the two electrochemically distinct areas at the disk must involve the bare platinum surface. A model of the electrode surface which accounts for these facts is shown in Figure 9, diagram V. One region of the electrode consists of bare platinum oxide exposed to the solution on which oxygen evolution can occur. At a second region of the electrode Bi(0) is oxidized to soluble Bi(II1). Both 0 2 and Bi(II1) can be collected at the ring electrode. The presence of IP-T’
proves that Bi(II1) is being stripped from the electrode at f1.25 V. RECEIVED for review January 24, 1972. Accepted May 22 1972. S.H.C. gratefully acknowledges the Samuel B. Silbert award. The support of the US.Air Force Office of Scientific Research under Grant No. AFOSR 70-1832 is also gratefully acknowledged.
Comparison of Internal Mixing and Vacuum Deposition Procedures for Calibrati ng ESCA Spectra William E. Swartz, Jr., Plato H. Watts, Jr., Judith C. Watts, J. W. Brasch, and Ellis R. Lippincott Center of Materials Research, Uniwrsity of Maryland, College Park, M d . 20742
Calibration of ESCA spectra is necessary because of the surface charge which builds up on the sample as a result of the photoejection process. The internal mixing and vacuum deposition techniques of calibrating ESCA spectra are compared. The comparison is based on correlations with ligand electronegativities, Mossbauer isomer shifts, and estimated atomic charges for the octahedral. tin complexes of formula [(CH3CH2)4N12SnX,Y,-,.
X-RAY PHOTOELECTRON SPECTROSCOPY is rapidly becoming recognized as a valuable analytical tool. One of the most serious problems facing users of ESCA centers around the question of sample charging and the development of methods to compensate for it. As electrons are ejected during the photoelectric process, a charge builds up on the surface of the sample. If the sample is a conductor, it can be put in direct electrical contact with the spectrometer, allowing electrical equilibrium to be reached. Thus, sample charging is of little or no consequence when the sample is a conductor. However, when the sample is a nonconductor, electrical equilibrium cannot be reached and the charge that builds up on the surface becomes a significant problem. This surface charge can be as large as several volts, depending on the electrical properties of the sample. Surface charging can cause significant error when chemical shifts in binding energies as small as tenths of an eV are being measured. Thus, before ESCA can realize its full potential as an analytical technique, a method or methods of accurately and reproducibly correcting for the surface charge must be established. The most common calibration method in the literature is the use of the carbon 1s electron line resulting from contamination of the sample by hydrocarbons (1). This method has several obvious limitations. Perhaps the most serious of these is that the nature of this contaminant is unknown. Thus, one cannot really be sure what the reference is. In addition, the contaminant can vary from laboratory to laboratory, or can be composed of several species. The latter can result in wide calibration lines. It also seems highly possible that this thin surface coating could be affected by the type of surface upon which it is deposited, irrespective of the surface charging ef(1) K. Siegbahn et al., “ESCA ‘Atomic, Molecular and Solid State Structure Studied by Means of Electron Spectroscopy’ ,” Almquist and Wiksells, Uppsala, 1967.
