V O L U M E 2 5 , NO. 5, M A Y 1 9 5 3
697
what on the relative range of thickness of the two platings. Under certain conditions, three plating layers may be considered and their thicknesses determined from curves similar to those of Figure 8 or perhaps more easily from 5- or 6-line nomographs. .i simpler met,hod of determining plating thickness is feasible where the plating element and the backing element differ appreciably in the energy required to excite their s-ray spectra. For instance, consider cadmium plating on iron; iron is excited a t about 7 kv. while cadmium requires about 26 kv. Thus when cadmium-plated steel is irradiated, the cadmium will not fluoresce appreciably but will absorb the fluorescent iron spectrum strongly. S o analyzer is required; t,he detector measures the total fluorescent radiation. For example, the total scattering from an unplated steel specimen tTas found to be 17,400 counts per second while a cadmium-plated specimen gave only 600 counts per second under the same conditions. This difference can be easily cdibratetl in ternis of cadmium thickness. FUTURE POSSlBILlTlES
Fui ther improvements in fluorescent spectroscopy are being made constantly. There are new x-ray tubes under development ii-ith thinner windows and capable of withstanding ten times the present tube currents. Improvement in analyzing crystals is certain as other and better crystals are gronn in large enough vctions. Exploitation of better geometry such as closer approach of the specimen to the crystal with the defining slits bet\yeen the crystal and detector rather than between the crystal and specimen has recently been shoa n to give appreciable increase in intensity without loss of resolution. .4lloning vertical divergence of the fluortlscent radiation by the use of vertical Soller dits rather than tubular collimators gives an increase in intensity almost proportional to the increase in vertical divergence (about a factor of 10) without appreciable loss of re-olution. Bending
crystals to spherical caps rather than cylindrical sections can be done with the alkali halides or aluminum and may lead to better intensity and resolution. Improvements in the detectors is being made with the use of scintillation crystals for short wave lengths and proportional counters with thin windows for long wave lengths. The next year should see some of these new developments in use and fluorescent spectroscopy applied to problems which are on the borderline of possibility a t present. LITERATURE CITED
( 1 ) Abbott, J. L., Iron .4ge, 162, 58, 121 (1948): Steel, 1 2 3 , 8 8 (1948). ( 2 ) Beeghly, H. F., J . Electrochem. SOC.,97, 152 (1952). (3) Birks. L. 8.. Rets. Scz. Instr,. 22. 891 (1951). , , i4j Birks; L. %,'and Brooks, E.'J., Ax.& CHEM.,22, 1017 (1950). (5) I b i d . , 23, 707 (1951). (6) Birks, L. S., and Brooks, E. J., Saval Research Laboratory R e p t . 3867 (1951). ( 7 ) Birks, L. S., Brooks, E. J., Friedman, H., and Roe, R. XI., ANAL.CHEX, 2 2 , 1 2 5 8 (1950). (8) Brissey, R. hl., I b i d . , 24, 1034 (1962). ( 9 ) Brissey, R. M . , presented before Pittsburgh Conference 011 ;inalytical Chemistry and Applied Spectroscopy, Use of X-Ray Techniques in -4nalytical Chemistry, Pittsburgh, Pa., March 1952. (10) Carl, H. F., presented before Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, Use of X-Ray Techniques in Analytical Chemistry, Pittsburgh, Pa., March 1952. i l l ) Despujols, ,J., J . p h y s . radium, 13, 3 l d (1952). (12) Friedman, H., and Birks, L. S., Rev. Sci. Instr., 19, 323 (1948). ( 1 3 ) Guinier. A , , private communication. (14) Hasler, XI. F., Signal Corps Contract R 3 6 - 0 3 9 ~ ~ 3 8 1 7(1949). 7 (15) Hevesy, G. \-on, "Chemical Analysis by X-Rays and Its Applications," p. 86, Sew Tork, McGraa-Hill Book Co., 1932. f l G ) Koh. P. K., and Caugherty, B., J . A p p l . Phys., 23, 427, 098 (1952). (17) Patrick, R. F.. J . A n i . Ceram. SOC.,35, 189 (1952). (18) Pellissier, G. E., E k e . M f g . , 4 9 , 124 (1952). RECEIVED for review S o r e i n b e r 19, 19.52. Sccepted January 6, 1953.
