For a powder source the material was ground fine with pestle
and mortar, placed loosely in a crucible and the surface smoothed.
Vacuum Deposited Thin Films
Using a Ruby Laser
Howard M. Smith and A. F. Turner
Howard M. Smith is at the Institute of Optics, University of
Rochester, Rochester, New York, and A. F . Turner is a t
Bausch & Lomb Inc., Rochester, New York.
Received 18 September 1964.
Partially sponsored by the U.S. Army Research Office, Dur-
ham, North Carolina. Paper TC18 presented at the Optical
Society of America meeting, Rochester, New York, October
1962.
The vaporization of materials in air by a laser is a familiar
phenomenon. I t suggests the use of a laser beam to evaporate
materials in vacuum for the deposition of optical thin films.
The success of this technique cannot be predicted a priori since
the constituent character of the vapor product, often visible in
air as a plume, is still problematical and its film-forming potential
unknown.
This work was undertaken to test empirically the technique by
applying it to a number of materials of interest in the vacuum
coating field. I t is found that many materials can be vaporized
in vacuum by a directed laser beam and condensed on substrates
as good optical films but t h a t others cannot, at least not under
the same conditions. Apparently each material will require
individual study. The work reported here is of a preliminary
survey nature only, but it shows the feasibility in many instances
of thin-film vacuum coating utilizing a laser external to the
vacuum chamber as the evaporation power source.
The evaporations were carried out in the 1 0 4 torr pressure
range using a vacuum coater with a 10-cm oil diffusion pump.
T h e laser equipment was the Trion LS-2 (Lear Siegler). A lens
focused the radiation from the T I R ruby rod through a window
in the bell jar onto the surface of the sample to be evaporated.
When cooled with liquid N 2 the rod furnished about 3 J per pulse.
I t was fired at a rate of 1 pulse/min. For an 0.5 beam divergence
and a 50-mm focal length lens, the minimum spot diameter was
0.5 mm. Assuming a 1 msec pulse the power density a t focus
was therefore 10G W/cm 2 . Q-switching by means of the Trion
MH-1 rotating mirror was used occasionally, as noted below, with
the rod uncooled. The estimated power density at focus was
then 108 W/cm 2 .
Most of films were evaporated from powdered materials con-
tained in small inclined crucibles in the bell jar. The powder
surfaces were at 45 to the horizontal laser beam. The electric
vector was in the plane of incidence. Some stibnite (Sb 2 S 3 ) films
also were evaporated from single crystals. Unheated glass, mica,
BaF 2 , or As2S3 substrate plates on which to deposit the films
were positioned in the bell jar 20 mm to 50 mm above the cru-
cibles. Because of this short throw the films were not of uniform
thickness and often showed lively interference colors.
Lateral motion of the lens allowed the focal spot to fall on the
powder surface where desired. The laser beam generally produced
a 1-mm to 2-mm diam crater at the impact point. After removal
from the bell jar little or no evidence of sintering or melting
around a crater could be observed under the microscope. All
traces of a crater disappeared on shaking the crucible slightly.
Table I lists those materials which in powder form gave opti-
cally satisfactory films as determined by simple inspection of their
interference colors or the quality of their specular reflectance.
The first column has the experimentally determined optical film
thickness, nd, in microns obtained per single laser pulse of
energy P 3 J, all film thicknesses being normalized to a throw
of 1 cm from sample to substrate under the assumption of the
inverse-square law. Experimental film thicknesses were esti-
mated visually on the basis of interference fringe counts. The
second column of Table I contains nominal refractive indices, n,
for thin films of the listed materials. Columns 4 and 5 give hand-
book values of melting points and boiling points.
The materials in Table I are listed in order of increasing diffi-
culty of producing evaporated films from them, as indicated by
the decreasing values of the entries in the first column. Their
sequence bears no obvious correlation with the melting points
and boiling points. I t will be noted t h a t even with the most
easily evaporated material, Sb2S3, for which nd/P = 0.35, eight
pulses are required for a quarter-wave film in the visible using a
throw of 4 cm.
The difference in the evaporation efficiency of the laser beam on
P b C l 2 and MoO 3 powders is noteworthy. Each can be evaporated
easily by conventional methods from resistance-heated boats.
Under the laser beam there was no difficulty in obtaining films of
MoO 3 . However, no films could be produced from the PbCl 2
powder until it was artificially darkened by the admixture of
carbon dust, the latter alone producing no film. Initially the
PbCl 2 powder was very white, the MoO 3 powder slightly greyish-
green. Thus the transfer of energy from the laser beam to the
powder samples appears rather critically dependent on small
amounts of absorption.
Q-switching had the opposite effect on the evaporation effi-
ciency of Ge and MoO 3 powders, allowing films to be evaporated
more easily from the former and less readily from the latter than
shown in Table I. This could be related to a lower heat conduc-
tivity of the MoO3, since Q-switching may restrict the amount of
material vaporized if the heat conductivity is insufficient for the
flow of thermal energy to an appreciable depth during the time of
one pulse. The superheated top layers, by not vaporizing rapidly
enough during the short Q pulse time, could shield the material
below from evaporating. 1
Films of the semiconductor compounds CdTe, PbTe, and ZnTe
all can be evaporated by conventional methods. The laser
beam, however, produced only rough and optically unsatisfactory
films of CdTe, having the appearance of being splashed on, al-
though good films of PbTe and ZnTe were obtained easily. X -
ray diffractometer lines of CdTe films were sharp, indicating
large crystallites, those of P b T e and ZnTe films were broad,
indicating fine crystallinity. Spectrophotometric transmittance
measurements on the ZnTe films showed the absorption edge at
0.8 as expected. I n PbTe films, the edge appeared to be at
about 1.5 instead of at the usual 3.5- position. For these very
thin films, this would not be in disagreement with published data 2
on the absorption coefficient which indicate a stepped curve, the
initial steep rise in absorption taking place at 3.5 , with a second
steep rise around 1.5 .
