem of extrinsic loss will be obtained
but some introduction to the subject is given here.
Extrinsic loss mechanisms determine the percentage of the theoretical
level of transparency achievable in a solid in practice. These mechanisms are
basically scatter and absorption arising from the raw material and the
fabrication process employed together with the chemistry of the material in
relation to certain specific impurities. For instance, pores and grain bound-
aries can be a particular problem causing scatter extending over several
microns in wavelength in a material hot pressed from powder, as described
for the Irtran materials in Chapter 4. Similar although less severe problems
can occur in the growth of material from the vapour, where growth defects
and grain boundary problems can cause scatter, as is the case with 8-12 l<m
quality ZnS discussed in Chapter 5. Absorption problems can be of a
general nature, i.e. stoichiometry problems can lead to general electronic
Extrinsic loss 17
12 - Ge0 2 in Germanium
10 Extrinsic oxygen in
- As 2 03 in selenide glass combination with
9 = Si 02 in silicon, sar with other elements
in Co. Lo. 2 54
e2- a
£ 7
~
c - 0-H bending in oxides and
• -, sulphides
"
>
c
3
6 Zn-H stretching in ZnS
5
Extrinsic co.rbon and
4 - H2 S in sulphide glo.ss hydrogen in combination
• C-H stretching,orgo.nic with other elements
3 surface conto.minants
- 0-H stretching in oxide
glass
2
Materials scientists put considerable effort into the synthesis and fabrica-
tion of bulk infrared optical materials and components sufficiently free of
extrinsic absorption for the intended applications. One could be excused for
thinking that having achieved these components the problems of bulk
extrinsic absorption are over but this has proved not to be so. Silica fibre
must rank as one of the most pure, if not the most pure, of the infrared
optical materials. However, Uesugi et a/ (1983) have found that if the
interior of a fibre optic cable is exposed to water and if electrolytic corro-
sion of metal takes place in the vicinity (for instance, if there are metal
strength members present in the cable) then the molecular hydrogen
18 Loss mechanisms in infrared optical materials
1.243 1.24
J.l96 J.l9
J.l67 J.l7
J.l27 1.13
1.080 1.09
i!
The needs of near infrared systems for optical materials can be fulfilled by
existing commercial optical glasses produced primarily for use in the uv
and visible region of the spectrum. The transmittance limitations set by
extrinsic impurities and intrinsic vibrational absorption in these glasses is
first of all discussed in this chapter in relation to chemical composition and
manufacturing techniques. The chapter then goes on to describe a number
of materials, which have been developed and produced for use in the mid
infrared. These are discussed in terms of their transmittance range and their
~I
extrinsic absorption in relation to their manufacture. During the 1950s and
early 1960s the need arose for materials suitable as airborne sensor windows
transmitting at 3-5 !LID wavelengths in thicknesses of about 3 mm. This
need continues today and has resulted in a range of materials options start-
ing with extended transmittance range glasses such as calcium aluminate
and germanate compositions and expanding to include a number of
crystalline fluorides, oxides and oxynitrides such as MgF2, Ah03, MgAhO•
and 5AlN.9Ah03. In addition to these window materials, infrared detec-
tion, imaging or laser systems require materials demonstrating good
3-5 ~tiD transmittance for the manufacture of other optical components.
Silicon available from the semiconductor industry has been used but
materials specially developed for the purpose are arsenic trisulphide glass,
Eastman Kodak Irtran hot pressed polycrystalline materials and crystalline
and vitreous fluorides produced from the melt. Materials made for the
8-12 !LID band such as germanium, chalcogenide glasses, ZnS and ZnSe
described in Chapter 4 and the halide materials described in Chapter 5 are
also very suitable for use in this mid infrared waveband spectral region.
Thus there are a large number of materials options available in the mid
infrared band and the majority of these are illustrated in terms of their
reciprocal dispersive power in relation to their refractive indices in figure
3.1. In addition their major physical properties are listed in tables 3.5
and 3.6.
20 Bulk materials for the near and mid infrared
r:::...r 3.0
E •Ge 30 As 13 Se 27 Te 30
..."- •Ge 30 As 13 Se 37 Te 20
AMTIR glass• : 1173 glass
-::;
l;j KRss• I • As2S3
~ ZnSe •-ZnS
.E
Q.t 2.0 oAgCl
>
Al203
~ e(sl • Spinel
• Fluoride gloss
•
•KCl •NoCl
"'"" • CaFz •
MgF2
1..~~~~~--~--~-----±~~~~~-L---L----~
1000 500 100 50 10
Reciprocal dispersive power (n 4 -1)f(n 3-n 5l
Figure 3.1 Reciprocal dispersive power (n 4 - l)/(n 3 - n5 ) plotted
against n• for a number of optical materials useful in the mid (3-5 I'm)
infrared.
3.1 Bulk Optical Glasses for the Near Infrared (0.75-3.0 I'm)
There are many visible band optical systems in the land, sea and air
environments whose effectiveness can be enhanced by the addition of night
vision (e.g. image intensifiers) and/or laser ranging capability (e.g. NdYAG
1.06 I'm, Ho YLF eye safe 2.06 I'm). These optical systems constructed from
bulk optical glasses are able to be upgraded to include the latter features
without the need to develop special new optical materials. This is because
most commercial optical glasses produced for use in the uv and visible
region of the spectrum exhibit effective transmittance in the 0. 75-2.5 I'm
spectral region i.e. in the near infrared and hence optical designers are able
to continue to use these familiar materials to achieve extra capabilities.
The silicate optical glasses, available from major suppliers in Europe,
USA and Japan, range in refractive index nct approximately from 1.40 to
2.05 ·and in reciprocal dispersive power Vct or Abbe value approximately
from 15 to 100 where Vct = (nct- 1)/(nr- nc). These are defined as crown or
flint glasses according to their properties i.e. glasses with nct > 1.6 and
Vct > 50 plus those with nct < 1.6 and Vct > 55 are classified as crown
glasses, the remainder are classified as flint glasses. Generally the refractive
indices are available to 1.0 I'm and in some cases beyond enabling near
infrared optics to be designed from these off-the-shelf high quality
materials: A plot of nd against Vct illustrating the range of glasses available
is shown in figure 3.2, where the growth in the number of commercially
available optical glasses in the past 100 years is seen. The refractive index
nct for Si02 glass is 1.4528 and this can only be decreased by the addition
Bulk optical glasses for the near infrared 21
v,
Figure 3.2 Reciprocal dispersive power (n•- 1)/(nr- n,) plotted
high resolution imaging systems. The many different glasses are used for
correcting the various lens aberrations to minimal values. For instance in
achromatic systems chromatic aberration is corrected at two wavelengths,
one in the red and another in the blue part of the visible spectrum, but the
spectrum in between is uncorrected. However in apochromatic systems this
secondary spectrum is also corrected perhaps at as many as ten wavelengths
to ensure minimal image curvature (Giiemeroth 1981, 1982). Some of the
recent optical glass compositions contain little or no traditional glass
forming oxides such as silica, germania or boric oxide but are based on
other glass forming oxides such as phosphorus pentoxide or tellurium
dioxide (Petrovskii 1978, Blair et a/1981) while fluorides and other oxides
such as Zr02, Ti02 and Nb20s replace some of the traditional modifying
or intermediate oxides (Giiemeroth 1983). Some of these more recent
chemical compositions are very corrosive and so to avoid composition
degradation and impurity pick up they are melted ln platinum or even gold
lined small tank furnaces. The uv cut-on edges of the optical glasses are
consistent with their energy gaps and impurity content and are not con-
sidered here since they have little effect on the near infrared transmittance.
The infrared cut-off in optical glasses is set by the first overtone of the
fundamental vibration with the highest frequency (Adams and Douglas
1959). In oxide glasses the latter is usually assigned to stretching vibrations
in X-0-X units where X is the major glass network forming cation e.g.
Si-'- 0- Si for silicate glasses. The assignment of these stretching vibrations
for the major glass forming oxides is as follows; B-0-B 1370 cm- 1 ,
P-0-P 1250 cm- 1 , Si-0-Si 1100 cm- 1 and Ge-0-Ge 910 cm- 1
(Spierings 1982). Thus the introduction of Ge02 into a silicate glass will
shift the infrared edge to a longer wavelength while B20 3 and P20s will shift
it to a shorter wavelength. However most optical glasses will transmit
usefully to between 2.0 and 3.0 JLm as illustrated in figure 3.3.
The major extrinsic loss mechanism in these materials is absorption due
to impurities in solution, particularly the transition elements (Weyl 1959)
and water in the form of OH (Spierings 1982). Extrinsic scatter is not a
problem since the glasses are produced to a high standard of optical
homogeneity with refractive index variations of only ± 1 x 10- 4 or
± 5 x 10- 6 in the case of premium quality material. In addition the products
are specially chosen to be free from macroscopic defects such as bubbles
and inclusions. The transition element impurities produce characteristic
absorption bands at specific wavelengths but in most cases these are very
broad affecting the transmittance at many wavelengths in the visible and
near infrared parts of the spectrum. For instance iron present in the reduced
form as Fe2+ exhibits an absorption centred around 1 J!m as shown in
figure 3.4 which affects the near infrared transmittance. Optical glasses are
not prepared to semiconductor or optical fibre standards of purity but care
is taken in the choice of raw materials and in some cases melts are made
Bulk optical glasses for the near infrared 23
80
:!:
2..
~
u 60
-c
0
·e
~
c 40 8 A
0
.=
20
0.4
~
0.3
-
c
-~
c.
~
~
0 0.2
~
<(
0.1
0
0.3 0.5 0.7 0.9 1.1
Wavelength (~m)
This results in the disappearance of the overtone band at 1.38 p.m and
instead a very broad absorption band of much lower intensity appears on
y y
1oor------
~eo
0 ""\/
300mm
g 60
·v; 30 mm 10 mm
·"
E
~
c
40
c
~ 20
1mm
0 ..__,__.....____.__
0.6 1.0 1.2 1.4 2.0 2.2 2.4 2.6 2.B
Wavelength { j.lm}
the long wavelength side of the overtone due to OH groups associated with
bridging oxygens through hydrogen bonding.
The manufacture of these high quality materials is accomplished by
careful choice of raw materials (e.g. oxides, carbonates, nitrates, chlorides
or fluorides) to avoid major extrinsic absorptions by transition elements,
and by specialist melting in refractory pots or small glass tank furnaces
constructed from refractory blocks and often lined with noble metal.
Heating is achieved by natural gas, oil or electricity. Care is sometimes
needed to avoid impurities from the fuel used in the heating process hence
muffle furnaces are often used to avoid direct contact of the molten glass
with oil or gaseous fuels. The powdered, premixed raw materials are fused
and reacted to form a crude glass which is then thermally conditioned to
remove major gas bubbles and straie. The molten glass is then further I,''
refined by physical stirring to achieve the required homogeneity followed by
temperature reduction and further thermal conditioning before the material
is cast into the familiar optical slab configuration by an intermittent or
continuous flow process and is finally annealed and cooled to ambient
temperature. Older technology involved cooling and annealing the complete
homogenised melt in its refractory pot prior to cutting it up carefully into
pieces. Further details of the glass manufacturing processes are available in
the literature (Gunther 1958).
Table 3.2 Extinction coefficients for vitreosil and spectrosil SiOz glass.
Fundamental
antisymmetric Combination tone
stretching of the fundamental First Second
vibration of and a vibration of overtone of overtone of
Assignment OH SiOz at 12.4 I'm fundamental fundamental
I
26 Bulk materials for the near and mid infrared
Thermal
Refractive index Density expansion Tensile Young's
Material (at x I'm) (kgm- 3) coefficient Knoop strength modulus
x 103 X 10- 6 hardness (MPa) (GPa)
1.0 2.0 3.0
3.2.1 Glasses
During the 1950s a requirement arose for transparent materials for use as
robust windows in airborne applications. To provide a solution to satisfy
this requirement, a number of materials approaches were examined amongst
polycrystalline semiconductors and ceramics discussed later, and amongst
glasses. Plate glass windows were used in some systems employing PbS
detectors operating in the region of 2-3 p.m but the need for 3-5 p.m
transmittance when employing InSb detectors eliminated silicate glass
windows because of the early Si-0 bond overtone cut-off illustrated in
figure 3.3. During the 1950s and early 1960s a number of glass systems e.g.
sulphide, bismuthate, antimonate, tellurite, aluminate and germanate were
examined in relation to this application. Some of the properties of these are
compared with silicate glass in table 3.4. The tellurite glasses (Stanworth
1952), the lead bismuthate glasses (Dumbaugh 1978) and the sulphide
glasses were found to possess insufficiently robust thermal and mechanical
properties and the antimonates (Hedden and King 1956) offered no advan-
tage over the more well known aluminate and germanate glasses. Thus the
latter two glass systems became prime candidates to satisfy this particular
need because of their enhanced thermal and mechanical properties and their
extended infrared transmittance. However, one sulphide glass, As2 S3 , was
put into production and found useful as an internal component material in
systems employing the mid infrared band and some of its properties are
given in tables 3.5 and 3.6. A full description of sulphide and other
chalcogenide glasses is given in Chapter 4. More recently a new system of
fluoride glasses based on ZrF• and HfF4 has been researched primarily for
use as mid infrared fibre component materials (Chapter 7). However these
new glasses demonstrate sufficient stability to be considered for bulk optical
applications and may in the future find use in 3-5 p.m systems since their
Materials for the mid infrared 27
Table 3.4 A comparison of the physical properties of infrared
. transmitting glasses.
the formulation of a calcium aluminate glass for 3-5 JLm infrared window
applications, many data were available on glass formation and basic physical
properties enabling research workers to concentrate on optimising glass
stability in relation to the infrared transmittance and chemical durability
and on overcoming the extrinsic OH absorption problems. Florence et a/
(1955) identified a problem of tendency towards devitrification when pouring
thick slabs of these glasses and of extrinsic absorption from OH as seen in
figure 3.6(a). They identified a composition of wt o/o A]z0 3 47, CaO 43, BaO
10 as the best glass of those examined for ease of production and showed
that dry air bubbling of the melt for 2 h was effective in reducing but not
eliminating the extrinsic OH absorption in useful window thicknesses of
glass. They also reported that these calcium aluminate glasses possessed
unusually high values of Young's modulus in comparison to normal silicate
glasses e.g. silicate glasses 54-80 GPa, calcium aluminate glasses around
105 GPa. Hafner eta/ (1958) reported that calcium aluminate glasses could
be made in quantities of 5500 g but that the surfaces of these materials
deteriorated significantly on exposure to 95% relative humidity at 49 o C for
24 h. However, antireflection coatings of MgF2 or SiO were found to be
completely effective in protecting the material from such moisture attack.
100
.._ 80
;i: --...,
\!
~ \I
E 60
=§
~
~
c
.=
40
20 I
'
1:
I
I
I
Ia) lb)
_i ~
1.0 3.0 6.0 0.4 1.0 3.0 6.0
Wavelength {).lm)
Barr and Stroud Ltd, UK are a major manufacturer of this type of glass
and offer two materials BS37 A containing a small quantity of Si02 and
BS39B containing no Si0 2. The latter transmits further into the infrared as
seen in figure 3.5(b) (Billard and Cornillault 1962, Barr and Stroud
technical leaflet C-1654). The uv cut-on edge and visible transmittance is
consistent with the likely energy gap and the known impurity content
(transition elements e.g. reduced oxides of Fe and Cu), while the infrared
Materials for the mid infrared 33
Germanate glasses
Germanium dioxide is a glass former and demonstrates a good transmit-
tance in the infrared to about 6 p.m (Cohen and Smith 1958) but compared
with silicate or aluminate glasses it is expensive and lacks robust mechanical
properties. Florence eta! (1955) and Blau (1955) studied the improvement
in physical properties resulting from additions of modifying oxides such
as PbO, La203, BaO and CaO while Fray and Nielsen (1961b) reported
germanate glasses containing PbO together with SnO or Ce0 2 or CdO or
LhO or Na20 or K20. Fray and Nielsen also reported that OH absorption
in these glasses could be removed by vacuum melting or by the addition of
fluorides amongst the raw materials. Small quanitites of germanate glasses
34 Bulk materials for the near and mid infrared
were produced in France, e.g. Sovirel VIR-3, and the USA during the 1960s
but due to the hig)ler cost and inferior mechanical properties of these
materials, the calcium aluminate glasses were more frequently used for
window applications. However, during the late 1960s and early 1970s
Dumbaugh (1973, 1975) reported improved germanate glasses with more
robust physical properties and also demonstrated OH extrinsic absorption
removal (Dumbaugh 1970) in this type of glass. This work has led to the
production of a very useful germanate glass composition, Corning code
9754, announced in advanced technology sales literature in 1970 and
recently described more fully by Dumbaugh (1981). The formulation of this ,_
calcium alurriino-germanate glass Ge02 33.0, Ah03 37.3, CaO 19.7, BaO
5.0, ZnO 5.0 overcame some of the problems of calcium aluminate glasses
whilst retaining sufficiently robust mechanical properties and adequate
infrared transmittance at only a little extra cost. Devitrification problems
tend to limit the scale of melting of the calcium aluminate materials to a few
kilograms and the glass quality although adequate for windows is unlikely
to reach first grade standards. The code 9754 glass is sufficiently stable to
be made in large quantities in a tank furnace to first grade optical glass
quality and the OH extrinsic absorption has been reduced to such an extent
as to allow the use of relatively thick components with little impairment in
the infrared transmittance as is seen in figure 3. 7. The short wavelength
transmittance of this glass is consistent with its likely energy gap and the·
long wavelength absorption edge is consistent with the data discussed in
§3.1. The optical properties of this material are given in table 3.5 and the
100
I
80 I
~ I
~
~
c
.E!
±
60
I
i
E
~
c
I
0 40 I
.= I
I
I
I
20 I
--------' I
thermal and mechanical properties are given in table 3.6. The rain erosion
resistance of this glass is likely to be good because of the high AhOJ and
CaO content but it is not likely to be as good as a calcium aluminate glass.
Fluoride glasses
The first fluoride glasses based on ZrF• were discovered in 1974 and since
then a considerable amount of literature has emerged on these glasses based
on ZrF4 and HfF4 • They have been synthesised from the melt utilising either
fluoridisation of mixed oxides or fluoride raw materials (Bendow and
Drexhage 1982). At first the technological development of these glasses was
hindered by their high tendency to crystallisation and their poor chemical
durability. Increasingly complex glass compositions utilising other fluorides
such as BaF2, ZnF2, YbFJ, ThF• and particularly AIF3 exhibited improved
physical properties to a point where they are being seriously considered as
materials for mid infrared fibre optic applications which is the prime
interest of research workers in this field (Chapter 7). However they are also
being researched for internal optical component applications in 3-5 pm
thermal band sensors. One of the major costs of optical components is in
generating the form from plane parallel blanks and in the final figuring and
polishing. Fluoride glass development has reached a point where these
materials can be considered for use in hot-pressing or hot forging processes
to produce preshaped and possibly optically finished components thus
offering the potential of cost reduction by avoiding the traditional machin-
ing operations involved in component production. Turk (1981) has
researched this hot pressing process for fluoride glass infrared optics. A
60"7o ZrF., 33% BaF2, 7% ThF• glass was used to investigate optical
forming by hot pressing and a 60% HfF4 , 33% BaF2, 5% LaF 3 , 2% AIF 3
glass was used for experiments in consolidating optical components from
smaller pieces. An optically finished cobalt bonded tungsten carbide die was
used to test the hot pressing of optical components. Pressing was done in
anargonatmosphereat315 °C(i.e. between T,and T,)and20x 106 Nm- 2
and an excellent duplication of the die surfaces was obtained. In the con-
solidation work it was found that the crystallisation temperature of the glass
was lowered with increased pressure so that in this case pieces were heated
to 340 °C and then the pressure was increased slowly to give a final con-
solidation pressure of 6.9 x 106 Nm- 2. Vitreous transparent components
I
were obtained but the boundaries between pieces were apparent. Only time
will tell whether these fluoride glasses will find a role as bulk optical com-
ponents in 3-5 I'm optical systems. Physical property data tend to be sparse
but the optical properties for the glass mole % 63 ZrF., 33 BaF2, 4 GdF4
are given in table 3.5 together with the transmittance shown in figure 3.8
(Mitachi 1982) and some indication of the general physical properties of
these glasses (Turk 1981) is given in table 3.6.
36 Bulk materials for the near and mid infrared
100
80
~
~
~
~
c 60
.,
.£
~
c
0
.= 40 -
20
components which could be pressed in multiple dies or cut from larger flat
discs. Further discussion of the hot pressing process is given in §4.4.1 where
8-12 I'm hot pressed components are discussed.
MgFz-lrtran 1
Magnesium fluoride has found extensive use as an airborne infrared
window material as it has excellent thermal and mechanical properties. It
is thus surprising to realise that it possesses a tetragonal (rutile) structure
rather than a cubic one but the birefringence is small, that is at 1.083 I'm
the difference in refractive index between the ordinary and extraordinary
rays is only 0.0116. Therefore the scatter caused by this is small compared
with that from the imperfect physical state of Irtran 1 as a result of being
manufactured by a hot pressing process. Much of the reduction in transmit-
tance at short wavelengths seen in figure 3.9 is likely to be caused by
the imperfect physical state of the window. The long wavelength cut-off
absorption edge is consistent with the published reststrahlen data of
Duncanson and Stevenson (1958). The transmittance shown in figure 3.9
indicates the most common absorption from OH stretching at 2. 75 I'm, OH
bending at 6.7 I'm and an oxyfluoride hydrogen complex at 5.0 I'm. Initially
raw material prepared for vapour deposition of antireflection coatings on
lenses was used in the preparation of hot pressed MgF2 and the extrinsic
absorption problems of OH and OFH complexes were recognised (Buckner
et a/ 1962). Later special powders were produced for hot pressing, for
instance by precipitation from anhydrous solutions of magnesium chloride
and hydrofluoric acid in methanol, which resulted in the almost complete
elimination of the extrinsic absorption bands in useful window thicknesses
100
80
~
0
I
60
~
~
§
I
.,
.;:
~ 40 I
c
c
o!= I
A
~ \
\. I
- -"
I
I
38 Bulk materials for the near and mid infrared
CaF3 -Irtran 3
This was the best visually transmitting Irtran material but still possessed a
hazy colourless appearance. It was available in flat blank sizes up to
!50 mm diameter and up to 13 mm in thickness. It found its main applica-
tion in spectroscopic cells and applications requiring visible as well as mid
infrared transmittance. The transmittance is seen in figure 3.10 (curve A)
and optical and other physical properties are given in tables 3.5 and 3.6.
MgO-Jrtran 5
Magnesium oxide components were hot pressed from a microcrystalline
powder in an inductively heated molybdenum or molybdenum alloy die at
a pressure of 276 X 106-448 X 106 N m- 2 and a temperature in the range
800-860 °C in a vacuum or in an inert gas atmosphere for 5-20 mins
(Carnal! and Hatch 1965). Some cracking of the MgO products was
experienced caused by bonding to the molybdenum mould parts. Tungsten
foil lining of the die cavity or coating with graphite were found to be
effective in preventing sticking. The material produced was virtually as
colourless as the CaF2 and was available in flat blank sizes of !50 mm
diameter and 9.5 mm thickness. Optical properties are given in table 3.5
Materials for the mid infrared 39
1ooro-.-~~--,--.--~,-,--,--,--,,-----,
80
;:.
~
./?'
~ 60
0
~
"'
~
0 40
!=
20
01~~~~---L--~~~~~~--~
0.5 1.0· 5.0 10.0 20.0
Wavelength I~m I
Figure 3.10 The transmittance of CaF2 Irtran 3, 3 mm thick (A) and
of MgO Irtran 5, 3 mm thick (B).
before pressure was applied in the hot pressing process the MgO plus 1% i
A family of Hall-Petch lines were obtained and the slope of these was
found to depend on the material surface finish. Additionally Rice {1972)
studied the strength of hot pressed MgO in relation to grain size, impurity
content and annealing. The strength of dense hot pressed MgO was found
to increase significantly as a result of slow annealing to about 1200 o C in
spite of some grairi growth during the process. The strength increased by
a factor of two i.e. to 579 x 106 N m - 2 at a grain size of 2 p.m. This increase
in strength and reduction in data scatter was attributed to the removal of
anion impurities i.e. carbonates trapped at grain boundaries. This work
illustrates the importance of surface finishing processes and impurity
contents particularly anion impurities in the manufacture of MgO
components if maximum strength is required.
essential where low loss material for high power applications is required.
Purification of the melt is accomplished both before and during melting by
the introduction of fluorocarbons according to equation (3 .2) or by the
pyrolysis of teflon according to equation (3.3)
2Ca0 + CF.(g) = 2CaF2 + C02(g)t (3.2)
2Ca0 + C2F4(g) = 2CaF2 + 2CO(g)t. (3.3)
This process has been found to be highly effective in removing extrinsic
absorption (Chernevskaya and Korneva 1972). Reactive atmosphere pro-
cessing (RAP), particularly useful for the growth of halides, is discussed in
more detail by Pastor and Arita (1975) who employed HF in helium carrier
gas as the RAP agent.
Magnesium fluoride
Monocrystalline magnesium fluoride produced by the Stockbarger process
'
is a useful material for polariser, waveplate and laser host applications ''
because of its excellent physical properties. However the birefringence of
MgF2 (RI Bray- RI Oray at 1.083 I'm is 0.0116) is detrimental to its use as a
lens and window material particularly where achromatisation is involved.
For these applications isotropic CaF2 or BaF2 is usually preferred to
the tetragonal MgF2. But since the latter does find use in some passive
infrared optical systems the transmittance of the material is shown in figure
3.11 (curve A) and the general physical properties for monocrystalline
material as reported in the literature by Hargreaves (1982) are listed in table
3.6.
100
~
~
~
u
80
60
\\\
-
~
.E
...
~
~
~
40
A B C 0
"'
,,
20
0
\ \
0.5 tO 5.0 10.0 20.0
Wavelength I ~m)
Figure 3.11 The transmittance of melt grown MgF2 2.1 mm thick (A),
CaF2 1.2 mm thick (B), SrFz 1.5 mm thick (C) and BaFz 1.5 mm thick
(D).
42 Bulk materials for the near and mid infrared
Calcium fluoride
Calcium fluoride grown by the Stockbarger or Bridgman techniques is
available in diameters ranging from 80-100 mm, and grown by the fusion
casting process is available with diameters up to about 250 mm. The
material transmits usefully in the range 0.2-8 I'm for a 2 mm thickness
(figure 3.11 (curve B)) and is routinely purified by the RAP process to be
free from major extrinsic absorption from oxide or oxide containing
species. Denham et a/ (1970) have investigated the optical, dielectric and
lattice properties of CaF2 together with SrF2 , BaFz, CdF2 and PbFz.
Results obtained from experimental studies on infrared and Raman spectra
yielded the fundamental transverse and longitudinal optical lattice mode
frequencies and the two-phonon absorption spectra. These data and
published elastic constant data were used to calculate phonon dispersion
curves, density of states and density of combined states functions.
It was demonstrated that the absorption coefficients of CaF2 , SrFz,
BaF2 and PbF2 in the wavelength region 100-300 em_, in the range of
ex= 5-15 x 103 em_, were almost entirely due to the one-phonon process
in which the positive lattice vibrates with respect to the negative lattice.
Some of the basic data for a temperature of 300 K are given in table 3. 7.
Data in the literature (Miles 1976) indicates that the bulk intrinsic
multiphonon level has been reached at the CO wavelength (5 .25 I'm) but
that there is some variability at HF (2.7 I'm) and DF (3.8 I'm) wavelengths.
This may be in part due to the inadequacies of the purification processes but
may also be partly due to variable surface absorption affecting laser
calorimeter measurements perhaps as a result of the slight water solubility
of CaF2 (1.7 x 10- 3 g/lOOg H 2 0 at 26 °C). The extent of the surface
absorption problem has recently been revealed by Braunstein et a/ (1980)
using infrared wavelength modulation spectroscopy. In this work it was
shown that physically absorbed surface species resulted in observable OH-,
C- H and C02 bands which were reduced or removed in a dry nitrogen
Materials for the mid infrared 43
atmosphere. These surface absorptions in the range !0- 4 -!0- 5 cm- 1 could
be detrimental in high power laser window applications (HF, DF) but would
be unlikely to be a problem for low power applications.
Calcium fluoride has been considered as a candidate for HF /DF laser
window applications because of its useful physical properties. However, it
has been shown to be susceptible to surface flaws and studies have indicated
that the improvement in fracture surface energy was about an order of
magnitude as the grain size was reduced from monocrystalline to !'
polycrystalline ( -10 J.tm) (Anderson et a/1978). Therefore, hot forging has
been suggested as a useful technique to improve the physical strength since
fusion cast material with a 10-20 mm grain size demonstrates a similar
grain size related flaw sensitivity to that of monocrystalline material. The
hot forging process isdescribed more fully in relation to alkali halides in
Chapter 5. During the deformation .process, mechanical constraint on the I
free periphery of the work piece is necessary to reduce tensile stresses below
the point where cracking occurs. An improved technique over a constrain-
ing ring is the use of hot isostatic forging using helium gas in the forming
chamber. Suitable forging conditions might be of the order of 750 ° C at a
helium gas pressure of 13.8 MPa at 250 ILmmin- 1 for 60"7o true strain. But
when the deformation exceeds 50% true strain optical scattering sites occur
in the bulk of the forged sample. Investigations of this phenomenon
{Anderson et a/1976) have revealed that uniform visual scatter occurs after
pressing (I 00) and ( 113) crystal directions, but striated veiling appears
to occur on slip planes within the bulk after pressing the (Ill) direction.
Electron micrograph studies have suggested that the veiling is caused by
microvoids formed by the coalescence of vacancies produced by dislocation
intersections and annihilations due to plastic deformation. Since these voids
. cause scattering of light in the visible spectrum, studies of their effects at
2.7 and 3.8 J.lm have been made and work to eliminate them by hot isostatic
pressing (HIP) was successful (Hopkins et a/1979). Pre and post HIP optical
absorption, scattering, homogeneity and stress birefringence measurements
were made. Samples of three crystal directions (100), (Ill) and (113)
wrapped in tantalum foil were heated to 750 °C at an argon pressure of
207 MPa for a period of I 0 h. Veiling was virtually eliminated to be
replaced by surface pits many of which were replicas of the striated veils.
Thus it was postulated that the veils migrated to the surface from the
bulk and therefore the bulk material collapsed upon itself. From the
measurements made in this work it was concluded that the HIP processing
slightly increased the absorption at 2. 7 and 3. 8 J.!m, that the optical
inhomogeneity at 633 nm increased after HIP processing due to stress
induced birefringence but still remained in the range I X 10- 6-3 X 10- 6,
that the visible scatter (0.6471 J.tm) after HIP processing was reduced in
spatial variation and decreased by a factor of 8 and that the infrared scatter
(3.39 J.tm) was largely unaffected in spatial variation or degree. These results
44 Bulk materials for the near and mid infrared
Strontium fluoride
CaFz and BaFz have been grown for infrared optical applications for a
number of years and have been reasonably well characterised. However,
recent interest in SrFz has mainly resulted from its potential use as a high
power laser window material (Miles 1976) and it is not yet as well
characterised. It is available grown by the Bridgman, Stockbarger or the
fusion casting techniques. The material transmits usefully in the range
0.2-9.0 I'm for a 2 mm thickness (figure 3.11) and is purified by the RAP
process to be free from major oxide impurities. Optical phonon data
(Denham et a/1970) for SrF2 are given in table 3. 7. Water solubility for this
material is 1.17 X 10- 2 g/100 g HzO at 20 °C. Surface absorption problems
similar to those discussed for CaF2 have been studied in SrF2 by Braunstein
eta/ (1980) and carbonates, C-H bonds and (OH-) have been shown to
be present on the surfaces of this material. The optical properties of SrFz
are listed in table 3.5. The refractive index data were derived from a three-
term Sellmeier type dispersion equation of the form shown in equation (3.4)
where
A1 = 0.678 058 94 AI= 0.003 HiS 55
Az = 0.371405 33 Ai = 0.01166622
A3 = 3.345284 Aj = 1592.541991
at a temperature of 20 °C. These data were measured by Dodge (1978) on
Materials for the mid infrared 45
Barium fluoride
Barium fluoride is available in excellent optical quality up to diameters of
the order of 150 mm grown by the Stockbarger process and in larger
diameters grown by the fusion casting process. Of the alkaline earth
fluorides, BaF2 transmits furthest into the infrared as can be seen in figure
3.11 (curve D) demonstrating a transmittance range for a 2 mm thickness
of about 0.2-10 pm. It is purified by the RAP process to be free from major
oxide impurities. Optical phonon data (Denham eta/ 1970) for BaF2 is given
in table 3.7. The water solubility of the material is 1.7 x 10- 1 g/IOOg H 20
at 10 °C and surface absorption problems due to carbonate and (OH-)
have been demonstrated by Braunstein et a/ (1980). Optical properties are
listed in table 3.5 in which the refractive index data are taken from Malitson
(1964). Data at other wavelengths can be derived from a three-term
Sillmeier type equation of the form given in (3.4) where
AI= 0.643 356 AI= 0.003 3396
A2 = 0.506 762 A~= 0.012 030
A 3 = 3.8261 AJ = 2151.70
at 25 °C. Other available physical properties are listed in table 3.6.
100
I
I
80 I
I
~
~
I
I
~
u
60 I
~
,g I
·e I
I
~
~
40
0
~
I
t- I
20 /
/
'
0
0.1 0.5 1.0 5.0 10.0
Wavelength (!J.m)
Vacuum
LHelium
Figure 3.13 Schematic diagram of the heat exchanger method for the
growth of AhO,; H =heater, I= insulation.
1.01,---....----,--,------.-----,,---.,
aoo•c
0.1
'E
-"
.!:!
c
:~ 0.01
:::
"
0
u
c
~
c.
~
0
~
"' 0.001
""'
12oo•c
0.0001
2 3 4 5 6
Wavelength l~ml
to 1.8 MNm- 312 at 600 °C but remained constant from 600-775 °C. The
rupture modulus of sapphire ha~ also been found to be temperature
dependent by Jackman and Roberts (1955) and Wachtman and Maxwell
(1959) falling to a minimum between 500 and 700 °C depending upon
crystal orientation and rising again to room temperature values at 1000 °C.
This increase in strength above 700 ° C has been attributed to a limited
amount of microscopic plastic deformation at stress concentrations. The
rain erosion resistance of sapphire is excellent in comparison with most
other infrared window materials. Single drop impact studies which can be
compared with those on MgFz, Si and spinel in this chapter and ZnS and
ZnSe in Chapter 4 have been made by Hackworth (1979). In this work it
was shown that the damage threshold for material oriented at 60 o to the c
axis was between 475 and 533 ms- 1 for a 2 mm drop size.
100
80 ' \\
I
~ I
:'i 60
c
_g I
-~
I
~
c
0
40 I
.= I
20
\
I
I
I
0 \
0.1 0.5 1.0 5.0 10.0
Wavelength l~ml
Figure 3.15 The transmittance .. of MgAhO• at room temperature
2.4 mm thick (full curve) and at 600 o C (broken curve).
Materials for the mid infrared 51
method. However Roy (1981) has taken the hot pressing fabrication
technique much closer to a viable component production capability. In this
synthesis technique the quality of the final product is critically dependent
upon the quality of the powder raw material. This was prepared by decom-
posing high purity alkoxides and calcining at 1000-1100 ° C to form the
spinel compound and to optimise the particle size. In order to produce clear
hot pressed material the heavy metal content of the powder needed to be
below 100 ppm and alcohol needed to be completely removed to avoid free
carbon in the components after vacuum hot pressing. Uniaxial pressing was
done in a graphite die lined with grafoil for a period of 3-5 h after an
outgassing period at 1250 o C. To obtain this visually clear material LiF was
probably added (Stewart et a/ 1981) thereby producing similar effects to
those reported for Irtran 5 type material. Polycrystalline material produced
by this process exhibits a transmittance (figure 3.15) which exceeds that of
sapphire beyond 4.5 p.m as can be seen by comparing figures 3.15 and 3.12.
The broken curve in figure 3.15 indicates the transmittance at a temperature
of 600 ° C. The total integrated forward scatter at 3.39 p.m from the bulk
and two surfaces of a number of IR domes was measured to be on average
2.36 ± 1.53 X 10- 2 • Other optical properties of hot pressed material (Roy
1981) are listed in table 3.5 and some additional physical properties
are given in table 3.6. The tensile strength is reported as 110 MNm- 2 ,
the compression strength as 2689 MN m - 2 and the bulk modulus as
192.6 GNm- 2 • Further information on properties is reported by Roy and
Hastert (1983).
Stewart and Bradt (1980a) have measured th~ fracture toughness of
vacuum hot pressed spinel in relation to grain size and temperature. At
room temperature Krc was found to be grain size independent with a value
of 1.90 ± 0.07 MNm- 312 over the range of grain size (5-38 p.m) studied.
This value is compatible with that for (111) monocrystalline material
(see below). Between room temperature and 900 ° C the Krc values were
found to decrease with increasing temperature at a rate, dKrc/dT of
-2.0 ± 1.1 X w-• MNm- 312 Oc-r with no grain size dependence.
Between 900 and 1400 °C the Krc values were found to decrease with
increasing temperature at the slightly increased rate, dKrc/dT, of
-2.5 ± 0.2x 10- 3 MNm- 312 °C- 1• No dependence on grain size was
observed. The authors found no obvious mechanism for the abrupt rate of
change of Krc with ter,nperature at 900 ° C and recommended additional
52 Bulk materials for the near and mid infrared
1200 o C. This compares reasonably well with the value reported by Roy
(1981) for Coors Porcelain material and listed in table 3.6. In a parallel
study Stewart and Bradt (1980b) investigated the fracture toughness of
monocrystalline spinel for the (100), (110) and (111) orientations from room
temperature to 1500 ° C. Two regions of fracture behaviour were observed;
a low temperature elastic region up to 900-1000 °C where K1c decreased
with increasing temperature, and an elevated temperature region where K1c
increased rapidly with increasing temperature. The elastic region was
explained by the decrease of elastic modulus with increasing temperature,
whereas the rapid increase of K1c at elevated temperature was attributed to
plastic flow in the vicinit,Y of the crack tip. The room temperature K1c data
found for the three crystal orientations were 1.18 ± 0.05 MNm- 312 (100),
1.54 ± 0.08 MNm- 312 (110) and 1.90 ± 0.06 MNm- 312 (111). Experimen-
tally derived values of dK1c/dTwere -2.5 x 10- 4 MNm- 312 °C (100) and
-1.7x 10- 4 MNm- 312 °C- 1 (111). Additionally Stewart et a! (1981)
demonstrated that the K1c values for the three crystal orientations followed
a linear dependence of elastic modulus.
Limited data on the thermo-structural evaluation of hot pressed spinel
domes have been reported by Strobel (1981). Tests on two domes showed
that they were capable of surviving heating conditions which had caused
thermal stress fracture in thirty previously tested magnesium fluoride
domes. Single water drop (2 mm) impact testing carried out by Hackworth
(1979) has indicated a damage threshold velocity of 396 ms- 1 for the (Ill)
orientation of monocrystalline spinel.
100
80
~
~
i'i 60
~
,g
••
~
~ 40
d
"'
20 I
I
I
0
0.2 0.5 1.0 5.0 10.0
Wavelength ().1ml
Figure 3.16 The transmittance of ALON 1.3 mm thick.
Applications utilising 3-5 p.m radiation, such as. thermal imaging, and ''
• I
requiring high resolution refracting optics need environmentally stable and ''
homogeneous materials from which to fabricate optical elements up to
about 150 mm diameter and several millimetres thick. In the 3-5 p.m
spectral region most ~aterials are sufficiently dispersive to necessitate the
54 Bulk materials for the near and mid infrared
Silicon
Over the last 30 years or sp, silicon has been developed as the world's major
semiconductor material. It is thus readily available in high quality and
quantity for use in infrared optical applications. The transmittance of
100 0 em n-type float zoned material is seen in figure 3.17. The transmit-
tance spectrum is free from major absorptions up to 6 !Lffi for thick samples
thus enabling many components to be designed into an imaging system
without fear of compromising the overall 3-5 !Lm transmittance of the
system. This would not be the case for the oxide materials which are only
useful in window thicknesses of a few millimetres as discussed earlier in this
chapter. The cut-on edge is consistent with the energy gap of 1.1 eV and the
lattice absorption bands of silicon extend from 6 !Lm to at least 30 !Lm .. The
latter have been investigated by Collins and Fan (1954) and Johnson (1959)
but they are not of any significance for 3-5 !Lffi applications.
100 '
:;"€
"- 60
~
v
~
0
\
:::
...
~
40
~
0
"' 2or
tions but n-type material would be preferred where high power require-
ments are a consideration. The effect of temperature on the free carrier I
electronic absorption at 4 I'm is insignificant until a temperature of about
.240 °C is reached, but the transmittance of a 3.mm thick sample of high
resistivity material is reduced from 52"7o at this temperature to 20"7o at a
temperature of 400 ° C. Thus there is no problem for optical components
inside systems, but the material would not be suitable for window applica-
tions where significant ( > 200 ° C) kinetic heating is anticipated. Extrinsic
impurity vibrational absorption is not a problem in this wavelength region.
For instance, interstitial oxygen in silicon gives rise to local mode
absorptions at 9 I'm and 19.5 I'm (Hrostowski and .Xaiser 1957) and the
vibrational absorption of carbon and carbon-oxygen complexes in silicon
occur beyond the 3-5 I'm region as discussed in detail by Newman and
Smith (1969).
The major optical properties of silicon are listed in table 3.5. The refrac-
tive index data were measured at the NPL, England, and they compare
favourably with those reported in the literature for wavelengths in the range
\1.3570-11.04 I'm by Villa (1972). Temperature coefficients of refractive
index data are taken from Hilton and Jones (1967). The optical homo-
geneity is likely to be good since high quality monocrystalline material for
substrate manufacture is routinely grown by the Czochralski technique
described in detail in Chapter 4. Equipment has been developed to pull
monocrystals up to 150 mm in diameter, a size most suitable for infrared
optical applications. The material is melted in a silica crucible inside a
carbon pot in vacuum. Corrosion of the silica crucible ensures that the
material contains some oxygen but this is unlikely to be a problem for
3-5 I'm optical applications as discussed previously.
Mechanical property data are much less well known for silicon than
electrical property data. This is because silicon has not been used for
demanding thermal or structural applications. Anthony and Hopkins (1981)
in utilising silicon for actively cooled cw laser mirror applications found it
necessary to measure some of the thermal and mechanical properties and
these are listed in table 3.6. On the basis of their test results, the tensile
strength was expected to be in excess of 35 MN m - 2 and Poisson's ratio was
found to be 0.20-0.28.
In earlier work on the fracture properties of silicon, St John (1975)
quoted Young's modulus as 155 GPa, Poisson's ratio as 0.215 and the
fracture toughness value quoted in table 3.6. Single drop (2 mm) rain
impact damage studies have been made by Hackworth (1979) for (100)
monocrystalline silicon. The damage threshold velocity was measured as
about 274 ms- 1 •
described in Chapter 5 are all useful in the 3-5 I'm spectral band. Physical
property data are given in these chapters but particular 3-5 I'm optical
data, where available, are listed for these materials in table 3.5. For the
alkali halides the data are taken from Li (1976), for KRS5 from Rodney and
Malitson (1956) and for AgCl from Tilton et a/ (1950).
80
~
r
"-
~
u 60
c
0
:::
·;;
~
c
0
40
.=
20
0
0.3 0.5 1.0 5.0 10.0
Wovelength l~ml
Figure 3.18 The transmittance of. Y203 12, Zr02 88 mole Ofo 1 mm
thick.
. I'
4
Bulk Optical Materials for the Far Infrared
4.0 Ge '
~
E
~As
g""30
. .,.....Ge-As-Se-Te glasses
r'
I ~'
I,CdTe '.,TI1173
•
KRSS
' - - .J
.
ZnSe
:ins
.
[sl
KBr NaCI
.
1.0!;;:--~--'-;~~~-~----:~~~--;';,.--"-~-~~--;;;i
1000 500 100 50 10
Reciprocal dispersive power {nw-1lf(n8-n12 }
air environment. Hence rain erosion, thermal shock and high temperature ''
transmittance properties become important in material selection.
Germanium becomes too absorbing above about 70 °C so that the most
favoured candidate materials are GaAs and ZnS. GaAs is useful to around
200 °C but is expensive and perhaps only likely to find limited use. ZnS is
therefore the most likely material to be used, but it does suffer from
transmittance limitations (8-10 l"m), particularly when hot. For i'
I'
applications requiring greater pass bandwidth, say 8-12 fLm, forward
looking infrared (FLIR) grade ZnSe would be suitable if its rain erosion
resistance could be improved. To retain wide band transmittance when hot
and to achieve a realistic rain erosion resistance, a composite window has
been suggested (Miles and Tustison 1979). This can take several forms but
is basically a substrate of FLIR grade ZnSe with either a layer of ZnS grown
onto it or a layer of ZnS or Si bonded onto it with chalcogenide glass. The
former is most likely to be successful if the technical manufacturing
problems can be overcome, thus achieving an 8-12 l"m transmittance
because of the reduced absorption of the composite structure, and the ZnS
exterior surface leading to satisfactory rain erosion resistance.
Thus Ge, GaAs, the chalcogenide glasses (Irtran ZnS, ZnSe, CdTe),
vapour grown ZnS and ZnSe are described in this chapter together with
present research on some rare earth ternary sulphide compounds, e.g.
CaLa2S., as possible second generation airborne windows.
4.1 Germanium
I
I
I
60 Bulk materials for the far infrared
0
B
0 \r
~
20 -
0
0.5 1.0 5 10 so
Wavelength (!J.m)
0.04
~ 0.20
~
c
w p -typ•
:g
'@ 0.10
2
-~ U.06
E-
0
~ 0.04
.c
<(
0.02 .
0.01
I .i
0
Resistivity (Qcml
figure 4.3 is always very much lower than the concentration necessary to
cause any detectable absorption by the dopant (Burnstein et al 1956).
The high level of purity needed to achieve the required electrical
characteristics ensures that bulk extrinsic optical absorption by impurities
is rarely of any significance. However, if the material were inexpertly
produced the electrical characteristics might well be achieved initially, but
extrinsic absorption at particular wavelengths or simply general absorption
might also occur. For example, oxygen contamination would result in
optical absorption at a wavelength of 11.7 I'm (Kaiser and Thurmond
1961). This contamination would only occur if there was a very high partial
pressure of oxygen in the gaseous ambient above the melt during the growth
of optical component material, since GeO and Ge02 formation and deposi-
tion on the cooler parts of the growth chamber would be likely to consume
a major part of the available oxygen. The curves shown in figure 4.4
Table 4.1
Transmittance (o/o) at
Resistivity
(0 em) 27 'c so 'c 10 'c
1.5 91.7 89.8 86.9
10.9 94.3 90.5 80.9
Germanium 63
Wavenumber (cm- 11
a1ro--,_~asro--,_~aw~-.--~9~
6.0
4.0
~ 2.0 f\
~ J\
8 1.0
c I
I 1
I
o I I
E-o.6 1 1
~ /
~ 0.4 / '
/ '' \
\
0.1 .·· \
\
\
0.1 L_-;;;!;;;;-;;-l;;;c---i~-;:;'o;;;--'----L;-,J \
11.00 11.76 11.50 11.10 11.0
Wavelength (!J.m)
0.05
0.01
I
0 20 40
Resistivity (Qcm)
During the early and mid 1970s most of the optical germanium was grown
in the polycrystalline form by the Stockbarger process. In this technique a
crucible, usually high purity carbon, containing the germanium together
with the n-type dopant is maintained at a constant temperature sufficiently
above the melting point of germanium to ensure that all of the contents are
molten, and then the crucible is physically lowered, through a temperature
gradient, into a second constant temperature zone maintained at a
temperature below the melting point (see figure 4.6(a)). After this lowering
process, in which the contents of the crucible solidify, the temperature is "
slowly reduced to that of the ambient. The material so produced consists
of many grains of the order of 10 mm or larger in dimensions, and can
be routinely grown in diameters of up to 300 mm. More recently the
Czochralski (or crystal pulling) technique has been used successfully for the
growth of large germanium crystals (Wilks 1959). In this growth configura-
tion, unlike that in Stockbarger growth, the growing crystal can be con-
tinually observed and is unconstrained mechanically as it grows and cools.
This has powerful advantages for the control of the crystal shape and
crystal perfection and has led to single crystal products which have
demonstrated superior optical performance over the polycrystalline
equivalent products. The process of crystal pulling is illustrated in figure
4.6(b). The germanium plus dopant is contained in a pure carbon crucible
which is heated resistively (in large commercial pullers) to above the melting
point. A pull rod or chain with a chuck containing a seed crystal at its lower
end is positioned above the centre of the melt surface. The seed crystal is
t
\..)
? <=J
{a) {b)
then dipped into the melt and the melt temperature adjusted until a
meniscus can be supported by the seed crystal. The pulling mehcanism is
then used to rotate and raise the seed crystal, and by carefully adjusting the
power supplied to the melt the crystal is grown to the desired diameter, up
to a maximum of approximately two thirds of the pot diameter. Thereafter
it is grown parallel sided until a crystal of sufficient length is produced or
until the melt is exhausted. The whole process is carried out inside a
chamber which permits the use of a controlled atmosphere, either gaseous
or vacuum, and which allows the crystal to be observed through suitable
windows. The material produced in this pulling process is either single
crystal or twinned material routinely grown in diameters up to 250 mm.
The Stockbarger process usually yields a single component blank from a
growth run although exceptionally a small number (1-4) of component
blanks can be cut from a deeper polycrystalline ingot. The pulling process
normally produces a ntbnber of component blanks (1-10) from one single
crystal ingot, with the exception of very large diameter material
(- 250 mm), where the crystal is usually only of sufficient length for one or
two component blanks at most. Thus for a large number of small diameter
components the pulling process is usually more economic and the quality
of the material is superior. Germanium required for optical components is
usually ordered in the form of shaped blanks. Thus the optical germanium
producer is able to saw his crystals into approximately plane parallel blanks
and then edge and shape them for shipping to component manufacturers,
whilst retaining the scrap germanium. This high value scrap germanium is
then either zone refined or recycled via GeC14 as already described.
i
Germanium 69
made that monocrystalline material was potentially of superior quality.
This led to a characterisation programme, both for polycrystalline and
monocrystalline germanium, to determine their relative merits in order to
achieve a material which could yield diffraction limited performance in
components. Initially the technique most frequently used to assess the
material was interferometry at 10.6 JLm.
