Temperature Coefficients of Resistance of Metallic Films in the Temperature

Range 25° to 600°C

Richard B. Belser and Walter H. Hicklin

Citation: Journal of Applied Physics 30, 313 (1959); doi: 10.1063/1.1735158

View online: http://dx.doi.org/10.1063/1.1735158
View Table of Contents: http://scitation.aip.org/content/aip/journal/jap/30/3?ver=pdfcov
Published by the AIP Publishing

Articles you may be interested in

Temperature coefficient of resistance in ultrathin metal films
J. Appl. Phys. 46, 2658 (1975); 10.1063/1.321899

Temperature Coefficient of Resistance in Thin Metal Films

J. Appl. Phys. 37, 3634 (1966); 10.1063/1.1708921

Mass Spectrometer Measurements of the Diffusion Coefficient of Hydrogen in Steel in the

Temperature Range of 25°–90°C
J. Appl. Phys. 29, 892 (1958); 10.1063/1.1723324

Compressibility of Gases at High Temperatures. III. The Second Virial Coefficient of Helium in the
Temperature Range 600°C to 1200°C
J. Chem. Phys. 18, 641 (1950); 10.1063/1.1747716

Compressibility of Gases at High Temperatures. II. The Second Virial Coefficient of Helium in the
Temperature Range 0°C to 600°C
J. Chem. Phys. 17, 751 (1949); 10.1063/1.1747394

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ]
IP: 129.49.170.188 On: Sat, 20 Dec 2014 14:10:04
 JOURNAL OF APPLIED PHYSICS VOLUME 30. NUMBER 3 MARCH. 1959

Temperature Coefficients of Resistance of Metallic Films* in the

Temperature Range 25° to 600°C
RICHARD B. BELSER AND WALTER H. HICKLIN
Engineering Experiment Station, Georgia Institute of Technology
(Received August 25, 1958)

The temperature coetlicients of resistance (TCR) of films of sputtered Au, Ir, Mo, Ni, Pd, Pt, Rh, Ta,
and W, of evaporated AI, Cr, Ti, and Zr, and of the alloy films Pt-Au, Pt-Ir, and Pt-Ni have been measured
in vacuo over the temperature range 25° to 600°C; the film thickness range was 75 to 2oooA. The TCR
values of films of the substantially pure metals were in the range i to i those of the respective bulk metals
whether deposited on glass, Vycor, or Stupalith substrates. The departure of the conditions of growth of
the film from those usual for the bulk metal, i.e., rapid cooling effects and impurities present, contributed
imperfections which reduced the TCR of the film. The TCR values of alloy films were low, 0.0004 per °C
for Au-Pt, and resistivities were 3 to 4 times that of either constituent. The TCR values of sputtered films
of Mo, W, and Ta and of evaporated Cr, Zr, and Ti were generally less than 0.0001 per °C. Electron diffrac-
tion examination of films of these latter metals revealed oxide inclusions in the films. The presence of the
oxide of the metal reduced the TCR and increased the R/sq of the film as compared to those of the pure
metal. Only films of gold, platinum, and iridium were corrosion resistant in air near 600°C. Overcoats of
evaporated SiO provided partial protection for the others. Powers up to 15 watts were dissipated by refrac-
tory metal films only i2 by ! in. and about 750 A thick. These high values contrasted with less than l w
for a gold film of similar dimensions.

INTRODUCTION In short, regardless of the method utilized, thin

A NUMBER of studies of the temperature coeffi-
cients of resistance of metal films have been
published by A. Fery,I.2 Van Itterbeek,a·4 Mostovetch,o·6
and others. Previous work has been summarized recently
by Blanc-Lapierre,7 Holland,s and Mayer. 9 However,
no single publication or collection has furnished the
data necessary to delineate the pattern of behavior of
metal films through the large temperature and thickness
ranges covered by these experiments, 25° to 600°C and
75 to 2000 A. This paper proposes to establish this
pattern.
BEHAVIOR OF THIN METAL FILMS
SUBJECTED TO HEATING
The authors have previously published papers and
reports outlining the electrical resistance behavior of
thin metal films subjected to annealing by heating and
the techniques utilized in depositing and measuring the
films. 1O-13 A more detailed report of the present work is
also available as a technical report. 14
* This research was sponsored by the Aeronautical Research
Laboratory of Wright Air Development Center under Contract
No. AF 33(616)--3870 of the U. S. Army Air Force.
1 A. Fery, J. phys. radium 4, 301 (1933).
2 A. Fery, J. phys. radium 9, 38 (1928).
3 A. Van Itterbeek and L. de Greve, Nature 156, 634 (1945).
4 A. Van Itterbeek and L. de Greve, Physica 11, 470 (1946).
6 N. Mostovetch, Ann. Physik 8, 61 (1953).
6 N. Mostovetch, Compt. rend. 234, 305 (1952).
7 A. Blanc-Lapierre and M. Perrot, Mem. sci. phys. acado sci
Paris 57 (1954).
8 L. Holland, Vacuum Deposition of Thin Films (John Wiley
and Sons, Inc., New York, 1956).
9 H. Mayer, Physik Dunner Schich/en (Wissenschaftliche
Verlagsgesellschaft M. B. H., Stuttgart, 1955), Vol. II.
10 R. B. Belser, J. Appl. Phys. 28, 109 (1957).
11 R. B. Belser and W. H. Hicklin, Rev. Sci. Instr. 27, 293
(1956).
12 R. B. Belser, Ceram. Age 66, 28 (1955).
13 R. B. Belser, Rev. Sci. Instr. 25, 180 (1954).
14 R. B. Belser and W. H. Hicklin, "Electrical conductivity
313
[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ]
IP: 129.49.170.188 On: Sat, 20 Dec 2014 14:10:04
metal films as usually deposited are in a state of strain
not associated with the bulk metal as it crystallizes
from the melt. This state is one which is thermo-
dynamically unstable; hence, changes in the film usually
occur with time and with exposure of the film to
temperatures above that experienced by the film during
its growth. In addition, films are subject to agglomer-
ation due to surface forces and to reaction with elements
of the environment in which they are heated. A knowl-
edge of these principles and their applicability to each
metal studied is essential in order to interpret resistance
versus temperature curves; and a further factor, as
will be shown, is the effect of inclusions such as foreign
atoms or chemical compounds of the metal under study.
SPECIMEN FABRICATION AND
MEASURING PROCEDURES
Tubes of fused quartz, Vycor, glass, and other
ceramics were used as substrates; their dimensions were
usually 392-in. diameter by 1-in. length. Substrates were
cleaned successively in solutions of chromic acid, a
detergent, and thoroughly rinsed with distilled water.
The layout of the films on the substrate is exhibited in
Fig. 1. DuPont Liquid Bright Platinum No. 6558 was
applied over an area extending 136 in. from each end
and subsequently fired at 745°C. Affixed to the plati-
nized areas were pigtail leads of lO-mil silver wire,
tightened by twisting and bonded to the platinum coat
with duPont silver paste No. 4756 subsequently fired
at 708°C.
After firing, the substrates were again cleaned by
immersion in chromic acid solution, a detergent, and an
alcohol rinse and were dried with hot air. They were
studies of metallic films," WADC Technical Report 57-..fJOO
(December, 1957),
 314 R. B. BELSER AND W. H. HICKLIN