Surface Analysis with the X-Ray Photoelectron Spectrometer RALPH G. STEINHARDT, JR.,
AND
EARL J. SERFASS, Lehigh 1-nirersity,Bethlehem, P a .
Because of the low energies of x-ray photoelectrons, their spectra are profoundly influenced by the chemical and physical properties of the surface from which they are emitted. A number of gold samples were coated with tarious numbers of monomolecular layers of barium stearate. The influence on the photoelectron spectrum of gold of even a single monomolecular layer of stearate was appreciable. Three effects on the gold edges were noted: energy displacement, spreading, and suppression. These effects were due to absorption and scattering by the barium stearate of the photoelectrons originating in the gold. The effect of increasing the roughness of a surface was to increase the average intensity of the spectrum without significantly altering the position and the shape of the characteristic edges.
P
REVIOUS work a i t h the x-ray photoelectron spectrometer ( 5 ) led to the ronclusion that this instrument was capahle of performing qualitative and quantitative analyses of the surfaces of solids. 4 s not all of the evidence leading to this conclusion mas direct, it was felt advisable to carry out additional confirmatory experiments. Furthermore, the effect of surface films and surface roughness had not been investigated experimentally. This communication describes and discusses these further experiments in the analysis of the surfaces of solids. I n order to study the manner in which a photoelectron spectrum is affected by the chemical nature of the surface from which emission occurs, and t o investigate the effect on these spectra of thin films supported on solid surfaces, a series of gold samples was
coated with various nunihers of monomolecular layers of barium stearate. Barium stearate \\as chosen because of the high precision with which the thickness of such films could be established. The spectra of these samples were compared using a gold sample coated with a single monolayer &s a reference standard. TWO effects should be observable as the thickness of the film increases:
i: eakened ~ ~in intensity. ~ ~ ~ ~ ~ ~ E , " , " distorted '~~ut~"
and
TT
Because the concentration of barium in the surface film was only about 18% and because the efficiency of photoelectron emission increases as about the fifth poaer of the atomic numher
ANALYTICAL CHEMISTRY
8698 the barium stearate experiments constituted a severe test of the instrument. The effect of surface roughness was investigated by means of a series of platinum samples of identical chemical composition which were ground to various degrees of roughness. The roughness was measured directly with a Brush analyzer, after which the photoelectron spectrum of each sample was determined. Severalsamplesof platinum black werealso prepared and the photoelectron spectrum of each sample was determined. Here, the average intensity of the spectra should increase as a linear function of surface area until the roughness becomes so great that electron trapping occurs and an upper intensity limit is approached. S o distortion or displacement effects should occur in this case.
employed effectively in establishing a relationship between roughness and photoelectron intensity.) EXPERIMENTAL RESULTS AND DISCUSSION
As noted above, one of the results to be expected from the stearate film experiments was that the intensity of the barium edge should increase as the thickness of the film was increased. However, primarily because of the limited resolution of the present instrument, and the relatively small amount of barium present, the data obtained did not show this intensity increase conclusively.
APPARATUS AND PROCEDURE
The spectrometer used was the same as that previously described (6),except that the spectrometer tube wm replaced by a chamber (Figure 1) having an adjustable slit system, rotatable sample holder, and increased signal-noise ratio. Furthermore, an improved stabilized magnet current control circuit (3) was designed and constructed. This circuit provided a source of less than 0.001 % ripple direct current continuously variable from 50 to 250 ma. No appreciable drift in current was observable during operation. These improvements have resulted in more precise and convenient operation of the spectrometer. By introducing several small refinements in technique, it was found possible to reduce the thickness of the window in the counting tube from about 25 to about 10 micrograms per sq. cm. Despite their small thickness, these windows are remarkably rugged. For example, the last two windows employed were in use 9 and 5 months, respectively, during which the spectrometer and counter were repeatedly opened to atmospheric pressure and re-evaruated.
E
Y
x
.-In CI
F c
s
U
-200
.480
Figure 2.
w’;r
T
Figure 1. Spectrometer C h a m b e r The apparatus and method used for the deposition of stearate monolayers xrere the same as those described by Blodgett and Langmuir (1,2), except that an automatic constant-speed dipping device was employed. The pH of the bath was 8.5, which resulted in the deposition of barium stearate-stearic acid in about a 9 to 1 ratio mixture (4). In order to eliminate the effect of the first layer, which was rubbed on, the standard itself was coated with a single layer. Omitting this single layer from further consideration. samples were prepared with the following numbers of monolayers deposited thereon: 0,2, 6, 12,24,50, and 100. The roughness samples were prepared using standard metallographic techniques, except that the grinding was carried out so that the grooves were parallel. The samples were mounted in the spectrometer with the grinding axis perpendicular to the direction of the electron beam. Brush surface analysis of these samples showed that they varied between 1.5 and 19 microinches root-mean-square average roughness. (Unfortunately, the distribution of these values was such that only four samples could be
.520
.560
1
Magnet Cur rent Peak Modification by Stearate Film
However, the effect of the stearate film on the gold photoelectron spectrum occurred as predicted. As an example, a portion of the gold spectrum is shown in Figure 2. The upper curve is that of pure gold, while the lower is that of pure gold plus 50 monomolecular layers (1240 A.) of barium stearate. Three effects of the film on the gold spectrum are immediately obvious: 1. The intensity is reduced considerably. 2. The electrons contributing to the edge are of lower energies. 3. Spreading of the peak occurs.
The decrease in intensity is clearly due to electron absorption, while the. energy decrease is that necessary for a photoelectron to pass through the stearate layer. The spreading of the peak is caused by electron scattering within the stearate layer. The effect of surface films is also demonstrated by the decrease in average photoelectron intensity as a function of film thickness (Figure 3). It will be noted, on interpolation, that even a single monomolecular surface layer has a definite and perceptible effect on the intensity. (The determination for 24 monolayers appears to be in error somewhat greater than is statistically probable. It has, however, been allowed to influence the curve as drawn, as no valid reason can be given for discarding the figure.) Primarily because of the difficulty of working with an iron electromagnet a t the low magnetic flux intensities which are necessarily involved in low energy electron spectrometry, these results must be considered as being of relatively low precision. For this
V O L U M E 25, NO. 5, M A Y 1 9 5 3
699
reason, no attempt has been made to correlate the alisolute energies and intensities with the filni thickness. (Instrumental revisions being carried out a t the present time should make it possible to obtain data of sufficient precision t o make such correlations significantly valuable.) I n evaluating t h r effect of roughness on average photoelrctron intensity, it i x necessary first to establish the actual surface area, A . of the saniple as a function of the root-mean-square average amplitude of roughness, R, as given by the Brush annl>-zrr. TWO assumptions ai’erequired:
propriate constant, the form of the equation of the cross-sectional curve in the present case was assumed as y = a sin w z. The second assumption was that the amplitude-period ratio was constant. a / P = k ~ . For the present case of a large number of grinding particles in which the period ii small compared to the totnl length of the curve, this assumption i- valid.
1. Form oi rquation of cross-sectional curve. 2 . Relationship hetween amplitude and period oi’ curvcx.
Becawe the Brush analyzer does not, take a true, root-nieansquare avrrage of the amplitude of a curve, hut siniplj. assumes a sine curve, anti multipliei: the average amplitude I)? the ap-
i :l;
\
-a
/
2 -as
\
-0.4
Figure 3.
2;
3;
4; ;3\;5
Number of Monomolecular Layers Effect of Stearate Film on Average Intensity
.440
.480
520
I
Magnet C u r r e n t Figure 5.
Effect of Roughness on Photoelectron Spectrum of Platinum
The arc length, 1, of a single cycle of the curve y = a sin w .r i.:
in which:
and E ( a ,
4)
is an elliptic integral of the second kind:
The second assumption states that: CllP = k, and therefore: aw =
A2
Thus, a is constant. and Eo1 this caw:
E
(a,
%)= k ,
Sine?, Tor any case:
.4 = k,l 11 .0
Relative Figure 4.
1.20
1.30
I n t e n s i t y (I)
Effect of Roughness on Average Intensity
and:
R
LJI
it is clear that : d
=
I..R for I? > 0
ANALYTICAL CHEMISTRY
700 That is, the actual area of the sample is a linear function of the roobmean-square average roughness value. Because the average photoelectron intensity, for sufficiently low values of R, is a linear function of the area, it follows that the average photoelectron intensity will be a linear function of R. The experimental confirmation of this is indicated in Figure 4. The relative intensity, I , obtained for platinum black was 1.63, which corresponds to a roughness value of 40 microinches. For this material the actual surface area is far greater than the value which would be obtained from Equation 8. It is probable that electron trapping is occurring in re-entrant cavities. Adsorption of foreign materials may also be of significance. One sample of platinum black showed values of Z which were considerably less than unity. Although only one sample of this nature was encountered, it is clear that the only assignable reason for this unexpectedly low emission value was a film of material of low atomic number adsorbed on the surface of the sample. The lack of distortion and displacement, however, of the characteristic platinum edges (Figures 5 and 6) indicates that electron trapping rather than film adsorption is the most probable reason for the relatively low value of Z noted in most determinations of platinum black spectra.
I
I
1
6fO
400
200
CONCLUSION
I t has been shown experimentally that x-ray photoelectron spectra are profoundly influenced by the chemical and physical nature of the surface under examination, Difficulties encountered have almost invariably been due to the use of an iron electromagnet a t low magnetic flux intensities and to the low intensity of photoelectron emission, which necessitates operations a t low resolution values. An electrostatic deflection spectrometer under construction a t the present time will eliminate the first and, because of radical changes in configuration, should considerably reduce the second difficulty.
Figure 6.
.400
.440
I
I
Magnet C u r r e n t Effect of Roughness on Photoelectron Spectrum of Platinum
ments; and to Katharine Blodgett, Research Laboratory, General Electric Co., for her kindness in personally introducing them to her techniques for the preparation and deposition of stearate monolayers.
ACKNOWLEDGMENT
LITERATURE CITED
The authors wish to express their grateful appreciation to Beckman Instruments, Inc., South Pasadena, Calif., for its financial support of the x-ray photoelectron spectrometer program; to Theodore Hailperin, Department of Mathematics, Lehigh University, for his assistance in mathematical problems encountered in the course of this work; to H. A. Liebhafsky, Research Laboratory, General Electric Co., Schenectady, N. Y., for his suggestion of the use of barium stearate monolayers in the film experi-
(1) Blodgett, K.B.,J . Am. Chem. Soc., 57, 1007 (1935). (2) Blodgett, K.B., and Langmuir, I., Phys. Rev., 51, 964 (1937). (3) Danner, W. E.,and Steinhardt. R. G., Jr., unpublished mste-
rial.
(4) Langmuir, I., and Schaeffer, V. J., J . Am. Chem. Soc., 58, 284 (1936). ( 5 ) Steinhardt, R. G., Jr., and Serfass, E. J., ANAL. CHEM.,23, l5S5 (1951). RECEIVED for review Kovember 19, 1952. Accepted January 10, 1953.
Lattice Parameters, Expansion Coefficients, and Atomic and Molecular Weights X -Ray Techniques fo r Precis ion Determination 31. E. STRAUMANIS, School of Mines and Metallurgy, University of Missouri, Rolla, M o .
P
RECISIONmeasurements of lattice parameters make it possible to decide quickly whether or not an observed coprecipitation is due to the formation of solid solutions, because this formation is always connected with the change of lattice parameters of the parent substance. Examples of this are: coprecipitation of cupric or cobalt ion by Zn[Hg(CNS)4](9, 19, 24, 31, 33, 3.4) and coprecipitation of barium nitrate by barium sulfate (36). Even the amounts coprecipitated can be determined by precise measurement of lattice parameters, if a solid solution forms. The accuracy is around i O . l % by weight ( 2 2 ) . To achieve more
accurate results, exposures of the samples a t constant teniperatures are necessary. Atomic weight determinations by the chemical method are difficult, and it is frequently desirable to check the values obtained. The x-ray method is convenient and very exact. The atomic weight of silicon was for a long time given as 28.1, rhich was changed to 28.06 in 1925, and in 1952 to the newly adopted weight of 28.09 (36). This weight is in perfect agreement with the weight 28.083 recently determined by x-ray methods (28). For the latter only minute amounts of a substance are necessary.
