107
Reaction of Evaporated Metal Films with Chlorine Gas Dependence of radical formation on benzoic acid concentration may be interpreted as follows. At low concentration of benzoic acid, the dimer concentration is small, while a t concentration greater than M , some coagulation and the self-quenching of the excited states of benzoic acid might be responsible for smaller radical yield. A correlation between the radical yields and the phosphorescence intensities in the slowly and rapidly cooled samples was obtained. This result might indicate that the triplet state of benzoic acid is responsible for the photosensitized radical formation in n-hexane. The monomer is known to give phosphorescence alone with a high quantum yield of 0.70,7 while the dimer emits both phosphorescence (ap = 0.46) and fluorescence (af = 0.25).6 The lifetime of the first triplet state of the dimer (3.1 sec) is longer than that of the monomer ( 2 . 2 sec). Therefore the lower quantum yield of the first triplet state of the dimer compared with that of the monomer is compensated for by its longer lifetime as a sensitizer involving triplet states. Another possibility may involve nature of the triplet state, i.e., (n,7r*)3 or ( 7 r , r * ) 3 . A further detailed study seems to be in order to clarify the nature of the excited states of the monomer and the dimer of benzoic acid. The effect of the cooling rate of solutions on the phosphorescence spectrum of aromatic molecules in isooctane was reported.8 Changes of the phosphorescence spectrum and the decay time of the triplet states were explained by the substantial difference of the crystal fields in which the
impurity molecules are located. The same explanation may also be applicable to our present case. On the other hand, it is also reported that the equilibrium between the monomer and the dimer of benzoic acid inclines toward the monomer side a t -196".'j In this case there is a possibility that the cooling rate can affect this equilibrium. In fact the phosphorescence spectrum of the rapidly. cooled sample was that of the dimer, while the slowly cooled sample gave the spectrum mainly of the monomer with some contribution from the dimer (see Figure 5). These arguments also support our interpretation that the dimer of benzoic acid is more efficient than the monomer form in the photosensitized radical formation in n-hexane.
Acknowledgment. The authors are indebted to the Sumitomo Chemical Co. for permission to publish these results. References and Notes (1) (2) (3) (4)
V. E. Kholmogorov, Russ. Chem. Rev., 38.164 (1969).
M. Ito. J. Mol. Spectrosc., 4, 144 (1960). T. Miwaand M. Koizumi, Bull. Chem. SOC.Jap., 38, 529 (1965). T. Takeshita, K . Tsuji, and T. Seiki, J. Polym. Sci.. Part A - 7 , 10, 2315 (1972). (5) T. Gillbro, P-0. Kinell, and A. Lund, J. Phys. Chem., 73, 4167 (1969). (6) H. Babaand M. Kitamura, J. Mol. Spectrosc., 41,302 (1,972). (7) J. 6. Birks, "Photophysics of Aromatic Molecules, Wiley-lnterscience, New York, N. Y . , 1970. (8) D. M . Grebenshchikov, N. A. Kovrizhnykh, and S. A. Kozlov, Opt. Spectrosc., 31, 214 (1971).
X-Ray Photoelectron Spectroscopic Study of the Reaction of Evaporated Metal Films with Chlorine Gas Kosaku Kishi* and Shigero lkeda Department of Chemistry. Faculty of Science. Osaka University, Toyonaka, Osaka. Japan (Received April 24. 7973: Revised Manuscript Received July 5. 7973)
The reaction of the evaporated metals iron, nickel, copper, palladium, silver, and gold with chlorine gas was investigated by X-ray photoelectron spectroscopy. Three types of chlorine species were observed on the surfaces as the surface reaction proceeded, namely the chlorine ion adsorbed at the metal surface, the ion of the surface metal chlorides, and Clz or C1 adsorbed on the chlorides.
Introduction Bonding character of adsorbent atoms on metal surfaces will vary as the localization of the atom from surrounding metal atoms in the course of a surface reaction. Investigation of such variations helps to elucidate the reaction mechanism and provides fundamental knowledge for analyses of corrosive, electrode, and catalytic reactions on metal surfaces. X-Ray photoelectron spectroscopy (XPS or ESCA) is a useful tool for the investigation of chemical reactions of solid surfaces since the effective sampling depth by XPS is smaller than 100 A. In our previous papers, the reac-
tions of evaporated iron' and nickel2 with 0 2 and H20 were studied by XPS. The surface oxides and the corresponding bulk oxides gave different photoelectron spectra and differed in their chemical reactivity with H2O. Two kinds of oxygen ions were shown to exist in the surface oxides of these metals. Similar differences in chemical reactivity were also observed by uv absorption spectra when acetylacetone was exposed to iron and manganese surfaces and bulk oxides.3 In the present paper, the reactions of evaporated iron, nickel, copper, palladium, silver, and gold with chlorine gas (which has higher reactivity than oxygen to noble The Journal of Physical Chemistry, Vol. 78, No. 2, 1974
108
Kosaku Kishi and Shigero lkeda
metals) were investigated by XPS in a continuing effort to clarify the difference in chemical states between surface and bulk compounds.