~THINF'LM
~OFMETAL
~ LIQUID BRIGHT
~ GOLD DR PLATINUM
(b)
FIG. 1. Construction of film element for temperature
coefficient of resistance measurement.
then inserted individually in the vacuum chamber for
plating by sputtering or evaporation.
The tubular substrate was held in a mask which
confined the plating to an area of ! by -hin. or to !
by -h in. and to an expanded end area overlapping the
platinized end films for about /6 in. This configuration
of the film was used for the bulk of the specimens
reported in this paper and thus the film being examined
consisted of 16 unit squares lz X :lz in. Exceptions were
films of chromium and titanium deposited on fused
quartz substrates (t in. diameter). These were -h in.
wide and consisted of 32 unit squares. An auxiliary
heater wire of 40-mil Nichrome extending axially
through the tubular substrate allowed the heating of its
surface to a known temperature. The substrate heater
power required to produce the desired temperature was
established by the use of a platinum film calibrated as
a resistance pyrometer. Substrates were heated to
temperatures in the range 350° to 600°C and cooled
before plating or maintained at a desired temperature
during plating. The plating time and rate were con-
trolled by monitoring of the resistance during plating.
The thickness of the deposited films was estimated by
visual inspection of light transmission, deposition time,
weights deposited, resistance per square measure-
ments,lO·t and by interferometric measurements in
some instances. Sections through films of small width
slightly above the substrate in order to monitor
resistance during plating. Thicknesses, in general, were
less than 1500 A and as low as 75 A.
Specimens were temperature cycled to 600°C in a
Vycor tube inserted in a hinged tube furnace in which
the atmosphere could be varied from a pressure of
2X 10-5 mm of mercury to atmospheric pressure; dry
argon was normally the residual gas. Measurements
were made at 25° or 50°C intervals dependent on the
rate of change of resistance; the resistances were moni-
tored by means of a Wheatstone bridge. Each specimen
was temperature cycled over the range 25° to 600°C
approximately five times unless failure intervened.
Temperature cycling rates were varied between 10 and
40 min up or down by using a Variac control to the
furnace. The usual time for a one way cycle was 15 to
20 min with about 10 min at peak temperature. The
hinged tube furnace could be removed for very rapid
cooling. The temperature was monitored with a chro-
mel-alumel thermocouple. On rapid cycles two operators
were required to track resistance and temperature
simultaneously.
METAL FILMS EXAMINED
Film specimens of 13 different metals and three alloys
of himetal films have been made. These fall generally
in the following classifications:
1. Noble metals-gold, platinum, iridium, rhodium,
palladium, and nickeLt
2. Refractory metals-molybdenum, tantalum, and
tungsten.
3. Semirefractory metals-chromium, titanium and
zirconium.
4. Alloys of platinum-platinum-gold, platinum-irid-
ium, and platinum-nickel.
5. Other metals-aluminum.
+200
1. Initial Resistanccs At Room Temperature:
+180 First Tesc-47.7 {) . I
+160 Second Test - 17,SO 2
+ 140 Third Test·IS.50 3
§+120
2. Room Temperature Resistance Afrer ~
Third Test: 19.10 V
~ +100

<"" "
(lz in.) examined by the interferometer were not 15 +80

rectangular but semicircular because of partial shadow-

ing of the large cathode source by the mask in a direc-
j :: 0
V
1/ "

tion other than parallel to the collimating slit. Beveling

+2 0

0 ~- -- - - ....
- - I
the sides of the slot improved the uniformity of the -20
0
thickness, but the metal mask was necessarily sited - 4 0 100 200 300 400 500 600
Temperature ("C)

t The authors have shown in the reference cited that sputtered
films of the noble metals, after annealing and in the thickness
range 200 to 1500 A, have resistivity values generally in the range
1.5 to 1.8Pb. Since thickness in cmXR/sq=p, film thickness of
films of these metals can be approximated from R/sq values.
In the thickness calculations corrections were made for the
semicircular cross section of the films examined.
FIG. 2. Resistance-versus-temperature characteristics of a
relatively thick film of sputtered gold (800 A-estimated)
temperature cycled in air.
t Not a noble metal but a transition metal related to rhodium
and iridium in electron configuration and chemical behavior.

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ]
IP: 129.49.170.188 On: Sat, 20 Dec 2014 14:10:04
 TCR OF METALIC FILMS 315

EFFECTS OF HEATING A THIN METAL FILM +300

T 1.1 I 1 1 1 I
+ 28or- 1. [Olnat Resistance:: 96.2 (1
+260 ~ 2. Unit Failed The Fint Test.
Annealing
+240
The physical effects of heating a thin metal film +220

deposited on a ceramic substrate are best exhibited by +200

I
-+180
films of a noble metal such as gold heated in vacuo. If.
-+160
I
In Fig. 2 will be observed a typical pattern of behavior ~
~ +140
1/
for a film deposited by sputtering onto a cool substrate j U +120
the thickness of this film was estimated as 800 A.
Curve No.1 of this figure displays the pattern during
+:~0
'" +60
I
.;
~

the first temperature cycle to 600°C. A temperature +4 0

J
coefficient of a sort is exhibited to about l50°C. There- +2 0
~I
after the effect of the annealing action on the resistance 0 ,.....
_2 0
is greater than the TCR effect until a temperature near _4 0
550°C is reached. Note that the resistance dropped to o 100 200 300 400 500 600 700
T empmcuc< l'C)
37.3% of its original value in the first cycle but the
temperature coefficient of resistance (TCR) was virtu- FIG. 4. Resistance-versus-temperature characteristics of a thin

ally the same for cycles 2 and 3, 0.0025 per °C compared film of sputtered gold temperature cycled in air. Rapid resistance
increase above 400°C denotes agglomeration.
with 0.0035 for bulk gold.
If now the film be deposited on a hot substrate
(500°C), the annealing cycle is eliminated and the reported by Faraday,15 Da C. Andrade,16 and one of the
pattern of Fig. 3 is obtained. The over-all TCR is authorslO among others. Consistent agglomeration of
virtually the same as for the specimen of Fig. 2. The films of the following metals of about 200 A thickness,
increasing slopes of the curves of Fig. 3 are not expli- when heated in vacuo, has been observed by the authors.
cable although the small increases in initial resistance Silver agglomerated at about 300°C, gold at about
indicate the possibility of some agglomeration occurring 450°C and platinum near 600°C. Agglomeration has
during each temperature cycle. also been observed for films of copper, tin, and indium.
Similar patterns of behavior were found for films of In general, agglomeration appears to occur for films,
platinum, palladium, iridium, and rhodium when heated of only a few hundred A thickness, at temperatures
in vacuo, except for small changes in the slope of the slightly above the recrystallization temperature of the
curve. Platinum and iridium films of a few hundred A metal of the film. Although electron micrographs of
thickness after annealing could be heated in air to films of some 30 metals have been examined after
600°C with very little irreversible resistance change. heating to 600°C in vacuo, agglomeration has not been
Electron diffraction studies of palladium thus heated ·observed for the others. These metals either were
exhibited lines ascribed to PdO. Rhodium films gave readily oxidized even in t'acuo, i.e., aluminum, iron,
unidentifiable new lines. Abrupt irreversible resistance manganese, or had high recrystallization temperatures,
increases for palladium films were observed at 400°C i.e., iridium, osmium, tantalum, molybdnum, rhenium,
and for rhodium at about SOO°C. These are ascribed to and tungsten. In Fig. 4 are recorded data for a 400 A
oxidation. gold film subjected to heating in vacuo. The film begins
Agglomeration to anneal at iS0°C as before but begins a sharp increase
in resistance at 350°C, culminating in failure at about
Very thin films of metal on glass, quartz, or ceramic 500°C. A platinum film of about 100 A thickness
substrates agglomerate when heated. This fact has been exhibited a sharp rise in resistance at 550°C. The abrupt
increases in resistance for these metals are ascribed to
+ 200 "r--r-;--,-,--,-..,-..,--,....-,r-r--r--.
+ 180 I I I I I agglomeration.
+ 160 f- Iniri~l Rcsjs{a~((;s Ac Room Temperature-
First Test- 11.50
~+1401- SccondTc" ..12.90 +-+-+-~-h4
Corrosion
:-+ 120H---+-+-+++-+--I--H-T/~+-I Of the various metals examined only platinum, gold,
§ + l00t--iH--+--+-+--+--+-+-+ ",H,,--+-+-4
(; + 80 f-t--f-+-+-+--+--i----\.4"/,~-l--_+___I and iridium films were able to withstand extended
~ + 60 hH-+-+-+--I-crl!C Z"+'--++-+-_+___1 heating in air at 600°C without severe damage. Silver
.1ii +40I-H-+-+·:;b"'lY::.. ... -~-+++-t---l
-+·
was not examined but it agglomerates so readily near
~ + 20 /--H~r::""'f-I:?-'!-" -/--+--+--1----1--+---1--1
""".- 300°C that its failure was virtually assured in the
thickness range of interest. In Fig. 5 are shown data
- -w O!:--'~IOO,,-'--::2":::UU-'-..,lW"'-U-'-.J..'OU-'--;OU OO.....J
L-.J.-6.J.. obtained for a nickel film deposited on a cool substrate
Tl:mpt:r,ltlln: Cc)
and heated in air, There was no annealing of the film
FIG. 3. Resistance-versus-temperature characteristics of a
sputtered gold film deposited on a hot (500°C) substrate and
temperature cycled in vacuo.
1: M. Faraday, Trans. Roy. Soc. (London) 147, 145 (1857).
I E. N. Da C. Andrade, Trans. Faraday Soc. 31, 1137 (1935)

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ]
IP: 129.49.170.188 On: Sat, 20 Dec 2014 14:10:04
 316 R. B. BELSER AND W. H. HICKLIN

+1 00 CORROSION PROTECTION
+280 1. Initial Resistance At Room
+2,6C
Temperature: 185.6 n f The damage to films by oxidation when heated in air
+2 40
2. The Unit F:ailed During Toe I severely limited prospective use of metal films in elec-
+2 20
FirsrTcst.
! I tronic components, and a satisfactory protective sys-
+200
II tem was sought. Silicon monoxide, silicon dioxide, and
ISO magnesium fluoride were examined for coating quality;
.60
/ silicon monoxide was chosen as the most satisfactory
140
120
II of these. It was deposited in vacuo by evaporation onto
100 the hot film (350°C) soon after it was fabricated
80 V although not necessarily before the system had been
60
/ opened to air.
40 Considerable protection was provided films of various
20 /
// metals as will be observed in the subsequently presented
0
experiments. However, the coating techniques or ma-
20
40
terials have not yet been developed to the desired state
100 200 00 400 500 GoO
Temperarure ("C)
of perfection.
A desirable effect of the coating, in addition to
FIG. 5. Resistance-versus-temperature characteristics of a corrosion protection, is retardation of agglomeration.
sputtered nickel film temperature cycled in air. Rapid resistance
increase above 250°C denotes oxidation. This effect is exhibited by the very thin platinum film
of Fig. 7. This film has an R/sq of 15 ohms compared
with an R/sq of 11.7 ohms for an uncoated film which
as would be expected; instead, a resistance increase was began to agglomerate at 550°C. This film withstood
registered. This increase is ascribed to corrosion. In 4 cycles to 600°C with little damage. Its thickness is in
Fig. 6 may be seen the behavior of a second film coated the neighborhood of 75 A; and its R/sq is the highest
with silicon monoxide and heated in an argon atmos- obtained for platinum films successfully withstanding
phere. The curve of cycle 1 exhibits some annealing temperature cycling to 600°C. The overcoating of
and in cycles 2 and 3 the normal TCR of the film is silicon monoxide reduces the surface forces tending to
shown. Sharp changes in TCR occur at the Curie destroy the film and this method is significant in that
temperature. The shape of this curve is also a character- it offers a technique of increasing the R/sq obtainable
istic of the TCR curve for bulk nickel according to with films of a specific metal The degree to which the
Potter,17 although the primary change in shape was R/sq can be increased by this method has not yet been
registered at 400°C in his data. By a comparison of determined.
Figs. 5 and 6 the temperature at which oxidation of a
EFFECTS OF FILM THICKNESS ON TEMPERATURE
nickel film heated in air becomes rather rapid can be COEFFICIENT OF RESISTANCE
estimated as approximately 250°C. This observation
In the preceding Fig. 7, the thinnest platinum film,
was subsequently confirmed by similar studies of 12
75 A (15 ohms/sq), tested exhibited a TCR of 0.0015
different nickel films and by a large number of long and a thicker one, 800 A (3 ohms/sq), was 0.0021. The
term aging tests of nickel films examined over the highest value obtained in tests of some 21 specimens
temperature range 175° to 400°C. on various substrates exhibited in Table I was a
TCR of 0.0025. The TCR values did not correlate
+200~~~-r~~~~~-,-,~
L inltia! Rc:sistano::s At Room Temper.Hurl"
well with thickness differences, although the thinner
Fi~t Tcst-IjQ9.0 n
+ 180 " V films had TCR values below 0.0020. Other variables,
+ 160 Second Test· 276.00 --+---+---+---#"'1'-+--1
_ + 140 Third Tt5(-270.0(2 3 r-,.,V for instance the choice of substrate, had as much effect
~ + 120H----+-++++-+/q-F---t-2+_+_ H as thickness variation. However, it is evident that
if + IOoH----+-+-f-+-t.l'/-+-+--H----+--+ control of thickness alone does not provide a satis-
6 T~H----+-++-h~~~~~~ factory method of obtaining films of a pure metal that
~ +Go
v might be useful as practical resistor elements of low
.:a +40
~ TCR.
+20
EFFECTS OF THE SUBSTRATE
2. The Third Tt:~t AgreeJ ~ith The Second Tcs[_
-20 Except For Minor Variations Above 350~C The effect of the substrate on the TCR of metal
-4°0 100 200 300 400 500 600
Temperature rC)
films deposited on substrates of different materials is
also exhibited in Table I. Films were deposited on
FIG. 6. Resistance-versus-temperature characteristics of a substrates of descending order of coefficient of expansion
sputtered nickel film deposited on a hot (350°C) substrate,
coated with silicon monoxide and temperature cycled in argon. varying from glass at 12 ppm;oC to Stupalith at
±O.OO ppm;oC. It will be noted that the maximum
17 H. H. Potter, Proc. Phys. Soc. (London) 49, 671 (1937). range encountered for films of similar R/sq, hence

