Principles and Applications of Emittance-
Independent Infrared Nondestructive Testing
Donald R. Green
Three emittance-independent ir nondestructive testing techniques are described. Two of these use the
dual scan ratio principle, and the other uses ir energy reflected from the test specimen surface to compen.
sate for emittance differences. Complete details on the theory behind the three techniques are given-
Instrumentation used to carry out tests and the results of two practical applications are described.
1. Introduction
Infrared nondestructive tests can be carried out by
injecting heat into the surface of a test specimen and
observing changes in the temperature with an ir radi-
ometer. The emittance* of the specimen determines the
rate at which it will emit ir energy at a given tempera-
ture, and, hence, it can influence the outcome of an ir
test. This is particularly true for tests on specimens
with a metal surface, since the emittance of bare metal is
low and is often higher in some localized areas than
in others, owing to oxidation, surface roughness, con-
tamination, nonuniform distribution of alloying ele-
ments, etc. Coatings can be used to provide the speci-
men surface with a high, uniform emittance. How-
ever, differences in coating thickness or bonding can
cause erroneous results In addition, application and
removal of the coating adds continuously to the ex-
pense of making tests.
Three emittance-independent ir nondestructive test-
ing techniques are described in the present article, with
emphasis on the supporting theory. These techniques
make use of approximations which, in some cases, re-
quire restriction of the range of specimen surface tem-
peratures generated and the interval of ir wavelengths
allowed to reach the detector. This is not too difficult
in practice, and several examples of practical applica-
tions, as well as the method of determining the required
test conditions are given in the sections below. Section
II mentions some general considerations on the art of
heat injection as applied to emittance-independent ir
The author is with Battelle Memorial Institute, Pacific North-
west Laboratory, Richland, Washington.
Received 1 April 1968.
* The term emittance is often used to describe the ratio of the
radiant emittance of a practical specimen to that of a blackbody.
In this case, the term ernissivity applies only to a specular surface
and is a special case of emittance.
testing. In Sec. III, the theory, general techniques,
and two methods of applying the dual scan ratio
method are described. Section IV describes the theory
and techniques of an ir test using reflective emittance
compensation. While not as generally applicable as the
dual scan technique, the reflective technique is simpler
to implement and is adequate in many nondestructive
testing applications.
II. Some General Considerations on
Heat Injection
Heating methods used in emittance-independent ir
tests should inject heat into the specimen at a rate that
is independent of the surface condition of the specimen
being tested. This ideal can be approximated in many
cases by hot gas or electrical heating methods. Elec-
trical current can be applied to conductors either by
direct contact or by induction heating. Dielectric
heating by electric displacement current can be used in
nonconductors by placing the test specimen in an in-
tense alternating electric field. However. variations in
the electrical properties of either dielectrics or conduc-
tors can cause differences in both heating rate and
emittance. Auxiliary electronic methods for checking
the uniformity of electrical properties should be used
with electrical heating to ensure that the heating will be
uniform. Induction heating is strongly dependent on
the spacing between the specimen surface and the in-
duction coil, and this is undesirable when testing speci-
mens having bumpy surfaces. However, induction
heating works very well for tests on smooth, electrically
homogeneous specimens that have a simple geometry,
such as planes or cylinders. '
Hot gas heating was used with the emittance-inde-
pendent techniques described in the present article.
This heating method was found to give uniform heat
input, that is, for all practical purposes, independent
of specimen emissivity. The rate of heat input with
this method is not critically dependent on the spacing
between the heat source and sample. Hence, the
September 1968 / Vol. 7, No. 9 / APPLIED OPTICS 1779
 1. 0 cm
1. 3 cm
6 KW WT -' , J, I_111
. --'
11=11.41'11'i]"11_7.
Wi
4Y
12 KW
33 KW
Fig. . Ieat transfer quality maps of an aluminum clad uranium
fuel element having 0.6-cm., 1.0-cm and 1.3-cm. diam core-to-
cladding bond defects. A plasma arc jet heat source was applied
at the powers indicated.
method is useful with samples that have somewhat
bumpy or irregular surfaces. One method of rapidly
heating gas is the so-called plasma arc jet, which uses
an electrical arc inside a cavity to produce a high tem-
perature plasma. The plasma is blown out through an
opening in the one end of the cavity by a stream of gas
that flows through the cavity. When inert gas is used
in the cavity, the plasma is chemically and electrically
neutral (in a macroscopic sense).
Another method of producing a hot gas stream is to
pass it through a length of heated tubing. Whereas
this method does not have the intense heating capacity
of the plasma jet, it is inexpensive and produces in-
tensities high enough for many purposes.
Location of the heated region on the surface of the
specimen is no problem when a plasma jet is used, since
the gas is highly luminous. With the heated tubing
method, however, placement is more difficult since the
gas is not visible. However, a conductive block coated
with thermographic phosphor (manufactured by the