Experimental Section Metals were evaporated onto a stainless steel plate from an electrically heated tungsten filament a t Torr in a reaction chamber as described elsewhere.' The evaporated metal film on the plate was then transferred in U ~ C U Ointo a sample chamber. After the spectrum of the metal was recorded, the surface was exposed to chlorine gas in the reaction chamber. Iron and nickel surfaces were exposed to the gas immediately after evaporation without recording the spectrum of the unreacted metal. The gas was then evacuated and the photoelectron spectra of the surface species formed were recorded. Copper was not used as a reaction chamber or sample holder material since the sample surface was covered immediately with CuCl sublimed from these copper surfaces after they were exposed to chlorine gas. During the measurement of the spectra, the vacuum of the sample chamber was maintained a t 10-7Torr. Spectra were recorded in 0.1-eV steps on a KES-X2001 (Kokusai Electric Co.) electron spectrometer using A1 Kcu X-ray radiation. The instrument was calibrated so that the difference between the photoelectron peak of Cls observed in the background spectrum and the Fermi level of palladium was 285.0 eV. The Fermi level was taken as the inflection point of the steep low-energy edge of the 4d band from palladium according to Baer, et aL4 Binding energies of other peaks were calculated using the Cls reference line. For the chlorides produced on the metals, charging effects were negligible under X-ray bombardment.
Results The C1 2p photoelectron lines from evaporated gold exposed to Clz are shown in Figure 1. The background spectrum for gold in the C1 2p region is indicated as (a). When gold was exposed to Clz a t a pressure of 0.5 Torr for 1 min, three peaks were observed at binding energies of 197.0, 199.4, and 201.1 eV as shown by (b). Chlorine 2p core levels appear as a doublet splitting a t ca. 1.7 eV (2p1,2 and 2 ~ 3ratio , ~ 1:2) due to spin-orbit coupling. Line b can be resolved into two doublets when the profile is analyzed on the basis of the spectral line of the metal chlo~ 2which are a t rides, in area ratio of ca. 5:4, the 2 ~ 3 of 199.4 and 197.0 eV, respectively, as shown by dotted lines in the lower part of the figure. Gold was exposed, in succession, to Clz a t a pressure of 10 Torr for 2 rnin (c) and for 10 min ( d ) . Lines e and f were obtained by subtracting (b) from (c), and (c) from ( d ) , respectively, in order to clarify spectral variations for each reaction step. Both lines show two maxima, around 199.3 and 201.1 eV, with a weak shoulder around 197.6 eV. When gold was exposed to Cl2 a t a pressure of 10 Torr for 10 min, the Au 4f5,2 and 4f712 lines appeared as very weak shoulders shifted +2.0 eV from those for the unreacted metal. In cases of milder reaction conditions the Au 4f lines were the same in shape as those for the fresh metal except for a decrease in counting rate. The C1 2p and the shifted Au 4f peaks decreased in intensity with time (during the spectral measurement), probably due to the decomposition of the surface species by X-ray irradiation. Therefore, the C1 2p or Au 4f line The Journal of Physical Chemistry, Vol. 78, No. 2, 1974
,,' 204
200
Binding Energy
'
'.
196
(eV)
Figure 1. Ci 2p photoelectron lines from evaporated gold: (a) exposed to CIz; (b) 0.5 Torr, 1 min; (c) 10 Torr, 2 min; (d) 10 min; (e) c minus b; (f) d minus c. The lower figure shows two doublets.
Binding Energy
(eV)
Figure 2. CI 2p photoelectron lines from evaporated silver exposed to CIz: (a) 0.1 Torr, 1 min; (b) 1 Torr, 2 min.
was recorded, as quickly as possible (within 20 min) after starting X-ray irradiation. The order of measurement was either C1 2p, Au 4f, C1 2p, or Au 4f, C1 2p, Au 4f. The spectra given in the figures are therefore the averaged ones. For the other metals described below, the same measurement procedure was used. When evaporated silver was exposed to Clz at a pressure of 0.1 Torr for 1 min, the C1 2~112and 2 ~ 3 1 2peaks were located a t 198.8 and 197.2 eV (Figure 2a). Silver was subsequently exposed to Cl2 a t a pressure of 1 Torr for 2 min. The C1 2p peaks shifted +0.5 eV to higher binding energy. The silver 3d peaks shifted to lower binding energies on oxidation with chlorine and no new peak was observed. When evaporated palladium was exposed to Clz at a pressure of 0.5 Torr for 30 sec, Pd 3d peaks were the same in shape as those for the unreacted metal except for a 5% decrease in counting rate. The C1 2p electrons gave two peaks a t 197.8 and 199.3 eV with a shoulder on the higher binding energy side of the peaks (Figure 3a). The C1 2p line was resolved into two doublets (dotted lines in the lower part of the figure) having the 2~312peak maxima a t 197.8 and 199.3 eV, respectively. The metal was subsequently exposed to Cl2 a t a pressure of 10 Torr for 1 min ( b ) and for 5 min (c). The C12p3,2 peak at 199.0 eV was intensified, changing the peak area ratio of the two C1 2p
109
Reaction of Evaporated Metal Films with Chlorine Gas
Figure 3. CI 2p photoelectron lines from evaporated palladium exposed to CI2; (a) 0.5 Torr, 30 sec; (b) 10 Torr, 1 min; (c) 10 Torr, 5 min; (d) b minus a; (e) c minus b. The lower figure shows two doublets.