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ]
IP: 129.49.170.188 On: Sat, 20 Dec 2014 14:10:04
 TCR OF METALIC FILMS 317

TABLE I. Properties of platinum films deposited on various substrates.

Substrate heat (OC)

Temperature
Film Substrate coefficient of
resistance Continuous Final Substrate expansion resistance Number
Unit (O/sq) Preheat heat heat material (ppm;oC) (Last test) of tests

Pt-SG-I0a 3.33 500 None 500 Soft glass 12.0 0.0024 3

Pt-SG-9 a 3.19 500 None 500 Soft glass 12.0 0.0023 3
Pt-SG-8 a 3.03 500 None 500 Soft glass 12.0 0.0024 3
Pt-ST-l b 3.93 350 350 500 Steatite 8.6 0.0021 2
Pt-AL-l b 3.39 350 350 500 Alumina 7.9 0.0021 3
Pt-7 a 4.15 500 None 500 Pyrex 3.6 0.0023 3
Pt_5 a 3.1 500 None 500 Pyrex 3.6 0.0024 3
Pt_V_Eb.c 15.6 350 350 None Vycor 0.8 0.0015 3
Pt-V-Db .• 7.27 350 350 None Vycor 0.8 0.0019 3
Pt_V_Cb.c 11.8 350 350 None Vycor 0.8 0.0017 2
Pt-V-Bb 8.46 350 350 None Vycor 0.8 0.0022 4
Pt-V-Ab 3.80 350 350 None Vycor 0.8 0.0025 4
pt-V-4a 4.42 500 500 None Vycor 0.8 0.0016 2
Pt-V-3 a 3.83 None None None Vycor 0.8 0.0015 3
pt-V-2a 4.98 None None None Vycor 0.8 0.0018 3
Pt-V-la 3.48 None None None Vycor 0.8 0.0020 3
Pt-FQ-l b 3.62 350 350 500 Fused quartz 0.5 0.0016 2
Pt-S-l b 3.08 350 350 500 Stupalith ±O.O 0.0021 3
Pt-17 a 5.28 500 500 None Stupalith ±O.O 0.0020 3
Pt-16 a 3.2 500 500 None Stupalith ±O.O 0.0021 3
Pt-15 a 1.81 500 500 None Stupalith ±O.O 0.0019 3

a Silver end bonds.

b Platinum end bonds.
e Silicon monoxide coated.

thickness, was only from 0.0021 to 0.0024 for substrates adapted to use as resistance thermometers and similar
of these two extremes of thermal expansion coefficient. films have been used for temperature indication in this
On the other hand, values of films on Vycor were work with excellent success.
generally low although in the range 0.0015 to 0.0025. The primary conditions, related to normal film
Presently the high variability observed is considered deposition and examination procedures, that affect the
rela ted to some effect of adsorption or surface area TCR of thin films of relatively pure metals have now
peculiar to the sintered structure of Vycor.§ No been discussed; remaining to be observed are the effects
explanation is offered for the low value of 0.0016 found of adding impurities to a film of a pure metal. Since
on fused quartz. impurities consisting of alloying elements or compounds
In spite of the fact that the data of Table I cover a such as oxides of the metal under study are relatively
considerable span of film thicknesses and substrate common, the effects of these will be examined in the
heating techniques, they clearly indicate that differences order named.
in the thermal coefficient of expansion of ceramic BIMETAL FILMS
substrates upon which films are deposited will not
greatly affect the TCR of a film, i.e., not sufficiently Films of different metals may be successively de-
to make this effect particularly useful in preparing posited by any of several methods and alloyed by
practical resistor elements of low TCR. Other attributes heating in vacuo to a temperature in the neighborhood
of substrate surface such as roughness, porosity, and +200
adsorbed materials present in or on the structure are of +18 0 - 1. Inici:tI Rt'sist:lllCCS At Room Tcmpcr:Hurc: f-
greater importance. +160
p"st Ttst·250.00 I I I I I
- $cwnJ Tes(· 235.·111
Worthy of note is that platinum films deposited on --+H 0 - Third T",· 236.30 _I . 1 1 .1 .. 1
~" 0 - 2. Room Temperarure
'-'+12 Afrer The
glass and Pyrex substrates gave particularly uniform g, Third Test: 2t\O.on
RCSISt:l.llCC f-
] +100 -
and regular TCR values in spite of some variation in u +80
2
film thickness. This was calculated consistently in the ~ ~ .. -,,;..- ::..~ 3-
1

range 0.00233 to 0.00237, the spread being even less ~

~ +4 0
+60
.. ~ ...-
..~~?'
than that indicated in Table I. The average for the +2 0
1",.:0 f"'"
five films is 0.00236. Hence these films may be readily 0

-2 0
§ Vycor is made of a high silicon glass by leaching out soluble -4 0
lOO 200 300 400 500 600
components of the glass, other than silica, sintering and shrinking Tcmpcr'l[urc rq
the porous structure formed in subsequent steps. Distinctly
different behavior for films of several metals, i.e., especially FIG. 7. Resistance-versus-temperature characteristics of a very
titanium and zirconium, were observed when the substrate was thin (about 75 A) platinum film deposited by sputtering on a hot
changed from Vycor to fused quartz. Unfortunately this obser- (500°C) substrate, coated with silicon monoxide and temperature
vation was only made near the end of the work reported. cycled in air.

