AES study of the interaction of dimethyl methylphosphonate

Sep 1, 1986 - ACS Legacy Archive. Note: In lieu of an abstract, this is the article's first page. Click to increase image size Free first page. View: ...
2 downloads 8 Views 563KB Size
J . Phys. Chem. 1986, 90, 4607-4611

4607

evolved on the Pt extreme and O2at the T i 0 2 end. A volume of 0.52 mL of H2 (identified by gas chromatography) evolved, and an H2/02 molar ratio of 2.4/1 was obtained. The deficiency in the amount of O2generated as compared to that expected from the decomposition of H 2 0 on T i 0 2 has been attributed to the formation of peroxide^.'^-'^

-,

. -

Number o f Panels

Figure 9. i,Ji=, vs. number of panels (1 cm2) in 1 M KOH.

where the two ends of the cell (KOH solutions [ l ] and [3]) are connected by a KOH salt bridge. The predicted behavior of this cell can be derived from addition of two A1 curves to yield A2 and the addition of 0 1 to H1 to yield OlHl (Figure 7). The rate of H2 and O2evolution can be estimated from the power curve derived by addition of these. The characteristics of a three-panel cell can similarly be derived by consideration of curves A3 and 0 2 H 1 . Note that the predicted i , for the two-panel cell is less than the radiation-limited current (isat),but for the three-panel cell is, = 0.85imt. A plot of isc/imtvs. number of panels, shown in Figure 9, suggests that, for conversion of solar energy to H2 and 02,there is little to be gained in using a configuration with more than three panels. However, to test the operation of a multiple-panel PEC cell for water splitting, a fivepanel cell, shown in Figure 3, was illuminated for 5 h with a Xe lamp (266 mW/cm2 incident). Gas evolution occurred at each end electrode, H2 being

Conclusions Vectorial charge transfer on bielectrode panels has been demonstrated. Means have been established for coupling these panels to produce higher driving forces than those available from systems with single semiconductor electrodes and for predicting the PEC characteristics from individual i-Y curves. The unassisted photolytic water-splitting reaction has been demonstrated. N o effort was made in these studies to optimize the behavior of the PEC cells through improvement of the T i 0 2 film, new interior redox couples, or better mass transport. Such improvements and alternative semiconductor materials are currently under investigation in these laboratories. Acknowledgment. The support of this research by the Gas Research Institute is gratefully acknowledged. Registry No. Ti02, 13463-67-7;KOH, 1310-58-3;Ti203,1344-54-3; TiO, 12137-20-1; H20, 7732-18-5; H2,1333-74-0;02,7782-44-7; Pt, 7440-06-4; Ti, 7440-32-6. (16) Duonghong, D.; Gratzel, M.J . Chem. SOC.,Chem. Commun. 1984. 23. 1597. (17) Muraki, H.; Saji, T.; Fujihara, M.; Aoyagui, S. J. Elecrroanal. Chem. 1984, 169, 319.

A TPD/AES Study of the Interaction of Dimethyl Methylphosphonate with a-Fe,03 and si02

M. A. Henderson, T. Jin, and J. M. White* Department of Chemistry, University of Texas, Austin, Texas 78712 (Received: February 14, 1986)

The interaction of dimethyl methylphosphonate(DMMP) dosed at 170 K onto Si02and a-Fe2O3was studied by temperature programmed desorption (TPD) and Auger electron spectroscopy (AES). On dehydrated S O 2 there was no DMMP decomposition and there were two DMMP TPD peaks, a multilayer state at 200-210 K and a monolayer state at 275 K. On hydrated S O 2 no more 10% of a monolayer of DMMP decomposed and the only detectable TPD products were methylphosphonate (MP) and methanol. On clean a-F%03multilayer DMMP was observed but no molecular peak corresponding to the monolayer. Decomposition led to C02, CO, CH30H, HCOOH, H2, H20, and an adsorbed phosphate species. The presence of the phosphate species did not completely inhibit the decomposition of subsequent doses of DMMP even after saturation of the AES P(114)/Fe(703) signal ratio at 0.45 to 0.55. The AES P(114)/Fe(703) signal ratio decreased at temperatures above 600 K. Migration of the phosphorus to an iron oxide-phosphate phase below the surface is proposed to explain the continued DMMP decomposition.