The stoichiometry of the laser beam-produced films in compari
son with the parent material is of interest. Semiquantitative
January 1965 / Vol. 4, No. 1 / APPLIED OPTICS 147
analyses of films of PbTe (0.2 thick) and ZnTe (0.4 thick) experimental sequence from Table I. The actual observed
were made by x-ray fluorescence. As far as the experiments thicknesses were, however, only 1% to 5 % of the computed,
could be carried, it was found that stoichiometrically the latter which is perhaps not surprising in view of the many experimental
definitely corresponded to its parent, while the P b T e film was uncertainties tending to produce a discrepancy in this direc
very probably the same as its parent. The composition of films tion. Careful yield determinations in the future could help in
produced from InAs powder was examined by x-ray diffraction answering the question whether laser beam evaporation is simply
with the E M X electron microprobe of Applied Research Labora thermal, depending only on the beam energy and not on its special
tories, Inc. Since this samples an area of only about 1 2, optical attributes.
individual particles blown up from the InAs powder onto the sub Thin film evaporation by means of a laser beam can offer
strate during the evaporation, commonly called spit, could be several unique advantages as the technique is developed further.
analyzed. On the basis of the detector counting rates for these I t will allow film formation under ultraclean vacuum conditions,
spit particles, the ratio of In to As was 2.5 times t h a t in an original e.g., by directing the beam through the wall of a thoroughly baked-
crystal of InAs, while in the film (0.2 thick) this ratio was re out and sealed-off bulb which needs to contain only the evaporant
versed with thirty-five times as much As compared to In as in the and a substrate to receive the film. Freedom from film contami
original. I t must be concluded that As was evaporated preferen nation is assured further by the surface evaporation characteristic
tially, and that the film is In-doped arsenic. of the beam. Because of the small beam divergence, the laser
I t was hoped that the laser beam technique would evaporate and associated equipment could be located at a considerable
stoichiometric films of some of the semiconductor compounds distance, an attractive possibility in radioactive areas. Once the
which ordinarily cause difficulty by conventional methods due to question of stoichiometry is resolved, applications of the laser
decomposition. Justification for this hope was by analogy with beam evaporation technique to vapor source instrumentation for
the flash method of producing films of some alloys by repeatedly solids in mass spectrometers also are apparent.
evaporating very small quantities to completion and relying on
We wish to thank L. Hadley, Colorado State University, for
recombination on the substrate of any decomposition products.
electron diffraction examinations of some of the films, and R. Roy
With the materials tried to date the laser beam technique has not
and E. W. White, The Pennsylvania State University, for the x-ray
been successful. Cu2O decomposed to give a film of Cu and no
fluorescence and E M X electron microprobe studies.
films could be produced from ZnO or InSb. InAs, discussed
above, decomposes badly under vacuum in a resistance-heated T a
boat to produce a grey deposit. The fact, however, t h a t with the Table I. Powdered Materials Producing Optically
laser clear films are obtained from InAs with as much as 3 % Satisfactory Films
concentration of I n relative to the original crystal, leads one to
suspect that success in stoichiometric films may depend on con
trolled laser-power operating conditions as has been shown neces
sary for laser welding. 3
Films of stibnite Sb 2 S 3 were produced from large single crystals
as well as from powder sources. The films are formed so easily
t h a t the entrance window into the vacuum chamber rapidly be
comes coated. Each successive laser pulse re-evaporates an
aperture in this coating so that it offers little obstruction to the
beam. The action of the beam on the stibnite crystal ejects a
white plume several millimeters long in the vacuum, as also ob
served from many of the powder samples. High-speed streak
camera pictures of the stibnite plume by G. Milne (University of
Rochester) showed that the generation of the plume was inter
mittent and followed the spiking of the ruby laser. This would
agree with the thermal spiking predicted for metals by Ready. 1
If the vaporization phenomenon is purely thermal, the amount
References
of material evaporated per absorbed joule of laser radiation can be 1. J. F . Ready, J. Opt. Soc. Am. 53, 514 (1963).
calculated from known specific heats and changes in enthalpy, 2. M. E. Lasser and H. Levinstein, Phys. Rev. 96, 47 (1954).
and from this the film thickness predicted for any given distance 3. G. W. Dunlap and D. L. Williams, Proc. Natl. Electron.
from the evaporant to substrate. Using published thermo- Conf., 18, 601 (1962).
dynamical data 4 this was done for Se, Te, and Ge, with the result 4. D . E . Gray, ed., A.I.P. Handbook, 2nd ed. (McGraw-Hill,
that the sequence of relative film thicknesses agreed with the New York, 1963).

148 APPLIED OPTICS / Vol. 4, No. 1 / January 1965