effects. Stec et al. (2) have observed large shifts in the C(1s) line when the hydrocarbon contaminant resided on compounds containing highly electronegative nearest-neighbor atoms. Additional complications may arise when the sample contains carbon. Normally, the C(1s) electrons from the sample will dominate the observed electron line; however, if the level of contamination is large enough, the resulting line can be very broad, and doublets may even result, depending on the nature of the carbons present. Thus, the use of the C(l s ) electron line originating from a hydrocarbon contaminant is a somewhat questionable procedure. Another method of calibration consists of homogeneously mixing a reference material into the powdered sample. Nordberg and coworkers (3) have had success by mixing graphite with their samples to use the C(1s) electron line from the graphite as a standard. Stec et al. (2) have mixed P b 3 0 4into their samples and find good agreement between arsenic binding energies measured relative to the Pf(4f) line and those measured in a separate experiment relative to the C(1s) line from residual carbon. Their results also agree with those of Hulett and Carlson (4) who referenced their arsenic-binding energies to the chlorine line originating from internally mixed KC1. Kumar et al. (5)have also had success in their study of palladium complexes by mixing graphite, sodium chloride, or palladium into their samples so that the graphite C(1s) line, the N a K L L Z 3( lD2)auger line or the palladium 3d line could be used for calibration. On the other hand, a large number of “ESCA users” have found that internally mixing of a reference material into the sample creates serious problems (6). They found that the mixing procedure gives rise to a large range (in eV) in the charging correction. In addition, they found that the results were sporadic, in that the necessary correction needed could be either positive or negative of the reference point. Thus, the “ESCA users” concluded that the internal mixing procedure for calibrating ESCA spectra was inadequate. (2) W. J. Stec, W. E. Morgan, R. G. Albridge, and J. R. Van Wazer, Inorg. Chem., 11, 219 (1972). (3) R. Nordberg, H. Brecht, R. G. Albridge, A. Fahlman, and J. R. Van Wazer, ibid., 9,2469 (1970). (4) L. D. Hulett and T. A. Carlson, Appl. Spectrosc., 25, 3 3 (1971). ( 5 ) G. Kumar, J. R. Blackburn, R. G. Albridge, W. E. Moddeman, and M. M. Jones, Inorg. Chem., 11,296 (1972). (6) ESCA Users Meeting, Summit, N.J., November 9, 1971.
ANALYTICAL CHEMISTRY, VOL. 44, NO. 12, OCTOBER 1972
2001
Table I. Tin (3d5l2)Binding Energies Relative to the Au(4hjz) and Mo(3dsiz)Electrons and the Comparison Parameters
No.
Compound
Sn(3ddEa relative to Au(4f7i2),eV
Av charging corr, eV
Sn(3dsiz)Eb relative to Mo(3dsiz),eV
1. K2SnF6 -0.3 489.6 f O . l d 488.8 i O . l d -0.4 487.1 i 0 . 2 487.8 f 0.1 2. (Et4N)nSnC14Fz -0.3 487.1 f 0.1 3. (Et4N)SnBraFz 487.7 f 0.4 4. (EtaN),SnCls -0.3 487.6 =k 0.1 487.6 i 0.2 -0.2 487.3 f 0.1 5. 487.7 =k 0 . 2 (EtaN)zSnC14Br2 -0.2 487.5 =k 0.1 6. (EtaN)ZSnCl& 487.9 i 0.1 -0.5 487.7 =k 0.1 487.7 i 0.1 7. (Et4N)zSnBr6 -0.4 487.2 i 0.1 8. (EtdN)2SnBroClZ 487.6 i 0.1 -0.3 487.1 f 0 . 1 487.5 =k 0.1 9. (Et4N)tSnBr41z -0.1 481.4 i 0 . 1 487.3 zt 0.2 10. (Et4N)zSnI4Cl2 -0.1 487.4 j=0.1 11. (Et4N)zSn14Brz 487.3 & 0 . 2 486.9 f 0 . 2 -0.2 481.2 zt 0.2 12. (Et4N)~Sn16 -0.2 13. KzSnBr6 488.2 i 0.2 487.6 f 0.1 486.5 & 0 . 2 486.4 i 0 . 2 ... 14. Sn(P) a Ref. 13. Ref. 16. Ref. 22. Experimentally determined value; see text for explanation of cation effect.
Hnatowich et al. (7) have shown that noble metals, chemically stable and thus characterizable when deposited in small amounts on nonconductors, come t o electrical equilibrium with the sample. By vacuum-depositing small amounts of gold and palladium on a nonconductor such as barium sulfate, they were able to show that the electron lines from gold (4fsiz4f712),palladium (3d3/~-3ds/z),and barium (3dSl2)shifted by equal amounts because of surface charging. This observation implies that the reference material and nonconducting sample are in electrical equilibrium. This electrical equilibrium would mean that the charging of the sample and deposited standard would be equivalent and thus, all the electron lines would shift in equal amounts because of surface charging. Therefore, deposition of metals such as gold and palladium on the surface of the sample has been suggested to be the most satisfactory method of calibrating ESCA spectra (6). Parameters such as ligand electronegativity (8, 9), Mossbauer isomer shifts (IO), and calculated atomic charges ( I , 11-16) have been shown t o correlate linearly with experimentally measured core-electron binding energies. Therefore, by comparing the degree of linearity observed between such parameters and binding energies measured by employing several calibration procedures, one can compare the calibration procedures. We wish t o report a comparison of the internal mixing and gold deposition calibration procedures (7) D. J. Hnatowich, J. Hudis, M. L. Perlman, and R. C. Ragaini, J. Appl. Phys., 42, 4883 (1971). (8) T. D. Thomas, J . Amer. Chem. SOC.,92,4184 (1970). (9) A. D. Baker, D. Betteridge, N. R. Kemp, and R. E. Kirby, Int. J. Mass Spectrom. Ion Phys., 4, 92 (1970). (10) M. Barber, P. Swift,.D. Cunningham, and M. J. Frazer, J. Chem. SOC.,D , 1970, 1338. (11) J. M. Hollander, D. M. Hendrickson, and W. L. Jolly, J Chem. Phys., 49, 3515 (1968). (12) M. Pelavin, D. M. Hendrickson, J. M. Hollander, and W. L. Jolly, J . Phys. Chem., 74, 1116 (1970). (13) D. M. Hendrickson, J. M. Hollander, and W . L. Jolly, Znorg. Chem., 8, 2642 (1969); 9, 612 (1970). (14) H. Basch and L. C. Snyder, Chem. Phys. Lett., 3, 333 (1969) (15) D. W. Davis, J. M. Hollander, D. A. Shirley, and T. D. Thomas, J. Chem. Phys., 52, 3295 (1970). (16) P. Finn, R. K. Pearson, J. M. Hollander, and W. L. Jolly, Znorg. Chem., 10,378 (1971). 2002
Av charging corr, eV -1.3 -1
.o
-0.4 -1.2 -0.9 -0.7 -0.9 -0.8 -0.7 -0.9 -0.9 -1
.o
-1.5 -0.5
-
ISa (cm/s)
6b
-2.46 -1.81 -1.57 -1.58 -1.43 -1.32 -1.26 -1.33 -1.14 -1.11 -1.01 -0.87 ... +O. 60
0.339 0.227 0.230 0.250 0.227 0.213 0.188 0.207 0.177 0.146 0.173 0.157 ... O.Oo0
4.10 3.40 3.27 3.15 3.08 2.98 2.95 3.12 2.85 2.83 2.75 2.65
...
2.15b
used in a systematic study of the tin (3daiz)binding energies in the tin hexahalide complexes, [(CH3CH2)4N]2SnX,Y6-,. EXPERIMENTAL
Apparatus. The photoelectron spectra were obtained with a Varian IEE-15 electron spectrometer equipped with a reaction chamber. Reagents. The tetraethylammonium tin salts were prepared by adding dropwise a stoichiometric amount of the appropriate tetraethylammonium halide dissolved in a 50 :50 mixture of chloroform and methanol to the appropriate tin tetrahalide compound dissolved in chloroform. After precipitation, the salts were filtered, washed with a 50:50 mixture of chloroform-methanol followed by anhydrous ether, and dried in vacuum. The preparations were all performed in a drybox (17). KzSnF6 and M o o 3 were purchased from the Research Organic/Inorganic Chemical Corporation and were used without further purification. Procedure. All of the spectra were obtained using the = 1253.6 eV) as the excitation Mg(Ka) X-ray line source. The work function of the spectrometer was set such that the A ~ ( 4 f ~electron ,~) binding energy obtained from a gold standard equaled 83.00 eV. With the analyzer voltage set to 100 volts, the width (FWHH) of the Au(4filz) electron line from the standard was 1.8 eV. The samples were run as powders dusted onto a backing of cellophane tape. For the internally mixed calibration, MOO, was mixed into the samples and the Mo(3dji2) electron line (Eb = 233.0 eV) was used as the reference. The Mo03/tin halide mixtures were homogenized using a Crescent Dental Manufacturing amalgamator. The ratio of sample to M o o 3 was maintained at approximately 3 :1. This ratio enabled intense Mo(3dsi2)and Sn(3dsiz)electron lines to be recorded in a minimum of time. The data for elemental tin relative t o Mo(3dsi2)was obtained by vacuum depositing the tin onto a sample of M o o 3in the reaction chamber. An attempt was made t o measure a binding energy for elemental tin by mixing some tin powder with Moo3. For each sample of powdered tin for which this was tried, only one Sn(3dsJ2)electron line was observed a t Eb = 490.0 eV. This binding energy is equivalent to that observed for a sample of Sn02. Thus, (17) C. A. Clausen and M. L. Good, Znorg. Chem., 9,220 (1970).
ANALYTICAL CHEMISTRY, VOL. 44, NO. 12, OCTOBER 1972
Table 11. Correlation Coefficients for Plots of Eb us. IS, %, and 6 Using the Au(4f712)and Mo(3d5,z) Standards Correlation coefficients Relative to Relative to Eb Eb Eb
US. US.
1s 6
US.
2,
Au(4fii 2)
Mo(3dsid
0.736 0.770 0.800
0.936 0.927 0.896
489 n
w
5n
488
0 c)
;4 8 1 a
it was necessary to vacuum-deposit the sample of elemental tin, The vacuum deposited standard used in this investigation was the gold (4f7/*)electron h e (Eb = 83.0 eV). The gold was vacuum-deposited from a hot tungsten filament onto the tin halides (containing no MOO,) in the reaction chamber. The elemental tin data relative to the Au(4f7p) reference was obtained by simultaneously depositing gold and tin onto the cellophane tape backing. The %Fez- data were obtained from the dipotassium salt while all other data were obtained from the ditetraethylammonium salts. Varying the cation in two otherwise identical compounds is expected to affect the electron binding energies of the atoms in the anion. This “cation effect” was compensated for in this case by comparing the Sn(3dsi*) binding energies obtained for K2SnBr6and [(CH3CH&N]2SnBrG. Since the only difference between the two compounds is the cation, any difference in binding energy must be attributed to the “cation effect.” The Sn(3dsi2) binding energies differed by 0.4 and 0.5 eV relative to the Mo(3ds/z) and Au(4f7l2),respectively, with that for the dipotassium salt being greater. One would expect similar differences in cations to have the same effect for all other pairs of salts. Therefore, by subtracting the appropriate amount from the binding energy obtained for K2SnF6, one can normalize the data to those of the ditetraethylammonium salt. This was done in plotting the data.
RESULTS The Sn(3dslz)electron binding energies measured relative to the Mo(3d5,s) and the Au(4f71~)electron lines are tabulated in Table I. Each reported energy is the average of at least three replicate measurements with the confidence limits established by the standard deviations. Also given in Table I are the average charging corrections for the tetraethylammonium tin halides using each standard. For a particular compound, the charging correction was constant to AO.1 eV using gold and 1 0 . 2 eV using molybdenum. The Mossbauer isomer shifts, estimated atomic charges on tin, and the average ligand electronegativities are also listed in Table I. The Sn(3dsiz)electron lines were 2.4 f 0.2 eV wide (FWHH) regardless of the method of standardization, while the Au(4f7pJ and M 0 ( 3 d ~ / ~electron ) lines were both 1.8 =t 0.2 eV wide (FWHH). The width of the Mo(3d5i2)electron line from pure M o o 3 and the Au(4f,ln) electron line from the gold standard were both 1.8 eV wide (FWHH). Therefore, it must be concluded that only singlet Mo(3dsiz) and Au (4f7pJ lines were observed from the samples. DISCUSSION Core-electron binding energies of a central atom have been shown to linearly correlate with the electronegativities of the bound ligands. Thomas (8) has found the C(1s) binding energies in the halomethanes to vary linearly with the sum of differences between the electronegativities of the ligands and that of hydrogen. Baker and coworkers ( 9 ) observed a linear relationship between the highest sigma (adiabatic)
486
1 2.00
-Y P
3.00
4.00
Figure 1. Sn(3ds/,) binding energies relative to M0(3ds/,) -I[. and Au(4fi,,) [_A-A-] us. average ligand electronegativity (X,) ionization potentials of HI, HBr, and HC1 and the electronegativities of the halides. Gelius et al. (18) have derived group electronegativities for organic functional groups using ESCA data and their results agree quite favorably with those arrived at by other means. Such correlations led us to believe that a linear relationship should exist between the tin (3d5/?) electron binding energies and the average Pauling electronegativities of the halide ligands in the [(CH3CH2)4N]2SnX6-nY , salts, Comparing the plot obtained using internally mixed standards with that using a vacuum-deposited standard should be a valid method of comparing the two calibration methods. Figure 1 is a plot of the Sn(3d512)binding energies in the tetraethylammonium tin halides measured relative to homogeneously mixed MOO3 and vacumm-deposited gold against the average Pauling electronegativities of the halide ligands. The drawn lines are the result of least squares analyses of the data. The equations of the lines are Eo = 0.934 487.4 and Eb = 1.09 484.1 for the data relative to Mo(3d51z) and Au(4f7;*), respectively. As seen in Table 11, the correlation coefficients are 0.800 and 0.896 for the A ~ ( 4 f , , ~and ) Mo(3dsIe) calibrations, respectively. The correlation using the internally mixed M o o 3 is considerably better than using vacuum-deposited gold. Chemical shifts in Mossbauer spectroscopy give a measure of the s-electron density at the nucleus; therefore, in a closely related series of compounds where stereochemistry and oxidation number of the central atom is constant, as in the tetraethylammonium tin halides, a relationship between Mossbauer chemical shift and core-electron binding energy should exist. Barber and coworkers (IO) have found a linear relationship between the tin (4d) electron binding energies and the Mossbauer isomer shifts for the series Y2Sn(ox)?; where ox = 8-quinolinolato, and Y = CHICHZ,CsH5, C1, Br, and I. The authors further point out that a linear relationship was found to exist between the sum of the Pauling electronegativities of Y and the Mossbauer isomer shifts (19), so that their work represents a further correlation between core-electron binding energy and ligand electronegativity. The tetraethylammonium tin halides meet the structural
w, +
x p+
(18) U. Gelius, P. F. Heden, J. Hedman, B. J. Lindberg, R. Manne, R. Nordberg, C. Nordling, and K. Siegbahn, Phys. Scr., 2, 70 (1970). (19) K. M. Ah, D. Cunningham, J. D. Donaldson, M. J. Frazer, and B. J. Senior, J. Chem. SOC.,A, 1969,2836.
ANALYTICAL CHEMISTRY, VOL. 44, NO. 12, OCTOBER 1972
2003
I
I
000
1.0
I
I
-2.0
-1.0
-3.0
(CIWS)
ISOMER SHIFT
Figure 2. Sn(3dr/,) binding energies relative to Mo(3d6/,) [-.-e-] and Au(4f~/,) [-A-A-] us. Mossbauer isomer shifts (Is)
linearly related to the calculated atomic charges. Huheey and Watts (22) have estimated the charges on the tin atom in the octahedral tin compounds investigated here. These charges reflect the charge induced on the tin atom by the halide ligands. These are by no means absolute charges, but are only relative charges for this series of compounds. Huheey and Watts (22) reported a linear correlation (correlation coefficient = 0.990) between these calculated charges and the Mossbauer isomer shifts observed for these compounds. This excellent correlation with experimentally determined parameters implies that the charges are good estimates of the partial atomic charges on the tin atoms. Therefore, it is very likely that a linear relationship should exist between these charges and the measured tin binding energies. The Sn(3d5,J binding energies relative to the Mo(3d512) and (Au4f712)electron lines are plotted in Figure 3. The lines are least square fits of the data. The equations of the lines are Eb = 5.706 486.4 and Eb = 6.196 486.2 for the data relative to MO(3d6,~)and Au(4f7,2), respectively. The correlation coefficients for the plot relative to Mo(3d5i2) is 0.927, while that relative to Au(4fTI2)is only 0.770. Again, the data relative to the internally mixed standard yield a much more linear correlation than do those relative to a deposited standard. The comparisons discussed above clearly imply that satisfactory standardization of ESCA spectra can be achieved by homogeneously mixing a standard material into the sample. In fact, it appears that such a mixing procedure may be as good as, if not better than, that where the standard is deposited on the surface. Hnatowich et al. (4) showed that deposition techniques enabled the standard to come to electrical equilibrium with the sample. Since our data using mixed-in standards correlate better in every instance, with sets of established data, we must assume that electrical equilibrium between standard and sample must also be reached using the mixing technique. Thus, the charge created on the surface by photoexcitation must necessarily be uniform. Vacuum deposition of gold does not occur evenly over the entire surface, but results in the deposition of “islands” of gold distributed randomly over the surface (23). The poorer correlations observed using the deposition technique may result from this random “island” character of the surface. When a standard is homogeneously mixed into the sample, it is implied that the ratio of sample to standard over a given area is constant. With deposition techniques, such is not the case. Thus, internally mixing in a standard may be a better way to reach electrical equilibrium. This study has shown that by homogeneously mixing a standard material into a sample, calibration of ESCA spectra can be achieved that may be better than other existing calibration techniques.
+
“t
A
4.9
I
0.00
I 0.10
I
I
0.2 0
I
I
0.30
I
I
0.40
Figure 3. Sn(3ds/,) binding energies relative to M0(3da/,) and Au(4fi/,) [-A-A-] us. estimated atomic charge (6)
-I[.
+
requirements defined by Barber and coworkers; therefore, one would expect a linear relationship to exist between the tin (3dsd binding energies and the Mossbauer isomer shifts measured for the compounds. Figure 2 is a plot of Mossbauer isomer shift us. the Sn(3d5iz) binding energies relative to both the Mo(3dsin) and Au(4f7,~)electron lines. Again, the lines are least square fits of the data. The equations of the lines are Eb = -0.639 IS 486.8 and Eb = -0.657 IS 486.5 for the data relative to MO(3dslz) and Au(4f7& respectively. From Table 11, it can be seen that the better fit is obtained using the internally mixed M o o B standard, the correlation coefficients being 0.736 and 0.936 for the Au (4f712) and Mo(3d51s)calibrations, respectively. A large volume of work has been reported on correlations between experimentally determined binding energies and theoretically calculated atomic charges. The methods employed in calculating the charges range from those using the partial ionic character of the bonds ( I ) , to EHMO (12) and CNDO (11, 13), to ab initio procedures (14,20, 21). In these correlations, the measured binding energies tend to be
RECEIVED for review February 28, 1972. Accepted June 15, 1972. Work supported in part by a grant from the Advanced Research Projects Agency, Department of Defense.
(20) M. Barber and D. T. Clark, J . Chem. Soc., D , 1970,23. (21) D. T. Clark and D. M. J. Lilley, Chem. Phys. Left., 9, 234 (1971).
(22) J. E. Huheey and J. C. Watts, Znorg. Chem., 10, 1553 (1971). (23) A. E. I. Scientific Apparatus Ltd., Manchester, United Kingdom, private communication, 1972.
+
2004
+
ANALYTICAL CHEMISTRY, VOL. 44, NO. 12, OCTOBER 1972