Introduction The purpose of this paper is to report recent findings in this laboratory for the interaction of dimethyl methylphosphonate (DMMP) with the surfaces of SOz and a-Fe203. In previous work involving the adsorption of D M M P on clean and carboncovered Rh( loo),' the bonding of DMMP to the Rh surface was assigned to interactions through the lone pair of electrons on the oxygen of -P=O. Only a small fraction of the molecularly adsorbed DMMP desorbed from an initially clean Rh( 100) surface during TPD experiments. The remainder decomposed to small molecules including H2,H 2 0 , CO, C 0 2 , CH4, and CH30H. In addition to these decomposition products, AES at 5 0 0 K detected the presence of P, C, and 0 on the surface following DMMP (1) Hegde, R.I.; Greenlief, C. M.; White, J. M.J. Phys. Chem. 1985,89, 2886.

decomposition. In contrast to the clean Rh(100) surface, no DMMP decomposition was observed on a carbon-covered Rh(100) surface. In related work, Weinberg and Templeton have recently studied the interaction of DMMP with A1203 using inelastic electron tunneling spectroscopy (IETS).2 They observed the stepwise decomposition of chemisorbed DMMP dosed at 200 K to methyl methylphosphonate (MMP) above 295 K and to methylphosphonate (MP) above 573 K, each step resulting in the loss of a methoxyl group. In this work, we have studied the interaction of DMMP with the surfaces of S O 2and a-Fe2O3 using temperature programmed desorption (TPD) and Auger electron spectroscopy (AES). (2) (a) Templeton, M.K.; Weinberg, W. H. J. Am. Chem. Soc. 1985, 207,

97. (b) Ibid. 1985, 107, 774.

0022-3654/86/2090-4607$01.50/00 1986 American Chemical Society

4608

The Journal of Physical Chemistry, Vol. 90, No. 19, 1986

Experimental Section The ultrahigh-vacuum (UHV) system for TPD and AES was similar to that described el~ewhere.~DMMP was dosed into the sample in one UHV chamber and, by means of a linear motion transfer rod, the sample was then moved into a separate UHV chamber for line-of-sight TPD and AES (using a single-pass CMA with primary beam energy of 3 keV and a spot size of about 1 mm in diameter). Both chambers operated with working base Torr. pressures of 5 X The substrate a-Fe203 was prepared by treating a 10 wt % solution of Fe(N03)3.9H20 with NHIOH which precipitated a-FeO(0H). This solid was filtered and washed several times with deionized water before being dried in an oven at 373 K for 24 h. The solid was then calcined in air at 800 f 25 K for 3 h, resulting in its conversion to a-Fe203(confirmed by X-ray powder diffraction) with a surface area of 16 f 2 m2/g. Silica was obtained from Cab-0-Si1 with a surface area of 200 m2/g. Thin samples were prepared by pressing the powders into a tantalum mesh (1 cm2) that had been spotwelded to two tantalum heating/cooling leads. Sample weights were between 10 and 20 mg. AES of these samples at various places on their surfaces gave no Ta signals indicating that the mesh was adequately covered by each oxide. Samples were heated by resistively heating the mesh, and the temperature was monitored by a chromel-alumel thermocouple spotwelded to the Ta mesh. Commercial spectroscopic grade DMMP was further purified by a series of freeze-pump-thaw cycles. At 298 K DMMP is a liquid with a vapor pressure of approximately 1.06 Torr.] Under these conditions DMMP vapor from a differentially pumped reservoir was dosed through a leak valve into the dosing chamber. Contributions in the TPD (decomposition products) due to the Ta mesh/leads were minimal on the basis of data obtained for DMMP on SiO, (see below). The pressure in the dosing chamber was maintained at 5 X lo-' Torr during DMMP exposures. Typical experiments involved dosing DMMP onto an oxide at 170 K followed by thermal desorption (with a heating ramp of 2 K/s) or AES analysis from 170 to 700 K. The desorption of DMMP was followed by monitoring the parent ion m / e 124. Assignments of desorbing species were based on detailed analysis of gas-phase fragmentation patterns of candidate molecules. In most cases coincidental m / e fragments from different species were distinguishable on the basis of their different temperature dependences in the TPD. The temperature difference between the Ta mesh and the oxides was small; the samples were carefully pressed to get good thermal contact, the heating rate was relatively slow, the mass of the Ta mesh was roughly equal to that of the oxide samples, and, most importantly, the DMMP mutlilayer desorption peaks were at the same temperature as those observed when DMMP was desorbed from a metal foil. Results and Discussion Silica. The adsorption of DMMP on dehydrated (by annealing at 750 K in UHV) SiO, at 170 K yielded two parent TPD peaks, the first corresponding to a multilayer state at 200 to 210 K and the second to a monolayer state at 275 K (Figure 1). The monolayer peak exhibited first-order desorption kinetics and, s-l, the activation energy assuming a preexponential factor of of desorption was ~ 1 6 . kcal/mol. 9 The monolayer state saturated at a DMMP exposure of about 30 langmuirs. Multilayer DMMP was observed before the monolayer state saturated because the particle nature of the oxide made some areas more accessible to the gas phase than others. A plot of the DMMP TPD peak area vs. the exposure time (Figure 2) was linear and intersected the origin. No decomposition products (such as CH,OH, CO,, CO, H 2 0 , or phosphorus-containing molecules) were observed in the TPD and no AES C(275) or P(120) signals were present at temperatures above the monolayer DMMP TPD peak. Thus, there was no detectable DMMP decomposition on the silica surface. The absence of measurable quantities of decomposition (3) Beck, D. D.; White, J. M. J . Phys. Chem. 1984, 88, 2764

Henderson et al.

400

200

600

TEMPERATURE ( K )

Figure 1. TPD of DMMP ( m / e 124) dosed at 170 K on SO2at various exposures: (a) 2.5, (b) 15, (c) 30, (d) 45, (e) 60 langmuirs.

DMM P/ S i0 2

h

v)

t z a

ai K

a " a w

a

U Y

a

w

a 0

n I-

%

fi

I

10

20

30

DMMP EXPOSURE (L) Figure 2. DMMP TPD peak area vs. exposure for DMMP dosed at 170 K on SO2.

products also indicated that the Ta leads did not contribute to the TPD results. The lack of any decomposition implies that the mode of bonding to the surface was weak. Since silica is relatively inert in terms of its acid/base properties, substitution of a methoxyl group by a surface oxygen or hydroxyl was unlikely. Figure 3 shows the AES spectra of clean silica at 700 K, of dehydrated silica at 273 K after a 60-langmuir exposure of DMMP a t 170 K, and of the same sample heated in UHV to 400 K. Charging of the oxide by the electron beam prevented AES measurements below 273 K. The line shape of the phosphorus signal in Figure 3b consisted of a single peak at 120 eV. This was of particular interest in light of results obtained on a-FezO3 (which will be discussed below). This line shape was assigned to P in molecular DMMP. The line shape was very similar to that of multilayer DMMP on oxidized polycrystalline Fe.4 Bernett et aL5 have observed that the AES line shape of phosphorus is R. I.; White, J. M.,manuscript in preparation. (5) Bernett, M.K.; Murday, J. S.;Turner, N. H. J . Electron Spectroc. Related Phenom. 1971, 12, 315. (4) Hegde,

The Journal of Physical Chemistry, Vol. 90, No. 19, I986 4609

Interaction of DMMP with a-Fe203 and SiOz

100

300

500

300

100

500

100

300

500

KINETIC E N E R G Y (eV)

Figure 3. AES of (a) clean S O 2 at 700 K, (b) Si02 at 273 K after exposure to 60 langmuirs of DMMP at 170 K, and (c) after heating to 400 K.

U

/

A

/

I 200

1 400

800

TEMPERATURE (K)

Figure 4. (a) TPD of

H20( m / e 18) from a 5-min exposure of S O z to atmosphere at 298 K. (b) TPD of D20( m / e 20) from a 2-min exposure of Si02to 1.0 Torr of D20 at 298 K. very sensitive to the degree of oxygen coordination to the phosphorus atom. Phosphorus species which are fully oxygen coordinated (as in a phosphate or phosphite) have line shapes consisting of a doublet at 96 and 114 eV, while P species which contain any P-C (as in DMMP) or P-N bonds possess phosphide-like line shapes with a single peak at 120 eV. The phosphide-like line shape of molecularly adsorbed DMMP on S i 0 2 supports these conclusions. Electron beam decomposition of adsorbed DMMP on SO2 was ruled out on the basis of the following two observations. (1) At a given temperature, the P line shape and intensity did not change with electron beam exposure, ruling out beam-induced losses of P and conversion to either a phosphate or phosphite. (2) The P(12O)/Si(76) and C(275)/Si(76) AES signal ratios taken as a function of temperature from a single spot attenuated to zero (Figure 3c) in parallel with the desorption of DMMP (Figure 1). If electron beam decomposition leaving P were important, the remaining species would not be expected to have the same thermal desorption characteristics as DMMP. Additional experiments were carried out on silica samples that had been hydrated at 298 K in an ambient atmosphere before exposure to DMMP. These silica samples exhibited a continual and intense water desorption starting at 300 K and extending above 700 K (Figure 4a). The D M M P TPD of these samples was identical with that of the dehydrated samples except that a small amount of methanol and methylphosphonate (MP) desorbed at 650 K. The small amount of decomposition probably resulted from the hydration of the P-O-CH3 bonds of DMMP by weakly adsorbed H20. A decrease in the D M M P peak area by about 8-1076 of the monolayer saturation coverage indicated that hydration had only a small effect. This small amount of decomposition required extensive hydration of the surface as no de-