,,I'
The most noticeable material defect was found to be the inhomogeneity
of refractive index. It was readily demonstrated, using literature data
(Moss 1959), that then-type doping used to achieve the required resistivity
of 5-40 n em had little effect on the refractive index homogeneity
(i.e. < I X 10- 4). It was therefore thought unlikely that this problem
originated from chemical composition differences, such as those which
occurred in multi-element chalcogenide glasses. It was considered that this
inhomogeneity amounting to between I X 10- 3 and I x 10- 4 , was probably
caused by strain imposed by the temperature profile present (Penning 1958)
during, and immediately after, solidification of large diameter mono- and
polycrystalline material. This was because the temperature gradients in
scaled up crystal growth equipment were likely to be much greater than in
the smaller equipment used for semiconductor applications. This proved to
be the cause of the problem. A typical example of a monocrystalline optical
blank exhibiting a refractive index homogeneity of the order of 3 x 10- 4 is
seen in figure 4.7(a). After annealing at 850 °C, the same blank was found
to exhibit a refractive index homogeneity of 6. 7 X 10- 5 (Gaskin and Lewis
1980) as seen in figure 4.7(b). The radial inhomogeneity seen in figure 4.7(a)
(a) (b)
to equally spaced intervals (a) before annealing, key is 0-3.117 x w-• in steps of I
0.5195 x w-• (b) after annealing, key is 0-0.6726 x w-• in steps of 0.1121 x w-•
I I
'
I'
70 Bulk materials for the far infrared
results from crystal rotation in the pulling process ·and is less damaging to
the overall imaging quality than that seen in polycrystalline material due to
its regular radial form: As the polycrystalline germanium is not rotated
during growth, its inhomogeneities are of a less symmetrical form. The
manufacturers of optical germanium have modified their techniques to
reduce the temperature gradients during growth and hence increase the
refractive index homogeneity of the product. This avoids any need for post
growth annealing, with its associated risk of conductivity type conversion
for material up to about 150 mm diameter.
During this work it was found that single crystal and polycrystalline
germanium could be readily produced with equally low and acceptable
refractive index variations of 0.0002 or less for 10 mm thick plane parallel
blanks. These were essentially equivalent in quality as characterised by
interferometry at 10.6 ,im (Lewis et a/1979). However, other tests showed
that monocrystalline material was likely to yield an overall superior perfor-
mance. The grain boundaries in polycrystalline material were found to be
lossy leading to transmittance non-uniformity, and a typical total loss of
about 1% in transmittance for a 10 mm thick blank. Additionally when
polycrystalline germanium lenses were examined by pupil-scanning equip-
ment to measure the transverse ray aberrations (TRA), a reproducible fine
structure was observed on the plots of data, and this was attributed to
localised refractive index variations in the region of crystal grain boundaries
(Haig et all976). To confirm this,lenses were constructed (Lewis et al1979)
from directionally solidified polycrystalline material and twinned
monocrystalline materiaL The transfer functions of these were measured by
pupil-scanning using the TRA and directly using a grating scanning instru-
ment (Williams 1974). The TRA plot of the lens from the twinned crystal
was ideal in form except for a deviation at a position precisely that of the
twin boundary, thus showing that TRA anomalies can be associated with
the existence of a grain boundary. In contrast, the TRA plot of the lens
made from polycrystalline germanium had a noticeable fine structure
superimposed upon the expected shape of the wavefront, and this was
assigned to the grain boundaries. The phenomena were not detected in
interferometric measurements since this technique provides the wavefront
directly and the fringe visualisation pyroelectric cameras used had low
spatial resolution (figure 4.8 (c)). The MTF plot for the polycrystalline lens
obtained from the TRA data was in excellent agreement with the MTF
obtained on a moire fringe grating scanning apparatus and was consistently
below the theoretical MTF plot. For instance, at 10 line pairs/mm, the
measured MTF was 15"7o less than that of the theoretical ·MTF. It was
reasoned that TRA deviations leading to loss in MTF performance were
caused by localised refractive index variations at grain boundaries. As
discussed previously, these were unlikely to be from compositional varia-
tions, even allowing for the excess dope incorporation at grain boundaries,
Germanium 71
. and were more likely-to be because of localised strain and/or lattice defects
at the boundaries. Hence it was concluded that in polycrystalline ger-
manium, dope accumulations resulting in increased resistivity, particularly at
grain boundaries, were responsible for reduced transmissions, as illustrated
in figures 4.8(a) and (b): Moreover the refractive index variations at grain
boundaries (figure 4.8(c)) were considered to be a result of localised strain
and/or lattice defects arising from the manufacturing process. While the
former conclusion is well established, the latter emerges from few experi-
ments and would benefit from further confirmation. Thus it is seen how the
ultimate optical performance of germanium is critically dependent on
I'
material quality, and it is not normally a trivial manufacturing problem to
produce material of the required standard routinely. Clearly monocrystalline
germanium is likely to offer superior performance over that of
polycrystalline material for requirements where diffraction limited perfor-
mance needs to be achieved.
(a) (b)
(c)
72 Bulk materials for the far infrared
Get 0.67 0.02 4.0054 4.0041 4.0032 4.0025 4.0019 4.0015 4.0012 858 3.90
GaAst 1.38 0.02 3.2867 3.2820 3.2769 3.2710 3.2648 3.2576 3.2495 104 1.48
Irtran 2 ZnS 3.60 0.29 2.2213 2.2107 2.1986 2.1846 2.1688 2.1508 - 23
Irtran 4 ZnSe 2.67 0.13 2.418 2.413 2.407 2.401 2.394 2.386 2.378 59 0.48
Irtran 6 CdTe 1.50 0.28 2.677 2.674 2.672 2.669 2.666 2.663 2.660 152 0.93
Early Raytrant ZnS 2.2235 2.2131 2.2010 2.1871 2.1713
Raytran product 3.60 0.24 - - 23 0.50
infoFmation 2.2228 2.2123 2.2002 2.186 2.170
Early Raytrant ZnSe ' 2.4178 2.4127 2.4070 2.4006 2.3935 2.3856 2.3768
Raytran product 2,67 0.0005 58 0.60
information 2.4173 2.4122 2.4065 2.4001 2.3930 2.3850 2.3762
1 -200 0
10'
creep
Diffusional
( Nabarro
Elastic regime
Homologous temperature
I
Figure 4.9 Deformation mechanism map for germanium.
deformation. The values of stress and temperature for which some of these
mechanisms control the deformation are seen in figure 4.9. Considerable
information on elastic constants of germanium has been provided by Fine
(1953, 1955), who has shown that Young's modulus has a marked orien-
tation dependence. At room temperature a value of 103 GPa was obtained
for the (100) direction and 156 GPa for the (Ill) direction. The lower value
is quoted in table 4.3 for general use, since it is likely that some
monocrystalline material used may be of unknown orientation. In the same
work, Poisson's ratio was determined as 0.279. Definitive data for rupture
modulus are difficult to obtain, since the results are markedly dependent on
the surface finish, and a large number of samples would need to be tested
over a reasonably long production run. Since this would be expensive much
more limited, but nevertheless useful, testing has been done. Metallurgie
Hoboken Overpelt are one of the world's major producers of ger-
manium and its products, and their technical literature quotes a value of
93 MPa for the rupture modulus. In recent work Goode (1977) has
measured the rupture modulus of bars of germanium, of dimensions
150 mm x 25 mm x 6 mm, with standardised surface finish, in a four-point
bending test. The twenty bars of polycrystalline germanium tested yielded
a rupture modulus value of 84 ± 9 MPa, while nine bars of monocrystalline
germanium gave a value of 95 ± 16 MPa. These results, not considered to
be significantly different, are in good agreement with the value quoted by
Hoboken and are given as a representative value in table 4.3. A hydraulic
bursting-pressure technique has been developed (Matthewson and Field
1980) to minimise the difficulties frequently encountered of edge breakages
in the four-point loading test method. This test method is likely to be
favoured if the results from it prove to be more representative of intrinsic
Table 4.3 General physical properties of 8-12 I'm transmitting materials.
Thermal
expansion Thermal Knoop Rupture Young's
Melting Density coefficient conductivity Specific heat hardness modulus modulus
Material point (C) (103 kg m- 3 ) (lo-• K- 1) (IO-'calcm- 1 , - 1 K- 1) (10- 2 calg- 1 'c- 1) (kgmm- 2 ) (MPa) (GPa)
values, and less affected by extrinsic effects, such as surface flaws. As ger-
manium may well be used on low speed aircraft and helicopters, some in-
dication of its rain erosion properties is relevant. Hooker (1977) has studied
the initiation of erosion in germanium at subsonic velocities. Specimens
were exposed to a standard 2.5 em h _, rainfall rate of 1.8 mm diameter
drops impacting at 222.5 ms- 1 • It was shown that the pits initiated were
followed by severe surface and subsurface cracking. The fractures occurred
by means of (111) cleavages and irregular fractures analogous to those
observed in bend tests. After 14 impacts/cm2 , each specimen contained one
region of detectable damage of dimensions 1-2 mm. The longest exposure
of 245 impacts/cm2 produced a 10% transmittance loss at 10 ,urn
wavelength. The effect of surface damage on the infrared optical perform-
ance of germanium is also discussed by Lewis and Jennings (1982). Some
other physical properti~s of germanium, such as density, hardness and
melting point taken from Metallurgie Hoboken technical literature, are also
given in table 4.3.
T!To at 10 I'm
Over the past two decades chalcogenide glasses have been researched to
assess their suitability as passive optical component materials for 8-12 I'm
wavelengths, and as active electronic device components in photocopying
and switching applications. The theoretical and experimental work done
as a result of this has led to a greater understanding of the range of glass
formation and the general physical properties of these materials.
Chalcogenide glasses are so named because they contain one or more of the
chalcogenide elements S, Se or Te, together with one or more of the
elements Ge, Si, As, Sb and a number of others. They are mainly covalently
bonded materials with room temperature resistivities of 103-10 13 Dcm. For
instance, As 2S3 has a resistivity of around 2 x 10 12 n em (130 o C), As2Se3
a resistivity of around 1.5 x 108 Dcm (130 °C), Sea resistivity of around
2 x 104 n em (120 °C) .and As 2SeTe 2 a resistivity of about 3.5 x 103 n em
80 Bulk materials for the jar infrared
(130 °C}. The conductivity activation energy of these glasses varies from
0..'-1.25 eV while the optical energy gap approximates to that of the
crystalline analogues where these exist (Edmond 1968). Before discussing
the optical properties, it is of interest to note how the band structure of the
chalcogenide glasses arises and how this differs from conventional semicon-
ductors. Materials possessing resistivities in the lower part of the range
(selenide and telluride glasses) are considered for electronic applica-
tions and this has created a major interest in their electronic properties and
conduction mechanisms beginning at the Leningrad school in the mid 1950s
to the mid 1960s. During this period the basic electrical properties were
established and considerable phase diagram information was obtained. A
start was made in understanding the chalcogenide electronic structure and
conduction mechanisms. Kolomiets (1964a) showed that the concepts of a
conduction and a valance band could be applied and that the gap, and hence
the conductivity, did not depend sensitively on composition, as it does in
the case of crystalline semiconductors. At the same time this phenomenon
was given a general explanation in chemical terms by suggesting that each
atom in the glass had the correct number of near neighbours to enable all
its electrons to be taken up in bonds. However, from the mid 1960s to the
mid 1970s there was a worldwide explosion of interest in threshold and
memory thin-film telluride glass electronic switches, particularly championed
by the USA (Ovshinsky 1968). Although this has not led to any major new
commercial exploitation, it has significantly advanced the theoretical
knowledge of conduction mechanisms in amorphous semiconductors, and
in particular in the chalcogenide glasses. At first the band structure model
assumed that tails of localised states extended into the gap at the band edges
acting as a continuous distribution of traps. Bonding defects were supposed
to give rise to localised energy states in the band gap (i.e. deep donor- and
acceptor-like levels pinning the Fermi level in the middle of the gap (Cohen
et a/ 1969, Matt et a/ 1975)). Then Anderson (1975) suggested that two
distinct electronic spectra existed in an amorphous material. One spectrum
consisted of extended states (normal bonding), the other consisted of a two-
. electron spectrum (defective bonding) which was strongly localised. The
two-electron spectrum led to the pinning of the Fermi level, to the material
being diamagnetic and to the observed high density of localised gap states
(10 19 em - 3) with their lack of optical absorption. This model was a large
step forward in understanding but failed to explain why double occupancy
was permitted in chalcogenides and not, for instance, in amorphous silicon
which can be doped in a manner similar to crystalline semiconductors.
Meanwhile, Kastner (1972) pointed out the importance of elements in which
the top of the valence band consists of lone pair (LP) orbitals, for instance
in selenium the p-orbitals which do not take part in a bond. Group VI and
group V elements form LP bonds but the group IV elements do not. Hence
here was an explanation (Kastner et a/1976) of why the electronic properties
Chalcogenide glasses 81
Temperature coefficient Rl
Glass (atomic o/o) n, n• n, v,_, ns nw n12 Vs-12 oo-' oc- 1
)
Ia)
40
{b)
...
~ '
w I
~ 0
~
~
:;;
F lei
40
Id)
Wavelength (l..lm)
Figure 4.10 Transmittance of: (a) sulphide glass atomic "loGe 30, As
20, S 50, 1.9 mm thick (fulJ.curve), extrinsic impurity absorptions due
to H 20, H2S oxide and carbon (broken curve); (b) selenide glass atomic
% Ge 34, As 8, Se 58, 1.8 mm thick (full curve), extrinsic impurity
absorptions due to oxide (broken curve); (c) Selenide-telluride glass
atomic% Ge 30, As 13, Se 27, Te 30, 2.3 mm thick (full curve), extrinsic
impurity absorptions due to oxide (broken curve); (d) telluride glass
atomic % Ge 10, As 50, Te 40, 1.6 mm thick (full curve), extrinsic
impurity absorptions due to oxide (broken curve).
Fuxi eta/ (1983) have reported devitrification and property studies in the
Ge-As-S glass system. They found that because of stable glass formation
in this system, devitrification was difficult mainly occurring at low As
Chalcogenide glasses 87
10 80
60
"~ 40
80 10
As 20 40 60 80 Ge
Figure 4.11 Glass forming region versus atomic "lo for Ge-As-S
glasses indicating the range of glass transition temperature T,.
10 0
w
w
w (a)
• I• (b)
-.-I J ~~~w m
.._. 1 . .-:·, ../
;~. / >.: 11
.. I
·, 1 '-,
\\
0
\[} I
. ..... ..... ~
40
w w ..,\. 1(···. '\ 11
\":j ··.\
0
II I • • ."1
\ I
'.
".._
\
-.,-
7 9 11 13 7 9 11 13 15
Wavelength (IJm)
batches too high for the required applications. Hence, emphasis was placed
on solving the extrinsic absorption problems associated with the sealed tube
process.
s. s.
20 0 20 80
~
.~\'v
40 'l."'~c
________
'~':
60
60 40 40
BO 10 80 20
I
Sb 10 40 60 BO ,, A< L---~,~,----T-40.---~60.---~,~,--~,.
Figure 4.13 Glass forming region Figure 4.14 Glass forming region
versus atomic "7o for Ge-Sb-Se glasses atomic "7o for Ge-As-Se glasses
illustrating the range of glass transition illustrating the range of glass transition
temperature T,. temperature T, and the thermal expan-
sion coefficient a.
1.6 +
+
'.
u A B
0
0
~
~
1.2 I
c
+ • i I
""'., 0.8
.
~
~
~
•• •
8
c
~ +.
~
5
~
D
i'
«
0 5 10 15 20
Oxygen content {ppm wtl
10.6 I'm· This is in broad agreement with the' data given above for
Ge-Sb-Se glasses and that below for Ge-As-Se-Te glasses.
Glasses Ge 28, Sb 12, Se 60; Ge 33, As 12, Se 55 (Hilton 1978) and
Ge 30, As 15, Se 55 have been produced in quantities of several kilograms
by the sealed tube process from semiconductor-grade raw materials. After
synthesis the glass boules are either annealed and cooled, or undergo further
homogenisation (Hilton 1970) before annealing and cooling to room
temperature. Components are either cut from the glass or heat slumped to
shape from slices of the glass. The refractive index homogeneity require-
ments are similar to those of germanium described previously. The main
problem in achieving these in chalcogenide glasses has been in maintaining
a homogeneous chemical composition during the cooling process which
follows the synthesis al)d mixing, as the composition of the vapour species
is not necessarily similar to that of the liquid glass. While this is not a
problem for experimental melts under 100 g, it becomes more serious for
several kilograms since the available vapour space is larger and the
temperature gradients tend to be greater. If any condensate with a different
chemical composition from the bulk is allowed to contaminate the liquid
glass after homogenisation during cooling then the viscosity of the glass is
low enough for limited intermixing and hence local compositional varia-
tions. These can lead to refractive index variations of 3 x 10- 3 or greater.
However, these problems have now been largely overcome (Worralll979).
applications. The glasses from the Ge-As-Te system (Savage 1971) possess
lower glass transition temperatures and are much less thermally stable than
those in the Ge-As-Se system (Savage and Nielsen 1966). Therefore high
tellurium substitutions would be expected to decrease their glass transition
temperatures and reduce their thermal stabilities. For this reason very stable
base glasses containing about 20-30"7o Ge and 10-30% As were initially
chosen. The area of investigation in the Ge-As-Se system is shown by the
dotted line in figure 4.14 where the ternary base glass thermal expansion i! :
coefficients and glass transition temperatures are also indicated. From
within this base glass area 40 glass melts were made, substituting up to 30%
Te for Se and then characterised by differential thermal analysis. On the
basis of the thermal properties and glass stability one base glass, Ge 30, As
13, Se 57 was chosen for investigation of optical properties and the results
are expected to be fairly typical of other quaternary glasses, containing
about 30% Ge and 10-15% As. It was found that under 30% Te could be
substituted for Se in this glass, but amounts of Te exceeding 30% caused
devitrification. Hence three Te substitutions of 10, 20 and 30% were made
for detailed optical property measurement. The basic physical properties of
the four glasses are listed in table 4.6.
A further extrinsic absorption problem was encountered with these
quaternary glasses. Te cannot be purified from the Te02 surface impurity
by thermal baking in vacuum, since the vapour pressure of the metal is
greater than that of the oxide. In this case an acid etching technique (Savage
et a/1980) was used which allowed glass of adequate purity to be made by
the sealed tube technique. Several melts of glass Ge 30, As 13, Se 27, Te
30 were analysed for oxygen by a gamma photon activation technique and
the measured range of within the 1-10 ppm wt oxygen content of the glass
was found to be given by
Y = O.D78 + 0.128x (4.3)
where x is the oxygen content in ppm wt and Y is the absorption coefficient
at 780 em - 1 • An absorption coefficient of 7 x 10---1:3 em - 1 was obtained for
this glass at 10.6 ,urn by laser calorimetry which correlates very well with the
values given previously for Ge 23.5, Sb 18.0, Se 58.5 and As 40, Se 60
glasses. The range of reciprocal dispersive powers is 110-185 and the refrac-
tive index varies from 2.47 to 2.86 at 10 ,urn when up to 30% Te is
substituted for Se (table 4.5). The glasses all possess very acceptable
physical characteristics for infrared optical applications. The optical
properties are compared to those of selenide glasses in figure 4.16 where it
is seen that a small family of optical glasses for use in the 8-12 ,urn spectral
region is achievable. The effect of 30% Te substitution on the short
wavelength end of the spectral window is to move the absorption edge from
0.6 to 0. 7 ,urn to 1.1-1.2 ,urn and to improve the transmission slightly at
longer infrared wavelengths (figure 4.10(c)). A good general review of
<)4 Bulk materials for the far infrared
2.9,-----.-----,------.----,------,
\--------
·16 '
I '
I
I
I
'
I '
\ •15 '
\ '
''
''
•5 '
•14 ' ' ,
•10'
''
11• •9 ' '
•4 •13 '
' ' •8
I
' ,.)_--- -- - - - - -- --
2.3'-----:0;;;----.;;:--------.;-,----------,,.;;,;----7,
~ ~ ~ m 100
Reciprocal dispersive power V8_12
mixed As 2 (Se Teh glasses including some optical properties data, is given
by Thornburg (1973) and this complements data already referred to on
As2S3 and As2Se3.
Te
--~
20 //' ) 80
/ I
I
(..... __ ......__,..;.
; /."
40 60
/.,..-~ ..\
/ \
I I
60 I
I
:,____ ,__
...... /
I
__
) 40
BO 20
P,As 20 40 60 BO Si,Ge
Figure 4.17 ~ Glass forming regions versus atomi~ "lo for Ge-As-Te
(broken curve), Ge-P-Te (dotted curve) and Si-As-Te (full curve)
glasses.
During the late 1940s and early 1950s germanium and silicon were
developed as semiconductor materials for device manufacture. As these
materials were gradually established the attention of some research workers
turned to other families of materials, first the III-V compounds such as
GaAs and then the II-VI compounds such as CdS, ZnS, ZnSe and CdTe.
One of the initial reas9ns for studying the II-VI compounds was their
96 Bulk materials for the jar in/fared
useful luminescence properties. Table 4.9 lists the vapour pressure at the
melting point for three of the most widely used II-VI materials.
The vapour pressure at the melting point of a II-VI compound is not
excessively high but does represent a severe problem for crystal growth from
the melt. Without suppression of the dissociation pressure, by one of the
components, or by a high inert gas pressure, the material is transported
through the vapour phase and condensed on any cooler region in the growth
system. This was more difficult to overcome with the technology available
in the 1950s than it would be now, so it is not surprising that the early
workers turned their attention to growth from the vapour. This can be
carried out at significantly lower temperatures than those shown in table
4.9, since both components of each II-VI compound possess high vapour
pressures, thus making growth by vapour transport possible with relatively
simple equipment. The basic requirement is a continuous supply of the
elements through the gas phase, either from dissociation of the preformed
compound or from separate sources. Dynamic and static techniques have
been developed. Frerichs (1946) first grew CdS by a dynamic technique
using a carrier gas to transport metal vapour from a boat of liquid
cadmium, and introducing H2S gas to this vapour stream in a region where
plates, ribbons and needles of CdS grew on the reaction tube walls. Essen-
tially a scaled-up and refined version of this technique is still used for the
growth of large area polycrystalline ZnS and ZnSe for infrared applica-
tions. Static techniques of crystal growth have also been developed in which
material is transported by diffusion through the gaseous phase. The source
is a compound powder or sintered solid in a high temperature region
{1550-1600 •c for ZnS) of a closed container, and crystal growth occurs
by dissociation to the vapour species and diffusion to a lower temperature
region (1457-1500 •c for ZnS). This method was reported by Greene eta/
(1958) and a further modification of it by Piper and Polich (1961). This
technique has since been widely used on a laboratory scale for the growth
of many monocrystalline II-VI compounds (Nitsche et a/ 1961). Many
II-IV compounds have been grown very successfully by iodine transport at
temperatures below 1000 o C for uses where iodine contamination is
unimportant. Meanwhile Fischer (1958, 1959) looked at melt growth for
ZnS and ZnSe and later for other II-VI materials using high pressure
II- VI compounds 97
autoclave equipment (Fischer 1963). Only CdTe can be grown without the
need for this expensive high pressure autoclave equipment because of its
relatively modest vapour pressure at its low melting point (1092 °C).
During the late 1950s and the early 1960s workers in the infrared field
became interested in some II-VI materials, such as ZnS, ZnSe and CdTe
for use as optical components in sizes up to 100 mm diameter. It is clear
from this description of II-VI compound growth development that the
vapour growth was just becoming established, and early melt-growth
experiments were just taking place at this time. However these only
produced small laboratory scale crystals so it is not surprising that materials
specialists turned their attention to the more familiar ceramics technique of
hot pressing for the first major production, thus avoiding the problems of
large scale vapour or melt growth. Fine powders of some II-IV com-
pounds, synthesised for earlier luminescence studies, were available as a
source of raw material for the hot pressing experiments. Sulphide powders
were normally prepared by precipitation in alkaline or acidic solutions
(Laverenz 1950). In the alkali process ZnO or ZnCh was dissolved in
aqueous ammonia solution and after several purification steps a zinc
complex was precipitated by HzS
Zn(NH3)4Ch + HzS + HzO->ZnS~ + 2N~Cl + 2NH40H. (4.4)
Using the acid process, zinc metal was dissolved in HzS04 solution and after
several purification steps ZnS was precipitated from the acid solution with
HzS
(4.5)
The average particle size was about 0.1 JLm. To synthesise ZnSe, a solution
of ZnSe0 3 was reduced by hydrazine to yield ZnSe.NzH4 precipitate. This
complex was then decomposed by reaction with acetic acid (Benzing et at
1958) or thermally decomposed to avoid acetate contamination (Gelling and
Haanstra 1961). 1
The main problem with all of these powder synthesis techniques is
contamination, particularly with oxides, which leads to extrinsic impurity
absorptions in the subsequently hot pressed material. A prime example of
this is S04 and S0 3 absorptions in ZnS. The necessity of working with such
fine particle size powders makes this task of attaining sufficient purity rather
difficult, but it is a key parameter in successfully manufacturing hot pressed
components.
established to avoid the setbacks associated with producing large size mono-
or polycrystalline components from materials possessing high melting
points and/or high vapour pressures. However, this method raised other
questions such as contamination, non-uniformity and scatter. Each
material started as a chemically pure grade powder (to avoid major extrinsic
impurity absorptions) which was compressed to shape in a mould while
being heated by induction. Athough the powders were pressed to near the
theoretical density so that there was no evidence of macroporosity, visual
and near IR (- 3 11m) scatter were still present at the end. Problems
occurred in maintaining fine powders in an uncontaminated state during the
pressing operation and absorption bands were sometimes present in the
spectra of the products. Removing the pressed pieces from the dies
sometimes resulted in cracking. To avoid this release agents, such as aqua
dag, boron nitride powder or graphite paper were used leading to possible
further contamination. However, the small grain size of the pressed pieces
(1-5 11m) meant that the mechanical strength was likely to be greater than
similar size single crystal components. This was shown to be a versatile
technique in principle, but it was best suited to the production of large
numbers of small flat components cut from a large diameter pressing
( -180 mm). The extrinsic contamination and scatter restricted the hot
pressed material to mainly low power component applications.
The mechanisms by which fine powders are densified to polycrystalline
bodies of near theoretical density have been described in the literature.
When a collection of powder particles of uniform composition is held at a
high temperature, any change of shape that the mass undergoes is termed
sintering. In the absence of any externally applied pressure such changes
occur as a result of surface tension because the surface free energy decreases
as the particles grow together and assume a more compact shape. The
transport of matter can take place by any of four mechanisms, viscous or
plastic flow, evaporation and condensation, volume diffusion, and surface
migration (Herring 1950). When external pressure is applied, as in hot
pressing, plastic deformation becomes the primary mechanism of compac-
tion with the other mechanisms being either of no or purely secondary
importance. This depends on the temperature in relation to the thermal and
mechanical properties of the material being pressed.
Zinc su/phide-lrtran 2
This material was pressed in tungsten or molybdenum dies in the
temperature range 800-870 °C at 232-309 MNm- 2 for 15 min (Eastman
Kodak Co BP 934,421). It was a translucent material, creamy beige to dark
green in colour, available in flat blank sizes up to 200 mm diameter and
25 mm thick. The transmittance and the emittance (Schleiger and Webb
1968) of the best quality material is shown in figure 4.18(a), together with
an indication of the most common extrinsic absorption caused by sulphate
II-:- VI compounds 99
I I
60 ,,
\ I (a)
"
I
. 40
I
I
........ __________ / I
...
~
~
0
c
,g (b)
·;;;
~
c 60 '\I I
I v
0
F
" 0
I
40
I
I
20
0
-. --·-- ·-. I
0.4 1.0 5 10 50
Waveler>Jth (llffi)
Zinc selenide-Irtran 4 I
Zinc selenide powder was pressed in a molybdenum die at 982 o C and
207 MNm- 2 for 5-60 min after prebaking in rotary-pump vacuum at
1121 °C for 15 min (Roy and Parsons 1965). This prebaking appeared to
be important in removing extrinsic impurities (Benecke 1971) but also
obviously resulted in some grain growth, thus increasing the grain size and
reducing the overall strength of the final product. This hot pressed material
had a transparent mid-brown colour exhibiting less scatter at visible
wavelengths than any of the other visually transmitting Irtran materials. It
was available in flat blank sizes up to 180 mm diameter and 13 mm thick.
The transmittance and the emittance (Schleiger and Webb 1968) of the best
quality material is shown in figure 4.18(b), together with an indication of
the most common extrinsic absorption apparently caused by sulphate at
9 JLm. The importance. of prebaking in obtaining good overall transmittance
100 Bulk materials for the jar infrared
60
B
Ia}
60
40
20
~ A
~
u
~ 0
,g
·s
~
lbl
~
0
60
"'
40
20
0
0.5 1.0 5 10 50
Wavelength (IJ.m)
Figure 4.19 (a) Transmittance of hot pressed ZnSe 1.9 mm thick (A)
illustrating the dramatic improvement after prebaking at about 1100 o C
(B). (b) Transmittance of hot pressed CdTe 3 mm thick.
Cadmium telluride-Irtran 6
Cadmium telluride powder was pressed in a molybdenum die at 650-850 o C
at 207 MNm -z for 30-45 min after presoaking at the above temperature for
10-30 min without pressure. This product was very dark brown and was
available in flat blank sizes, up to 150 mm diameter and 6 mm thick, also
up to 75 mm diameter and 13 mm thick. The transmittance range of the
best quality material is shown in figure 4.20(b) and some of the optical and
mechanical properties are listed in tables 4.2 and 4.3, respectively.
plates with larger overall area and higher imaging quality than was possible
with the hot pressing process was clearly preferred. Vapour growth was
singled out, since much more work had been done on this technique than
on the melt growth of II-VI materials. Also large area vapour deposition
of carbon parts for rocket motors and the nuclear industry had been
achieved in CVD equipments constructed to chemical engineering stand-
ards. Thus all of the ideas and sufficient practical experience was available
to move into the large scale CVD growth of II-VI materials. Many CVD
processes are preferably carried out at less than atmospheric pressure,
I
I,
resulting in an increase in the diffusivity of the gaseous species such that
surface reaction tends to be the rate-determining step in the synthesis
(Bryant 1972). This low pressure CVD process has been used, for example,
in the deposition of carbon and alumina (Schaffer 1965) and was also
applied to the II-VI materials (Miles 1973). Around 1970, workers at the
Raytheon Company Research Division, Massachusetts, USA, made a very
significant breakthrough in II-VI material growth by cvo which has since
become a well established chemical engineering technique used for ZnS and
ZnSe. Essentially, the dynamic growth technique first used by Frerichs
(1946) for CdS was adapted to suit ZnSe. The reactants, H 2Se gas and Zn
vapour, transported from a large liquid Zn reservoir by an inert gas, were
ducted into a vertical rectangular growth chamber whose walls acted as the
substrates for large area polycrystalline deposition, I em or more in
thickness. This innovative use of wall deposits as the main product neatly
overcame the usual problem of unwanted wall deposits when the substrates
are separately mounted inside a cvo growth chamber. On cooling at the
end of a production run the 'flat plate wall deposits' were removed from
the substrates for grinding and polishing into components. Furthermore,
plates from a recessed-growth chamber allowed shaped components such as
missile domes to be grown in this versatile equipment. The development of
this process was quite rapid, Pappis (1971) reported that usable ZnSe and
CdTe had not been grown by CVD, but results 1 on ZnS indicated that
suitable IR material could be made. The ZnS product suffered from low
angle scatter caused by the presence of pores and this was very obvious from
the transmission curves reported at that time. A year later Pappis et a/
(1972) reported that high optical quality ZnSe with measured absorption
coefficients between 0.004 and 0.007 em - I had been made, and that the ZnS
material had been greatly improved. The scatter loss in the ZnS had been
reduced by growing zinc rich material which exhibited a ZnH2 stretching
absorption at 6.0 p.m. Deposits of ZnS and ZnSe up to 13 mm thick and
430 x 600 mm 2 in area had been made, and physical property data for this
material were reported. Miles (1974) stated that equipment was available
to manufacture one metre square deposits and many physical properties
results were given. Pappis et a/ (1976) indicated that properties such as
spectral transmittance,. hardness, grain size, flexural strength and image
102 Bulk materials for the jar infrared
t
•k
Dllllp box
H,
H
H1S
•
Zn
"'t
Figure 4.20 Schematic representation of an industrial scale CVD plant
for the growth of ZnS and ZnSe. H is the heater and Zn metal is the
hatched area.
II- VI compounds 103
80
.
60
/
/
/
/ --
-~
~
\ I
l_t
;-
"'
Ia I
-
I
40
, --
I
I
1:=--
,_
,_'-
~ 20 1-
'-- L
:':
~
,g
·e;
0
.
~
\b)
~
0
"' 6Of-
{
-
40
20
0 I
0.3 1.0 5 ' 10 50
Wavelength l~m I
Figure 4.21 (a) Transmittance of CVD ZnS 6 mm thick: multispectral
material (full curve); 8-12 I'm infrared quality material (broken curve);
horizontal bars represent typical product variation due to scatter. (b)
Transmittance of cvo ZnSe 10 mm thick.
it exhibited visual scatter and strong Zn- H infrared absorption at 6.2 I'm.
Later material was Jess dark and exhibited less visual scatter. The spread of
transmittance of Raytheon production ZnS due to scatter is indicated by the
horizontal bars in figure 4.2l(a) and is taken from sales literature. Similar
literature for CVD Inc ZnS products indicates a rather better transmittance
shown by the broken curve in figure 4.2l(a). This curve is reasonably typical
of current good imaging quality 8-12 I'm infrared-grade ZnS exhibiting
Zn-H absorption.
Table 4.10 Critical point analysis: two-phonon summation spectrum in cubic ZnS.
•
106 Bulk materials for the far infrared
~so
.
_..-· ~-----·-
:;; ~- ------·
!a 40
60 ./.··
·"
~20 ./·
!{
0 2 4
Total absorbance at 393 nm {arbitrary units)
The significant optical properties of 8-12 I'm quality ZnS are given in
table 4.2. The early Raytran refractive index data were obtained on dark
orange brown material at the NPL while the product information is taken
from Schott information (1982a) for current Raytran material ·and similar
figures are available for CVD Inc material. Both sets of data show higher
values of refractive index than the Irtran material, which is consistent with
its less dense hot pressed physical form. There is no information on
refractive index variation with the different products in terms of scatter and
1
colour. However, in table 5.2 data are given for the visually clear multi-
spectral Raytran material (Cleartran) and it is seen that the refractive index
varies by only about 0.0002 between this material and the opposite extreme,
the very dark brown 8-12 I'm quality material (table 4.2) when measured
on the same equipment. Hence it would appear that any product variation
of the 8-12 I'm quality material is unlikely to affect the far infrared refrac-
tive index except perhaps in the fourth decimal place. The absorption at
10.0 I'm (Klein et a/1979) and the temperature coefficient of refractive index
are also given in table 4.2. The image spoiling properties of ZnS are of
considr.rable importance and published information indicates that these are
satisfactory (Connolly et a/ 1979). It is reported that standard 8-12 I'm
quality ZnS made by CVD Inc shows a refractive index homogeneity of
under 100 ppm (Donadio et a! 1981) and this correlates with Raytheon
108 Bulk materials for the far infrared
the entire grain boundary may be impacted. In these conditions the grain
boundary mechanical properties, influenced by the fabrication technique,
may be important in relation to rain erosion to a larger extent than might
be indicated by the bulk mechanical strength.
In later studies Hackworth (1979) showed that in single 2.0 mm diameter
water drop impact studies fracture was initiated in vapour deposited ZnS
at slightly below 175 m s - 1 • Tests were also conducted in a small drop rain-
field with a mean drop diameter of 0. 75 mm and a rainfall rate of
1.0 cmh- 1 • At 90° impact angle the threshold velocity for damage to occur
was slightly less than 192 ms- 1 • Table 4.11 presents representative samples
of the results generated by the experiment to investigate velocity and impact
angle in the small drop rainfield. The data illustrate three general features
observed for all of the specimens:
(i) the transmittance at shorter wavelengths was more sensitive to
damage,
(ii) the loss of transmittance was approximately linear with exposure
for all wavelengths,
(iii) the loss of transmittance at short and long wavelengths were
closely related.
Table 4.11 Representative data from small drop rainfield experiments with ZnS.
17c 222 90 0 71 70 72 71
6 67 68 70 69
12 63 66 68 67
18 I 58 64 67 65
30 51 58 61 60
7c 340 45 0 69 69 72 71
10 67 68 71 70
20 64 68 70 70
30 61 65 68 68
40 54 59 62 62
Work on liquid jet impact damage on ZnS was done by van der Zwaag
and Field (1982). In this work the high velocity rain drop impact was
stimulated by firing a liquid jet at stationary ZnS samples using the
technique originally develop~d by Bowden and Brunton (1961). The CVD
110 Bulk mater"ials for the far infrared
Raytheon ZnS specimens used in this work were shown to possess a colum-
nar grain structure of average dimensions 2 x 2 X 25 1Lm 3 • The axes of the
columnar grains were normal to the specimen surface i.e. the plane of
deposition. The ( 100) direction was normal to the surface of the specimens
and hence parallel to the axes of the columnar grains. The critical stress
intensity factor K 1c was measured using the Vickers indentation technique
developed by Lawn and Fuller (1975). The average value determined was
K1c = 0.75 ± 0.05 MNm- 312 (fracture surface energy 3.4 Jm- 2 ) and this
agrees reasonably well with other data of Adler and Hooker (1978b)
K1c = 0.67 MN m - 312 (four-point bending technique) and Williams and
Evans (1973) K1c = 1.0 MNm- 312 (double torsion technique). Before
impact testing, the ZnS specimens (50 mm diameter, 3 mm thick) were
bonded and acoustically ,matched to a rigidly mounted 10 mm thick ZnS
specimen in order to red~ce the effects of reflected stress waves on impact
damage. A hydraulic bursting disc technique was used to measure the
fracture stress of the specimens after impact and to quantify the impact
damage. The average fracture stress of eight unimpacted specimens and
eight impacted but undamaged specimens was 76 MPa with a standard
deviation of 5 MPa. This is much lower than the value of 100 MPa reported
by Adler and Hooker (1978a) and was attributed to a poor surface finish
on the specimens as received. No reduction in strength or visible damage
occurred up to 125 m s - 1 for specimens impacted with 4 mm diameter
drops. From these results the calculated threshold velocity for 2 mm drop
impacts was 175 ± 5 ms- 1 which agrees with the results of Hackworth
(1979). For specimens impacted at 150-600 ms- 1 the fracture stress was
always lower than the initial value and was typically in the range
36-44 MPa. The decrease in fracture stress was particularly large over the
velocity range 125-175 ms- 1 • From 200 to 600 ms- 1 the residual strength
still decreased continuously with impact but at a much lower rate. Examina-
tion of the damage features revealed that there were two types of pre-
existing defects: a high density of very small defects, which were responsible
for the large number of relatively small cracks observed in the damaged
area, and much larger defects, which led to very large final crack sizes. The
second type of defect was most likely to be caused by grinding, polishing
and handling damage, but the first type of defect was classed as a grain size
related phenomenon. Although the effect of grain size on rain erosion
damage in ZnS has not yet been studied, it is likely that the threshold
velocity will depend on grain size. In general the smaller the grain size the
higher the impact velocity required for crack growth but the work of Adler
and Hooker (1978a,b) on hot pressed and vapour grown ZnS must be
considered. However this effect of grain size on rain erosion resistance
would be well worth studying in ZnS, since Hackworth and Kocher (1978)
have shown that the rain erosion resistance of CVD ZnSe increases markedly
with decreasing grain size. Also Lewis eta! (1984c) have recently reported
II- VI compounds Ill
a study relating the grain size of vapour grown ZnS to Vickers hardness and
fracture toughness measured by the indentation method. In this work the
fracture toughness was found to reach a maximum value between 5 and
10 p.m grain size, peaking at the high value of 0.8 MN m -J/2 at around
8 p.m. Thus there may be im optimum grain size for minimising rain erosion
damage of ZnS since it has been suggested that a high value of fracture
toughness is instrumental in retarding crack initiation (Evans eta/ 1980).
Van der Zwaag and Field (1982) have not claimed that there is complete
agreement between jet and drop impact damage because of the different
loading conditions i.e. the flow pattern of liquid over the surface leads to
differences in the crack profiles, the deformation in the undamaged zone
and the total extent of the damage. However they consider their work shows
that the very simple and inexpensive jet technique can be used to obtain a
good estimate of drop impact damage over a wide velocity range. Further
information on the single drop liquid impact and Vickers hardness in rela-
tion to grain size for ZnS is reported by Field eta/ (1983). Case and Evans
(1983) discuss the water impact velocity dependence of the crack growth and
Corney and Pippett (1983) report the effect of the erosion duration on the
10 p.m transmittance of ZnS for multidrop impact in a simulated rainfield. I.I
CVD zinc selenide
The transmittance of ZnSe is shown in figure 4.21(b). The cut-on absorp-
tion edge is consistent with the energy gap of 2.67 eV and the far infrared
cut-off edge occurs as a result of multiphonon absorption. Mitra (1966)
studied optically active multiphonon processes in hot pressed and
monocrystalline ZnSe. The infrared reflection, transmittance and Raman
spectra were investigated, and critical point phonon frequencies were
obtained from an analysis of the spectra. Also Bendow et a/ (1977) per-
formed experiments and theoretical investigations of multiphonon infrared
spectra in the three- and four-phonon regimes of cvn ZnSe over a range of
frequencies and temperatures. It was concluded t}lat the principal factor
determining the spectral shape was the density of phonon states and that
selection rules played a relatively minor role in the many-phonon regime.
Zinc selenide is a visually clear yellow polycrystalline material (grain size
-70 p.m) transmitting in the range 0.6-15 p.m (a 10 mm thick sample),
essentially free of extrinsic impurity absorptions. It can be produced in sizes
up to 750 X 1000 mm2 and thicknesses up to 38 mm. Product sales
literature quotes the absorption coefficient at 10.6 p.m to be 5 x 10- 4 em - 1
but Lipson (1977) quotes values of 8-18 X 10- 4 cm- 1, implying the
presence of non-uniformly distributed bulk or surface absorption. In this
work Fourier transmittance spectroscopy and laser calorimetry were used to
characterise extrinsic bulk and surface absorption. The main identifiable
extrinsic bulk absorption was zinc hydride, whose free diatomic molecule
has a vibrational mode. at 1608 em - 1 • When present in the ZnSe the zinc
112 Bulk materials for the far infrared
8-12 I'm thermal band sensors will probably play a major role in future
target acquisition and weapon aiming devices, surveillance equipment,
alerting devices and missile homing seekers in the military field
(Papayoanou 1982). The windows of equipment deployed on aircraft and
guided weapons will be subjected to harsh environmental conditions, such
as rain erosion and thermal shock caused by aerodynamic heating. Thus the
thermal and mechanical properties of the window material must be near
optimum if the windows are to survive. It has been demonstrated that
germanium is an unsuitable window material where major aerodynamic
heating is likely to occur, since free-electron absorption increases markedly
above 70 °C. GaAs was seen to be more advantageous retaining good
transmission up to about 200 °C. However, for applications up to several
hundred degrees centigrade, ZnS is the most suitable of the present genera-
tion 8-12 J'ID transmitting materials from thermal and mechanical con-
siderations (Whitney 1976). It will withstand moderate rain erosion
conditions, particularly at small angles of incidence, for reasonably long
periods. However it is insufficiently resistant to survive undamaged for
more than a few minutes in the most severe rain erosion conditions of I in
rain per hour at 223.5 ms- 1 at normal angle of impact (Hackworth 1979).
The thermal and mechanical properties of ZnS are also likely to be inade-
quate for the most demanding of future requirements. Hence there is a need
for an advanced optical window material for 8-12 I'm with similar
transmittance characteristics to ZnS but with enhanced thermal and
mechanical properties. In general this is a difficult problem since materials
with strong chemical bonds tend to exhibit good thermal and mechanical
Advanced optical window materials 115
Table 4.12 Some known structures of ABzS• compounds, where T is the thorium
phosphide one, S the spinel and 0 the orthorhombic.
Sc y La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu
Yb
Eu T T T 0 0 0 0 0 0 0
Zn s 0 0 0
Cd s s T s s s s s
Pb T.
Mg s 0 s
Ca 0 T T T T T T T T 0 0 0 0 0
Sr 0 T T T T T 0 0 0 0 0 0 0
Ba 0 T T T 0 0 0 0 0 0 0 0 0
116 Bulk materials for the jar infrared
frequencies for the spinel structure and the thorium phosphide structure
AB 2 S4 compounds are reported as about 447 em- 1. and 266 em- 1 , respec-
tively, (Provenzano 1976) indicating adequate 8-12 I'm transmittance for
bulk material uses. Data on the vibrational spectra of thorium phosphide
compounds are provided by Provenzano et a/ (1977) and on spinal cubic
sulphide compounds by Boldish and White (1978).
The reason for exploring these materials is to assess the thermal and
mechanical properties (thermal expansion, thermal conductivity, hardness,
Young's modulus, fracture toughness and stress) in order to assess their
potential as advanced optical windows. While hardness, thermal expansion
and Young's modulus can be measured on small samples, fracture stress
measurements require samples of about 25 mm diameter and at least 3 mm
thick, either to be used in a bursting disc technique or to cut bars from
a miniature three- or fm;r-point breaking technique. So that the results can
be statistically significant, tests on at least 10-20 specimens are needed.
Thus a preparative technique is required which can yield material of
sufficient uniformity and size to make such measurements. The melting
points of these materials are high (- 2000 ° C), they are volatile close to
their melting points and a number of them are likely to dissociate at these
high temperatures. Hence growth from the melt is difficult and can only be
achieved in small thick-walled sealed tantalum or tungsten containers.
Although some useful results could perhaps be obtained on monocrystalline
material, the experimental laboratory technique by which melting is
accomplished does not normally yield large enough area material for
mechanical property assessment. A laboratory vapour growth technique
such as that for ZnS and ZnSe (Miles 1976) would be very expensive and
technically demanding to set up for rare earth ternary sulphides. Having
established that the transmittance window is adequate for these ternary
sulphides, then the hot pressing technique (Pearlman et a/ 1973) is quite
acceptable providing each material is pure enough to avoid second phase
strengthening. For example, the thermal and mechanical properties of hot
pressed Irtran ZnS and ZnSe were very similar to the more recently pro-
duced vapour grown material, the major difference being in their optical
quality.
In early work on this family of materials (White et a/1981, Chess eta/
1983a) the sulphioe powers for hot pressing were made by firing mixed
carbonates and oxides, such as CaC0 3 and La2 0 3 , to make CaLa2 S4 at a
temperature of 950-1100 °C in flowing H 2S for periods of three to seven
days. Exact weighing and thus stoichiometry control was difficult because
of the hydroscopic nature of La2 0 3 (Chess et a/1983a). In later work (Lewis
et a/!983), nitrate solutions of the alkaline- and rare earths were separately
made up, assayed and then mixed in appropriate proportions to yield the
stoichiometric oxide mixture after evaporation and firing in air or nitrogen.
The mixed oxide was then immediately converted into ternary sulphide by
Advanced optical window materials 117
firing in HzS or HzS/Nz. This technique was crucial in obtaining a very fine
particulate ternary sulphide which had fully reacted at a low enough firing
temperature to avoid grain growth and sintering. A rapid decomposition of
solution technique (RDS) (Lewis et a/ 1983), where the mixed nitrate
solution is rapidly heated from cold to give a honeycomb of mixed oxide,
yielded satisfactory material. An evaporative decomposition of solution
technique (EDS) (Chess et a/ 1983a,b, Lewis et a/ 1983) in which nitrate
solution was sprayed through a hot furnace to yield fine particulate mixed
oxide was also very successful in generating oxide material for conversion
to ternary sulphide. The sulphide optical ceramics were then prepared by
hot pressing either in an inert gaseous atmosphere, or in a vacuum inside
a titanium-zirconium-molybdenum (TZM) die or a high density graphite
one at a temperature of 1300-1500 °C and a pressure of 48-85 MNm -z.
Alternatively the cold pressed sulphide powder was sintered in an H 2 S
atmosphere for 5 hat 1400 °C and then hot isostatically pressed (HIP) for
30 mm at 1450 °C at an argon pressure of 0.17-0.21 GPa (Beswick eta/
1983). A number of materials have been prepared for physical property
assessment by Beswick et a/ (1983), Walker and Wood (1984) (CaLazS•)
and by Lewis et a/ (1983) (SrLazS•, MgSczS4, CdErzS• plus LazS3 for
comparison purposes). It is too early to predict the long term viability of
any of these compounds as advanced infrared window materials, but initial
property data listed in table 4.13 are encouraging except for the high
coefficient of thermal expansion and low thermal conductivity data. Most
research has been done on CaLa2S4 and in figure 4.24(a) the optical absorp-
tion edge of this material is shown, and in figure 4.24(b) (broken curve) the
impurity dominated infrared transmittance of early hot pressed samples is
seen (Lewis eta/ 1983). Oxygen contamination leads to S04 (9.1 I'm) and
S03 (10.8 JLm) absorptions, and an infrared opaque phase present because
of non-stoichiometry may well account for the reduced transmittance at
Table 4.13 Some provisional physical property data on 'll!r1y samples of rare earth
ternary sulphide materials.
t Lewis et a/ (1983).
t Lewis eta/ (1984a).
§ Saunders et a/ (1984).
118 Bulk materials for the far infrared
Wavelength (nm)
300 500 1000 1500 2000
'
lal
,'-0\
lbl
I
60
...il
~
/
/ I
I
I
I
c 40 I I I
E I \ I
~~
/ \ I
c I \I
0
/
,= 20
I v
0
0.5 1.0 5 10 ' 50
Wavelength l~ml
The most robust and chemically inert materials which are useful in the
0.4-12 ,urn spectral region are multispectral ZnS, ZnSe and diamond.
Diamond, which is rare and expensive, has been used for window applica-
tions on space vehicles, but is not generally applicable in terrestial applica-
tions. ZnS and ZnSe have recently become available as high quality off-the-
shelf optical materials, and are hence important for applications where the
moisture sensitive and less robust alkali halide materials are unacceptable.
However, chlorides, bromides and iodides are amongst the most versatile
of all the infrared optical materials, offering transparency from the uv to
60 ,urn. They are also the most well known, the cheapest and easiest
materials to synthesise, and have been used for many years in laboratory
equipment. Recent work in many fields, on the reduction of optical loss, the
improvement of mechanical strength by hot forging, and the development
of protective coatings to reduce or eliminate moisture attack (Chapter 9),
has made possible the wider usage of these materials in thermal equipments.
They are also the only widely available materials offering useful trans-
parency between about 12 and 60 ,urn. Many of the fluorides have been
discussed in Chapter 3 but one of the least well kno~n of them, and perhaps
the most likely to be considered as multispectral because of its extended far
infrared transmittance, is PbF2. Since this material has been considered as
a lens element (Aurin 1983) for visible to far infrared multispectral applica-
tions it is discussed here, together with other multispectral materials.
5.1 Diamond
80
~
"- 60
~
c
,g
·e 40
~
c
0
"' 20
0
2 5 10 so 100
Wavelength !~ml
0.319-0.315 2 TO 0.316
0.302 TO+LO 0.302
0.281 TO+LA 0.281
0.267 LO+LA 0.267
0.251 TO+TA 0.251
0.244 LO+TA 0.237
Temperature
Material 0.4 0.7 1.0 3.0 4.0 5.0 8.0 10.0 12.0 20.0 40.0 60.0 °C 0.5 0.6 10.6 pm
I
Multispectral CVD materials 125
the specimen and the reaction chamber. The duration of the treatment was
determined by the thickness of the specimen and also by its initial optical
quality. The less visually transmitting samples required a longer treatment
time to achieve a predetermined level of optical transparency, but an upper
limit was determined by the amount of grain growth that took place.
Temperatures in the range 700-1050 ° C and argon pressures in the range
34-205 MNm- 2 have been used on specimens ranging from 4-15 mm in
thickness for times from 3 to 36 h. In particular, a 6 mm thick sample of
CVD ZnS was processed at 990 °C and 34 MNm- 2 argon pressure for 3 h.
The apparent absorption coefficient measurements for this sample, seen in
table 5.4, were calculated by dividing the fraction of the absorbed light by
the thickness of the specimen, and this included surface contribution to the
absorption.
5.3 Halides
The halide optical crystals are grown from the melt, and in general the
technology has developed along two paths. The first is the production of
standard off-the-shelf crystals from reagent grade quality material by
traditional melt growth techniques and the second is the more limited
production of laser window quality crystals from high purity raw material
by the reactive atmosphere processing (RAP) melt growth techniques. The
simplest method of growing NaCl and KCl is by the Kyropoulos (1926) or
the Stober (1925) technique. The Kyropoulos technique is illustrated in
figure 5.2(a). The material is melted in a refractory crucible by a cylindrical
heating coil surrounded by thermal insulation. A crystal is grown by lower-
ing a water cooled seed into the melt. The power to the heater is then redu-
ced. A crystal progressively grows on the end of the seed, but not in contact
with the walls of the pot. At the end of the cycle the pot is reusable in a
subsequent growth run (Menzies 1952). The Stober technique illustrated in
figure 5.2(b) uses a Stockbarger-type crucible in a furnace with three
heaters, a cyclindrical one to melt the material, and upper and lower ones
to establish a temperature gradient in the melt. Crystallisation occurs at the
Halides 127
base of the crucible while power to the lower heater is reduced and a freez-
ing isotherm is swept through the melt as the power to the main cylindrical
heater is reduced. Where high purity material is not required these furnaces
are not sealed from the ambient atmosphere, but are either 'plugged' or
bled with inert gas to prevent major ingress of air. If higher purity is re-
quired, or expensive halides (thallium, caesium or silver) are being grown,
then the Czochralski or Stockbarger methods described in §4.1.2 are
employed. The former is inside a sealed system using an inert gas, whilst
the latter is in a sealed pyrex or silica glass ampoule in a vacuum ambient.
For instance, the need for COz laser window components up to 450 mm
diameter made from NaCI has led to the use of the Czochralski technique
rather than the traditional Stockbarger technique, in order to allow better
control of crystal orientation and perfection in these large sizes. Nestor et
a/ (1979) have described a resistance heated facility for growing crystals up
to 500 mm diameter and 91 kg in weight using 50 X 50 mm 2 seeds and pull
rates of 0.5-2.0 mm h - l out of a silica crucible 58 em in diameter and
91 em deep.
(b)
Figure 5.2 An illustration of (a) the Kyropoulos growth technique, (b) the
Stober growth technique.
I
During the 1970s a need arose for very low loss laser window material for
C0 2 lasers and much effort was put into purifying KCI to meet this require-
ment. When pure raw materials are used in the growth of KCI the major
extrinsic absorber is oxygen in its various combinations. Growth in an inert
atmosphere was not enough to eliminate infrared absorbing trace oxide
impurities and thus RAP was used. In the RAP process, crystal growth was
carried out using inert gas containing a halogen compound (e.g. CCL,) and
purification was achieved in this case according to
CCL,+ 2KOH ->2KCI + COz + 2HCI. (5.1)
More expensive compounds such as AgCI, KRS5 and Csl tend to be needed
in smaller quantities, and these materials can be purified by an alternative
128 Bulk multispectral materials
RAP process. The material can be premelted and bubbled with halogen gas,
e.g. HI in the case of Csi, in its glass growth ampoule before this is sealed.
The material can then be grown by the normal Stockbarger process. This
avoids the need for specialist crystal growth equipment tolerant to hydrogen
halide gaseous atmospheres, and perhaps tailored to individual materials.
Hence existing simple vacuum furnaces with carbon heaters can be used to
manufacture high purity material.
For many years halides have been used commercially in non-stringent
laboratory applications, and components such as prisms, lenses and
windows are quite common inside spectrophotometers and other similar
laboratory equipment. During the late 1960s and early 1970s halides were
considered as refracting components for environments outside the
laboratory, and achromatic lens designs using these materials were
suggested by Strong (19'72). Although these materials are readily available
and comparatively cheap they have not been extensively used for two main
reasons. Firstly, the mechanical strength of the halides is low and the
monocrystalline form cleaves rather easily under impact (Sprackling 1976)
and secondly, the chemical durability leaves much to be desired as many of
the halides are water soluble. However the drive to develop low loss C02
laser windows during the 1970s (Deutsch 1973), the drive to develop cheaper
thermal band lens systems during the late 1970s and early 1980s coupled
with that to develop 10 JLm optical fibres during the same period (Chapter
7) has stimulated further research and development into this class of
materials. The problems of scale-up of the crystal synthesis techniques and
reduction of the extrinsic absorption have been solved sufficiently for these
materials to be considered for many thermal applications. Progress towards
overcoming the second of these problems through the achievement of pro-
tective coatings resistant to moisture attack has been good, and is reviewed
in Chapter 9. In order to solve the first problem, research has been directed
primarily at hot forging of halides to yield polycrystalline material. The
fracture of halide crystals results from dislocation processes (Sprack ling
1976) and thus strengthening occurs as in metals, when dislocation mobility
is inhibited. Becher and Rice (1973) reported that the yield stress, ay of KCl
is increased by tailoring the grain size of polycrystalline material according
to the relationship (5.2), first reported by Petch (1953)
ay =a+ k d- 112 (5.2)
where a and k are constants and d is the grain size.
The work of Carnahan eta! (1961) on AgCl and Stokes (1966) on NaCl
also suggests this type of behaviour. Since low loss monocrystalline KCl was
readily available Becher and Rice (1973) chose to deform and recrystallise
this raw material by hot forging, since such an approach was considered the
most suitable in avoiding problems of grain boundary contamination and
porosity. Cleaved KCI monocrystals of aspect ratio 2: 1 were water polished
I
Halides 129
and press forged along the ( 100) axis at a constant ram speed, achieving
plastic strains of 70-850Jo in the temperature range 150-250 °C. Pyrolytic
graphite foil was used as a lubricant between the specimen and the loading
ram. The use of forging temperatures in the lower part of this range, and
rapid cooling to 100 ° C after forging was found to be effective in reducing
the grain size. Approximately one order of magnitude increase in yield
stress was achieved, i.e. 32 MPa for 5 fLm grain size material compared with
about 4 MPa for single crystal material. NaCI forgings showed similar
increases to 24 MPa. An increase of the order of a factor of two in fracture
toughness was also reported for KCl (Becher and Rice 1972). The strength
against grain size behaviour of hot forged KCl as reported by Becher and
Rice is shown in figure 5.3. In further studies of this hot forging technique,
Anderson eta/ (1973) constrained the billet while it was being forged with
an annealed copper tube in order to exert a compressive hoop stress on the
periphery of the deforming billet. Hence this inhibited edge crack initiation
which had been a problem with unconstrained forging. Anderson (1978)
compared the stress-strain curves of constrained and unconstrained press-
ings (figure 5.4) showing the constrained billets deformed at higher stresses
than the unconstrained. Since the hot forging process has been shown to be
able to deform and recrystallise halide material and improve its physical
strength, it has been taken a stage further and combined with optical blank
formation. Strong (1974) disclosed a method of hot forging infrared optical
elements from melt grown bulk material. Chrome-plated steel dies or steel
ones containing an accurately figured glass liner were employed to make
KCl, NaCl, KBr, CsBr, KI or Csi components. In addition, pairs of optical
30
I
10
Figure 5.3 Strength in terms of yield stress plotted against grain size
for press forged KCI.
I
130 Bulk multispectral materials
8
150 '(
Constrained
6
~
~
~
200'(
.1=
~
150 '(
200°( }
0 0.5 1.0
True strain
Figure 5.4 A comparison of the stress-strain curves of constrained
and unconstrained KCI during hot forging.
Table 5.5 Specification for 450 mm diameter 10.6 I'm polycrystalline NaCI
windows.
Parameter Specification
10 0
A
80 B ,\
\
0
f
"'
0
E c 0 A B c 0 E
20
0 1,\
0.2 0.4 1.0 5 10 50 100
WavelenQth {!lm l
I
Figure 5.5 An illustration of the transmittance capability of several halides: A,
NaCl 10 mm thick; B, KCl 10 mm thick; C, AgCl I mm thick; D, KRS5 I mm
thick; E, Csl 5 mm thick.
The need for high power laser windows during the 1970s stimulated
research on the absorption processes in alkali halides. Deutsch (1973)
reported absorption coefficient data on thick samples using a differential
technique with a dual beam spectrophotometer. Measurements were made
in the region of multiphonon absorption and bulk absorption coefficients
at 10.6 I'm were extrapolated from those data. To improve the measure-
ment accuracy of absorption coefficients less than w- 3 em- 1 , adiabatic
laser colorimetry has been employed and found to be extremely useful
134 Bulk multispectral materials
Material 3 ww(cm- 1 )
NaCl 795
KCl 615
KBr 489
CsBr 342
Csl 270
AgCI 567
(Deutsch 197 5). Data on RAP grown KCI and KBr has been listed by Miles
(1976) for HF, DF and C02 wavelengths, as seen in table 5.8. Rowe and
Harrington (1976) and Alien and Harrington (1978) have reported more
detailed data on the measurement of absorption coefficients in KCI and
KBr, and in KCI and NaCl, respectively. In the former work Rowe and
Harrington (1976) reported that the loss in KCl at 10.6 !Lm was two or three
times greater than the predicted intrinsic multiphonon absorption of
8 x 10- 5 em -l, and that for KBr was two orders of magnitude greater than
the predicted intrinsic loss of 2 x 10- 7 cm- 1 at 10.6 !Lm. Both materials
exhibited an unidentified surface and bulk extrinsic absorption centred at
9.6 !Lm which was shown to be essentially independent of temperature.
Allen and Harrington (1978) reported the achievement of intrinsic bulk
absorption in KCl and in NaCl of I x 10- 3 em -l at 10.6 /Lm, and that
surface extrinsic absorption was the predominant residual loss in these
materials. The absorption data of table 5.2 for KRS5 and diamond are
taken from Deutsch (1975), and those for AgCl are taken from Sahagian
and Pitha (1972). General thermal and mechanical properties of the halide
materials are listed in table 5.3. The data for diamond are taken from Field
(1979), those for ZnS and ZnSe from commercial sales literature, those for
the alkali halides, KRS5 and the silver halides from Billard and Cornillault
(1962), Li (1976) and commercial sales literature. The mechanical strength
and deformability of these materials is of great interest from the
Table 5.8 Optical absorption in RAP grown alkali halides at HF, DF and C02
laser wavelengths.
Material 2.7 I'm (HF) 3.8 I'm (DF) 10.6 JLm (C02)
f
6
Laser Damage in Bulk Low Loss Infrared
Optical Materials
The satisfactory operation of high power lasers largely depends upon the
performance of optical components such as mirrors, windows, output
couplers, beamsplitters and lenses. Thus laser damage of these components
is an important issue. But what is laser damage? This is usually considered
to be some irreversible change which has taken place in an optical compo-
nent, thus degrading the optical performance of a laser system. It may be
useful to consider a wider definition such as a change which degrades the
performance of a laser system and thus to include reversible effects such as
thermal distortion. It is necessary to understand the underlaying principles
which govern these changes in order to provide a basis for quantitatively
assessing the potential performance of optics made from specific materials,
and for the characterisation of damaged optical materials. Much work has
been done on this topic, and a large amount of literature has been generated
as illustrated by reference to the Boulder Symposia on Optical Materials for
High Power Lasers which have been held annually in Colorado since 1969.
The proceedings of these meetings have been published as NBS Special
Publications and have become the standard reference documents on laser
induced damage in optical materials (see References). However, interpre-
tation of these data is not easy, since there are a large number of variables
and experimental problems such as the laser wavelength, continuous wave
(CW) or single shot operation, variation of properties with temperature and
the difficulty of measuring the energy density at the damage site. Since the
laser damage of optical components is such a wide subject, it has been
decided, of necessity, to limit discussion of it for the purposes of this text.
It is thus the limited aim of this chapter to indicate the competing
mechanisms in laser damage of mainly refracting components, and to
discuss examples of laser damage for some materials suitable for use at 1.06
and 10.6 JLm. A full treatment of the subject is given by Wood (1985).
Mechanisms 137
6.1 Mechanisms
~ 300
's
~ 0.2
·o
:::•
8
c
~ 0.1
~
0
~
~
O.O:i;;;---;;';;;---¢-,;--~,------~.
300 320 340 360 380
Temperature (Kl
Hack and Neuroth (1980, 1982) have reported the surface and internal
damage thresholds of optical glasses, using a 3 ns pulsed 1.06 JLm NdY AG
laser with a slightly focused beam diameter of 2 mm at the sample plane.
The most intense central portion of the beam, 0.6 mm diameter, was
responsible for the damage. Surface damage was investigated on 5 mm
thick samples and internal damage on 20 mm thick samples. Damage was
confirmed using an optical interference microscope (Nomarski). Each
specimen was irradiated with decreasing energy in steps of 1OOJo, until no
damage was observed after four shots at the same energy, and each shot was
aimed at a different spot on the sample. The surface damage had the
appearance of small pits probably caused by absorption centres in the pol-
ished surface layer. Threshold damage values were found to depend upon
the polishing method rather than the chemical composition of the glass.
Polishing on a plastic lap with Ce0 2 gave the highest damage threshold and
with 80% Zr0 2 and 20% Fe20 3 the lowest threshold. All the samples were
subsequently polished with Zr02 + Fe20 3 to give typical lowest values. For
all the glasses tested the surface damage threshold occurred between 15 and
21 J em- 2 at the rear face of the specimens. At the front surface a plasma
is created in air which protects the material by absorption of the incident
radiation, but at the rear surface the plasma occurs inside the material thus
increasing the absorbed power density and causing more damage (Boling et
a! 1973). Internal damage was point-like and thread-like, each defect
appearing in the same glass at variable thresholds. Point-like damage occur-
red somewhere along the beam whereas thread-like damage began near the
centre of the specimen and ran to the exit surface of the specimen. With
increasing energy density the thread-like damage started closer to the entry
surface. A constriction of the laser beam occurs at high power densities due
to self-focusing (Kelley 1965), and this constriction occurs earlier when the
non-linear refractive index of the glass is higher. Thus the threshold values
for thread-like damage are compared with non-linear refractive index values
in table 6.1. No correlation with any physical property was found for point-
like damage, but these were considered most likely to be caused by minute
platinum particles or crystals present in the glass. Temple eta! (1979) report
that the surface damage threshold for normally published Si0 2 glass ranges
from 10-20 J cm- 2 in agreement with the data given above on more complex
optical glasses. However, when these surfaces were C02 laser polished, the
damage resistance to 1 ns pulses of 1.06 JLm radiation increased to values
in the range 23-55 J em- 2 indicating the importance of good surface
finishing.
Laser damage in glasses at I. 06 p.m 143
Table 6.1 A comparison of non-linear refractive index and threshold values for
damage for various glasses.
FK-51 0.69 7 47
FK-52 0.73 11 25
PK-51 0.86 8 26
FK-5 0.91 7 > 47
PSK-50 1.03 > 46 46
BK-3 1.06 29 36
PK-2 1.13 14 > 48
BK-1 1.14 >44 > 44
BK-7 1.15 > 49 > 49
TiK-I 1.16 > 45 45
BaLK-3 1.27 >41 41
K-5 1.31 34 24
BaK-2 1.37 43 >44
ZK-1 1.40 31 46
PSK-52 1.44 14 29
KF-6 !.56 20 >26
PSK-53 !.58 22 26
KZF-2 1.65 >37 32
SK-16 1.71 38 43
LaK-21 1.82 35 41
LaKN-7 1.95 11 36
SSK-2 2.07 41 41
LLF-1 2.09 34 41
LaK-8 2.59 11 19
BaF-4 2.63 41 41
LF-5 2.73 >42 30
'' 18
LaFN-3 3.11 15
TiF-4 3.18 > 23 23
BaSF-1 3.33 >29 22
F-3 3.45 >31 27
BaSF-52 3.77 8 >27
LaF-21 3.78 8 26
F-7 3.80 >42 24
LaF-22 5.35 19 22
LaSF-5 5.80 8 18
TiSF-1 5.89 > 25 25
SF-6 9.90 7 7
. SF-57 12.02 > 24 8
SF-59 19.20. 7 9
144 Laser damage
Germanium
The bulk absorption in pure germanium results from the intrinsic carrier
concentration according to the temperature of the material. There is a large
population of intrinsic carriers at room temperature and thermal runaway
occurs at temperatures above about 55 o C. Thermal runaway is associated
with an exponential rise in the number of free electrons and holes with
increasing temperature, and a corresponding rise in optical absorption.
When the rate at which a germanium component loses heat is less than the
rate at which energy is being gained by absorption of laser radiation, the
temperature of the component increases. For any given cooling situation
there is a power density above which the component temperature rises
ending in a region of rapid increase known as thermal runaway. The result
is the fracturing or local surface melting of the component. This phenom-
Laser damage in materials at 10. 6 p.m 145
enon was discussed by Young (1971), and a thermal model which examines
the temperature dependence of thermal conductivity has been presented by
Wilner et at (l982). A physical description of the effects of thermal runaway
has been reported by Willis and Emmony (1975) for a number of etalons
used as output mirrors of a pulsed COz TEA laser. The damage on the
inside cavity face was always greater in extent and depth than that on the
outside or exit face of each etalon. Two regular orthogonal pattern intervals
occurred as a result of the surface melting of germanium following the
formation of avalanche currents in response to the electric field of the
incident laser radiation. Willis and Emmony proposed that after avalanche
formation, linear melting occurred together with constructive interference
between the incident radiation and the field of the induced current doublet,
followed by initiation of new avalanches conforming to a regular array. The
near field within the germanium gave rise to broadside pattern growth
• I
(roughly 1. 7 p.m spacing) whilst a surface wave produced new avalanches
and/or increased damage at a spacing close to the free space wavelength of
10.6 p.m. Similar patterns were also observed in silicon.
The bulk absorption coefficient of germanium can be varied by doping.
Material with p-type conductivity is more absorbing than material with
n-type conductivity (see Chapter 4). Standard material for room temper-
ature operation with the lowest absorption coefficient ( < 0.02 em- 1 ) is
5-40 0 em n-type. However, the high temperature limit of the exhaustion
range can be increased by making the material more strongly n-type at the
expense of an increased room temperature absorption. Thus 3 0 em n-type
material offers an absorption coefficient lower than 0.1 em- 1 up to about
77 ° C, a value roughly 20 ° C.higher than the point of equivalent absorption
of I 0 0 em material. Thus there is the possibility of adjusting the resistivity
and therefore the absorption coefficient within limits to suit the application.
Wood et al (1982b) have shown that free-carrier absorption theory is
adequate to explain the absorption of bulk intrinsic and doped germanium
following the work of Capron and Brill (1973) and Bishop and Gibson
(1973). I
The surface absorption of germanium is an equally serious problem since
high surface absorption can enhance thermal runaway. Wood et al (1982b)
reported a series of results on a well characterised series of germanium
samples. C0 2 laser calorimetry was performed in ambient laboratory condi-
tions on samples of different thicknesses and it was shown that there was
a surface contribution to the total absorption. When measurements were
made in a vacuum calorimeter, it was shown that there was a removable
surface absorption contribution which returned on exposure of the samples
to air. The removable surface contribution measured varied from
0.05-0.7% per surface and was attributed to water or hydrocarbons. A
25 mm diameter germanium sample was heated and the water was removed 'I
'
I I
from its surfaces and collected in a molecular sieve. It was calculated that I
! !
146 Laser damage
Ge 600
GaAs 100
ZnSe 800
KCI 100-1000
Gallium arsenide
Libenson et a/ (1981) have investigated the surface damage of [ 111] gallium
arsenide mono crystalline material upon irradiation with 150-200 ns pulses
from a TEA C0 2 laser. The final stages of beam erosion were seen to have
a thermal character as illustrated by the formation of loose deposits of
arsenic oxides on adjacent surfaces. Libenson et a/ proposed that the beam
erosion was due to heating following thermal instability. The initiation of
thermal instability is likely to be caused by a local increase in the surface
absorption coefficient leading to a temperature increase and the loss of the
volatile component, arsenic. Once this has happened a further Increase in
the temperature occurs because of absorption by the gallium rich material
in the irradiated region. Thus the laser beam erosion of gallium arsenide
was considered as a sequence involving local heating at an inhomogeneity,
Laser damage in materials at 10.6 p,m 147
Zinc selenide
Kompaneits et a/ (1981) have investigated the relationship between bulk
damage threshold and the defect structure of melt grown ZnSe crystals. The
main defect in melt grown ZnSe resulted from deviation from stoichiometry
of the chemical composition of the melts and crystals. Studies of annea-
ling on the defect structure indicated that the solubility of Se in solid ZnSe
decreased with the fall in temperature. Thus cooling was accompanied by
precipitation of the ZnSe-Se solid solution, yielding Se and gaseous Se
microparticles at sites such as deformed bonds and dislocations. It was con-
sidered that carbon, oxygen and hydrogen were also associated with the
microparticles. The results of bulk damage threshold testing revealed that
optical breakdown was caused by the thermoelastic damage at sub-micron
Se particles containing the above impurities. Material grown from the
vapour at no more than half the melting point should contain very few, if
any, of this type of defect. For instance Leung et a/ (1978) investigated
relatively early CVD material (1974-5) and found that inclusions near the
surface or in the bulk played an important role in the damage mechanism.
Later, improved material was found to be the first ZnSe which could be
damaged on the surface and not in the bulk. Wood et a/ (1982a) have
reviewed studies of impurities in the bulk and at grain boundaries and their
likely effect on absorption (see also Chapter 4). The very pure ZnSe now
available is useful in laser optics and Patel (1977) has assessed this material
as well as Ge, GaAs and the alkali halides for use as laser windows, where
a pressure differential of 1 atm exists across them. From optical considera-
tions ZnSe emerged as the best window material. This was in spite of the
fact that KCI has a lower absorption coefficient and Ge and GaAs are
mechanically stronger. Detrio et a/ (1979a) have studied the laser power
dependence of the temperature distribution, optical absorption· and strain
for specimens of ZnSe with absorption, coefficients in the range
0.02-0.006 em -I. The response of ZnSe to an axisymmetric COz laser
beam was predicted by a mathematical model valid for powers up to
700 W. C02 laser single shot damage threshold for current material is listed
in table 6.2.
148 Laser damage
Potassium chloride
Some of the problems affecting the laser damage threshold of KCl are
illustrated in work reported by Allen et a/ (1974). In this, laser damage
thresholds of reactive atmosphere processing (RAP) grown KCl were
measured as a function of surface and bulk processing techniques. Both
monocrystalline and hot forged material polished by conventional and HCl
chemical polish techniques were measured for bulk and surface damage
thresholds. These threshold values were correlated with absorption, Auger,
low energy electron diffraction (LEED) and SEM data. The RAP boules were
essentially monocrystalline but contained some low angle grain boundaries
and the hot forged samples possessed an average grain size of 5-I 0 f'm.
Conventionally polished samples were found to be very sensitive to humi-
dity and it was thought that the surface quality was probably a limiting
factor, effectively masking the improved bulk matrial quality of the RAP
grown material. A much improved surface quality was obtained after con-
ventionally polished samples were chemically polished in concentrated HCl
for 1-2 min followed by rinsing in isopropyl alcohol and a final cleaning
in a freon TF vapour-degreaser. The scratches and surface imperfections
seen by optical microscopy before this treatment were entirely absent after
it, even when examined in an SEM at 18 000 X magnification. The total
absorption coefficients measured by C02 laser colorimetry were reduced by
a factor of 8-10 for monocrystalline and 2-10 for hot forged material. This
was attributed to less surface absorption as a result of the chemical
polishing. The chemically treated surfaces were also less humidity sensitive
and caused less scattering. Moreover they exhibited less oxygen and carbon
contamination on Auger analysis and much sharper more ordered LEED
patterns. Both the surface perfection and degree of contamination were
improved further after 6 min of argon ion sputter cleaning (800 eV,
10 f'A em- 2). Damage measurements were conducted using a pulsed TEA
C02 laser, whose pulse consisted of 0.6 f!S of individual 2 ns longitudinal
mode spikes. The spot size on the samples was in the range 60-70 f'm. The
tests were made in an evacuated chamber equipped with an X- Y translation
stage. The exit surface damage site on mechanically polished surfaces was
found to consist of a central melted portion surrounded by a smooth ring
with shallow cleavage cracks radiating outwards. This damage occurred at
a low threshold value of 120-580 MW em- 2 whereas for etch polished
samples damage occurred at an order of magnitude higher level of
550-4500 MW em- 2. In the latter case the damage sites were several
millimetres in diameter and consisted of radiating cleavage cracks. Bulk
damage was found to occur at power densities of 4200 up to ·more than
7800 MW em- 2. This work clearly illustrates the importance of surface
perfection and purity in achieving high laser damage thresholds. Work
reported by Vora et a/ (1978) on tunable C02 laser calorimetry
Laser damage in materials at 10.6 p.m 149
I
7
Infrared Optical Fibres
I
Infrared optical fibres 151
near infrared at 0.8-0.9 fLm, at 1.3 I'm and at 1.55 fLm. Monomode fibres
operating at 1.3 and 1.55 I'm have been developed to the point where
monomode optical communications are also now possible over very long
distances. Having overcome most of the synthesis problems of near infrared
fibres, researchers have recently turned their interest to mid and far infrared
fibre optics (Klocek 1982). The discoveries of mid infrared fluorozirconate
glasses (Poulain et a/1977) and fluorohafnate glasses (Drexhage et a/1980)
have opened up the possibility of fibres operating at around 4 I'm with
much lower theoretical attenuation than silicate fibres, thus offering the
potential of many kilometre repeater less links. There is also interest in far
infrared fibres for image transfer and co~ laser power transmission,
although here research is hampered by the lack of stable glass compositions
exhibiting low enough loss. Thus crystalline fibres and hollow core fibres are
being considered for operation in this waveband (Harrington 1981).
As described in Chapter 2, the loss in a II\aterial at short wavelengths is
set by the electronic absorption edge and also by Rayleigh scatter which
decreases as }, - 4 • This total loss therefore decreases as the wavelength
increases. However a crossover point is reached where the total loss rises
again because the dominance of the phonon absorption edge increases
proportionately with increasing wavelength. This is illustrated for three
materials in figure 7.1 (Harrington 1981). For optical communciations
requiring high performance waveguides, it is preferable to operate at a
wavelength where the material dispersion is near zero and where the loss is
minimal, i.e. at the V region in the plots of figure 7 .1. However for short
distances, and power and image transmission applications, minimum
material loss is the main criterion. To illustrate the effect of the material
anion on this minimum loss, Gannon (1980) calculated the reststrahlen
frequencies for a number of materials and assumed an anharmonic multi-
phonon mechanism. While these data are not completely valid for atomic
KCI KilS5
10-~'o-.5.L.L.U..C1.co-O-~~~__,5!-,_.0~~10;c--2;e;0,-~'--;"50
Wavelength (Jlm l
100 dB km1
10 dBkm-'--- -
'
Oxides
10~' ~------~~----~--~~~~-L~~-7-~----------~
1.0 3.0 10 13 so
Woveler1Jih lpm)
Cladding
Li ht ---- ---- ---- -----
9 - a1 Core
ray -- ---- ---- ---
(a)
1
3~m
- - - - - , __
t.....-...l
125 ~m
(b)
~
I:
~
_ __ , -60~m
125 ~m
(c)
_
I
u
i---1
125 ~
ld)
I
.
- - - - n mr=1
Thus more modes can be guided if either the core radius, or the refractive
index difference between core and cladding, is large. It is possible to con-
struct a fibre so that only one mode is transmitted, and this type of fibre,
illustrated in figure 7.3(c), is called monomode. The normalised frequency
V ,of equation (7 .2) for monomode operation should be less than 2.405
(Marcuse 1972). Signal distortion in these fibres results from dispersion in
the glass, and from the fact that the group velocity of the mode is dependent
on frequency, even in the absence of material dispersion (Dyott and Stern
1970). These effects are small for small bandwidths, hence narrow band-
width lasers are used in long length monomode communications links. The
material dispersion for Si0 2 glass, the major constituent of near infrared
fibres, tends to zero at about 1.3 11m (Payne and Gambling 1975, Fleming
1978) which also corresponds to the minimum optica11oss wavelength. For
high capacity long length data transmission systems, monomode fibres are
154 Infrared optical fibres
7.2 Oxide Glass Fibres for the Near Infrared (0.75-2.5 fLm)
[1as
1n
( L; -Gosout
,----- L
,--
~
Furnace
c___-, ~ --!-
~ ,------1 Diffusion
~ _fngth
Fibre
Figure 7.5 Cross section of a double crucible used for fibre drawing
from molten glass.
crucibles having nozzles at their bases and the clad fibre is drawn directly
from the nozzles. Step index fibre from several glass systems has been
produced by this technique and these glass systems are listed in table 7 .1.
Uchida eta/ (1970) produced the first graded index fibre by an ion exchange
technique. Since then the double crucible fibre drawing technique has been
modified to yield graded index fibre. Concentric double crucibles are
charged with low loss core glass rod (centre), and cladding glass rod
(outside) made by the melt process, and the clad glass fibre is drawn from
the nozzle at the base. The graded index profile is obtained by diffusion or
ion exchange of mobile ions across the core cladding interface within the
molten glass, between the inner and the outer crucible nozzle. This diffusion
length L is tailored to produce the required profile for the glass systems
being used. The diffusion profile depends upon the diffusion coefficients of
the ions, the time allowed for diffusion (i.e. the diffusion length, L, in the
crucible in relation to the fibre drawing speed) and the radial distance over
which the diffusion has to take place (Dyott and Brain 1974). The double
crucible technique yields a relatively stable graded index profile over the
whole length of the pulled fibre provided that the temperature profiles are
kept constant, because the diffusion relies on the geometry of the system
and on the bulk glass properties. The graded index profile produced by this
diffusion technique deviates from the required profile so that in practice
pulse broadening of the order of 1 ns km- 1, equivalent to about
500 MHz km bandwidth, occurs in all the optimised systems listed in table
7.2 (Beales and Day 1980).
A novel melt synthesis technique has been developed by Macedo and
Litovitz (1976). This includes a purification stage so that the loss in the fibre
Table 7.1
Numerical Loss
Core glass Cladding glass aperture (NA}- (dB km- 1) Reference
is less sensitive to the purity of the melting environment and the raw
material. Alkali borosilicate glass of low alkali content ( -1 OOJo) is melted
and converted into rods. These rods are then phase separated by heating to
a temperature above T,, the glass transition temperature. The transition
metal impurities from the starting materials and the melting environment
tend to ·segregate into an interconnected boron rich phase which is
subsequently dissolved in acid, and a silica skeletal phase. The pores of this
etched silica skeletal phase are back filled with dopant in order to modify
the refractive index. Finally, the outer surface of the rod is leached of
dopant to create a cladding layer of lower refractive index. The resulting
preform is then dried out and pulled into a fibre. Macedo et a/ (1976) have
reported losses of around 15 dBkm- 1 at 0.8 !Lm for step index fibre with
an NA of 0.28 made by this technique, and Bamford and Loukes (1979)
reported a loss as low as 6.5 dB km _,. Further detail on this type of process
is given by de Panafieu et al (1980):
Table 7.2
Pulse
Diffusion Loss dispersion
mechanism Base glass (dBkm- 1 ) (nskm- 1 ) Reference
I'
I '
160 Infrared optical fibres
The vapour deposition technique has been used 'for many years for the
commercial production of pure silica glass for scientific and engineering
purposes. This product is usually prepared as large boules, many kilograms
in weight, of the order of 100 em long and 10 em or more in diameter for
instance, by the flame hydrolysis of SiHC4. This is a process similar to the
Verneuil technique for crystal growth (§3.2.4), except that the vapour feed
stock is used instead of powdered material. This type of material usually
contains of the order of 1000 ppm of H20, and thus for fibre applications
is only useful at around 0.85 p;m since OH absorption eliminates operation
at longer wavelengths. To overcome this problem and hence create a much
wider operational window, the CVD process has been modified by the
utilisation of SiC4 feedstock in a plasma torch instead of a hydrogen flame,
and thus major OH impurity incorporation is avoided. Material can be
produced by this process ' with H 20 contents of under 2 ppm. Winterburn
(1967) has patented this process but no details have been published for
commercial-in-confidence reasons. However, Nassau and Shiever (1975)
have published a detailed study of a laboratory scale equipment used to
produce material which, when drawn into fibre form, has exhibited a loss of
around 4.5 dB km - l at 1.06 p;m. The plasma process used by these authors
is illustrated in figure 7.6. An oxygen plasma is generated inside a water
cooled silica tube by an external radio frequency coil. SiC4 and oxygen are
fed into the base of the plasma, and Si02 is formed on a rotating mandrel
according to
SiC4 + 02-+ Si02 + 2Ch. (7.6)
The growth equipment needs to be very well controlled since fluctuations
in the process can lead to powder particles or SiC4 droplets and thus to
bubble formation. If the process is operated so that ambient air is excluded
from the deposition zone by shielding with dry gas, then low OH contents
are possible in the product. However, if ambient air is allowed to interact
with the process at a critical point then the relative humidity of the air can
influence and possibly even control the OH content of the product up to
several tens of ppm. Nevertheless, if the process is operated with care and
attention to critical detail, very good quality, high purity, essentially OH
free glass suitable for fibre optic applications can be produced.
Another vapour technique which can be used to synthesise large
quantities of material for fibre optic applications is that first used by Hyde
(1942). In this technique material is generated by hydrolysing halide
vapours in a methane-oxygen flame and deposited as a 'soot' onto a
rotating and traversing alumina rod. The material adheres to the rod as
partially sintered glass, and a cylindrical porous glass preform is grown
from sequentially deposited thin layers, as shown in figure 7. 7. The vapour
composition is changed during deposition to produce a step or graded index
profile across the diameter of the boule (Schultz 1974). At the completion
Oxide glass fibres for n(!ar infrared 161
of the deposition the alumina rod can be removed (due to its higher thermal
coefficient of expansion) and the porous material is zone sintered at about
1500 °C to yield a clear glass preform which is subsequently drawn into
fibres or possibly into a series of rods for fibre pulling. This method yields
fibre with a high OH content and this can be reduced by utilising chlorine
as a dehydrating agent during the zone sintering (DeLuca 1976). Preforms
yielding about 10 km of fibre have been prepared, for instance, with a core
of Si02-Ge02-B203-P20s and cladding of B203-Si02 (Schultz 1973).
Gas feed for plasma
j
II
II
0
I 0~\
/
0
0 1Plasmo.t
I
120 00 OCI 0
I
I
0 I
I
I
I o
I
I
I I
I
I
Sict4 -
- ' I f--
o,• 't {1
T~:. .~"·
Burner
}\
Figure 7.6 Schematic diagram of the Figure 7. 7 The flame hydrolysis tech-
radio frequency plasma synthesis process nique for fibre glass preform synthesis.
for bulk Si02 glass.
_ Transparent
preform
-Carbon furnace
Porous_
preform
SiCl4 GeCl 4
POCt 3 02+H 2
Figure 7.8 Vapour phase axial deposition (VAD) of glass for fibre
applications.
The plasma technique described for the production of bulk water free
silica has been used in the inside silica tube deposition process by Kuppers
et a/ (1976). Instead of an oxyhydrogen flame, a microwave cavity
(2.45 GHz) was used to traverse the tube and maintain a non-isothermal
plasma within it at a pressure of 1-100 mm pressure. However, a stationary
furnace around the tube was used to maintain a base temperature of about
1000 °C. Other workers (Irven and Robinson 1979) have used an induc-
tively coupled radio frequency plasma, operating at a frequency in the range
3-6 MHz or 27 MHz at atmospheric pressure, to produce a homogeneous
Oxide glass fibres for near infrared 163
gas phase reaction. This high temperature discharge causes fusion of the
deposited material and an enhanced reaction rate is achieved. After
synthesis the preforms are collapsed in the normal manner to yield a rod
for fibre drawing.
The preforms which result from all these vapour processes are normally
drawn into fibre by feeding them through a vertical carbon or Zr0 2 furnace,
followed by in-line primary plastic coating to avoid contamination and
strength reduction during secondary coating and cabling. More detail of all
of these deposition techniques, together with consideration of their relative
merits and the results achieved is discussed in an excellent review by Beales
and Day (1980).
Flame
'kpour- ~Preform
Figure 7.9 The inside tube cvo deposition process for fibre preform
synthesis.
Thus, the short wavelength luminescence is absorbed by the fibre, and any
near infrared component can be filtered. Therefore, luminescence effects
can be avoided relatively easily in systems, and they are not very important.
On the other hand, attenuation is of major significance. It must be
emphasised that attenuation resulting from radiation damage is a dynamic
process. That is, along with colour centre formation, the emptying of holes
and electrons out of these centres causes a concurrent recovery (Friebele and
Griscom 1979), and the observed attenuation is the difference between these
two processes. The recovery is usually thermally activated but can be partly
optically activated depending upon the glass composition. Thus the
measured attenuation is influenced by temperature and the intensity of light
travelling in the fibre. The dose rate, the dose level and the nature of the
irradiation are important, as are the signal launch conditions and the radia-
tion history of the fibre under test since hardening can occur in some fibres.
The most important parameters are listed in table 7.3, and figure 7.10
illustrates the general wavelength dependence of the induced attenuation
\o,Javelength (!Jm)
0.8 0.6 0.5 0.4 0.3 0.2
! i
2 3 4 5 6
Photon energy (eV)
Figure 7.10 General form of the 7 irradiation induced attenuation
plotted against wavelength for bulk synthetic silica.
166 Infrared optical fibres
when bulk fused silica is y irradiated. From this it can be seen that, in
general, the attenuation is less severe at longer wavelengths (e.g. at 1.3 and
1.55 ,urn) than at shorter ones (e.g. at 0.8-0.9 ,urn) and this results from the
absorption peak tail near 2.0 eV which is probably caused by holes trapped
on a non-bridging oxygen site (Evans and Sigel 1975). The fibre absorption
spectrum is more complex than this, involving additional transient and/or
longer term peaks than are shown in figure 7.10, but the general trend
of lower induced attenuation at longer wavelengths also holds for fibres.
The realisation of this trend aids assimilation of numerical data listed for
different wavelength values. Basically, the experimental apparatus for the
measurement of radiation induced attenuation consists of a light source
(white light plus monochromator, LED or laser), launch optics, fibre tails to
the length of fibre under test, a detector or detectors and electronics for
signal processing and pr¢senting or recording the data in the desired form.
Thermoluminescent dosimeters attached to the test fibre record the total
dose received in the case of y irradiation. A more detailed description of
the type of apparatus used is given by Friebele (1979) and Rosiewicz et a/
(1980). Measurements can be made at a single wavelength or at a number
simultaneously on a transient timescale, about 1 ms to 1 min, and/or on a
longer timescale of about one minute to many hours. The most commonly
used y radiation sources are Co-60 for continuous and long term impulse
testing, and high intensity flash x-ray equipment operated in the
bremsstrahlung mode for transient effects during impulse testing. A pulsed
reactor facility is used for neutron irradiation as described by Share and
Wasilik (1979). These authors also found that neutron effects at levels
encountered normally are as a result of electron ionisation rather than
atomic displacement. Thus the effects of nuclear irradiation in terms of
0
For some data this is stated or can be inferred, but for other data it is not
so and care is needed in using other data for more than a general guide. A
very good general review of the subject for multimode fibres is given by
Friebele (1979) and other recent useful information is given in SPIE, 296
1981, and SPIE, 404 !983.
Friebele (1979) reports that there are two basic fibre types which are
obviously the most useful for long lengths in radiation environments. These
are pure silica core fibres with polymer cladding (Pes) or fluorosilicate glass
cladding (FLCS), and silica core fibres doped with Ge0 2 and P20s and clad
with borosilicate-type material. More complex glass fibres and plastic fibres
are generally only useful in short lengths. The induced attenuation
behaviour of pure silica core fibre at 0.82-0.85 ,urn after impulse testing
with 'Y radiation is reviewed by Friebele (1979). The main feature of this
behaviour is transient attenuation followed by complete recovery after a
few minutes. The transient attenuation magnitude is variable depending
upon the fibre source and is markedly affected by the OH content (Barnes
and Wiczer 1981), radiation hardening and the light level travelling in the
fibre (West and Lenham 1982). It is in fact reduced by high light levels,
previous radiation hardening and a moderate OH content. PCS fibre is
reported to be generally the most radiation resistant (Sigel et a/ 1979), and
Friebele (1979) states that it is excellent for systems operating around
0.85 ,urn and requiring moderate fibre length runs, with downtimes in the
I 0 ,us to I min range following a moderate dose of about I 0 3 rads 'Y.
The most common telecommunications fibre is the germania doped silica
core fibre which has been manufactured with very low intrinsic loss and
offers useful transmittance at 0.85, 1.3 and 1.55 ,urn. F~iebele eta! (1980)
have reported that pulsed 'Y irradiation induced a large transient attenuation
in some members of this fibre class, and a large permanent one in others.
It would be of interest to use this more versatile type of fibre in radiation
environments and thus a number of reports of radiation studies on these
fibres have appeared (Share and Wasilik 1979, Friebele et a/1980, Rosiewicz
et a/ 1980). Friebele et at (1980) reported that within the limits studied
(13-31 wt OJo), the Ge02 content of the fibre had no influence on the radia-
tion sensitivity and neither had the OH content any significant effect on the
induced attenuation at 0.82 and 1.3 ,urn with concentrations from 3 to
160 ppm. Thus it appears that the OH content can be minimised to achieve
low intrinsic attenuation, whilst the Ge02 content of the core, and
hence the refractive index, can be tailored to achieve the required NA
without affecting the radiation response. The transient attenuation after
pulsed 'Y irradiation was found to be high at 0.82 ,urn and, although reduced
by a factor of about 100, was still significant at 1.3 ,urn. However the long
term recovery of these binary core fibres was good at room temperature.
The response of this class of fibres in the temperature range -77-100 °C
has been studied by Share and Wasilik (1979). Friebele et a! (1980) have
168 Infrared optical fibres
Table 7.4 Losses in dB km - 1 krad - 1 at transient and longer times for fibres after
impulse exposure to -y radiation.
Time
Material w-' s w- 1 s 5 min 30 min
Binary Si 02-Ge02 58 10 5 4.4
Ternary Si02-Ge02-P20s with
high P20s 16 14 13 12
Ternary Si02-Ge02-P20s with
low P20s 15 11 9 7
Once the optical attenuation in fibres had been reduced to levels which
were useful from an applications point of view, the attention of a number
of workers concerned with the practical deployment of fibre optic systems
turned to strength considerations (Olshansky and Maurer 1976). This is
because the brittle miture of glass can result in the failure of a communica-
tions system which might not be readily accessible for repair (e.g. a cable
Oxide glass fibres for near infrared 169
under the sea). Fibres in lengths up to about 10 km must not contain any
flaws which could result in failure under the conditions of service envisaged.
In general, fibres are required to retain high breaking strain ( > 1o/o) over
these long lengths for up to 20 years in their working environment. The total
flaw population includes intrinsic as well as extrinsic flaws. In practice it is
the extrinsic flaws created in the fibre surface during manufacture which
need to be eliminated or reduced to controlled low levels to maximise fibre
strength. A crack 1 I'm deep is likely to cause failure at 1% strain, thus
illustrating the magnitude of the problem. Fibres in deployed cables are
likely to be under some stress and to be subjected to some degree of
moisture attack, thus allowing flaws to grow by stress corrosion (Justice
1978). Thus even smaller flaws can be dangerous since they can grow to a
critical size by stress corrosion mechanisms, resulting in delayed failure.
Hence, having produced an ultra pure low loss material, it must be drawn
down to a fibre in an ultra clean· environment under rigid control to
minimise the size and distribution of surface flaws. Having achieved this
surface perfection, it must be preserved by coating the surface, normally a
plastic applied in line, to avoid any possibility of surface degradation during
subsequent cable manufacture (Ramsay et al 1982). Further research is
being carried out into improved primary protective coating materials such
as metals, silicon nitride and amorphous carbon (Chapter 9). Ritter (1978)
has reviewed the probability of fatigue failure in silica glass fibres in terms
of fracture mechanics. The data are illustrated by a design diagram. Such
a diagram for a fibre in a particular environment gives the probability of
failure for a given lifetime and applied stress, and the proof stress necessary
to ensure a minimum lifetime at a given stress. By employing this type of
approach, and using ultra clean drawing conditions together with good
coating practice, a gradual improvement in the strength of fibres has been
reported (Kurkjian and Rast 1981). The probability of a 5 km length of
fibre passing a 0.7 GPa proof test has increased from 0.1% (Maurer 1975)
to 10-90% (Maurer 1980) over the last few years, depending upon the
degree of care taken during fibre preparation and ptocessing. An important
step in obtaining long lengths of high strength fibre might be to fusion splice
together shorter lengths of proof tested material. It has been reported
(Kurkjian 1981) that fusion splices can be made with a failure probability
of 10% at 1.4 GPa and a median strength of 2.8 GPa. Most of the telecom-
munications fibre made by the vapour process is likely to possess an invar-
iant outer composition of Si0 2 or a high percentage Si0 2 glass, and thus
the major thrust towards strength increase is to improve and maintain the
already improved glass surface quality (Ramsay et at 1982). However, the
low melting multicomponent glasses made by the double crucible process
offer significantly different strength properties to those of silica. The
presence of alkali metal oxides in such glasses is likely to influence the
chemical durability and fatigue of the material. France et al (1983) have
170 Infrared optical fibres
l~ml
10 9 8 7 6
10
~ 1.0
~
8
5
"'~
~
~
<( 0.1
0.01,';;;;------;::!;:,:---:+,,----:;f,;;;--l
1000 1200 1400 1600
Frequency(cm-1)
low losses are not achieved, then researchers may well turn their attention
more seriously to sulphide and selenide glasses. Shibata et a/ (1980b)
have shown that optical losses in Ge-P-S glass may be as low as
10- 10 -10- 2 dB km - 1 at 5.5 ,urn. Dianov (1982) has considered the Rayleigh
scattering and the infrared absorption loss of As2S, and As 2Se, glasses and
predicted aminimum loss of around 0.05 dB km- 1 in the wavelength region
of 4-5 ,urn. CO laser calorimetry of high purity bulk glass indicated the
absorption losses in these glasses to be around 70 dB km - 1 • Miyashita and
Terunuma (1982) have drawn unclad fibre from high purity rods of As 2S3
glass and have found that the optical loss was limited by impurity absorp-
tion (e.g. SH at 4.1 ,urn) but nevertheless a loss of 170 dBkm- 1 was
measured at 5.25 ,urn. Kanamori eta/ (1983) have reported fibre losses of
64 dB km - 1 at 2.4 ,urn for As2S3 glass and 290 dB km- 1 at 3.4 ,urn forGe 5,
As 38, Se 57 glass. Katzir and Arieli (1982) indicate that losses of I dB m - 1
at 5 ,urn have been measured in a Ge-S glass. Thus it remains to be seen
whether there is enough interest in these chalcogenide glasses for mid
infrared applications to allow their theoretical loss levels to be realised.
This may well be best achieved by the utilisation of vapour deposition
techniques.
prolonged melting. Argon gas carrying GeC4, SbCls and Se2Clz was passed
through a lead glass substrate tube of composition mole OJo Si0 2 57,
Na,O + KzO 12, PbO 30 and dimensions 12-13 mm internal diameter,
14 mm external diameter. Heating to 600 ° C during deposition was done by
a traversing oxyhydrogen burner. After a glass of composition mole OJo Ge
28, Sb 12, Se 60 was deposited in the tube it was collapsed and drawn into
a conventional fibre or hollow core fibre at 800 o C. The hollow core fibre
had a measured loss of 0.7 dBm- 1 at 10.6 p.m and the conventional fibre
had a loss of 0.1 dB m - 1 • When a core glass of composition mole OJo Ge 28,
Sb 12, Se 60 and a cladding of composition mole OJo Ge 23.5, Sb 12, Se 64.5
were deposited in a lead glass tube and drawn into a solid fibre as before,
the measured loss at 10.6 p.m was 0.01 dB m - 1 • This patent claim represents
a breakthrough in reducing the loss in chalcogenide glass. If this achieve-
ment can be repeated for a commercial cabled product, selenide glass fibres
are very promising for far infrared applications. Glasses other than
chalcogenides offering far infrared transmittance are possible and the major
source of these is the halides. It has been demonstrated that fluoride glasses
are suitable for the mid infrared but their transparency is insufficient to
cover the far infrared. Thus it is the chlorides, bromides, and iodides which
offer adequate far infrared transparency for fibre applications. The
development of these glasses is in its infancy since until recently they have
only been of academic interest, and their likely properties (i.e. physically
weak and lacking in chemical durability) have not offered much attraction.
However a number of workers have now taken up the challenge and are
investigating the potential of these materials for 8-12 p.m applications.
ZnCh glass has been known for a number of years (Maier 1925) but it
is an extremely hygroscopic material sometimes difficult to prepare and
crystallises easily due to residual water (Goldstein and Nakonecznyj 1965).
Schultz (1957) reported a number of binary glasses based on ZnClz contain-
ing roughly 50 mole OJo KCl, KBr, or KI with the ZnClz-KI system being
reported as the most stable. Savage (1982) also reported binary glasses with
ZnCiz and up to 25 mole OJo PdClz, up to 300Jo Cdh and up to 70Jo CdBr2.
The glass transition temperature of these materials was reported to be in the
range 60-122 °C. Van Uitert and Wemple (1978) seriously considered that
ZnCiz glass demonstrated potential as an optical fibre material. Intrinsic
absorption, scatter and extrinsic absorption (when pure) were expected to
be small. However, the extreme hygroscopicity of the known ZnClz glasses
is a problem and, unless this can be overcome, is likely to exclude them as
serious candidates for fibre applications. Hu eta/ (1983) has reported that
ZnBr2 is vitreous and transparent to about 20 p.m but like ZnClz it exhibits
poor chemical durability.
Over the years other halide glasses have been reported, AgCl, AgBr, Agi
and PbBr2 (Sun 1946), PbClz-RaClz (Mellor 1929), SnCb-Pbh (Winter
1957) and TIC! (Moynihan 1971). More recently Angell and Ziegler (1982)
reported glasses based on BiCb in the binary BiCb-KCl system in the region
178 Infrared optical fibres
of 60-80 mole o/o BiCh, and have gone on (Ziegler and Angell 1982) to
measure the optical properties of these and glasses modified by the addition
of NaCI, TIC! and PbCh, and other glasses in the PbCb-TICI-BiBr 3
system. The refractive indices are in the range nd 1.96-2.22 and disper-
sions in the range Vd 8-14.9. However the glass transition temperatures of
these glasses reported in the range 25-45 o C are too low for fibre applica-
tions. Cooper and Angell (1983) have reported glasses based on Cdlz, such
as Cdlz-Csl-KI, but these show relatively poor moisture resistance and Tg
values are very low in the range 10-35 °C. Hu and Mackenzie (1982) have
also reported new glasses in the ThC4-NaCI-KCI system transparent to
around 14 JLm. The Tg of one of these glasses, mole% NaCI 30, KCl 30,
ThC4 40, is 130 °C and Matecki eta/ (1983) reported a Tg of 170 °C for
a glass of composition ~ole% CdCh 50, BaCh 40, NaC110, both offering
more promise for fibre applications. A useful review of halide glass forma-
tion is given by Baldwin et al (1981) and Mackenzie (1983). Much very
interesting scientific work is being done to discover new glasses amongst the
halides, but only time will tell if any of them are sufficiently robust in all
senses of the word to be seriously considered for fibre applications.
t
Most users would prefer to have a clad fibre, and Anderson {1981) has
taken the extrusion process for AgCl and AgBr a stage further to provide
one. At 10.6 ~tm the refractive index of AgBr is 2.0 and that of AgCl, 1.98.
This means that an AgBr core AgCl clad fibre possesses anNA of 0.28. This
is suitable for many signal and image transmission applications. A core of
AgBr and a tube of AgCl are first preformed in a press, to align the axis
of the core and tube in the direction of extrusion. The resulting preform
billet is then extruded to yield an AgBr core AgCl clad fibre. No details of
the fibre properties were given, but Takahashi eta/ (1983b) have reported
a loss of 0.22 dB m -I for a fibre made by a very similar technique.
Of the polycrystalline fibres recently researched, KRS5 clad in a loose
polymer tube stands out as being the most successful for short length near
term applications and has been offered commercially as Kristen 5 by Horiba
180 Infrared optical fibres
200
Extrusion temperature ( 0 ( l
Figure 7.12 Grain size against extrusion temperature for AgCI poly-
crystalline fibres, grain size can be varied between the two full curves.
250 and 500 I'm in diameter possessing an average grain size of 4 I'm. The
total attenuation coefficient was found to vary as A- 2 in contrast to the
Rayleigh ).. -• dependence in silica fibre. A model was developed which
showed that the ).. - 2 dependence resulted from the combination of bulk
scattering from large scale optical thin imperfections and surface scattering
and absorption. It was concluded that the most likely source of this
scattering was residual strain and poor surface quality from the extrusion
fabrication process. Thus in spite of the fact that KRS5 polycrystalline fibre
is the first successful fibre optic waveguide for the 8-12 I'm region, it is clear
from the above, that there are very serious loss problems which will
probably be difficult to overcome. Also, the extrusion technique is only
applicable to materials possessing appropriate mechanical properties such
as the silver and thallium halides.
Melt i'jelt
Fibre Fibre
Figure 7.13 Modified pulling down method for the growth of mono-
crystalline fibres. 1
a melt container in which a raw material rod was fused, a capillary through
which the melt flowed into a shaper, and the shaper which controlled the
cross sectional profile of the fibre. The length of the capillary was defined
so that the heat flux from the main heater could be isolated from the shaper,
which possessed its own independently controlled heater in order to achieve
a steep thermal gradient at its base. For KRS5 the diameter of the melt con-
tainer was 1 em, that of the capillary was 70-120 l"m and that of the shaper,
0.8 or 1 mm. The length of the shaper was 0.5 em and the length of the
capillary 4 em. Okamura et a/ (1980) went on to use this MPD technique to
grow Csl fibres 0. 7-1.0 mm diameter and up to 1.5 m long at a growth rate
of 5 to 6 mm min - 1 • It was found that the microstriations present on the
surface of the fibre caused weak scattering at 0.63 JLm. Mimura eta/ (1981)
used the MPD technique to grow CsBr fibre up to 1. 5 m in length and
0.7-2 mm in diameter at a growth rate of 5 to 10 mmmin- 1 • The total loss
. of this fibre at 10.6 l"m was measured as 5 dB m - 1 • As a result of the initial
success of the MPD technique, Mimura eta/ (1982) have further developed
it and used it to grow the additional materials KCl, KBr and KCl-KBr. For
the potassium compounds the grown fibres showed irregular surfaces and
square like cross sectional shapes due to preferred growth along the ( 100)
axis. In addition the fibres were quite brittle due to cleavage fracture along
the (100) plane. Of the materials discussed here, CsBr fibres were shown to
offer the best combination of optical loss and mechanical strength. Thus the
fabrication of CsBr fibre and the study of its properties was looked into
further. The growth crucible was simplified and made from platinum as
illustrated in figure 7.13(b). The nozzle was 4 em long and 0.4 mm internal
diameter, containing a Pt-Rh needle to adjust the liquid flow through it.
The crucible and nozzle were separately RF heated by two independent
induction units. The apparatus was contained in an inert gas atmosphere
and the fibre was drawn down by an endless belt puller. Although the CsBr
Optical fibres for far infrared 183
then total internal reflection occurs and radiation is guided within the
hollow core (Hidaka et a/ 1982). The lattice absorption, {3, centred at w0
occurs in the region of 1000 em -I for oxides and the imaginary part K of
the complex refractive index is related to {3 according to
{3 = (2wjc)K (7.7)
where cis the velocity of light. The relationship between the real part n, and
the imaginary part, K, of the complex refractive index can be expressed as
2 I~ w;K(w;)
n,(w)= 1 +- 2 2 dw;. (7.8)
71" 0 Wi- W
The need for advanced optical systems in the visible as well as in the infra-
red, is resulting in more demanding specifications for the optical surface
finish quality. This, together with a demand for a high quality but rapid
186 Specialist sample preparation, characterisation and testing
patterns are usually digitised and analysed with the aid of a mini-computer.
This type of instrument is described by Williams (1975) and Gaskin and
Lewis (1980), and a schematic diagram of that of Gaskin and Lewis is
shown in figure 8.1. Having established the basic homogeneity of a new
material by interferometry, the next requirement is to test the components
made from it for imaging quality. This is usually done by line spread
function (LSF) or modulation transfer function (MTF) instruments and a
typical instrument is shown in figure 8.2. LSF measures the blur on the
image of a sharply defined object, such as a thin hot wire or a line slit
source. It is a mathematical representation of the intensity of the image as
a function of distance in the image plane. The MTF of a lens system, at a
given spatial frequency, may be defined as the ratio of the modulation or
contrast in the image of a sine wave grating of that frequency, to the
'
P/V
camera
C02 laser
0
Sample under test
Figure 8.1 A schematic diagram of a modified Trope! interferometer
of working aperture !50 mm (Gaskin and Lewis 1980).
Target generator
Analyser unit
Vacuum
t
t
In-Out
Nanovoltmeter
Figure 8.3 A schematic diagram of a vacuum laser calorimeter (Skolnik 1975).
190 Specialist sample preparation, characterisation and testing
{3 = mCp
LPy
X -{!!...-[(dTgain) + (dT1o")
n +1 dt r, dt r,
J {3L <€. 1 (8. I)
where E is the Young's modulus, vis Poisson's ratio, K is the thermal con-
ductivity, a. is the linear coefficient of thermal expansion and a the rupture
modulus.
Examples of situations where thermal shock can be a problem are high
power laser windows and infrared domes or windows in air and space
environments. A water quench test is commonly used to test bar specimens
of materials by heating them to particular temperatures and quenching
them in a water bath through a range of temperature gradients. The
strength retained after quenching, measured in a flexure test of the bar
specimens, is related to AT. The strength declines rapidly at a threshold ''
i:
value of ATe, and the thermal shock resistance is related to ATe and the
i I
retained strength at values of AT> ATe. Superficially this test appears to
be simple but quantitative interpretation is difficult. This type of test has , I
I .
I '
I I
!
192 Specialist sample preparation, characterisation and testing
8. 3. 2 Fracture stress
The science of fracture mechanics stems from the work of Griffith (1920)
and it is now generally accepted that the failure-of brittle materials is con-
trolled by the mechanics of crack growth from small flaws (Lawn 1983), and
that fatigue and stress corrosion can cause very significant strength reduc-
tion (Fuller et a/1983). In order to develop an adequate knowledge of these
mechanisms and provide a useful structural design technology for the
relatively weak and specialist materials used in infrared optical applications,
it is necessary to be able to evaluate the strengths of these optical ceramics
in their different flaw states definitively. Traditional four-point bar breaking
techniques are subject to edge failures to a large degree and a considerable
number of specimens need to be consumed to establish any reliable data.
Recently this situation has been much improved by the development of a
hydraulic pressure loading test first used in the British Glass Industry for
the strength testing of plate glass (Bowles 1973), and further developed by
Matthewson and Field (1980) and also used by Shetty eta/ (1983). This test
makes use of disc specimens which are commonly available for research and
production materials alike, since they are used for transmittance and
absorption measurements, and perhaps also for thermal shock
measurements as discussed in §8.3.1. Thus the disc of varying diameter is
becoming the standard test piece shape for the measurement of many
physical properties. The apparatus, procedure and calculation of results for ,/
fracture stress measurement is described by Matthewson and Field (1980).
The apparatus consists of a steel pressure vessel in which the specimen
support ring is of hardened alloy with the surface in contact with the
specimen ground accurately flat. Pressure is generated by a hydraulic
system separated from the specimen by a neoprene diaphragm, as illustrated
in figure 8.4. Specimen diameters can be 50 mm or 25 mm depending upon
Oil
Pressure____.
the availability of the material under test. About six specimens are necessary
for each data point, and before testing the specimens are covered with self-
adhesive tape to retain the fragments after fracture and to act as a gasket
between each specimen and the support ring. The attractions of this method
for routine strength evaluation are the elimination of edge failures and
stress concentrations common in mechanical loading systems, the sim-
plicity of the theory and the easy applicability of Weibull statistics. The
environmental conditions and the stressing rate are also easily controlled.
assumes that the cracks do not grow after indel)tation so that if post-
indentation slow crack growth does occur, the results will be subject to an
error dependent on the rate of crack growth and the time of measurement
of the crack length after indentation. The working range of indenter load
must be selected to ensure that the crack pattern is well developed (Co ;;;> 2a,
Co being the post indentation cracksize, and a the size of the hardness
impression) but not so large that chipping occurs. It is also important that
the test surfaces contain no pre-existing stresses prior to indentation. Figure
8.5 shows a schematic indentation fracture system produced by the Vickers
indenter, and the basic formula for the determination of the fracture
toughness, K1c, is
where~~
is a calibration constant (Lawn et a/ 1980) determined as 0.016 ± 0.004
(Anstis et a/1981), His the hardness, E is Young's modulus and Pis the
peak load. In further work Chantikul eta/ (1981) evaluated the indentation
technique for strength test pre-cracking to provide a dominant flaw in test
piece materials used in conventional strength testing. This method can be
used for materials which do not yield well defined radial indentation crack
patterns, but many more specimens are required to provide the data.
pagating into the elastic zone. Radial and lateral cracks occur as illustrated
in figure 8.6, and it is the lateral cracks which are most important in terms
of transmittance and erosion (Evans et a/ 1978). Such cracks initiate
whenever the impact velocity exceeds a critical value, Vc, given by
- i: '
'
(8.4)
I
where K1c is the fracture toughness, His the hardness of the solid, m is the
'I
mass of the particle and }q is a material independent coefficient. Clearly the
'I
fracture toughness is very important in relation to inhibiting fracture. The
detrimental effect of high hardness is due to the direct proportionality
between the hardness and the amplitude of the residual stress field which
dictates the crack driving force (Lawn and Evans 1977). Once cracking has
been initiated then erosion follows under the influence of continued impact
according to an erosion rate, e, derived by Evans (1981) and consistent with
practical evaluations where I
, I
(8.5) I I
I'
196 Specialist sample preparation, characterisation and testing
Plastic zone
J
Figure 8.6 A sectional representation of an impact site, showing the
plastic zone together with lateral and radial cracks (Evans 1981).
and rain erosion resistance has only recently begun to be forged, actual
water impact testing of materials is very important.
The testing of small samples takes two forms. In the first, rain impact by
single water drops is simulated using water jets, fired at a range of velocities
from a compressed gas gun, followed by quantitative assessment of the
damage caused by means of residual strength measurements. In the second,
a standard rainfield is simulated by spray nozzles resulting in multidrop
impact inside a rotating arm facility. In this type of testing the damage is
assessed visually or spectrophotometrically in relation to exposure time in
the equipment. Testing of large samples is usually carried out by mounting
· actual window components on a rocket sledge and firing it along a rail,
through a section containing spray nozzles simulating a multidrop rainfield.
!
Impact
_____(
Wata- jet )
Bowden and Brunton (1961) first developed high velocity liquid jet
impact equipment to fire water jets at stationary specimens. Field et a/
(1979) showed that it was possible to obtain a reasonably accurate simula-
tion of drop impact by means of a water jet, and placed this jet test method
on a quantitative basis. In this method an airgun slug is fired into a stainless
steel chamber containing a small quanitity of water sealed in by a neoprene
disc. The projectile and neoprene drive forward as a piston and extrude the
water through a narrow orifice. The ratio of water jet velocity to projectile
velocity is typically 3 to 5. A 0.4 mm nozzle produces a jet which simulates
2 mm drops for velocities in the range 300-600 ins- 1 • Residual strength
testing after impact is achieved by means of the hydraulic fracture stress
measurement technique described in §8.3.2. Field et a/ (1983) discuss the
liquid jet impact damage and residual strength curves of several infrared
optical materials, and report threshold velocities for rain impact damage for
2 mm drops as 170 ms- 1 for ZnS, 205 ms- 1 for Ge and 455 ms- 1 for
AJ,0 3 (basal plane). Water jet impact testing is clearly very useful and
instructive in determining damage threshold velocities for research and
production materials alike.
There are a number of rotating arm rain erosion test facilities in the
world; some are subsonic and others subsonic and supersonic. Most
equipment expose samples to a multidrop rainfield but at least one has the
facility for single drop impact testing (Hackworth 1982). A description of
198 Specialist sample preparation, characterisation and testing
Cooling fins
0
Rotating arm
a facility capable of reaching speeds of Mach 2.0 has been given by Foulke
(1981) (figure 8.8). This consists of a 1.22 m radius arm, rotating in a 3 m
diameter steel chamber and carrying a 25.4 mm diameter sample. At Mach
2.0 the pressure in the chamber is reduced to 1/3 atmosphere to reduce the
power requirements of the drive motor. The cooling fins shown in figure 8.8
are used to reduce the temperature in the chamber during high speed rota-
tion, and in addition control the induced turbulence. The angle of impact
can be adjusted and the rainfall is created by 30 hypodermic needles
shrouded by 16 mm diameter brass tubes to avoid shock wave disturbances
and chamber turbulence. The drop size and rain rate are controlled by the
water pressure and needle size. The standard operating conditions produce
12.7 mmh- 1 rain rate with a 2.0 mm drop size. Such equipment can give
a very rapid assessment and comparison of samples of both research and
production materials. Results for infrared optical materials obtained on
I '
I '
'
..
I
,,I' I
'
9
Optical Coatings
where R is the reflectivity of one surface and n the refractive index of the
material for that particular wavelength. For instance, the transmittance of
a plane parallel plate glass window, with refractive index nd = 1.5 is 92"7o,
while that of a similar plane parallel plate of germanium of refractive index
I'I
Optical coatings 201
n10 = 4.0 is 47"7o. This means that the transmittance of a seven element un-
coated lens made from glass is likely to be of the order of 500Jo, whilst that
of a similar uncoated germanium lens is likely to be roughly 0.5%. These
transmittance values take no account of material absorption or scatter. This
illustrates the magnitude of the problem where many elements are required
in an optical system and where the refractive indices of these elements are
high. Thus antireflection coatings are essential in modern optical systems in
order to achieve a usable image.
Historically the subject has developed from the last century when
Rayleigh (1887) observed that old glass plates had a lower reflectance than
new ones. Taylor (1896, 5th edition 1983) noticed this effect on lenses and
went on to develop an etching technique to produce an artificial surface
tarnish on lenses. At this time, and during the very early part of the present
century, it was assumed that the decrease in the reflection seen on glass
samples was due solely to a change in Fresnel reflection coefficient. Bauer
(1934) correctly deduced that the reduction in reflectance was due to an
interference phenomenon. In addition, suitable vacuum pump oils were
made available at this time enabling the technology of vacuum deposition
to be developed. The technology proved to be ideally suited to the produc-
tion of these interference coatings. Thus in the 1930s the theory, the
technology and the realisation of the importance of the phenomenon were
put together to yield the first practical antireflection coatings. During the
last 50 years, the design and practical realisation of coatings for the visible
and infrared have developed to the point where antireflection coatings are
indispensibie to present-day optical technology. The general theoretical
aspects of optical coatings design are well established and are given in the
literature, but detail on the actual design and deposition of state of the art
coatings are not well reported for commercial-in-confidence reasons. The
theory of optical loss in bulk materials is now becoming well known. This
has been related to loss measurements of materials, and in many cases to
the morphology, impurity and defect states in tnaterials because of the
extensive characterisation which has taken place: In comparison, coatings
technology is still a black art to those outside the industry because detailed
materials characterisation, with reference to optical and environmental
performance, has only recently been carried out and reported to any great
extent (Bennett 1980). Current optical coatings, while offering good per-
formance for low power applications, demonstrate a very high optical loss
coefficient in relation to their corresponding bulk materials. It has now been
recognised that this high loss results from impurities, defects and major
morphological problems some of which are likely to be overcome, but only
when rigorous characterisation of the coatings has been carried out in rela-
tion to deposition conditions and physical performance. The need for low
loss laser coatings during the past few years has emphasised this problem
and a number of workers have recently published information on the
202 Optical coatings
There are several vacuum techniques which can be used to synthesise anti-
reflection coatings on a variety of flat and curved substrates. Amongst these
, I
are evaporation (e.g. thermal, electron beam, ion plating and laser),
; I
204 Optical coatings
sputtering (e.g. AC, RF, magnetron and reactive) and glow discharge
techniques (Vossen and Kern 1978) illustrated in figure 9.1. Others such as
ion implantation, ion exchange, CVD and deposition from solution have
not come into general use for coating infrared optical materials.
Substrate Substrate
2-SkVnegative
I
RT-200 °( RT-200 °(
I
k???i??/1
Source Source
(a) (b)
Substrgte Substrate
0-200Vnegative
T
RT-200 °( zoo•cI
I
/ ....... - ...... , \ Ar+
{Plasma
~.----- ......;
Si02
T • T
1-5 kV negative 1 kV negative
RF Rf
cathode cathode
(c) ldl
Figure 9.1 Vacuum techniques for the synthesis of antireflection
coatings at certain pressures and particle energies: (a) thermal evapor-
ation at 10- 6 Pa and 0.1 eV; (b) ion plating at 10- 2 Pa and
10-100 eV; (c) sputtering at 10- 2 Pa and 10-20 eV; (d) glow
discharge at w- 2 Pa and 10-200 eV (Green and Lettington 1981).
The most well developed and widely used of these is evaporation in high
vacuum because of the flexibility, uniformity and control it can offer.
However, the structural perfection and packing of the films so produced
often leaves much to be desired, as discussed in §9.4, and the bonding to
the substrate can sometimes be weak. This can be overcome to some extent
with very good substrate cleaning and the use of thin bonding layers. Dur-
ing electron beam evaporation at, say, 10- 3 to 10-s torr working pressure,
Manufacturing technology 205
the particle kinetic energy at the source is about 0.1-0.2 eV. Magnetron
sputtering at a working pressure of 10- 1-10- 2 torr offers particle kinetic
energies of the order of 10-20 eV, while the ion impact at the substrate can
be further enhanced in RF plasma techniques since the particle kinetic
energies can be the order of 10-200 eV or more. Thus it can be seen why
the bonding to the substrate of thermally and electron beam evaporated
coatings can be poor, and how it can be improved by resorting to sputter-
ing, ion plating, or RF plasma deposition. In these techniques, the surface
being coated is subjected to ion bombardment before and during deposi-
. tion. This enhances the bonding at the interfaces of the antireflection
coating layers and encourages the removal from the substrate of very ·
loosely bonded deposited material. The techniques of ion plating and
sputtering have not come into extensive use for coating deposition on
infrared optical materials. However, RF excited plasma deposition has
recently assumed some prominence for the deposition of anti-abrasion, or
protective antireflection coatings, to the exterior windows or elements of
thermal systems where a high degree of bonding to the substrate is essential.
The general evaporation technique for the deposition of thin films is
reported by Holland (1961), Powell et a! (1966), MacLeod (1985) and
Chopra (1969). Substrates are usually cleaned ultrasonically and then
loaded into the dome of a vacuum coating unit which is subsequently
evacuated. The shuttered electron beam or thermal evaporation sources
situated in the base of the unit are energised, and the rotating dome is
heated to achieve the required substrate temperature for deposition. When
the conditions are established, the sequence of layers is deposited by control
of source temperatures, and shutter operation aided by quartz crystar
monitoring. The quality of the source materials is crucial to the achievement
of good quality films. The purity, gas content and grain size are important
in avoiding gas outbursts and sputtering during the evaporation process.
Hass et a! (!959) simply discuss the preparation of rare earth oxides and
fluorides, but a large number of materials suitable for evaporation are now
commercially available. Coleman (1973) and Sites•"et a! (1983) reported the
deposition of a number of optical films of oxides and fluorides by the
sputtering technique (Stuart 1983). In this technique a substrate may be
suspended above a solid cathode target of the coating material in a vacuum
system into which is leaked argon gas. Wh~n high voltage (usually RF) is
applied to the cathode and the pressure is adjusted to 10- 1-10- 2 torr a
discharge is set up in the system and argon ions strike the target, thereby
removing small particles of the target material which then adhere to the
substrate forming the required coating. Reactive gases may also be used in
the vacuum system to adjust the stoichiometry of the coatings. Davy and
Hanak (1974) have reported the deposition of dielectric films by an ion
plating process (Mattox 1973). Here ion bombardment is applied to the
substrate before and during evaporation.
206 Optical coatings
Ion plating is usually done in an inert gas discharge similar to that used
· in sputtering, except that in this case the substrate is made the high voltage
sputtering cathode. For a coating to be deposited it is necessary for the
deposition rate resulting from the evaporation to be greater than the
backsputtering caused by ion bombardment. Recently the RF excited
plasma deposition process has been used commercially for the deposition
of amorphous carbon coatings. This technique is described by Holland
(1981) and Green and Lettington (1981). The Holland technique consists of
leaking a hydrocarbon gas, such as butane, into a glow discharge chamber
with one RF electrode grounded and the other capacitively coupled to an RF
source in the megahertz range. This serves as the substrate carrier with a net
negative bias. In the discharge region the hydrocarbon is ionised, and the
positively charged particles are accelerated towards the substrate to form an
amorphous carbon coa{ing. Conventional sputtering equipment can be
readily utilised for this glow discharge process. It is adequate for the syn-
thesis of hard coatings on metal, semiconductor or insulating substrates,
such as aluminium, silicon, and germanium, but tends to yield films with
a high compressive stress (Holland and Ojha 1978). This is because, in order
to achieve an adequate deposition rate, a high potential has to be applied
or developed at the cathode and hence the substrate is subjected to high
energy particle bombardment. This maintains a significant backsputtering
rate and tends to yield a highly stressed coating. The Green and Lettington
(1981) technique uses similar equipment except that, in this case the genera-.
tion of ions and the deposition of the coating are controlled independently.
This is reported to allow the deposition of amorphous carbon with much
reduced induced stress thus allowing thick low strain coatings to be pro-
duced. Using this method, butane is leaked (- 10- 2 torr) into a glow
discharge chamber with a capacitively-coupled water cooled RF (e.g. 1 kV,
13 MHz) cathode serving as the ion generating source. The substrate is
mounted on the second electrode which can be cooled or heated (usually
200 o C in the case of Ge) independently of the cathode, and to which may
be applied a negative AC potential (0 to -200 V). The positive ions formed
in the plasma strike the cathode forming a deposit there, but a significant
number of neutral carbon atoms leave the cathode and strike the substrate
to form the required layer. The bias voltage when applied to the substrate.
affects an ion plating enhancement of the deposition. Thus, in this tech-
nique a much greater degree of control and flexibility of the deposition is
reported to be attained.
In general, fluorides and many other halides can easily be evaporated and
condensed stoichiometrically. They have been used extensively in the
synthesis of antireflection coatings. However, some fluorides and all
chlorides, bromides and iodides are soluble in water, which makes them
impractical as coating materials. Two of the most important and extensively
208 Optical coatings
employed fluorides are MgF2 and ThF4. In mass spectrometer studies of the
evaporation process of dielectric materials, Pulker and Jung (1969) found
that MgF2 evaporates practically without decomposition. Ritter and
Hoffman (1969) investigated the effect of substrate temperature on the con-
densation and the films of MgF2. Films deposited on unheated substrates
possessed a porous structure with a calulated packing density of 0.80-0.84.
On exposure to air these porous films took up water which led to an increase
in the refractive index, and therefore an increase in the optical thickness.
At substrate temperatures of 190, 280 and 340 o C the packing densities were
found to be 0.895, 0.935 and 0.955, respectively. The mechanical and
chemical properties of the films were dependent on the packing density.
Those with a high packing density, put down at a substrate temperature of
around 300 o C, were extremely hard and adherent, and such films are used
as single layer coatings,' or as low index films in multilayer antireflection
coatings. ThF4 layers, transparent from 0.2-15 p.m, have been used for a
number of years in coatings on infrared optical materials, particularly in
combination with ZnS in multilayer coatings. This fluoride is mechanically
quite stable and films with high packing densities can be achieved (Heit-
mann and Koppelmann 1967). Often the raw material is relatively impure
I
and contains ThOF2, but this is not a major problem for non-critical
I applications, since above 1000 °C reaction (9.2) occurs and the residual
Th0 2 does not evaporate until the temperature is greater than 2000 o C.
2ThOF2-+ ThF4 + Th02 (9.2)
Ritter (1975) references many useful data on these and other fluorides used
to make interference coatings.
Of the chalcogenide compounds, ZnS has become one of the most useful
coating materials, particularly for germanium since it provides a very
convenient single layer antireflection coating. The environmental stability
and adhesion of ZnS layers depends on the substrate cleaning and its
. temperature during deposition. Germanium components are usually glow
discharge cleaned prior to evaporation, and deposition occurs at a substrate
temperature of 150 °C. Films deposited in this manner are reported to
withstand exposure to moisture, several hours boiling in 50Jo NaCl solution
and repeated washing without damage (Cox and Hass 1958). ZnS evaporates .
effectively at around 1200 o C and dissociates into Zn and S (Pulker and
Jung 1969). There may be a danger of depositing non-stoichiometric films
because of this dissociation, particularly at high substrate temperatures
where the condensation coefficient of Zn is decreased (Ritter and Hoffman
1969). Useful properties of evaporated ZnS films are reported by Preisinger
and Pulker (1974).
The suggestion by Black and Wales (1968) that chalcogenide glasses have
much to offer has been taken up in research. Mixed glasses tend to be
unsuitable for conventional thermal evaporation because of non-
'-
The most widely used optical thin film deposition technique is that of
vacuum evaporation from thermal or electron beam sources. While this is
a relatively simple technology, there are nevertheless a very large number
of interacting deposition parameters. The interpretation of the optical and
environmental properties of the films in relation to these parameters is thus
very difficult. Historically these difficulties stem from the Jack of layer
characterisation other than optical, so that there has been a general lack of
understanding of the film physical structure, how this affects the optical and
I
210 Optical coatings
9.4.1 Microstructure
The major relevance of microstructure to optical coatings technology was
initially not appreciated. However, it was recognised that dielectric layers
possessed a structure which contained voids. The term packing density was
created to measure the void content, and there is much discussion in' the
literature in relation to this term for particular films.
p k" d "t _ Volume of deposited material ( _)
ac mg ensi y- TotaI vo Iume of t he Iayer ("meIu d"mg vm"d s) · 9 3
Koch (1965) first considered the microstructure of optical films when he
studied the optical drift of MgF2 films caused by water absorption in the
grain structure, i.e. it was assumed into the pores. Since then advances in
electron microscopy have allowed the determination of the structural
features of vapour deposited optical thin films. One of the most crucial
parameters in determining the microstructure is the substrate temperature
Layer characterisation 211
which is typically very low (of the order of a few hundred centigrade) in
relation to the melting points of the evaporated materials. It is not surpri-
sing therefore that the microstructure is rather poor quality, being very
columnar containing many pores which act as sinks for impurities and water
vapour. Vook (1982) has reviewed the various modes of thin film growth,
the most relevant being the Valmer-Weber model of three-dimensional
island formation. This is promoted during the nucleation and coalescence
stage if the incoming vapour species have insufficient energy for significant
surface diffusion on the substrate. This can readily occur in thermal
evaporation if the substrate temperature is low. In this situation, the
probability of void formation in the films, particularly at grain boundaries,
is high (Nakahara 1977, 1979). In a review by Nakahara (1979) the aging
phenomenon, the variation in film properties over a period of time, is
attributed to the annihilation of non-equilibrium defects by mobile vacan-
cies. Impurities such as hydrocarbons water vapour and other compound
species are also a problem if they are insoluble in the coating material, since
they then accumulate at the grain boundaries altering the surface energy of
these interfaces. For instance, this can influence the mechanical stress in the
films as discussed later.
It has been shown by Movchan and Demshishin ( 1969) that the ratio of
substrate temperature to the melting point of the evaporant, T,fTm is an
important parameter in determining the structure of both metal and
dielectric films. When values of this ratio are less than 0.45, u~ually the case
in optical film deposition, the structure of the films is found to be columnar
with the columns running in the direction of growth. Also Vincett et al
(1977) proposed a critical substrate temperature of 0.33 of the boiling point
of the evaporant at which both polycrystalline and epitaxial films exhibit
optimum properties. They offered the explanation that at this temperature
the disordered or glassy regions in the depositing film can just boil against
the impinging vapour flux resulting in improved structural perfection. Thus
it can be seen that the substrate temperatures normally encountered during
thin film deposition are unlikely to lead to high quality polycrystalline struc-
tures, or indeed to high quality amorphous films, since columnar structures
have also been reported in aGe (Swab et a/1980, Dirks and Leamy 1977).
The determination of the structure of thin films has been aided by
advances in electron microscopy. Such a technique is necessary since film
thicknesses are under 1 JLm, and the structural features of interest are
smaller. The replication technique offers the required resolution of under
5 nm or less and has been described by Guenther and Pulker (1976) and
Guenther (198!c). Direct observation by SEM of thicker coatings is possible
in the case of metal films, but dielectric films are more difficult since a con-
ducting coating must be applied to them to avoid charging effects. Some of
the parameters applicable to the technique of electron microscopy are given
212 Optical coatings
Thicker films tend to exhibit larger diameter columns than thinner ones
and MacLeod (1982) has presented a solution originally provided by Ross
and Messier (1981). The larger columns in thicker films are thought to
consist of bundles of chain-like sub-units, and hence there is thought to be
a microstructure within the more obvious microstructure of the film itself.
In addition to this columnar microstructure, macroscopic nodular defects
become visible to the eye as a result of light scattering (Guenther 198la,b).
These defects are individual growth distortions consisting of inverted
parabolic cones originating either at the substrate surface, or at the coating
material spatters incorporated into the growing films. These nodules can be
a particular problem since they are likely to be broken out of a coating
Layer characterisation 213
Table 9.3 Critical factors for columnar and nodular growth. j'
Columnar growth Nodular growth
instance ZnS was found to deposit with a compressive stress and MgF2 was
found to develop a very high tensile stress. It was shown that the defor-
mation of a multilayer coating was proportional to the product of the
average stress and the film thickness, but that the average stress in a multi-
layer could not be predicted from the stresses found for the individual single
layers. Stress in dielectric films has been studied by Pulker (1979b) and
Pulker and Maser (1979) who described a grain boundary model for an
approximately quantitative treatment of the intrinsic stress of crystalline
films. As a coating nucleates, the crystallites often grow as isolated randomly
oriented aggregates. At a later stage these aggregates coalesce in two stages.
The aggregates are enlarged until the gap between adjacent crystallites is very
small. It is assumed that the spacing between atOII\S across the gap can have
any value between a and 2a, where a is the lattice constant of the conden-
sate. Then the interatomic forces acting across the gap cause a constrained
relaxation of the top layer of. each surface forming a grain boundary. The
gap spacing then decreases by A to between a- A and 2a- A. The relaxa-
tion is constrained due to the adherence of each crystallite to the substrate
surface. Thus the energy due to elimination of free surface at grain boun-
daries produces volume strain energy. Using this model it can be seen that
moisture and impurities segregating at grain boundaries have a strong
influence on the surface energy and therefore on the stress. For instance
CaF2 (1.20/o) or ZnF2 (0.8%) impurity in MgF2 layers reduced the tensile
stress. by a factor of two. Control of the stress in films by selected impurity
addition, or by mixing of materials developing opposite stresses is impor-
tant, not only in avoiding gross effects but also since the stress in films has
been correlated with laser damage thresholds (Austin et a/ 1973).
When multilayer antireflection coatings are to be used in laser systems,
the optical loss is of prime importance. For example water absorption in
coatings can be a particular problem for HF lasers operating at 2.8 I'm.
Temple (1979) has studied the surface, bulk and interface absorption on
wedge shaped films at 2.8 I'm. In further work Donovan et a/ (1979).
correlated hydrogen concentrations and profiles, measured using a nuclear
resonant technique, with these calorimetric measurements. The values of
absorption calculated assuming that the hydrogen was present as water in
the films, agreed in concentration and distribution with the absorption.
values measured by calorimetry. The results showed that H 20 was a major
source of contamination and absorption in a variety of coating materials.
ThF4 films were found to absorb large quantities of H 20 when they were
exposed to air. ZnS and As2Se3 films were found to contain fewer than
10 ppm H20 and were effective encapsulants for ThF4 films reducing the
H20 from the order of 20 ppm. Materials with less affinity for H 20, such
as ZnS, As2Se3, Ah03 and Si02 were found to be contaminated predomi-
nantly at air/film or film/substrate interfaces rather than through their bulk.
Lusk (1982) also studied ThF4 films by scanning AES, x-ray photoelectron
Recent developments for specific applications 217
material has been reported by Vora and Moravec (1981). The layers were '
examined in a TEM and it was found that the films were predominantly
amorphous, even when annealed in UHV up to a temperature of 600 °C
(Ojha and Holland (1977) report that films become graphitic on heating to
750 °C in argon). The degree of crystallinity of the films was observed to
vary with deposition parameters in a manner which was not understood.
·The films contained several cubic crystalline phases, in agreement with the
findings of Weissmantel et al (1980). SIMS survey work on films produced
by RF plasma decomposition indicated only carbon, hydrogen and oxygen.
The oxygen and OH were present mainly in the surface. Large fragments
of C and H were evident as C2, C2H2, C3, C3H, C4, C4H, Cs and C6. Since
the hydrogen in the films did not show major infrared activity, Vora and
Moravec (1981) were not able to determine how these fragments were
bonded into the films. Be·cause of the low degree of crystallinity it was not
found possible to establish any correlation between the SIMS data and the
crystalline phases seen in the SEM investigations.
The success of a-C as a protective antireflection coating rests, not only
on its excellent intrinsic mechanical properties and adhesion because of the
deposition method, but also on its amorphous structure, which is not likely
to offer the same opportunity for impurity and moisture ingress as the badly
packed columnar crystalline structures of most antireflection coatings. It is
likely, therefore, that there will be further developments of this type of
coating in terms of different materials with refractive indices suited to a
greater range of substrates, and perhaps multilayers based on them
(Gautherin and Weissmantel 1978). However, the absorption due to
hydrogen bonded in these layers as H- X (e.g. X= C) is likely to restrict
their use to thin layers in low power situations, unless it can be reduced to
exceedingly low levels. It is not yet absolutely c~ear whether hydrogen is
essential in obtaining the amorphous structure, or whether it is present
merely as a consequence of employing hydrogen containing gases in the
glow processes; but a number of amorphous materials containing hydrogen
have been reported. For instance, Anderson and Spear (1977) have reported
that hydrogen containing amorphous silicon carbide, germanium carbide
and silicon nitride can be produced by a glow discharge process, and within
certain limits over a usefully wide composition range around the
stoichiometric value. Le Contellec eta/ (1979) and Yoshihara eta/ (1981)
have grown and measured general properties of amorphous Si-C-H films,
Catherine and Turban (1980) have studied the infrared absorption proper-
ties of Si-C-Hand Ge-C-H amorphous films. Weissmantel (1979) and
Turban and Catherine (1976) have studied Si-N-H films. A number of
220 Optical coatings
depends upon their ability to tolerate and dissipa,te the heat created by
absorption of the laser energy passing through them. The two parameters
of prime importance are absorption and thermal conductivity. The thermal
conductivity of materials tends to be insensitive to the method of synthesis
so that the performance of laser optics depends upon their absorption
characteristics. Proper cleaning of the coating-substrate interface is essen-
tial, since the component polishing process may leave residual contaminants
embedded and adsorbed in the surface. In addition the coating deposition
chamber pumpdown and outgassing cycles are important in avoiding a
greater than necessary surface hydrocarbon and water concentration. This
contamination problem could also occur at coating-coating interfaces if
any time delay occurs between the multilayer depositions. When the
finished coating is exposed to the atmosphere the porous columnar structure
fills with water thus incJieasing its absorption coefficient. The presence of
water in coatings is considered to be one of the major problems in current
thin film technology (Glass and Guenther 1977).
A link between absorption and damage in high power laser coatings has
been established by Kuster and Ebert (1979). Layers of Ab0 3, BeO, MgO,
Hf02, Zr02, Nd203, Ce02, Ti02 and Si02 were electron gun evaporated in
thicknesses of 0.5 !Lm on to suprasil 1 glass substrates. Damage thresholds
were obtained with 800 ns pulses from an unstable resonator-type Nd 3+
glass laser with a Gaussian far field intensity profile. The occurrence of
damage was determined by electronic registration of laser induced
scattering. The absorption of the layers was measured in the temperature
range 600-2000 °C. It was found that the absorption coefficients showed
a strong exponential temperature dependence, and were inversely propor-
tional to damage thresholds near the destruction temperature. The latter
occurred far below the melting points of the oxides. Pawiewicz et al (1979)
studied the damage properties at 1064 nm of stoichiometric oxide coatings
being mainly Ti02 deposited by reactive sputtering. High damage
thresholds of 7-10 J em- 2 were obtained for Ti02. Correlation of damage
with grain size indicated that glassy coatings had the highest damage
resistance. Damage occurs in a multilayer coating when the local electric
field intensity reaches a critical value in the coating material most prone to
damage. Apfel (1977) has suggested modifying the design of a multilayer to
reduce the electric field in the critical layer, but of course this inevitably
means it is increased in an adjacent layer. The limit of this technique will
be determined by the laser damage threshold within the stronger of the layer
materials.
Sparks (1977) has suggested a near term development programme aimed
at the achievement of lower loss and higher laser damage threshold coatings
at 10.6 !Lm. He suggests using only ultrapure materials and depositing them
by sputtering or UHV techniques. The materials should have a bulk absorp-
tion of under 0.5 em -I, be non-hygroscopic and show high packing density,
A new approach, particularly for ultra low loss coatings 223
) problems discussed in §9.4 make the realisation of very high laser damage
resistance in coatings produced by current technology extremely difficult.
This evidence, although slender at present, does point the way towards
the UHV approach for high technology coatings. Thus a systematic optical
and materials characterisation study of non moisture sensitive, ultra high
purity coatings, deposited in UHV equipment, taking rigorous precautions
to reduce water and other deleterious gaseous impurities, would be well
worthwhile. In addition, a study of the glassy morphological approach for
at least some if not all of the films in multilayers would be well worth
investigation as a means of preventing moisture ingress during service. This
science based approach would be likely to lead to much less lossy, more
damage resistant and environmentally stable coatings.
co,
spectrum
Scanning
Auger ion
analysis gun
facility
6
Power
,
me:ter
Chopper
detector
,.
Figure 9.2 A UHV equipment for research into the synthesis of low loss, high
damage resistance laser coatings.
9. 7. I Specifications
Optical coatings are used in civil and military applications. All of these
coatings must pass some accepted quality assurance specification to enable
the users to have confidence in the commercial products on offer. In the
absence of agreed and well publicised specifications for coatings on infrared
226 Optical coatings
100 5
I
100
\
I I'
80 . ;;-
\
I
I
I
~ ~
90 I 3-
• ~ 60 . .E I
\ I
I
'- ""
.E
~
.Q
:g
,g
~ 40 . -
.=0
.\
I
\ I
I 2].
&!
\ I
20 ~ - 80
\
\_/I
I
0 •
6 8
··-- -
10 12 '
14
'
16 7 10 11 11
0
13
Wavelength ( !l m) Wavelength [Jlm}
Figure 9.3 Antireflection coating for Figure 9.4 Rugged antireflection coat-
Amtir I chalcogenide glass both sides of ing for outside surfaces of germanium
substrate coated with high efficiency optical components interfacing with
coating; transmittance (full curve), stringent environments, rear surface of
reflectance (broken curve). substrate coated with high efficiency
coating; transmittance (full curve), reflec-
tance (broken curve).
8-12 p.m rugged antireflection coating for germanium (os type) (figure 9.4)
Spectral characteristics: average transmittance 8.0-11.5 p.m 87"7o when
substrate rear surface is coated with high efficiency coating. Conforms
to humidity, abrasion, salt spray and salt solution tests of MIL-C-675A
and humidity, adhesion, salt spray, temperature cycle and hardness
tests of MIL-M-13508B. In addition the coating will withstand 60 000
wipes of a windscreen wiper blade loaded to 20 g using a sand
(DEF-STAN-07-55C)-water mixture.
5 100 5
-~
~
4
~
-I \ 4
"-- I
~'" 90 \
~ 90~
3 "--
c:
'"c:
u
I
"= I I 0
:;::
.E I I
~
I
~ \I I /-
-
u
2 .21 "§
~
I
c: I
._g I I
80 \ , __ ,,. , - ,', II "''" ..=" 80 I
I
I
I
1
,...._/
.•••
I
I
.
'. _.( .....
, --- .... , ,_ /
I
I
7~---8~~9~~1~0~11~12~1~ 7 8 910111213
0
Wavelength l~m) Wavelength l~m)
Figure 9.5 Antireflection coating for Figure 9.6 High efficiency antireflec-
germanium component modular inter- tion coating for germanium components
face surfaces, rear surface of substrate inside surfaces; both sides of substrate
coated with high efficiency coating; coated with high efficiency coating;
transmittance (full curve), reflectance transmittance (full curve); reflectance
(broken curve' (broken curve).
8-12 JLm high efficiency antireflection coating for germanium (IS type)
(figure 9. 6)
Spectral characteristics: average transmittance 7.5-11.5 JLm greater than
98% both sides of substrate coated. Conforms to humidity test of MIL-
C-675A and humidity, adhesion and hardness test of MIL-M-13508B.
Commercial coatings-standards and specifications 231
];==::J3 3.5
Wavelength {!lm)
4
7 8 9 10 12
Wavelength (!lm l
Figure 9.7 Antireflection coating for multispectral quality zinc
sulphide, both sides of substrate coated.
100r-~----.---.---.-·--~---,
·~·
, 80 / ~
E
1
.= 601-
8 10 12
wavelength !1-1ml Wo.velengl'h (11m]
Figure 9.8 Antireflection coating for Figure 9.9 High efficiency antireflec-
8-12 ~tm quality zinc sulphide, both tion coating for zinc selenide, both sides
sides of substrate coated. of substrate coated; transmittance (full
curve), reflectance (broken curve).
232 Optical coatings
References
1 Introduction
Adams I AuCain R T and Wolff G A 1962 J. Electro. Chern. Soc. 109 1050-54
Adams R V 1961 Phys. Chern. Glasses 2 39-49
Adams R V and Douglas R W 1959 J. Soc. Glass Techno!. 43 147 T
Anderson R, Koepke B and Bernal G E 1976 NBS Spec. Pub/. 462 87-94
Anderson R, Skogman R, Ready J and Bennett J 1978 NBS Spec. Pub/. 541 70-7
Anthony F M and Hopkins A K 1981 Emerging Optical Materials SPIE291 196-203
Becher P F 1977 Ceram. Bull. 56 1015 and 1017
Bendow B and Drexhage M G 1982 Opt. Eng. 21 118-21
Billard P and Cornillault J 1962 Acta Electron. 6 Spec. Infrared Cahier No 3
Billig E 1956 Proc. R. Soc. 235 37-55
Blair G E, Greco E J, DeJager D and Wylot J M 1981 Emerging Optical Materials
SPIE 291 70-9
Blau H H 1955 USP 2,701,208
Bradt R C, Dulberg J L and Tressler R E 1976 Acta Metal/. 24 529-34
Braunstein R, Kim R K and Braunstein M 1980 Laser induced damage in optical
materials NBS Spec. Pub/. 620 29-43
Bridgman P W 1925 Proc. Am. Acad. Arts Sci. 60 305
Buckner D A, Hafner H C and Kreidl N J 1962 JAm. Ceram. Soc. 45 435-8
Campbell I E and Sherwood E M (ed.) 1967 High Temperature Materials and
Technology (New York: Wiley) p 142
Carnal! E Jr 1967 Mater. Res. Bull. 2 1075-86
Carnal! E Jr and Hatch S E 1965 Eastman Kodak Co BP 1011 826
References 235
5 Bulk Multispectral Materials for the Visible, Near, Mid and Far Infrared
and Materials for Use Beyond 12 I'm
Saul R S and Williams T L 1978 Refractive index data for optical materials Sira
Institute Ltd
Shrader E F and Bastien G 1979 Los Alvmos Conf on Optics '79 SPIE 190
Smakula A 1962 Opt. Acta 9 205-22
Smith S D, Hardy J R and Mitchell E W J 1962 Proc. Conf Physics Semi-
conductors (Exeter) pp 529-34
Sprackling M T 1976 The Plastic Deformation of Simple Ionic Crystals (New York:
Academic)
Stober F 1925 Z. Krist. 61 299
Stokes R J and Li C H 1963 Materials Science Research vol. 1 ed. H H Stadelmaier
and W W Austin (New York: Plenum) pp 133-57
Stokes R J 1966 Proc. Brit. Ceram. Soc. 6 189-207
Straughan V E 1979 Los Alamos Conf. on Optics '79 SPIE 190
Straughan V E and Krus D J 1981 High Power Lasers and Applications SPIE 270
11-8
Strong J D 1972 USP 3,674,330
- - 1974 USP 3,794,704
Tilton L W, Plyler E K and Stephens R E 1950 J. Opt. Soc. Am. 40 540-3
Willingham C Band Pappis J 1982 UK Patent Application GB 2090237 A Raytheon
Co USA
I,
,,
i
248 References
References 249
Sakuragi S 1982 Advances in Infrared Fibers SPIE 320
Savage J A 1982 First Int. Symp. on Halide and Other Non-Oxide Glasses,
Cambridge, England (Society of Glass Technology.)
Schultz P C 1973 Am. Ceram. Soc. Bull. 52 383
- - 1974 USP 3,826,560
Scott B and Rawson H 1973 Glass Techno!. 14 115
ShareS and Wasilik J 1979 IEEE Trans. Nuc!. Sci. NS-26 4802-7
Shibata S, Kanamori T, Mitachi Sand ManabeT 1980a Mater. Res. Bull. 15 129-37
Shibata S, Terunuma Y and Manabe T 1980b Japan J. App/. Phys. 19 L603-5
Shultz I 1957 Naturwiss. 44 536
·Sigel G H Jr 1983 Lasers Appl. May 49-50
Sigel G H Jr, Friebele E J, Gingerich ME and Hayden L M 1979 IEEE Trans. Nuc/.
Sci. NS-26
Snitzer E 1959 J. Opt. Soc. Am. 49 1128
Snyder A W 1969 IEEE Trans. Microwave Theory Tech. MTT-17 1130-8
Sparks M G and DeShazer L G 1981 Infrared Fibers SPIE 266 3-9
Sudo S, Kawachi M, Edahiro T, Izawa T, Shiada T and Gotoh H 1978 Electron.
Lett. 14 534
Sun K H 1946 Glass Ind. 27 552, 580
- - 1949 USP 2,466,507 USP 2,466,508 and USP 2,466,509
- - 1979 Glass Techno/. 20 36
Sun K H and Huggins M L 1950 USP 2,511,224
Takahashi H, Sugimato I and Sato T 1982 Electron. Lett. 18 398-9
Takahashi K, Yoshida Nand Yokota M 1983b. 4th Int. Conf. on Integrated Optics
and Optical Fibre Communication, Tokyo
Takahashi S, Kanamori T, Terunuma Y and Miyashita T 1983a 4th Int. Conf. on
Integrated Optics and Optical Fibre Communication, Tokyo
Takahashi S and Kawashima T 1977 Int. Conf. on Integrated Optics and Optical
Fibre Communications, Tokyo p 621
Taylor R L 1983 Laser Focus May 60
Tebo A R 1983 Electro-Optics June 41-6
Tran DC, Burk M J and Sigel G H Jr 1983 Second Int. Symp. on Halide Glasses
Rensselaer, New York
Tran D C, Fisher C F and Sigel G H Jr 1982 Electron,. Lett. 18 657-8
Turk R R 1982 Advances in Infrared Fibers, SPIE 320
Uchida T, Furukawa M, Kitano I, Koizumi K and Matsumura H 1970 IEEE J.
Quantum Electron. QE-6 606
Van Uitert L G, Pinnow D A, Williams J C, Rich T C, Jaeger R E and Gradkiewicz
W H 1973 Mater. Res. Bull. 8 469-76
Van Uitert L G and Wemple S H 1978 Appl. Phys. Lett. 33 57-9
Weber M J, Layne C B, Saroyan R A and Milan D 1976 Opt. Commun. 18
171
Weinberg M C, Neilson G F and Smith G L 1983 J. Non-Cryst. Solids 56 45-50
West R H and Lenham A P 1982 Electron. Lett. 18 483-4
Winter A 1957 J. Am. Ceram. Soc. 40 54· ':I
Winterburn J A 1967 BP 1,061,042
Yamazaki T and Yashiyagawa M 1977 Int. Con/. on Integrated Optics and Optical
Fibre Communication,, Tokyo p 617
250 References
Zief M and Speights R 1972 Ultrapurity- Methods and Techniques (New York:
Marcel Dekker)
Ziegler D C and Angell C A 1982 Appl. Opt. 21 2096-8
9 Optical Coatings
References 253
..
Index
I;