U. S. Radium Company) can be placed in the
gas stream to make the heated region visible. The
phosphor fluoresces under uv light. When it is heated,
the fluorescence is extinguished so that a dark spot ap-
pears in the heated region.
The quantity of heat applied during a test is also
important. Too much heat will cause overheating of
the test specimen and can cause more severe effects of
emittance on the test results. For example, Fig. 1
shows the output of an ir radiometer during tests using
llasma arc heating on ui aluminum-clad uranium
nuclear fuel element having 1.3-cm, 1.0-cm, and 0.6-cm
flaws in the core-to-cladding bond. The radiometer
output wVasa 2-lfHz carrier that was amplitude modu-
Iated by the specimen surface temperature and emit-
1780 APPLIED OPTICS / Vol. 7, No. 9 / September 1968
0. 6 cm
tance. The sawtooth variations in amplitude over the
entire length of the recordings are due to emittance
differences, whereas the sharp pips in three locations
were due to the bond defects as indicated in the figure.
Note that the ratio of defect signal amplitude to emit-
tance signal amplitude is better for a 12-kW power
input than for either a 6-kW or 33-kW power input to
the plasma jet. Whenever possible, empirical selection
of the heat injection intensity is desirable, since theo-
retical determination is extremely difficult for all but the
simplest defect and sample geometries. Ordinarily,
natural defects will have complicated shapes, and dif-
ferent heat conductances will exist at different locations
within the defects.
The time between application of the heat and de-
tection of the surface temperature is another important
factor. Obviously, sufficient time must be allowed for
an appreciable temperature change to occur at the
flaws within a test specimen or they cannot possibly
have an effect on the surface temperature. On the
other hand, if only a thin coating covers the flaws, the
surface temperatures will rise rapidly and destroy the
coating if too much heat is applied. After the heat is
removed, the thin coating will cool rapidly over the
flaws, since its heat capacitance is low compared with
the rate of heat loss to the surroundings. It might
appear that heat could be applied to the entire surface
of the sample to solve the dilemma of too short vs too
long a heating time. However, the intense heat flux
required to produce temperature differences on the sur-
face over low resistance flaws would then cause over-
heating of the entire sample. The obvious solution is to
apply the heat in a localized region. The radiometer
must then be located so it views the specimen surface
during heating if flaws are expected near the surface,
and at some time after heat application if flaws are ex-
pected deeper under the surface. If the general region
in which flaws would occur is unknown, several tests
may be required to provide optimum conditions for
detecting several different possible types of flaws.
111. Dual-Scan Ratio Technique
Sections III.A-C describe the theory and two practi-
cal approaches used in the dual scan ratio technique.
Basic theory for both of these approaches is the same,
and it is described i Sec. III.A. Section III.B de-
scribes an analog ratio approach that was applied to
nuclear fuel elements. In Sec. III.C, a digital ap-
proach is described that was applied to testing the bond
between a thin niobium carbide protective coating and
graphite.
A. Theory and General Method of the
Dual Scan Ratio Technique
Several assumptions and approximations arc neces-
sary to reduce the theory of the dual scan ratio tech-
nique from the rigid idealism of a mathematical model
to a useful technique that can be applied to test practi-
cal samples. These are mentioned at the appropriate
points in the text and are summarized near the end of a theoretical derivation of equations needed in the dual
this section. scan ratio technique. The definition of D* (T,fAf)
Derivation of the theory was carried out for a de- is given as 3
tector of incident ir power, since the figures of merit for
D*(T,f,Af) = (Af V8 /H- V0 ), (3)
detectors are ordinarily expressed in terms of power
rather than photon flux. However, it is shown at the where H is the rms value of the sinusoidally chopped
end of the derivation that the method is emittance- irradiance at the detector, A is the detector area, V7 is
independent, regardless of whether a power detector or a the rms noise voltage, V, is the rms signal voltage, and
photon detector is assumed. Af is the electronic bandwidth of the circuits following
Let us assume the detector to be a semiconducting the detector. The quantities T, f, and Af in the paren-
bolometer, comprising a very thin layer (10 u) of semi- theses following D* mean that a given value of de-
conducting material on the end of a slab of sapphire. tectivity is valid only for a given target temperature,
In addition, assume that the opposite end of the sap- chopping frequency, and electrical bandwidth. Solving
phire slab is in perfect contact with a slab of some Eq. (3) for Vs we obtain
other material and that the incoming ir is sinusoidally
chopped at a high enough frequency that the tempera- = D*(T,f,2,Af)Hl '2Va
(4)
ture at the far end of the entire assembly does not
change appreciably. Then the sinusoidal component Hence, for a given detector in a giver radiometer,
of the semiconducting layer temperature, assuming that
variations in radiant losses due to this component are V. = D*('T1,f,,Af)HG, (5)
negligible relative to conduction of heat into the sap- where G is a constant. The value of D* (T,.fAf) is
phire, is given by2 given by3
T = HZ, (1)
fHx(TI)D*, (fXAf)dX
where H is the total ir power absorbed at the near end
and Z is the thermal surface impedance given by3 D*(T1,f1,Af)=- ()- (6)

1 (coshdk + sinhdk"
f H('1 )dX
(2)
Ki\sinhd4 + 0 coshd()' where Hx(T 1 ) is the rms value of the sinusoidally
where t = Ki'(7rfASp) '(1+j); 0 = (K2S 2p2/ chopped irradiance at the detector per unit wavelength,
KiSipj.) ; d is the thickness of the sapphire slab; centered at wavelength X,and D*x (fAf) is the spectral
K,, S, and p are the thermal conductivity, specific detectivity. Note that the denominator of Eq. (6)
heat, and density, respectively, of the sapphire; K 2 , is actually just the rms value of the total irradiance H
S2, and 2 are the thermal conductivity, specific heat, at the detector. Thus, combining Eqs. (5) and (6)
and density, respectively, of the second slab; and f we obtain
is the chopping frequency. Examining Eqs. (1) and
(2) we see that Z is fixed for a given bolometer and that V = GE Hx()D*(f,,Af)dX (7)
the amplitude of the sinusoidal temperature component
of the semiconducting layer is directly proportional to Assuming that all components of the radiometer are at
the sinusoidally chopped ir power absorbed by it. For the same temperature, and that the sample is wvarmerthan
small temperature changes, the output voltage of a the radiometer, the rms spectral irradiance value at the
voltage divider, using a semiconducting bolometer as letector is given by
the variable element, is very nearly linear. Assuming Hx(T) = I,(T,) - H(YC), (8)
an exponential temperature dependence of the bolom-
eter resistance, it can be shown that using a series re- where JJ8x(T,) is the irradiance component originating
sistance equal to the bolometer resistance gives a volt- from the sample surface at temperature T, and passing
age output linear to within less than 1% for temperature through the chopper, and IICA(T') is the irradiance
changes up to several degrees centigrade. Changes in component originating from the chopper blade itself
bolometer temperature due to the chopped ir will be in at temperature T. (These quantities are actually
terms of microdegrees. Hence, the voltage out of a phasors but are 1800 out of phase under the conditions
bolometer (of a type that is commercially available) is specified and thus only the rms values of their real com-
directly proportional to the sinusoidal ir radiation ponients need be considered.) Combining Eqs. (7)
power per unit area falling on it; the bolometer is thus and (), we obtain
an ideal detector of irradiance.
In practical detectors, we must take into account the V. = G (T,) - H,.x(T,)ID*x(f,,Af)dX
variations in detection efficiency with wavelength.
The detectivity D* (T,J,Af) is often given by manu-
facturers as a figure of merit for their detectors. 3 Since - G f lH8x(T.)D*(f,Af)dX
we have shown that commercially available bolometers
give a signal output that is directly proportional to the fo,~~~~
-J H(lTc)D*X(fiLf)dXI. (9))
irradiance at the detector, we can use the detectivity in