200

300

400

TEMPERATURE ( K )

Figure 5. TPD of DMMP ( m / e 124) dosed at 170 K on a-Fe203at various exposures: (a) 2.5, (b) 15, (c) 45, (d) 60 langmuirs.

composition was observed when it was partially hydrated by D 2 0 exposures of 2 min at pressures from 1 X lo6 to 1 Torr. The hydration in these cases was limited to dissociated and/or strongly chemisorbed water molecules (Figure 4b) that were not able to hydrolyze the P-O-CH3 bonds. In no case did hydration significantly either poison or enhance DMMP adsorption. Thus, DMMP adsorption on hydrated silica must involve either displacing weakly adsorbed water or adsorption at remotely separate sites not influenced by preadsorbed water. Clean a-Fe2O3. In contrast to results obtained on silica, DMMP exposed to clean a-Fe203 yielded only one DMMP TPD peak located in the multilayer desorption region of DMMP/Si02 at 200 to 210 K (Figure 5 ) . N o molecular peak was observed for exposures of less than 2.5 langmuirs at 170 K (total decomposition), while at exposures above 30 langmuirs the molecular desorption peak increased rapidly. Under no conditions was a distinct higher temperature monolayer peak observed. Various DMMP decomposition products were observed in the TPD including C 0 2 (586 K), CO (586 K), H2 (595 K), C H 3 0 H (600 K), H 2 0 (>600 K), and HCOOH (550 K), the first three being the most significant. However, no phosphorus-containing decomposition products were observed. Because of cooling limitations, analysis of decomposition products below 170 K was not possible. There were signs of both methanol and water desorption below 190 K, indicating that some DMMP decomposition may occur upon adsorption at 170 K. Similar observations have been made for DMMP adsorbed on oxidized polycrystalline iron! The ability of iron oxide to oxidize adsorbed C H 3 0 H to C02, CO, and H2 is well-known. Zavakil et aL6 have shown that adsorbed C H 3 0 H or H 2 C 0 on a-Fe2O3 is completely oxidized (by lattice oxygen) above 493 K to CO, C02,and HZ In the case of CH30H, very small quantities of H 2 C 0 are observed after long periods of heating. Iron oxide does not oxidize methanol to formic acid unless it is mixed with a more acidic oxide like Moo3.’ The presence of significant amounts of HCOOH as a DMMP decomposition product is thus very interesting. Figure 6 shows the TPD peak areas for both DMMP (m/e 124) and HCOOH (m/e 46) vs. DMMP exposure on clean a-Fe203. Peak areas for the other decomposition products could not be determined because of their location at the end of the temperature ramp. The HCOOH TPD peak area levels off at the same exposure (about 40 langmuirs) where the DMMP TPD peak area begins to increase rapidly due to distinct multilayer formation. The AES spectra of clean a-Fe2O3 and a-Fe2O3 heated to 247 K after exposure to 60 langmuirs of DMMP at 170 K are shown in Figure 7. As with S O 2 ,charging of the oxide by the electron beam prevented AES measurements below 247 K. The P line shape differs from that of DMMP/Si02 (Figure 3). The two peaks at 96 and 114 eV are indicative of a fully oxygen-coordinated (6) Novakova, J.; Jim, P.; Zavadil, V. J . Catal. 1971, 21, 143. (7) Ai, M. J . Catal. 1978, 54, 426.

Henderson et al.

The Journal of Physical Chemistry, Vol. 90, No. 19, 1986

4610

1.0

0MMPh-Fe2O3

I

DMMP ( m / e = 1 2 4 )

0

1

P(114)/Fe(703)

h

m

k Z

3

ai

n a v a w a a

O L , , 250

' 300

'

a

400

'. 500

-- ,

800

_ ,

700

TEMPERATURE (K)

Y

U w

Figure 8. AES peak-to-peak height ratios vs. sample temperature for a 60 langmuir exposure of DMMP on cu-Fe,03 at 170 K.

a Q

n I-

SCHEME I

20

60

40

DMMP EXPOSURE (L)

Figure 6. DMMP and HCOOH TPD peak areas vs. exposure for DMMP dosed at 170 K on a-Fe203. ,

,

.