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ]
IP: 129.49.170.188 On: Sat, 20 Dec 2014 14:10:04
 318 R. B. BELSER AND W. H. HICKLIN

FILMS OF THE REFRACTORY METALS

Films of the refractory metals molybdenum, tantalum
and tungsten may be readily deposited by sputtering~
but they are subject to oxidation at relatively low
temperatures. Tantalum and molybdenum films heated
in air exhibited rapid irreversible resistance increases
~ +60 •• 2 at about 400°C. The resistance increase is ascribed to
§ +40 <. .....' .
'" +20 oxidation since electron microscope studies of films
similarly treated did not disclose agglomeration and
- 20'I-----H---+-+-+-+-+-++--+-+-+--I electron diffraction studies disclosed a high percentage
- 400 100 200 )00 400 )00 600 of the oxides of the respective metals.
Temperature (""C)
In contrast to this, observe in Fig. 9 the behavior of
FIG. 8. Resistance-versus-temperature characteristics of a film a molybdenum film temperature cycled in vacuo. The
composed of successively deposited layers of sputtered platinum TCR of this film is <0.0001. A tantalum film gave a
and sputtered gold temperature cycled in vacuo. Note that similarly low TCR with a slightly negative slope.
alloying increased the film resistivity and decreased the temper-
ature coefficient of resistance. Examinations of similar films of molybdenum and
tantalum by the electron diffraction technique revealed
the presence in the respective films of the Mo0 2 and
of the recrystallization temperature of the specific M0 0 compounds of molybdenum and the Ta 0.
2 a 2
metals, and frequently at a lower temperature.1B·11 compound of tantalum. It was evident that oxides of
Metal pairs examined were Pt-Au, Pt-Ir, and Pt-Ni. the metals were being formed and deposited with the
Films were deposited successively by sputtering and metals during the sputtering of the film.
relative pen;entages were estimated from sputtering Of particular interest are the high R/sq values
rates and deposition time for each metal. In Fig. 8 are obtained with films of the refractory metals. Values in
plotted the data obtained for the pair Pt+Au deposited the hundreds of ohms/ sq were common and a maximum
on a cool Vycor substrate in the order named. Thick- value of 4225 ohms/sq was registered for a tungsten
nesses were in the range 300 to 400 A each. film. These high values are important in miniaturization
Whereas, normally, annealing effects would be ex- of electronic components.
pected to be present, alloying appears to begin instead A second attribute of the refractory metals is a
at about 200°C. The resistance versus temperature high recrystallization temperature in the range 600° to
curve underwent an irreversible increase and tempera- 1200°C. Such high temperatures reduce the likelihood
ture cycles 2, 3, and 4 gave a repeatable TCR of 0.0007. of damage from agglomeration and increase power
This TCR value is approximately t that of a film of dissipation abilities; these properties are superior for
either metal in a relatively pure state. Subsequently films of the refractory metals. Unfortunately, films of
this film was baked in argon for one hour at 600°C at these metals require protection from the atmosphere for
a pressure of one-half atmosphere. Its TCR dropped to successful operation as resistors above approximately
0.00045. Its room temperature resistance increased by 300°C. Overcoating or hermetic sealing processes are
a factor of 3.2 in spite of the fact that the films of the
+2oor-~~~~~-r-r-r-r-r-ro
pure metals alone normally would have annealed to a + 180 I. l~i[ial Resistances At Room Temperature: -r--
FIrst Test-47().O n
value about one-half the original value. Hence a + 160 Second Test - 466.00 -r-
resistivity increase of approximately six times the pure __ + 140 2. Room Tc:mpcraturc Rcsim.ncc After The -!------
~+ 120 Second Test. 466 .){1 ,--I-
metal films in electrical parallel was experienced. The 1

~ + 100 I--t--I~ -I-f---+-+-+--r-+-+---+----t-l

R/sq value as a result was high, 39.5 ohms/sq, finally, 6 + 80 I .t--t-'-j---j--+-+--t-t--I
compared with a maximum of about 15 ohms/sq that
was obtained with platinum and 5/ohms/sq for films
~5 +40
+ 60

of gold only. '" +20 2

1-
Temperature coefficients of resistance of 0.0007 and -20
0.0002 were the lowest obtained with the bimetal films -40
0 100 200 )00 400 )00 600
Pt-Ir and Pt-Ni, respectively, in the alloy percentages Temperature CC)
used.
FIG. 9. Resistance-versus-temperature characteristics of a
sputtered molybdenum film deposited on a hot (500°C) substrate
18 R. B. Belser, "Aging study of metal plating on quartz and temperature cycled in vacuo.
crystals," Engineering Experiment Station, Georgia Institute of
Technology, Interim Report No.7, U. S. Army SCEL Contract ~ Cathodes were cut from sheets of the respective metals
No. DA-36-039-SC-42453 (May, 1952-April, 1954). obtained from A. D. Mackay, Inc., New York City, and were of
II A discussion of the alloying of thin metal films is currently the best technical grade obtainable from that company. Arc cast
being submitted as a separate paper. Other discussions may be cathodes of Mo and Ta were subsequently obtained but no data
found in a technical report, reference 18, by one of the authors. from these were available for this paper.

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ]
IP: 129.49.170.188 On: Sat, 20 Dec 2014 14:10:04
 TCR OF METALIC FILMS 319

necessary in order to make use of the potentialities of +2oo""-"--"--"'-"'-"'-"'-"'-"'-"'-"'-T""TI

I. Imtial Rcsisr.lnces At Room TcmpcLHurc
First Tcst-6:no{l
these films in the electronic field. +180
+160 Second Te<;t - ?6(U{l -+--+--+--+-+-+-+
~+140 Thircl Tc<;t-H5,5{l
SEMIREFRACTORY METALS ----- +120 Fourth T~~[-31'iO{l ---t---ll--j-I--I--t--i
g, 2. The TClllp::r.HlllT W.1S. HdJ At 600°C ?uri~r---­
;; +100
Efforts to deposit titanium and zirconium at. a Tllc Sl"lot1d TN To Corn pi etc The Aging
~ +80 Of The Film t~1--
satisfactory rate by sputtering were unsuccessful m ~ +60 3. The Fourth 1\·-,£ Agrw.l \\'ilb The Tl,1(f,ll---
initial attempts and no suitable chromium cathode ~as ~ +-10
l\~t \\.'ltl1ll1 ± I",: ClInl ~,J( SIi()\\l1 : - t---::....---
available; hence films of these metals were deposIted +20
, : I I I: 1--1-- I
by evaporation under the same conditions of masking ,~~~~;l~~~r-~_~~,_I ___ I--
I ,--- --. ".- - 1
and heating utilized for the previously discussed films.** -'0 -i'-
il:::-L~.....L......,~L7,;:;-l
-4oL-l-L-L..,L,,--L..,J
i -

Films of zirconium deposited on a hot substrate (350°C) o 100 200 )UO IWD 500 600
TempuJ.(lIJ"l· rC)
were far superior in appearance and adherence to those
deposited on a relatively cool substrate. . FIG. 11. Resistance-versus-temperature characteristics of an
Figure 10 shows the data for an evaporated chromIUm evaporated chromium film deposited on a hot (350°C) su?strate,
coated with silicon monoxide and temperature cycled III air.
film temperature cycled in air. Although the film
underwent annealing during the first cycle, in spite of
A film of zirconium overcoated with silicon monoxide
the fact that it was deposited on a substrate maintained
exhibited the interesting behavior shown in Fig. 12.
at 350°C subsequent cycles underwent successive
Evaporated films of this metal in the bare state con-
increases in resistance suggesting effects of oxidation.
sistently underwent increases in resistance of about
Tests of several similar specimens indicated that
SO% on initial exposure to air. If heated in air abrupt
oxidation became relatively rapid at approximately
irreversible resistance changes began at approximately
550°C. The TCR in the range 25° to 250°C was nearly
350°C.
negligible and increased to 0.0007 over the range 250°
Films of titanium behaved somewhat similarly to
to SOO°C.
those of zirconium. Figure 13 displays a typical pattern.
The benefits to be derived from overcoating a
Even in vacuo the resistance increased when heated.
chromium film with silicon monoxide, evaporated onto
When this film was overcoated with silicon monoxide
it while heated to about 350°C in vacuo, may be noted
and heated in air to 600°C for three cycles and subse-
in the data of Fig. 11. Here is exhibited what appears
quently held at 600°C for 7 min, the resistance of the
to be the repeatable behavior of a chromium film. The
film was reduced to 44.S% of its original value. The
film underwent two successive annealing cycles and it
TCR increased from a slightly negative value to
is apparent that complete annea~ng must req~ire
+0.00075. A second titanium film was overcoated with
treatment for some minutes at 600 C. The annealmg
SiO before temperature cycling and heated to 900°C.
temperature is in agreement with data on chromium
In the range 600° to 900°C the film underwent a
films previously reported by one of the authors. A very
reduction in resistance and the final resistance was
thin chromium film, of 100 ohms/sq and a thickness of
approximately 85% of the original value. Its TCR on
200 to 400 A, exhibited after annealing a behavior very
the cooling curve from 900°C to room temperature
similar to that of the film just discussed.
was 0.0005.
+6....-____________________~-,
+ 200 1. initial Resistances At R~m Temperature: ~_ + 'j 1. N~~e The .Exp.tnJeJ Rcsist.111Ce Change Scale. ~
+180 FJrScT<:st-118tO FIfth Tcst-I2750 2. II1I[(al Reslst-tIlC(S At Room Temperature:
+160 SCLOnJTcst-S600 SlxchTcsr-I331(} -t-- +4 First Tcst-16000 Third Tesr.141S0 r-
Third Tcst-93S0 Sevellth Tcsr-18WO_t-- +3 Se(olld Test· I jllO Fourth Tes[-I·i59(l t--
...-.+140 FounhTcst-11890 EighrhTcsr-20nn t---- + 2 3. The fourth Tc~t AgrceJ With The ThirJ Test _
~ + 120 2. T<:stS Four Through Eight Agreed With Test +I Within ±O.2X, The Curve For The FOUTth Tesf_
~ + 100 Three Except For The Initial Rcsimncc r--
8 + 80 Changes Noted. The Curves Arc NO{ Shown

~ + GOf-e- --
-

3 ~ -1 tjI5:Nf~~(SEh~"'E'n~tl__~::t~I·_:;:~:~t-=t-:t-"""~_l,.,...-c..:t~
2
~ + 401--l-l-l--l--l--j--j-j-t-t..,-tV;S;·P21
..-__1--.-_t-...,+~-±~;;;fo_-f9---t1
'" + 201-l--l--l--1--
r
';; - 21--l--l--t--f--11--i-i'- f-\.+--+-+--t--i

-
_3 I
~
~ -4 -1-++-','\,+++-1
~. _5 -I~I--t- _._+-+-'--+-+-+-+---1
-20,1-+-+-+-+-+-+-+-l~+-i-+-t-1
~ -6t- -- I--t- --f--.f--I--!---I--+.-+-+.-1
-400 100 200 300 400 500 600 -7
Temperarure rC)
_sl--l-I--I--I.----j----- -- r- ,-- ~+-I-t--t--i
FIG. 10. Resistance-versus-temperature characteristics of an -91-1--j--t-+·~ l--+-+-+--J.1- +'.. I ~''--'--
,
evaporated chromium film deposited on a hot (350°C) substrate -lOt- _ . I - - -~ -~~ ~ --r--+-+-+-i
and temperature cycled in air. -lll-I-- \

** Dr. H. Basseches of the Bell Telephone Laboratories reported -12 0 lOU 200 .lUU 100
TCllljlcr.trllrC ( C)
500 600

in a private conversation that he was able to sputter titanium

by pumping the sputtering chamber to 1(}6 mm of mercury FIG. 12. Resistance-versus-temperature characteristics of an
before increasing the residual pressure of argon to 3 X 1(}2 mm evaporated zirconium film deposited on a hot (350°C) su?strate,
of mercury for the sputtering operation. coated with silicon monoxide and temperature cycled III air.