September 1968/ Vol. 7, No. 9 / APPLIED OPTICS 1781

 5
-I
-2
.3
-4
-5
26 30 34 38 42 46
Temperature,'c
Fig. 2. Radiometer output E vs temperature
having three different emittances.
But the value of H(T 8) is related to the spectral
radiant emittance of the sample by a constant that
depends upon the light gathering power and efficiency
of the radiometer optics and the position of the radi-
ometer relative to the sample surface. The Planck
radiation law states that the total spectral radiant emit-
tance of a blackbody sample at temperature T is
Wx = C1X 5[exp(C 2 /XT,) -
where C and C2 are constants. For simplicity, we
first derive the radiometer output voltages for the case
in which the sample is hot enough that reflections of
external sources are negligible. This result is then ex-
tended to the more general case in which reflections are
important. Using Eq. (10),
H8 x(T.) = Mexfx.,
where M'Jis a constant and e,,is the spectral goniometric
sample surface emittance. The angle of inclination of
a particular sample surface element being viewed by
the radiometer influences the value of H 8x(T8). How-
ever, this angle has a fixed value for any element of the
sample surface, just as the emittance for that area can
be considered fixed. For this reason, both the angular
and emittance dependence of H 8x(T,) are contained in
the quantity ex.
The right-hand term of Eq. (9) is a constant for a
given radiometer at a constant temperature, since both
Hx(T,) and D*x(frAf) are then constants. Using this
information and combining Eqs. (9) and (11), we obtain
Vd = G[WJ' exD*x(Jf,,Af)WsxdX
- N
where N, is a constant for a given radiometer at a con-
stant temperature. Assuming that the specimen tem-
l)eratures and interval of ir wavelengths are restricted to
a small enough region, ex is essentially constant and we
may write
V = C3ef D*x(fhAf)VfxdX - C = C3 ef(T) - C,
1782 APPLIED OPTICS / Vol. 7, No. 9 / September 1968
where is the goniometric emittance, C and C3 are con-
stants, and the wavelength interval is taken care of by
For the more general case, where the sample surface
reflects an appreciable amount of radiation from the
environment, Eq. (11) becomes
led
m //d
Emittance Hx(TsTB) = M[eXWV. + (1 - e)C4WXB]
= I[eX(W,, - C4WXB) + C4WXBI, (14)
where C4 is a constant and WXB is the radiant emittance
Low Emittance
_ of the environment around the sample. Here it has
I I 58 been assumed that the absorptance is equal to the
50 54 58 emittance, that the environment radiates as a black-
body, and that the sample temperature change during
testina is small. Since W- is a constant. E. (12)
of samples becoms__
becomes _ ----- - -- - I-
V = G{iMIf exD*x(fiAf)[Wxs - C 4WXB]dX
+ fl MD*XC4WXVdX - Nc} = GMIeF(Ts) - C. (15)
This equation is of the same form as Eq. (13) and the
assumptions regarding the wavelength dependence of
the emittance are the same.
To carry out an emittance-independent dual scan
ratio test, heat is applied to the test specimen surface.
Each increment of area on the surface is viewed by a
1-', (10)
radiometer either during heating or shortly after re-
moval of the heat. The temperature over a flaw in the
specimen is higher at this time than it would be over a
sound region since a flaw impedes the inward flow of
heat from the surface of the specimen. After removal
of the heat, the difference between surface tempera-
tures over the flaw and over sound regions of the speci-
(11)
men decreases until the surface temperatures are nearly
the same everywhere. Using either Eq. (13) or (14)
and forming a ratio of the voltage outputs of the radi-
ometer when it is looking at the surface temperatures
early and late in the temperature transient of a particu-
lar increment of the surface area, the result is
(E 1 + C)/(ED + C) = f(TA)/f(TD) = E. (16)
Here, EA and ED are the values of V corresponding to
sample temperatures TA and TD, and E0 is a value which
is independent of emittance and gives an index of the
thermal quality of the sample. Obviously, the emit-
tance must be identical in the numerator and denomina-
tor so that it will cancel. Hence, EA and ED must be
the output of the radiometer when it is viewing the
same area on the sample surface.
(12) Because of their importance, it is worthwhile to
emphasize the following assumptions used in the der-
ivations: (1) the change in sample temperature is re-
stricted to a small enough value that the emittance can be
considered constant; (2) the wavelength interval of the
infrared irradiance at the detector is restricted to a small
enough value that the sample emittance can be considered
nearly constant over the entire interval; (3) the radiometer
(13) temperature is constant and cooler than the sample te!n-
perature at any time during a test; and (4)the same area of
 4.0 as is expected for the samples to be tested. Figure 2
o Low Emiltance
shows experimental values taken with a Barnes model
A MediumEmittance RST-1 radiometer on samples having emissivities be-
3. 5
High Emittance tween about 0.06 and 0.95. (The output voltage goes
negative for some sample temperatures because of an
offset voltage added to the signal in the preamplifier of
3.0
the radiometer.)
The values of Eo, computed from Eq. (16) using the
o 2.5 0
data in Fig. 2, are shown in Fig. 3 as a function of TD for
0 0
0 TA 54C TA values of 54 C and 44 C. If there were no errors,
the values of Eo for all emittance values would be the
2.0 same at any given TA and TD. Note, however, that the
TA 44 C
curves separate for different emissivities at a TD value of
approximately 390C; this is the minimum acceptable
1.5 value for TD during a test. The reason the curves sepa-
1.0~~~~~
rate is that a low value of TD for a low emissivity sample
gives a low E, and this increases the effect of noise on the
output Eo. In this particular example, the value of C
35 40 45 50
was about 4.7. As the values of ED approach negative
TE. °C
4.7, a small error in the denominator causes increas-
Fig. 3. Eo vs TD for samples at TA = 54 0 C and 44 C.
0 ingly larger errors in Eo; extreme accuracy in measure-
ment of ED is required as TD approaches the radiometer
reference temperature.
The value of TA is important, since it is influenced by
the sample is viewed at different times to provide values of the condition of the sample. (Recall that TA is de-
EA and ED used in Eq. (16). tected at a time when temperatures over flaws are
The above derivation could have just as easily been higher than over sound regions, whereas TD is detected
carried out for a detector that gives an output voltage after the temperature differenceshave leveled out. TD
directly proportional to the number of photons per unit will, therefore, be about the same in a bad sample as in a
time falling on the detector. In this case, however, perfect one.) The dependence of Eo on TA, for the
D* should more properly be defined in terms of photon worst case TD value of 390 C, is shown in Fig. 4. Note
flux rather than power. In addition, the radiant emit- that there is no critical limit on the range of TA.
tance given by Eq. (15) would be replaced by the num- The value of C can be determined empirically by
ber of photons radiated per unit time per unit source making scans at two temperatures on a standard having
area per unit spectral interval. (The expression for two or more areas of different emissivities. These scans
this quantity can easily be obtained by dividing Eq. give output voltages ElA, E2A, ElD, and E2D, where the
(15) by h/X, where h is Planck's constant.) The result subscripts A and D refer to high and low temperatures,
is that we obtain an emittance-independent ratio of a and the subscripts 1 and 2 refer to high and low emit-
different temperature function, viz., tances, respectively. The entire standard temperature
(EA + C)/(ED + C) = g(TA)/g(TD) = EO, (17) is uniform at either of the two temperatures selected and
thus T1A = T2A and T1D = T 2D. Using Eq. (16),
where the quantities have meanings that correspond to (18)
(EIA + C)/(EID + C) = (E2A + C)/(E 2 D + C),
their counterparts in Eq. (16). Thus we have shown
that a photon detector could be used as easily as a power
detector to obtain the emittance-independent output.
Practical methods for obtaining the emittance in- 3.0
dependent test output, either continuously or on a dis-
crete point by point basis, are described in the two
sections below. 2. 5

Method of Determining the Proper Temperature Ranges

and the Value of C Eo 2.0
Selection of the proper range of sample temperatures,
particularly for TD, is important to ensure valid test
results. The temperature range can easily be selected 1.5

by choosing the proper initial temperature and heating

rate for the sample. The first step in selecting the
1.0
temperature range is to calibrate the radiometer output 40 45 50 55
as a function of sample temperature for several different TA
0
C
emissivities. Actual values of the sample emissivities
are not important, but they should cover as wide a range Fig. 4. E vs TA for samples at TD = 39°C.

September 1968 / Vol. 7, No. 9 / APPLIED OPTICS 1783

 through a bond to a 0.75 -mm thick cladding of XS001
aluminum alloy. A general description of the method
and results is given below. Further details of the
method can be found elsewhere.4
Heat was injected during the tests with a 1-cm diam
plasma jet operated at a power of 15 kW. This heat
source was intense enough to allow a scanning speed of
1.2 m/sec along an overlapping helical path around the
cylindrical fuel elements. The time required to test a
single 4-cm diam, 20-cm long fuel element was approxi-
mately 15 sec. Metallographic studies on the surfaces
of the fuel elements showed that no changes in structure
or in the distribution of alloying agents occurred as a
result of the intense heat injection. The plasma jet
heat injection method has the advantage that small
changes in spacing between the heat source and surface
of the test object have little influenceon the rate of heat
injection, provided proper spacing between the plasma
head and test object, is used.' One disadvantage of the
method, however, is that several seconds must be al-
lowed for the jet to come to equilibrium before each test
Fig. . Infrared heat transfer testing of an aluminum-clad run. (Turning off the jet between runs is desirable to
uranium fuel element. prolong the life of the plasma forming electrodes.)

EIA
E2D
EID
2A
Scanning of each fuel element was accomplished by
simultaneously rotating it ad moving the heat injec-
where an E value with a given subscript corresponds to a tion head and two identical radiometers along its length
sample temperature with the same subscript. Solving so that the entire surface of the fuel element was scanned
Eq. (18) for C, along a spiral path. Infrared entering each radiometer
was chopped at 2 kHz ad focused on an indium anti-
C= B1 )E2 , _-EIA2D-. (19) monide photoelectromagnetic detector by means of an
+ - - f/0.5 germanium lens. The scanning mechanism was a
The thermal conductivity of the standard must be high "9 in." screw cutting lathe with the heat injection head
enough that the emittarnce difference will not cause an
appreciable difference in temperature. When the emit-
tance of one of the areas is very low, it is difficult to ob-
tain an accurate value of C, and caution is necessary to
ensure accurate values of E, and ED. Drift in the out-
put of even a high quality research radiometer is great
enough to require constant checking and recalculation of
C if the internal temperature of the radiometer is riot
regulated. A standard having areas of high and low
emittance can be included with each batch of samples
tested to allow determination of C. The standard can
be heated to T.] and cooled to TD at the same time as the
samples. This method of continuously determining C
is particularly useful when the data are automatically
processed by high speed computer. For this reason, it
was used in the digital technique described below.
B. Analog Dual Scan Application
The dual scan ratio principles were applied to develop
ani analog technique for testing the uniformity of ther-
mal conductance in nuclear fuel elements. Results of
tests using this technique showed that there were dif-
ferences in the conductance of heat from the interior to
the outer surfaces of the fuel elements. Such differ-
ences, in high performance fuel elements operating at
the limit of their designed power production, could cause EA ED E,
overheating in localized regions of low heat conductance. Fig. 6. Infrared heat transfer map of a fuel element with
In the type of fuel element used in the studies described tempera painted spots on surface. Eo is the emissivity-indepen-
in this section, heat is conducted from the uranium core dent instrumeni output.

1784 APPILIED OPTICS / Vol. 7, No. 9 / September 1968

 tape delay and second radiometer, and the resulting
signals were fed into a high speed analog ratio circuit to
give a continuous real-time solution of Eq. (17) for the
emittance-independent output. Selection of the value
of C simply required increasing the amount of constant
voltage added to the signal until the Eo output was con-
0.6 cm stant for scans over tempera painted spots.
Defect Figure 6 shows a time recording of the EA, ED, and Eo
outputs during a scan on a flawless fuel element having a
/Signal
large tempera painted spot on one side. A increase in
EA and ED deflects their traces to the left and an increase
in E0 would deflect its trace to the right. Both the EA
signal out of the first radiometer and the ED signal out of
the second radiometer showed large fluctuations as the
radiometers passed over the high emittance spot. The
emittance-independent output Eo, however, was nearly
constant.
Figure 7 shows the outputs during a test on a fuel ele-
ment having artificially produced 1.0-cm diam and 1.3-
cm diam defects in the core to cladding bond. One side
of the XS001 alloy cladding on this fuel element had
been highly polished, while the other side was left in the
natural oxidizedstate. Therefore, the emittance of this
EA ED E, EA ED E, fuel element was about 0.10 on one side, and about 0.4
on the other. The effect of this emittance difference
Fig. 7. Infrared heat transfer maps of 1.3-cm and 1.0-ihe effect can be seen in Fig. 7 as a regular variation in the EA and
mica produced defects in fuel elements QC-20 showing t
of the polished, low emissivity area. Eo is the emissiv it-inde- ED signal over the entire length of the recording.
pendent instrument output. Here, as in the previous case, the emittance-independent
output Eo is not appreciably influenced b y the emittance
and radiometers mounted on its tool carriage, a, shown differences. However, the Eo signal variations due to
in Fig. 5. To begin a test, the lathe rotati in aind the bond defect can be clearly seen.
plasma jet were started simultaneously and iere al-
lowed several seconds to reach equilibrium. . water

cooled copper shutter that covered the jet ou tlet re-

mained closed during this initial warmup peri()d. At
the end of the warmup period, the shutter was opened
and the lead screw of the lathe was engaged to c Lrry out
the scaln.
The two radiometers were located downstrea m from
the heat injection point so that the first one vieived the
fuel element surface about 25 msec after injectio n of the
heat, and the second one viewed the surface abo ut 1 see
later. The lens barrel of the radiometer neal rest the
heat injection head was water cooled to prever Lt over-
heating. As mentioned in Sec. III.A, computnation of
the emittance-independent output requires ti gnat the
ratio in Eqs. (16) and (17) be taken when the radiom-
eters are viewing the same increment of area. 'I'his was
accomplished in the analog technique by using an ad-
justable Ampex continuous loop time delay unit Ito delay
the output of the first radiometer until the secoiid radi-
ometer reached the appropriate location on the fuel ele-
ment surface. Adjustment of the time delay annd selec-
tion of the value of C required for the emittani 3e-inde-
pendent ratio was carried out by making test setans on a
uniform fuel element having several high emitta rice sur-
A B C
face regions produced with black tempera pain- t. The
time delay was adjusted until pips in the delayed .output Fig. 8. Fuel elements with surface ir emissivity variations:
occurred simultaneously with pips that occurre I in the (A) fuel element QC-20 with highly polished area on surface;
output of the second radiometer as the paint spoia passed (B) fuel element QC-17 with dented areas on surface; (C) fuel
under it. A constant C was added to the outpul -s of the element M-6 with black tempera painted area on surface.

September 1968 / Vol. 7, No. 9 / APPLIED OPTICS 1785


I
I
I
I II I
EA ED EA ED
A B
Fig. 9. Infrared heat transfer maps of two areas oni fuel elem ent
QC-22 showing (A) similar variation in uncompensated adisom-
eter outputt E, but (B) a difference in the behavior of emissiv ty-
independent output E.
Fuel elements containing artificially produced defe cts
and wide variations in surface condition are shown in
Fig. S. These fuel elements were used in extensive te sts
that experimentally confirmed the emittance in(le-
pendence of the analog dual scan ratio technique fo r a
wide range of emittance differences, including samp les
that had bumpy surfaces, such as the middle fuel le-
merit showni in Fig. S.4
Results of a test on an unmodified normal producti ,on
fuel element are shown in Fig. 9. The EA trace, whiich
is the output one would obtain with an ordinary sin le
radiometer testing method, shows a small variation in
both Figs. 9(A) and (B). The Eo trace in Fig. 9( A)
shows a small variation indicating an actual difference in
the heat conductance. However, the Eo trace in F ig.
9(B), which was for a different location in the same f iel
element, shows almost no variation. This means ti hat
the E, variation for part of the fuel element was alms)st
entirely due to natural emittance differences, whern,as
the variation for another part of the same fuel elemEent
was due to nonuniform conductance. Other examp les
similar to this one, as well as a few large natural flaxVs,
were found in tests on normal production fuel element tS.
C. Digital Dual Scan Application
Tests at discrete locations on bonds between graph ite
and a thin niobium carbide protective coating requirecIa
nondestructive testing method that did not require t he
use of couplants between the test specimen and the te:st-
inig device. An ir digital dual scan ratio method le-
veloped for this purpose was demonstrated to be stIc-
cessful in detecting bond defects and porosity appro: i-
mately 0.0002 cm thick. These results were confirm ed
by independent destructive analyses carried out at L1OS
Alamos Scientific aboratory. The main features a and
1786 APPLIED OPTICS / Vol. 7, No. 9 / September 1968
results of the digital dual scan method are described
below. Further details are given in a previous publica-
tion.5
A functional diagram of the equipment required to
carry out the digital dual scan tests is shown in Fig. 10.
The initial temperature of the test specimens was ad-
justed to a constant, uniform value prior to the tests by
placing the specimens close to a preheater. Heat injec-
tion was carried out during scanning by projecting hot
helium from a nozzle onto the surface of the test speci-
mens. Heating of the helium was accomplished by
passing it through a length of Inconel tubing that was
heated by passing a large electrical current through it.
A Barnes Engineering Company model RST1 ir radiom-
eter was positioned so that its field of view occupied a
0.3-cm diam spot on the surface of the test specimens in
the center of the heat injection region. This unit had a
noise equivalent temperature of about 0.05'C with a
bandwidth of 10 Hz. Positions of the preheater, heat
injection nozzle, and radiometer were fixed. Scans
were made bv running the test specimenspast the heat
source and radiometer field of view at a uniform rate of
about 2.5 cm/sec. In this way, temperatures along a
straight line on the surfaces of the test specimens during
heating were observed b the radiometer. After com-
pletion of the first scan, the test specimens were returned
to the preheater for a time sufficient to allow decay of
temperature differences produced during the scan. The
heating gas was then turned off and a second scan was
made. Values of EA and ED needed to compute the
emittance-independent ratio in Eq. (16) were simply the
outputs of the radiometer during the first and second
scans, respectively. A high and low emittance standard
was included, with each batch of samples scanned to
permit determination of C.
Output signals from the radiometer were fed into a
digital voltmeter and then into a digital printer capable
of printing five 6-digit readings per second. A wheel,
arranged as shown in Fig. 10, generated one printout
command pulse at the input of the digital voltmeter for
every 0.3-cm interval of scan length along the specimen.
Printout positions along the EA scan coincided with
those along the ED scan within 0.012 cm. Data from
the digital printer were punched into cards and fed into
a Univac 1107 computer, which automatically solved
60
602
Radiometer
Fig. 10. Emittance-independent ir bond test.
 11. A continuous strip chart record of the radiometer
MMUMMONNIMUMM output during a dual scan is shown in the middle of Fig.
11. The EA trace was generated in the first scan with
2 ~~~ffim1 the heating gas on, and the ED trace was generated i a
2 later scan with the heating gas off as described earlier.
0 Times along the abscissa are marked with the cor-
0 responding number of digital printouts accomplished
-2
up to each time. Computer solutions for the emittance-
2 independent ratio Eo for discrete points along the sam-
ple are plotted at the bottom of Fig. 11. Note that
-4
I1 5 10 15 20 although the defect signals i the ordinary radiometer
output EA are completely masked by the emittance
differences, they are clearly visible in the emittance-
Emittance- 0.6 cm Unbond independent output Eo. To give further confirmation
Independent of the method, the Krylon painted regions on the stan-
Plot 1 0.3 cm Unbo d
4 dard were removed and a single spot painted on one end.
1:;
0. 16 cm Pit The results of a subsequent test are shown in Fig. 12.
3 Note that while the EA and trace is much different in
this case than in the previous one, the emittance-inde-
2
I pendent output Eo is about the same.
I I Results of a dual scan ratio test on the standard and
0 5 10 15 20 two practical niobium carbide coated graphite test
Voltmeter Print Out Point specimens are shown in Fig. 13. The lowest Eo value
Fig. 11. Emittance-independent bond test of standard that for the standard was chosen as the threshold unbond
has three built-in defects and randomly spaced high emittance alarm level, since it represents the best bond conduct-
areas. Top: photograph of standard; middle: radiometer ance in the standard. Note that one of the practical
output; bottom: emittance-independenit plot. test specimens had Eo values greater than the threshold,
whereas the other had E0 values below the threshold.
Eq. (19) for the value of C, using data from the high and Metallographic results from the bonds in these test
low emittance standard, and then used the values of C, specimens are pictured in Fig. 14(A), which shows the
EA, and ED to solve Eq. (16) for the value of Eo for each bond in specimen 1, and Fig. 14(B), which shows all inn-
point along the scan on the test specimens. Outputs
from the test could have been recorded on punched tape
for direct submission to the computer. However, the will III -- rerr

data cards were more economical for the small number

of test specimens in the present study since a longer
computer setup time would have been required for a ;i J. Mt$
2
4 i
A1i.Ma~I' :
<v0
KI XX ., , :.. 7 [...va_ni2a
punched tape input.
The high-low emittance standard also contained some 2
X . :W.. ,. ^Wt0. .0u:.~!: ': + , e
7re{{
I ;v~ v
rVI-eHzv
artificially produced bond defects at known locations
that served as monitors of the test sensitivity during I
scans ol each batch of specimens. The standard was Z
t 0o EAScan
IE
constructed by bonding a 0.012-cm thick stainless steel
sheet to a massive piece of graphite with W. T. Bean
type RTC epoxy. Bond defects were produced by em- -4
bedding 0.3-cm and 0.6-cm square pieces of 0.002-cm 0 5 10 15 20

thick Teflon in the bond layer. A defect was also manu-

factured by placing a 0.16-cm pit approximately 0.08 cm Emitt an ce- 0.6 cm Unbond
deep in the graphite surface. This pit was filled with Independent
Plot l
epoxy to produce a low conductance region in the graph-
ite. Inadvertent bonding differences were also detected 0.3 cm Unbond|

in the standard during subsequent tests, and care was

required in selecting well-bonded regions to be used as
high and low emittance standards for obtaining C. The 2 I3 I
high emittance regions were produced by spraying flat
0 5 10 15 20
black Krylon paint on small areas, and the low emit-
tance regions were produced by leaving the stainless Voltmeter Print Out Point

steel coating in its natural shiny state. Fig. 12. Emittance-independent bond test of standard that has
To demonstrate the emittance independence of the three defects and a high emittance area. Top: photograph of
digital technique, six high emittance regions were pro- standard; middle: radiometer output; bottom; emittance-
duced on the surface of the standard, as shown in Fig. independent plot.

September 1968 / Vol. 7, No. 9 / APPLIED OPTICS 1787

 5
side of the fuel element. As the fuel element rotated
0.6 cm Unbond
during scanning, the high and low emissivity sides were
o.3- U n bond d
4 alternately presented to the radiometer.

in~~~
o. 16c Pit The lower trace in Fig. 15 shows the radiometer out-
Unbond
Threshold Level put with reflective emissivity compensation during an
identical scan on the same fuel element. Note that all
defect signals are now visible, including those for the
3
0.6-cm diam defect, and that almost io signal variations
occur due to emissivity differences.
Specimen 1 Specimen 2
|. - Standard 1. .1. Reflective compensation for emissivity is based upon
the fact that the reflectivity of a surface is related to its
0 5 10 15 20 25 30 3.! emissivity by7
Voltmeter Print Out Point
YX = (1 - EX), (20)
Fig. 1:3. Emittance-independent plot of data resulting from
bond tests on niobium carbide-coated graphite.
where -yl,is the spectral reflectivity and i,is the spectral
emissivity. Consider ann ir radiometer as being ar-
ranged so that it is receiving both emitted and reflected
ir energy from the surface of a test specimen. Also, let
us assume that the reflected energy originates from an
0,..00 mm ,g
auxiliary source intense enough that reflections of en-
ergy from other sources (at room temperature) are
ii~M
negligible. If the ir is restricted to a sufficiently narrow
wavelength interval, or if the emissivity of the test
specimen is not too wavelength dependent, the irradi-
A B
ance at the detector is approximately
Fig. 14. Niobium carbide-coated graphite (X7-50): (A) well R = Keyf(Ti) + K2ef(T,), (21)
bonded; (B) partially bonded.
where Kn and K2 are constants, T, is the temperature of
the auxiliary source, T, is the surface temperature of the
bond in specimen 2. Tests and destructive examina- test specimen, and e is the emissivity of the test speci-
tions o other practical specimens indicated that un- men. Under the test conditions stated, we can use Eqs.
bonds and bond porosity could be detected in most cases (20) and (21) to obtain
with the digital dual scan ir technique.
R (1 - e) [Kif(Ti)] + eKf2(Ts). (22)
IV. Reflective Compensation for Emittance
It is possible to adjust the auxiliary source so that
Differences
Kif(TI) = K 2f(T 0) for any given sample. Then,
Reflection of ir emitted by an auxilliary source has
been used to minimize the effects that emittance differ- I
R K 2f(7'). (23)
ences have on ir tests.' This method, while not as gen- This expression indicates that rioappreciable change in
erally applicable as the ratio technique described in the the irradiance at the detector will occur due to emis-
previous sections, is much simpler to implement. The sivity changes alone. However, an emissivity-depeii-
method must be used on samples for which the gonio- dent change will occur when the sample temperature
metric distribution of the radiance is constant. The changes.
terms emittance and emissivity are used interchange-
ably in the discussion that follows. L.O -
0.6 - .3 .

The scanning method, the single ir radiometer re-

quired, and the heat source for experiments ol the reflec-
tive method were identical to those used i the analog
dual scan technique described earlier. However, the
output of the radiometer was in the form of a 2-kHz car- Wfthout
Compenatio
--- aviy

rier that was amplitude modulated by the ir irradiance

at the detector. Figure 15 (top) shows a portion of a
strip chart recording of the radiometer output during a
test o a 3.7-cm diam aluminum clad uranium fuel ele-
ment without emissivity compensation. Amplitude
variations due to 1.3-cm, 1.0-cm, and 0.6-cm diam Wach R..n..-
E-Cv-iiy
artificially produced bond defects can be seen. Note Compenstion

that the 0.6-cm diam defect would not have been de- Fig. 1. Heat transfer quality maps of altnminum-clad uranium
tectable above the sawtoothlike signal variations that fuel element having 0.6-cm, 1.0-cm, and 1.3-cm diam core-to-clad-
can be seen over the entire length of the trace. These ding bond defects. Plasma arc jet heat source was applied at 12-
sawtooth variations were due to a high emissivity oi one kW power input.

1788 APPLIED OPTICS / Vol. 7, No. 9 / September 1968