,

,

.

,

. , . . . . . .

,

a

b

x

500-600 K

OH(a), O(a) 0

200

400

600

800

200

400

600

COa, CO, CH30H, HCOOH, H20, H2

800

KINETIC ENERGY ( e V )

Figure 7. AES of (a) clean cu-Fe,O, at 723 K and (b) a-Fe,03 at 247 K after exposure to 60 langmuirs of DMMP at 170 K. r 5 0

phosphorus species, as in a phosphate or ph~sphite.~ As mentioned previously, the phosphide-like line shape of molecular DMMP (Figure 3b) is attributable to the methyl ligand. It appears that upon heating to 247 K the P-CH, bond was oxidized to a fully oxygen-coordinated phosphorus species, the oxidizing agent being the iron oxide surface. The possibility that electron beam decomposition resulted in the altered phosphorus line shape is precluded on the basis of results mentioned previously for DMMP on silica. Additionally, Hedge and White have shown that electron-beam-induced decomposition of multilayer and chemisorbed DMMP on oxidized polycrystalline Fe is minimal or nonexi~tent.~ Thus the presence of the doublet in the phosphorus AES line shape results from oxidation of the P-CH, bond and not from electron beam damage. Hegde and White have also observed the same P line shape change upon heating DMMP adsorbed on oxidized polycrystalline Fe at 100 to 400 K. These results are in contrast to those obtained by Weinberg and Templeton' for DMMP adsorbed on A1203where the P-CH3 bond remained intact even to >573 K. Evidently, A1,03 does not possess

P

7 / / //// the capacity to oxidize this bond at low temperatures. A plot of the C(275)/Fe(703) and P(114)/Fe(703) AES signal ratios taken from the same spot on the sample vs. temperature for a clean a-Fe20, sample exposed to a 60-langmuir DMMP dose at 170 K is shown in Figure 8. The C(275)/Fe(703) signal ratio dropped to zero at about the same temperature that desorption of all carbon-containing D M M P decomposition products began in the TPD. At significantly higher temperatures (600-650 K), the P( 114)/Fe(703) ratio decreased from about 0.5 to 0.25. These observation are consistent with Scheme I. Adsorption of DMMP at 170 K on clean a-Fe203followed by heating to 250 K resulted in decomposition that included the oxidation of the P-CH, bond (as evidenced by the AES line shape change). Around 600 K there was further decomposition and desorption of all carbon-containing

The Journal of Physical Chemistry, Vol. 90, No. 19, 1986 4611

Interaction of D M M P with a-Fe203and SiO,

. . . . . . . b

l a

‘120 eV

L 200

400

4

710 K

800

KINETIC E N E R G Y ( e V )

Figure 9. (a) AES of cu-Fe203(P) at 250 K after exposure to 30 langmuirs of D M M P at 170 K. (b) P AES line shape changes at various temperatures for the sample in (a).

species in the form of C H 3 0 H , C 0 2 ,CO, and HCOOH, some of which were oxidized by lattice oxygen. Above 650 K phosphorus migrated into the bulk of the oxide (since no phosphorus species was observed to desorb other than DMMP) causing a decrease in the P(114)/Fe(703) signal ratio. Repeating multilayer DMMP TPD evidenced decomposition despite the presence of the phosphate species from previous exposures. Thus, the surface was partially “cleaned” by the migration of phosphorus into the bulk of the iron oxide and surface sites were regenerated for further D M M P adsorption and decomposition. By comparison, Tanabe et a1.,8 using IR, have shown that adsorbed SO3 on a-Fe203 migrates into the oxide and forms bulk sulfate upon heating above 573 K. It is known that the chemistry of phosphates and sulfates on a-FeO(0H) is very ~ i m i l a r it ; ~is quite reasonable then to suggest that bulk phosphate is formed on a-FezO3. After TPD/AES experiments of over 400 langmuir total DMMP exposure in 12 cycles, the P(114)/Fe(703) AES signal ratio became saturated at 0.45 to 0.55 corresponding to a P/Fe atomic ratio of about 1/5.1° This signal ratio then remained relatively constant even after treatment with 1 X lod Torr of 0, or H 2 at 700 K for prolonged periods. The AES P(114)/0(503) signal ratio indicated an atomic P/O ratio of about 1/8. At this point the TPD/AES properties of the sample changed (as discussed below), and the sample lost its characteristic rust color and turned black. W e will refer to this sample as phosphated and denote this by a-Fe203(P). Phosphated a-Fe2O3. After saturation of the AES P( 114)/ Fe(703) signal ratio the TPD characteristics of DMMP/a-Fe203 exhibited a major change. A very broad DMMP monolayer peak (8) Jin, T.; Yamaguchi, T.; Tanabe, K., submitted to J . Phys. Chem. (9) (a) Parfitt, R. L.; Russell, J. D.; Farmer, V. C. J. Chem. SOC.,Faraday Trans 1 1976, 72, 1082. (b) Parfitt, R. L.; Smart, R. S. C. J . Chem. SOC. Faraday Trans 1 1977, 73, 796. (10) AES sensitivity factors for a 3-keV primary electron beam were obtained from Handbook of Auger Electron Spectroscopy, 2nd ed, Davis, L. E., et al., Ed.; Physical Electronics Industries Inc: Eden Prairie, MN, 1976. Using these factors provides only a semiquantitativeanalysis of atomic ratios since factors such as chemical state may alter their absolute values.

centered at about 410 K was present in the TPD. This monolayer state was saturated by a 15-langmuir dose of DMMP. Smaller quantities of HCOOH, C 0 2 ,CO, H2, H 2 0 , and C H 3 0 H in the TPD indicated that some DMMP decomposition persisted but only at about 20% of its original level. A plot of the DMMP TPD peak area vs. exposure intersected near the origin rather than at high exposure as observed for DMMP/a-Fe203 (Figure 6). Thus, the P-saturated surface had lost most of its ability to oxidize the P-CH3 bond or hydrate the P-0-CH3 bonds. Figure 9 shows the AES of a-Fe203(P) at 250 K after a 30langmuir dose of DMMP at 170 K. The line shape changes in the P signal upon heating support the idea that there are two chemically different phosphorus species present. The signals at 94 and 114 eV correspond to the phosphate species at or below the surface, and the signal at 120 eV corresponds to chemisorbed DMMP. The latter disappeared when the sample was heated above 573 K in accordance with TPD data for the monolayer state mentioned above. The line shape of this chemisorbed DMMP was very similar to that for DMMP/Si02 (Figure 3). Similar observations were made by Shafrin and Murday” after exposure of tricresyl phosphate (TCP) to steel bearings at 383 K for various time periods. They observed phosphide which gave an AES signal at 119 eV similar to chemisorbed DMMP on Si02 and aFe203(P),and surface phosphate which gave AES signals at 94 and 111 eV similar to decomposed DMMP on a-FezO3.

Summary DMMP on S O 2 . No DMMP decomposition occurred on dehydrated SO2. There were two DMMP TPD peaks which corresponded to multilayer (200-205 K) and monolayer (275 K) desorption states. On heavily hydrated SiO, not more than 8-10% of the chemisorbed DMMP decomposed yielding M P and CH30H. DMMP on a-Fe203.On iron oxide surfaces monolayer amounts of DMMP decomposed completely and at high temperatures an iron oxidephosphate phase forms which does not strongly inhibit the decomposition of subsequent exposures of DMMP. The decomposition products include C H 3 0 H , C 0 2 , CO, H,, H 2 0 , HCOOH, and a carbon-free phosphorus-containing residue on the surface above 600 K. AES P( 114) line shape suggest that initial decomposition occurred below 250 K in the form of oxidation of the P-CH, bond by lattice oxygen. After desorption of all carbon-containing decomposition products (which occurred below 600 K) the phosphorus migrated into the bulk of the oxide leaving the surface available for decomposition of subsequent DMMP exposures. DMMP on a-Fe203(P).Continued DMMP exposure to the a-Fe2O3 surface followed by heating above 600 K saturated the AES P(114)/Fe(703) signal ratio at 0.45 to 0.55. At this point the extent of DMMP decomposition decreased and a broad DMMP monolayer state was present in the TPD. Analysis of the AES P line shape was able to differentiate between the P due to adsorbed DMMP and the phosphate within the oxide. Acknowledgment. This work was supported in part by the U S . Army Research Office. Registry No. DMMP, 756-79-6; S O 2 , 763 1-86-9; Fe203,1309-37-1. (1 1 ) Shafrin, E. G.; Murday, J. S. J . Vacuum Sci. Technol. 1977, 14, 246.