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ]
IP: 129.49.170.188 On: Sat, 20 Dec 2014 14:10:04
 320 R. B. BELSER AND W. H. HICKLIN

values. These included specifically films of molybdenum,

FIG. 13. Resistance-versus-temperature characteristics of an
evaporated titanium film deposited on a hot (350°C) fused
quartz substrate and temperature cycled in vacuo.
A life test of a titanium film resistor similar to that
of Fig. 13, but on a Vycor substrate, was made. The
film survived a life test of 262 hr in the temperature
range 200° to 600°C. The last 72 hr were at 600°C.
The film dropped in resistance approximately 6% about
midway through the test but had returned to its
original value by the end of the test. During the last
72 hr at 600°C it increased in value only 2%.
The changes observed in the resistances and TCR
values of titanium films may be ascribed to oxidation
and reduction, to gas adsorption and desorption, to
high temperature annealing effects subsequent to some
oxidation , or to combinations of these effects. Since
. the
large resistance decreases were observed only m the
presence of SiO overcoat possible reduction of a
component of the film by the SiO appears plausible.
Oxides of the respective metals were found by
electron diffraction examinations to be impurities in
films of all the metals exhibiting markedly low TCR
TABLE II. Power dissipated by films of various metals'
Melting Test conditions
temper- Resist- Power Ambient
ature of ance of dissi- temper-
Metal metal film pation Atmos- ature
film (OC) (O/sq) (w) phere (OC)
tantalum, tungsten, chromium, titanium, and zir-
conium. Chromium was the least active of the six
metals in this respect. The presence of the oxide of the
metal reduced the TCR and increased the resistivity
of the film as compared to those of the pure metal.
Hence it is probable that variations observed between
films of the same metal could be attributed to the
relative percentage of the oxide versus metal in the
specific film. However, one may not rule out the
possible presence of other compounds such as nitrides
in the case of titanium as indicated by unidentified
electron diffraction lines.
The percentage of the oxide present in a film can be
controlled, to a degree, by control of sputtering or
evaporation conditions and the treatment the substrate
receives prior to deposition or that the film and sub-
strate receive subsequent to deposition. As deposited,
however, all of the listed metals gave markedly low
TCR values. The apparent adaptability of films of the
semirefractory metals to protection by overcoating
suggests that with further development they should
provide materials for reliable high temperature resistor
units.
POWER DISSIPATION ABILITms
The power dissipation abilities of thin metal films are
related to the nature of the substrate, the configuration
of the film with respect to it, the environment, and the
metal of the film.
A few power dissipation tests of films were run under
various conditions of atmosphere and ambient temper-
ature as exhibited in Table II. These tests suggested
that the intrinsic power dissipation abilities of the
metal films were related to their melting temperatures
(or to their recrystallization temperatures) and that
films of the refractory metals would dissipate power
satisfactorily at relatively high rates per unit area. In
order to attain this desirable behavior, however, films
of these metals must be protected from corrosion.
Remark::;

Au 1063 3.5 0.20 Air 600 Film failed RESISTANCE PER SQUARE
Au-Pt 57.8 0.26 Air 600 Film failed
Au-Pt 15.3 2.9 Air 600 Film failed In this study of films of various metals each metal or
Zr 1830 156.0 2.4 Air 450 Film failed
(SiO coated) alloy film has exhibited a maximum resistance per
Zr 1830 94.2 5.4 Air 450 Film failed square beyond which agglomeration or corrosion de-
(SiO coated) stroyed the film during successive temperature cycling
Cr 1890 30.5 2.6 Air 600 Film failed
Rh 1966 11.9 4.03 Vacuum 100 End bonds to 600°C. By protection of the film with an overcoat of
failed silicon monoxide this limit may be extended to a degree
Pt-Ir 41.8 11.2 Air 25 Film failed as yet undetermined. In Table III are tabulated the
Ir 2454 24.4 7.5 Air 25 End bonds
failed limits of resistance per square measured for each metal
Mo 2622 99.0 10.0 Argon 600 Resistance during this study. The upper limits which would survive
decreased
66% temperature cycling vary from a few ohms per square
Mo 2622 38.8 9.2 Argon 600 Resistance for gold films to more than 4000 ohms per square for
decreased
41.7% tungsten. The metal oxides included in films of the
Mo 2622 20.4 17.25 Argon 600 Film failed refractory and semi refractory metals as prepared con-
tributed to the high R/sq values and the low TCR
• All film resistive elements were 1/2 X 1/32 in. except that of Cr which
was 1/2 XI /64 in. A typical film specimen is shown in Fig. 1. values found for them. The high R/sq values, in addi-

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ]
IP: 129.49.170.188 On: Sat, 20 Dec 2014 14:10:04
 TCR OF METALIC F I L M S 321
TABLE III. Variation of resistances per square and temperature coefficients of resistance of the metal films examined.

Number Resistance per square (ohms) Temperature coefficients Ratio

of films Before annealing After annealing of resistance Bulk metal- TCR (highest value)
examined Lowest Highest Lowest Highest Lowest Highest (O-lOO·C) TCR (bulk)

Films of noble or seminoble metals

Au(14) 1.60 22.2 0.45 4.95 0.0016 0.0028 0.0034 0.82
Pt(19) 8.10 8.7 1.80 15.6 0.0015 0.0025 0.0039 0.64
Ir(l3) 8.7 12.8 2.78 42.5 0.0008 0.0018 0.0040 0.45
Rh(19) 10.0 17.3 1.31 15.8 0.0018 0.0020 0.0046 0.43
Pd(19) 1.24 20.3 1.68 20.8 b 0.0001 0.0023 0.0037 0.62
Ni(12) 11.6 28.5 16.8 41.0 0.0030 0.0050 0.0064 0.78
Binary alloy films
Pt-Ir(6) 7.4 19.3 6.1 14.0 0.0007 0.0018 0.0012 0.67
(Ir 10%)
Pt-Au(2) 11.2 12.3 12.0 43.5 0.0005 0.0008
Pt-Ni(3) 25.3 31.8 43.7 0.0002 0.0009
Semirefractory metals
Cr(6) 19.5 172.5 9.8 62.0 <0.0001 0.0006 None listed
Ti(S)d 12.3 67.1 15.9 59.5· <0.0001 0.0007 0.0054 0.13
Zr(l1)e 24.4 134.4 ••• 0 <0.0001 <0.0001 0.0044 0.02
Refractory metals
Mo(8)! 13.9 99.5 17.2 49.0 <0.0001 0.0002 0.0033 0.06
Ta(20)& 25.6 768.0 ••• c <0.0001 <0.0001 0.0031 0.03
W(6). 64.9 4390.0 63.9 4225.0- -0.0002 <0.0001 0.0048 0.02
Other metals
AI(l) 0.41 0.36 0.0028 0.0043 0.65

• TCR values listed in Smithell, Metals Reference Book (Butterworth's Scientific Publications, Ltd., London, 1949).
b Palladium has peculiar gas adsorption properties which interfered with annealing observations.
• Films of these metals did not undergo appreciable reductions in resistance during annealing treatments in HCuo and rose in resistance because of their
high activity in relation to gas adsorption and their susceptibilities to oxidation. Tungsten and tantalum have recrystallization temperatures above the
temperature of examination.
d Annealed only above 600°C.
• Gas adsorption and oxidation prevented annealing to an appreciable degree.
f Annealed only during power dissipation tests.
• No annealing at 600°C.

tion-to the relatively high power dissipation abilities
and low TCR values previously discussed, are desirable
features for films to be used as resistors in electronic
applications.
OVERCOATING
The retardation of the corrosion of thin metal films
by overcoating with silicon monoxide has been exhibited
in the data presented, The retardation of film agglomer-
ation, hence, the possibility of examining thinner films
at higher temperatures, has also been indicated. The
choice of possible film material for examination or
utilization has thus been broadened by overcoating.
CONCLUSIONS
The temperature coefficients of resistance of thin
metal films depart from those of the bulk metal because
of the imperfection of the films intrinsic in their method
of deposition and growth.
Imperfections consist of two general types, those
which may be annealed out and those which may not.
In the first category are vacancies, dislocations and
occluded gas, and in the second are foreign atoms,
refractory oxides or other compounds, and imperfections
intrinsic to the film-substrate and film-environment
interfaces. Since the interface influences must extend
for several atomic diameters from either surface their
importance increases as the thickness of the film is
diminished.
The resistance versus temperature curve for an
unannealed film reveals the influences of annealing,
corrosion, agglomeration, and alloying in addition to
the normal temperature coefficient of resistance of the
metal of the film. If the effects of the other influences
are limited by methods outlined the TCR of a film of
gold or platinum may be shown to approach that of the
bulk metal and those of less noble metals will approach
it to a lesser degree. The less noble metals are so prone
to oxidation or other chemical combination during
evaporation, sputtering, handling or by contact with
adsorbed gases driven off from substrates during subse-
quent temperature cycling that measurements of their
temperature coefficients of resistance reflect a high
degree of impurity in spite of the employment of
preventive measures. In addition, the high recrystal-
lization temperatures of some of the metals studied,
i.e., iridium, molybdenum, tantalum, and tungsten,
prevented complete annealing of these metals during
temperature cycling to 600°C.
By knowledge and control of the cited factors
affecting the TCR of the films, however, one is able to
prepare films of low TCR, high R/sq, high power
dissipation and improved stability over specified
temperature ranges. Films of these properties are
desirable in electronic applications.

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ]
IP: 129.49.170.188 On: Sat, 20 Dec 2014 14:10:04
 322 R. B. BELSER AND W. H. HICKLIN

In conclusion, the primary factors affecting the ACKNOWLEDGMENTS

electrical resistances of thin metal films in the temper-
ature range 25° to 600°C have been delineated. By
judicious use of the methods explained informative
studies of many metals and their alloys may be made,
the practical utilization of thin metal films in industry
can be expanded and present devices incorporating
thin metal films can be improved with regard to
stability and reliability.
Acknowledgments are made to Mr. John L. Brown
and Mr. James W. Johnson of the Engineering Experi-
ment Station of the Georgia Institute of Technology
for their assistance in preparation and interpretation
of the electron diffraction studies reported, to Mr.
James O. Darnell for his assistance in the experimental
work, and to the Misses Dorothy A. Brine and Virginia
A. Rouse for their assistance in preparation of this
paper in its final form.

JOURNAL OF APPLIED PHYSICS VOLUME 30. NUMBER 3 MARCH, 1959

Density Change in Silicon upon Melting

R. A. LOGAN AND W. L. BOND
Bell Telephone Laboratories, Inc., Murray Hill, New Jersey
(Received July 29, 1958)

The density of both solid and liquid silicon has been measured in the vicinity of its melting point. From
the average of several determinations, the increase in density upon melting is 9±1%.

y observation of the mass of the sprout formed at
B the surface upon freezing a known mass of silicon,
a density change of 1O±1% has been reported.! The
measurement of density change is not feasible by con-
ventional techniques due to the reactivity of silicon.
Furthermore, quartz vessels become soft at the melting
point of silicon (1415°C).
A technique has been devised to measure the density
of molten silicon in a more direct way with a precision
of approximately 1%. This is done by observation of
the volume of a small known mass of silicon celntained
in a calibrated quartz capillary of sufficiently heavy
wall to provide a dimensionally stable container during
the melting. The density of solid silicon at the melting
point was deduced from the lattice constant, obtained
by extrapolation of x-ray diffraction data taken at
temperatures up to 950°C.
To determine the density in the molten state, silicon
samples were heated in air in quartz capillary tubing,
20 to 40 mils i.d. and about 300 mils o.d. The average
bore diameter, over a !-in. length of tubing, was com-
puted from the length of a known mass of Hg. In the
samples of tubing used (approx 2 in. in length), the
diameter was constant within 1%. To observe the
effect of heating on the diameter of the capillary, a
tube was heated in air at 1450°C for 15 min. On re-
measuring the diameter it was found to have changed
by about 1%.
Samples of silicon, about 0.5 in. in length, were
ground and etched to fit inside a capillary of known
bore. The sample was then weighed. The capillary con-
I H. von Wartenberg, Naturwissenschaften 36, 373 (1949).
taining the sample was placed on an alundum platform
inside a tube furnace in air at a temperature of about
1425°C. The sample was discarded unless the silicon
melted and formed a single, bubble-free column.
The length of the column was readily observed by
using a reading microscope, and this length was stable
during the run, indicating a negligible reaction with the
silica. The microscope was calibrated in each run by
observing the total length of the quartz tube, which
had been measured with a micrometer. The microscope
calibrations were averaged. Corrections due to the
expansion of the quartz are negligible.
From the length, diameter, and mass of the molten
silicon, the density was calculated. The results for five
successful runs, as described above, give an average
density of molten silicon at 1425°C of 2.49 glcc with
an average deviation of ±0.02 gl cc.
The density of solid silicon at high temperature was
determined from x-ray diffraction of powdered silicon,
using a hot camera technique which has already been
described. 2 The cell size was measured at temperatures
up to 950°C. By extrapolation of these results to the
melting point of silicon, one obtains a density of 2.2861
glcc at 1415°C.
Hence, the density of silicon increased by 9± 1%
upon melting.
ACKNOWLEDGMENTS
We wish to thank A. J. Peters and J. Andrus for
assistance in performing the experiments.
2 W. L. Bond, Rev. Sci. Instr. 29,654 (1958).

[This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to ]
IP: 129.49.170.188 On: Sat, 20 Dec 2014 14:10:04