Figure 5. CI 2p photoelectron lines from evaporated nickel exposed to CI2: (a) 0.5 Torr, 30 sec; ( b ) 10 Torr, 2 min; (c) 10 Torr, 10 min; (d) b minus a; (e) c minus d .
1 Pd 3d 0
c
B
0,
.-c
E
6
344
340
Binding Energy
336
864
860
856
852
Binding Energy (eV)
(eV) Figure 4. Pd 3d photoelectron lines from the same samples as Figure 3. Minimum and maximum counts/5 sec are written in Daren theses.
Figure 6. Ni 2 p 3 p photoelectron lines from the same samples of Figure 5: - - - , NiCI2. Minimum and maximum counts/5 sec are written in parentheses.
lines, and new P d 3d peaks appeared, shifted +2.5 eV from the unreacted ones (Figure 4b,c). When copper metal was exposed to Cl2 at a pressure of 0.5 Torr for 10 sec, the C1 2p electron counting rate became very high. A difference between maximum and minimum counts of 7900 counts/5 sec was observed, compared with about 2000 counts for other metals. The copper 2p312 peak did not shift and had the same peak width as the metal alone, while the spectrum in the valence electron region changed completely to that of CuC1. Figures 5 and 6 were obtained from evaporated nickel exposed, in succession, to Cl2 at a pressure of 0.5 Torr for 30 sec (a), 10 Torr for 2 min (b), and 10 Torr for 10 min (c). The C1 2p line of (a) shows two maxima a t 198.1 and 199.5 eV. The intensity of the left (199.5-eV) peak is too strong to consider the two peaks as a doublet for one kind of chlorine species, indicative of the presence of another doublet on the high-energy side. Lines d and e of Figure 5 were obtained by subtracting (a) from (b), and (b) from (c), respectively. The maximum of (d) is at 198.7 eV, being different from that for (e) a t 199.7 eV and indicating adsorption of a different kind of chlorine species for
the two reaction steps, in contrast to (e) and (f) of Figure 1 for gold where the two maxima are at the same binding energy of 199.3 eV. The Ni 2p3/2 lines of (a), (b), and (c) of Figure 6 indicate the appearance of a shifted peak (+3.3 eV from that for the metal). In (b) and (c), a satellite peak was observed at 862.3 eV which can be assigned to the peak due to “shake up” in nickel chloride. The peak a t 859.8 eV in (a) is due to plasmon energy losses for metallic nickeL4 The dotted curve shows the spectrum for NiClz powder. When evaporated iron was exposed to Clz a t a pressure of 10 Torr for 10 sec, a shifted Fe 2~312peak appeared, +3.3 eV from the unreacted peak. The spectral variation in the C1 2p lines were similar to that in the case of nickel. The binding energies of the C1 2 ~ 3 ~and 2 the respective metal core electron peaks observed above are listed in part in Tables I and 11, with the peak intensities (counts/ 5 sec) for C1 2 ~ 3 1 2obtained by subtracting the background counting rate from the peak maximum counting rate. The corresponding binding energies for the bulk compounds are also listed. The Journal ofPhysical Chemistry, Vol. 78, No. 2, 1974
110
Kosaku Kishi and Shigero lkeda
TABLE I: Binding Energies (eV) of C1 2p3 Peaks for Metals Exposed to Cli ~
Fe
After slight exposure t o Cl? After further exposure t o Cl? Bulk compounds
"
Ni
CU
Pd
Ae
198.4
197.8
197.2
(800)('
(600)
(800)
199.4
Au
197 . O (600) 199.4
199 . 0 (4700) 199.9
198.3 (3200) 199.7
198.6 (7900)
199 .o (2300) 197.8
197.7 (3600)
199.4 (2100) 197 . O
199 . 0 (FeCl?) 199.2 (FeCI:,)
198.5 (NiCI?)
198.6 (CuCl) 199.6 (CUCl2)
198.7 (PdCl?)
197.9 (AgCl)
198.6 (AuCI) 199 .o (NaAuCI4)
Counting rates in the parenthescs were obtained by subtracting minimum counts from maximum ones (countsi5 sec).
TABLE 11: Chemical S h i f t s f r o m Metals feV) Fe2p7 .
Metals exposed to c1, Bulk compounds
Cu 2pl >
Ni2p3
P d 3d5
Ag 3ds / %
+3.3
+3.3
0
+2.5
-0.3
+3.4
+3.6 (NiCI?)
-0.3 (CUCI) +3.0 (CUCI?)
$2.6 (PdClr)
-0.3
(FeCI?) +4.7 (FeCl: