Photoelectron Spectra of Phosphorus Halides, Alkyl Phosphites and Phosphates, Organo-Phosphorus Pesticides, and Related Compounds D. Betteridge,’ M. Thompson,* A. D. Baker,3 and N. R. Kemp4 Department of Chemistry, University College of Swansea, Singleton Park, Swansea SA2 BPP, Wales, U.K. The photoelectron spectra of 15 phosphorus-containing compounds, including 3 pesticides, have been obtained. During these measurements, instrument memory and side-reaction effects were observed. Several of the spectra have been interpreted with the aid of simple molecular orbital theory. The spectra of the compounds are sufficiently different to allow qualitative identification. The analytical significance of the results is discussed.
USEOF ORGANO-PHOSPHORUS COMPOUNDS as pesticides is increasing because of their relatively facile degradation compared with organo-chlorine pesticides. Gas-liquid chromatography has so far been the major analytical method employed for the determination of organo-phosphorus pesticide residues ( I , 2). Because of their importance, we decided to investigate them as part of our study of the applicability of UV-photoelectron spectroscopy to the analysis of compounds of agricultural significance (3, 4). Their comparative involatility (3) and molecular complexity have made the investigation difficult, but we have obtained the photoelectron spectra of several pesticides and related phosphorus-containing compounds. The spectra have been interpreted in a simplified manner by making use of arguments based on the symmetry of phosphorus and ligand orbitals (5), and by comparing the spectra with those of molecules of less complexity, such as the simple phosphorus halides. This approach has been used successfully by Price et al. (6) for simple molecules in general. We have extended these arguments to molecules of a more complex nature by using appropriate approximations. As yet no complete theoretical treatment of these molecules is available, so at present it is not possible to compare the simple approach with the rigorous one, but the method does appear to be useful. During this work, several complications arose due to memory effects and side-reactions in the spectrometer and these are discussed. The resolution of a spectrum of a mixture into the sum of the spectra of its components was achieved and the implications of this are considered. EXPERIMENTAL The compounds studied were obtained from Albright and Wilson (Mfg) and Shell (Woodstock Agricultural Research To whom communications concerning this paper should be addressed. * Present address, Department of Chemistry, University of Technology, Loughborough, LEI1 3TU, U.K. Present address, Department of Chemistry, Queens College of the City University of New York, Flushing, N.Y., 11367. Present address, British Council, Djakarta, Indonesia. (1) D. C . Abbot and H. Egan, Analyst (London),92, 475 (1967). (2) J. Askew, J. H. Ruzicka, and B. B. Wheals, ibid., 95,275 (1969) (3) A. D. Baker, D. Betteridge, N. R. Kemp, and R. E. Kirby, ANAL.CHEM.,42, 1064 (1970). (4) Zbid., 43, 375 (1971). ( 5 ) D. J. Royer, “Bonding Theory,” McGraw-Hill, London, 1968. (6) A. W. Potts, H. J. Lernpka, D. G. Streets, and W. C . Price, Phil. Trans. Roy. SOC.A , 268, 59 (1970).
_i
7
1$
Figure 1. The 584-A excited photoelectron spectra of simple phosphorus halides. Spectra A and B are of PC15contaminated with HCI and POC13. Spectrum C corresponds to A stripped of HCI and PCIJ
Centre) and purified where appropriate. Purities were checked by infrared and mass spectrometry. Photoelectron spectra were obtained on a Perkin-Elmer PSI 5 spectrometer, modified as described previously ( 3 ) . Samples which were not volatile enough to be introduced via the standard inlet system were introduced by a probe approximately 10 cm from the target chamber. The experiments were carried out at room temperature and the sample pressure adjusted to ca. 0.07 Torr to give optimum spectral conditions. RESULTS AND DISCUSSION Spectra of Mixtures and Memory Effects. Spectrum A, Figure 1, is of a sample of supposedly pure PC15. On evacuation of the compound for a period of four hours in the spectrometer, spectrum B was obtained. Further evacuation overnight resulted in a constant spectrum which was taken to be that of the pure material. Clearly this treatment has gradually removed several peaks from A, those associated with HCl being immediately recognized. In order to distinguish other possible contaminant(s), the spectra of HC1 and pure PC15 were stripped from A using a simple hand calculation. The resulting spectrum C is without doubt identifiable as that of P0Cl3 (Figure 1). These results show that PE spectra are additive and further illustrate the analytical potential of the
ANALYTICAL CHEMISTRY, VOL. 44, NO. 12, OCTOBER 1972
* 2005
10
13
16
19 “ 7
10
13
16
19
-
Figure 2. The 584-A excited photoelectron spectra of alkyl phosphites and phosphates. Spectrum A is of product of instrument reaction between (CH30)3Pand PBr3/PBr, 7
I
1
I
I
10
13
16
19
Figure 4. The 584-A excited photoelectron spectra of phenyl dichloro phosphine (top), tris(dimethy1amino) phosphine oxide (center), and methyl phosphoric dichloride (bottom)
10
13
16
19
”
7
10
13
16
19
Figure 3. The 584-A excited photoelectron spectra of other pesticides dichlorvos (top right), butonate (center right), and disulfoton (bottom right). Spectra A, B, and C are of impure dichlorvos
technique (7). Additionally, it is evident that computerized spectrum stripping (or curve resolving) techniques will aid in the identification of mixtures of compounds just as they have in other areas of spectroscopic analysis (8). The spectrum (A, Figure 2) of trimethyl phosphite obtained four days after an attempt to record that of PBr5 is quite different from the actual spectrum of (CH30)3P. The two sharp peaks in A at 10.54 and 10.86 eV (ionization potential scale) correspond to the spin-orbit split bromine lone-pair ionization potential of CH3Br. The reaction of the phosphite with PBr, (PBr5 being dissociated) had obviously taken place in the instrument even though the spectrometer had been evacuated for the above mentioned time to “remove” contamination due to PBr5. That one of the products of the reaction is indeed CH3Br was demonstrated by monitoring the vapors evolved from a liquid mixture of (CH30)3Pand PBr3 with the spec(7) D. Betteridge and A. D. Baker, ANAL.CHEM.,42 (l), 43A (1970). (8) C . W. Childs, P. S. Hallman, and D. D. Perin, Talanta, 16,629 (1969). 2006
0
trometer. These results confirm the danger to spectral analysis from instrument memory effects and side-reactions. Spectra A, B, and C, Figure 3, are the result of gradual evacuation of a sample of the pesticide dichlorvos. The spectra suggest that more volatile contaminant(s) have been removed from the pesticide. (At present these contaminant(s) have not been identified.) The spectrum of pure dichlorvos (Figure 3) contains bromine lone-pair peaks due to a CH3Br memory effect associated with the (CH30)3P-PBr3reaction mentioned above. Again a reasonable period of time was allowed for the “cleaning” of the instrument by evacuation. Interpretation of Spectra. The spectra of the compounds studied are shown in Figures 1 to 4. The vertical ionization potentials measured from the spectra are given in Table I. No attempt has been made to measure these values to an accuracy of greater than 0.1 eV. ORIGINOF GROSSFEATURES IN THE SPECTRA.Both tervalent and pentavalent phosphorus compounds were examined. The former possess a formally unshared pair of electrons [P 3p] which would be expected to give rise to a band near ionization potential 9 or 10 eV. Such a band would be expected to be absent in pentavalent phosphorus compounds. The other details of the spectra will depend upon the atoms to which phosphorus is bonded. From experience with the spectra of other molecules, it can be predicted that broad bands due to the sigma bonding orbitals will occur in the region from 21 eV down to about 10-12 eV. “Lone pair” electrons in substituent atoms will give rise to more sharply defined bands in the region below 12 eV. Compounds with more than one substituent carrying formally lone-pair electrons will be expected to give rise to spectra in which bands appear due to in-phase and out-of-phase combinations of the lonepair orbitals as discussed elsewhere (5). These combination bands will be roughly symmetrically disposed over a small region (about 2 eV) either side of the position expected for an uninfluenced lone pair. These general rules account for the gross features of the spectra of the compounds studied, which are now considered in more detail.
ANALYTICAL CHEMISTRY, VOL. 44, NO. 12, OCTOBER 1972
Table I. Vertical Ionization Potentials (eV) of Phosphorus Compounds Studied Band 1st 2nd 3rd 4th 5th 6th
Compounds Phosphorus trichloride 10.5 Phosphorus tribromide 10.0 Phosphorus oxychloride 12.0 Phosphorus pentachloride 10.7 Methyl phosphonic dichloride 11.4 Phenyl dichloro phosphine 9.7 tris-(Dimethylamino) phosphine oxide 8.7 Trimethyl phosphite 9.0 Trimethyl phosphate 10.8 Dimethyl phosphite 11 .o Triethyl phosphite 8.8 10.4 Triethyl phosphate Dichlorvos 9.4 Disulfoton 9.0 10.3 Butonate = Exceptionally broad composite band
11.7 10.8 12.4 11.3 12.4 11.6
12.0 11.2 13.0 11.5 12.9 12.4
13.0 11.9 13.5 12.1 13.2 12.8
14.3 13.1 13.9 13.1 15.0a 13.1
9.1 10.6 11.3 11.8 10.1 10.9 12.3” 9.3 11.1
9.6 11.1 11.9 12.9 10.6 11.5 13.4 11.5 11.8
11.0 11.8 12.6 14.3 12.0= 12.P 14.6 11.9 12.7
13.4a 12.4 12.9 15.6 13.3 13.2 15.4 14.0a 14.1
SIMPLEPHOSPHORUS HALIDES(Figure 1). The spectra of PC13 and PBr3 closely resemble those reported previously (6). The spectrum of PoC1, has not been published, although Hillier and Saunders (9) have compared ab initio calculations of orbital ionization potentials with those obtained from a PE spectrum. (These calculations were also performed for PC13.) The symmetry of the orbitals to be considered in the case of PCls ( D 3 J are as follows: P - s(A1’), pz (A*”), pz (E’), py (E’), dzz (El’), dzz-uz (E’), dzz dzz--yZ dz, (E’), dzz (E”), dyz (E”); C1 (linear combination of p orbitals) -2 of AI’, A*” and E’ (a), and Azf,A2“,2 of E’ and 2 of E” ( T ) . Appropriate combination of u and T orbitals results in the possibility of ten PE spectral bands. On the crude assumption that the higher energy u orbitals do not interact with ir orbitals, ten molecular orbitals can be represented: P-C1 bonding (a) -2 of A1’, A*’’ and E‘, and chlorine nonbonding ( x ) , Az’, A2“, 2 of E‘ and 2 of E” symmetry. A reasonable ordering of certain of these orbitals can be obtained by reference to the peak areas of the seven observable bands in the spectrum of PClj. This ordering according to symmetry type, with no attempt to distinguish between E’ and E’’ may be correlated with the observed ionization potentials, eV, (Table I) as follows: 10.7(E), 11.3(E), 11.5(&’), 12.1(E), 13.1 (indistinguishable E Azs), 13.6(E), and 16.0(Alf). The remaining orbitals from which ionizations are not detected in the spectrum are, therefore, u and of symmetry AI’ and A*”. Similar ionizations from u orbitals of A1 symmetry in PC13 and Poc13 (at 18.85 and 19.48 eV, respectively) are barely seen in our spectra of these compounds. Thus, the spectrum is very similar to that of PC13with additional bands in the 10.7-13.1 eV region due to chlorine orbitals of essentially nonboliding character and additional bands in the P-CI u-bonding region above 14 eV. OTHER SIMPLE PHOSPHORUS COMPOUNDS (Figure 4). Methyl phosphonic dichloride can be considered to be a severely perturbed POC13 molecule (C3& with a methyl group replacing one chlorine atom. One might expect such a substitution to result in, (i) a general lowering of the ionization potentials because of the inductive effect of the methyl group, (ii) the disappearance of bands associated with the displaced chlorine atom (both nonbonding at about ionization potential 13 eV and u-bonding in the approximate range 15-20 eV), and (iii) the appearance of bands associated with the methyl group
+
(9) J. H. Hillier and V. R. Saunders, Cliem. Commim., 1970, 1510.
7th
15.2 14.2 15.5 13.6
16.6 16.0
14.4
15.3
18.9
13.0 14.4 17.4 15.6a 16.3” 15.1 15.2
-__ 8th
16.8 16.ga
17.7
16.5 16.4
molecular orbitals at about 13 eV (A) and 15-17 eV (E). In the spectrum of Poc13 over the range 12.0-13.9 eV, there are five bands corresponding to orbitals of AI, A?, and 3 of E symmetry. With CH3POC12we expect to lose two contributions of nonbonding orbitals because of replacement of the chlorine atom, leaving nonbonding orbitals of either 2 of A type and 2 of E type, or 1 of A type and 3 of E type symmetry. Experimentally, the first four bands of the PE spectrum of CH3POC12are of the appropriate peak area to give an ordering of nonbonding orbitals of (Table I): 11.4 (E type), 12.4 (E type), 12.9 (A type), and 13.2 (A type). As expected, this range of values is shifted to slightly lower ionization potential compared to the corresponding range for POC13 (12.0-1 3.9 eV). As with Poc13, it should be emphasized that there are undoubtedly P 3p,-O 2p, and P 3d,-0 2p, contributions to the CH3POC12nonbonding orbitals. The broad band over the ionization potential range 14-18 eV corresponds to components of P-0, P-Cl, and C-H u-bonding. The peak at 14.9 eV has the characteristic shape of the methyl E band, but it is not possible to assign the other peaks with certainty or to account for the P-C o-component of the spectrum. In the spectrum of phenyl dichloro phosphine we might expect bands due to ionization from the 7r2 and r 3orbitals of the benzene component of the molecule (split due to monosubstitution of the benzene ring) (9), and a peak corresponding to the phosphorus lone-pair orbital all in the ionization potential range 9--10 eV. Experimentally, the spectrum clearly exhibits only one peak at 9.7 eV; therefore, we conclude that the bands listed above are all superimposed. It is not surprising that the n-r3 splitting is not observed since substituents in benzene of an electron-withdrawing nature do not exhibit this effect (9), and the PC12group can probably be placed in this category. The broad band with sharpish spikes from 11.O to 13.7 eV is undoubtedly associated with the second and third bands of benzene and chlorine nonbonding components. The width of this band relative to benzene and other chlorinated benzenes (IO) coupled with the sharp peaks suggests the chlorine bands are split, but it is not clear whether this is solely due to C1-C1 interactions, or to C1- T(benzene) interactions, or both. The bands at higher ionization potential than 14.0 eV correspond to benzene, P-C, and P-C1 u orbitals. (10) A. D. Baker, D. P. May, and D. W. Turner, J. Chem. SOC., 1963, 1250.
ANALYTICAL CHEMISTRY, VOL. 44, NO. 12, OCTOBER 1972
2007
It often appears that a spectrum is to a first approximation the sum of its parts. For example, the spectrum of chlorobenzene is very similar to that of benzene with additional chlorine lone-pair peaks. (There are of course important differences, but the similarities are evident.) We wondered whether any quantitative support could be found for the view that the spectrum PhPC1, in the region 9-14 eV could be considered as a superimposition of benzene and PClz spectra. (The bands in the hypothetical-PClz spectrum expected over this energy range being one phosphorus lone-pair and four chlorine nonbonding lone-pair bands.) Comparison of the spectra of PhPClz and benzene indicates that the band of ionization potential 14.4 eV in the PhPC12 spectrum has the same shape as the band of the same energy in the benzene spectrum. The other bands, as noted above, seemed to be “mixed” benzene and -PCh bands. The band at 14.4 eV was therefore taken as due to “pure” benzene and used to normalize the peak areas of the two spectra. Then the “benzene component” of each peak in the PhPC12 spectrum was subtracted and the residual area was taken to be the -PCl2 component. The result, expressed in electron pairs, was for the 9.7 eV band, 1.5 for benzene and 1.O for -PCl2, and for the 11.O to 13.7 eV band, 3.0 for benzene and 4.0 for -PCl,. This agreement with what is known for benzene and deduced above for PhPClz is surprisingly good considering the grossness of the assumptions underlying the calculation. It suggests that such computations may have some value in helping to interpret the spectra of complex molecules. The spectrum of tris (dimethy1amino)phosphine oxide (Figure 4) shows some expected features. The broad intense band in the region 12.3-18.0 eV must derive from the six methyl groups and three P-0 u orbitals of the molecule. The band at 11.O eV, by analogy with Poc13, is probably the oxygen 2 p lone pair with P - 0 , 2 pr-2p,, and 3dr-2p, contributions, and possibly with methyl (E) and nitrogen lone pairs being mixed in as well. The three peaks at 8.6, 9.1, and 9.5 eV of approximately equal area are more difficult to interpret because although they fall in the region expected for nitrogen lone pairs ( I I ) , symmetry arguments suggest that three such lone pairs should give rise to two peaks, A E, of intensity ratio 1 :2. The spectrum remained constant through a period of prolonged evacuation which suggests the “extra” peak is not due to impurities in the sample and this view is confirmed by the NMR and mass spectra. There are three possible explanations: (i) The PE spectrum of dimethylamine, produced in situ by the photolysis of [CCH&N], P=O, is superimposed on the spectrum of the sample molecule. The mass spectrum shows that the main fragmentation route is via (CH&N+ and CH,NH+ so that if 21.21 eV photons were able to bring about similar fragmentations, the amines would be present in the target chamber despite evacuation. The nitrogen lone pair ionization could account for either the 8.6 or 9.5 eV peak, depending whether dimethylamine or methylamine were the main product, and the methyl peaks would be part of the broad band in the 12.3-18 eV region. However, the constancy of the 8.5-10.6 eV peak intensities and their approximate equality do not support this explanation. (ii) The P-N u orbitals have a lower 1 p than the oxygen 2 p lone pair orbital, in which case there would be five bands to be accounted for in the region 8-12 eV-viz, N 2 p lone pair (A + E), P-N (A E), and 0 2p lone pair whose intensities would be in proportion 1 :2 :1 :2 :1, but not necessarily in that
+
+
(11) A. B. Cornford, D. C. Frost, F. G . Herring, and C. A. Mcdowell, Can. J . Chem., 49, 1135 (1971). 2008
order. This hypothesis requires one more band than so far considered and this may be the shoulder at 10.5 eV. There is not a PE spectrum presently available which helps in the unambiguous assignment of a P-N IP and the extent of band overlap precludes reliable assignment of intensities so that it is equally possible to make the P-N assignments to the 12.3-18 eV. (iii) Professor E. Heilbronner has suggested that the methyl groups may interact to such an extent that the molecule does not have C,. symmetry, and the distortion will lead to an effective splitting of the E orbitals so that three peaks would be expected. This proposal is straightforward and consistent with the other symmetry-based arguments. Bock and Fuss have obtained the spectrum of the corresponding boron compound [(CH3)?N],B(12). It clearly has two N lone pair peaks in the approximate ratio 2 :1. The separation between them is 1.62 eV, and from this it is deduced that the molecule has a propeller shape to reduce interactions between the methyl groups. Even allowing for the differences between phosphorus and boron, and for the addition of an oxygen atom, one might have expected the spectra of the two compounds to be similar. Models suggest that the propeller form is a stable one for the phosphorus compound. Nevertheless this hypothesis will be more fully investigated, since if true it will be an example of the PE spectrum providing evidence of distortion which is not apparent in the NMR spectrum. On balance, although inclining towards iii, we feel that it is best to await the results of further investigations before deciding between these hypotheses. ALKYLPHOSPHITES AND PHOSPHATES (Figure 2). At first sight the spectra of the alkyl phosphites and phosphates seem complex. The most distinctive feature is the phosphorus lonepair peak at about 9 eV which is present in the spectra of the phosphites and absent in those of the phosphates and dimethylphosphite (which does not have a phosphorus lone-pair). Characteristic alkyl bands are also prominent. However, the detailed consideration of trimethyl phosphite, which follows, shows the spectra contain much more information. Trimethyl phosphite can be considered as a derivative of methanol in which the methoxy component is locked in C3. symmetry by the phosphorus atom and the 0-H bond of methanol is replaced by a P-0 bond; the phosphorus has a lone-pair of electrons. If we imagined the molecular orbitals to be completely localized, we would expect to see ionization from the following orbitals: phosphorus lone-pair (l), oxygen lone-pairs (3), P-0 u (3), C-0 u (3), and C-H u (9). Now while it is generally true that molecular orbitals are delocalized they may often be considered as having a predominant character. Thus in methanol we have previously identified bands at about 11.0, 12.7, 15.0, and 17.0 eV as being essentially derived from oxygen lone-pair, C-0 u , CH3(A1 E), and 0-H, respectively (4,the methyl orbitals of AI and E symmetry being derived from the Tz orbitals of methane) (6), but only the E component is seen at ionization potential 15.0 eV in the spectrum of methanol. In (CH30),P the 0-H bond is absent, but we would expect the other appropriate groups of three orbitals and C3”symmetry to give rise to two orbitals of A and E symmetry. Thus, in the PE spectrum we would expect to see two bands with an area ratio of 1:2 (A E) at about 11.0 eV, a similar pair at 12.7 eV, and four with intensity E, A E) at about 15 eV. The P-0 u ratio 1 :2 :1:2 (A orbital in POCl, has an ionization potential of 15.4 eV (8), but one would expect the corresponding band in (CH30)3Pto be
+
+
+
+
(12) H. Bock and W. Fuss, Chem. Ber., 104,1687 (1971).
ANALYTICAL CHEMISTRY, VOL. 44, NO. 12, OCTOBER 1972
less due to the inductive effect of the methyl group. The three P-0 u orbitals would also be split into two bands of intensity ratio 1 : 2 , and the phosphorus lone-pair would result in a single peak. In this way, all of the available electrons are accounted for, and it is evident that, although several overlapping bands are to be expected, it should be possible to distinguish several spectral features. It is not strictly possible to relate peak areas with the number of electrons in the orbital being ionized, because (i) for a given electron-energy analyzer, the observed intensity for a given electron flux is dependent upon the energy of the electrons being analyzed (13), and (ii) ionization cross sections are not constant. Nevertheless, peak areas do provide a guide to orbital occupancy especially if the energy range being considered is limited and the orbitals considered have comparable character (14). In the 9-14 eV range of the spectrum of (CH30)8Pthe areas of the bands centered at 9.0, 10.6, 11.1, 12.4, and 13.0 eV, respectively, are in the ratio 1 :2:4:2:1. We make the tentative assignments to them, respectively, of phosphorus lone-pair (A), oxygen lone-pair (E), oxygen lonepair (A) and C-0 u (E A), P-0 u (E), and P-0 u (A). The bands above 15 eV are derived from the methyl groups, but in view of the extent of overlap and the relative insensitivity of the analyzer in the region, no detailed assignment can be attempted. We would expect the overall spectral differences of trimethylphosphate compared to the phosphite to resemble those of P0Cl3 relative to PC13, i.e., loss of a peak due to the phosphorus lone-pair and a gain of one due to a P-0 orbital, and band(s) corresponding to the extra oxygen nonbonding orbitals which also have minor contributions from P 3d,-0 2p, and P3p,-O 2p, bonding. Experimentally, the spectrum of (CH30)3PO does show the loss of the phosphite phosphorus lone-pair, and although its general shape is very similar to the spectrum of (CH30)3P,there are additional bands at 10.8 and 14.4 eV which we assign to oxygen nonbonding (with minor P-0 7r-bonding contributions) and P-0 u, respectively. In addition, the spectrum of (CH30)BP0is at overall ionization potential approximately 1 eV higher than that of (CH30)3P, as expected because of the introduction of an electronegative oxygen atom (compare POC13 and PCI3). As with the spectrum of ethanol compared to that of methane (4), the spectrum of triethyl phosphate is more complex than that of the corresponding methyl compound due to extra bands corresponding to C-C u and C-H u orbitals. However, the following are shown in the spectrum of (C2HS0)3P0, (i) the loss of the band due to phosphorus lone-pair at 9.8 eV in (C2H50)3Pand (ii) an extra band at 10.4 eV compared to (C2H50)3Pwhich we assign to the oxygen nonbonding orbital with usually 7r-bonding contributions. In the spectrum of dimethyl phosphite, only the absence of a band due to phosphorus lone-pair can be distinguished. PESTICIDES (Figure 3). Dichlorvos, disulfoton, and butonate are molecules of greater complexity than the ones discussed thus far. Therefore, the appropriate full interpretations are not possible ; however, using the previous arguments some deductions can be made, and an increased library of information concerning simpler molecules will undoubtedly contribute further to these deductions. We would expect the spectrum of dichlorvos to consist of a mixture of spectra due to perturbed (CH30)3 PO(OCH3
+
(13) D. W. Turner, C. Baker, A. D. Baker, and C. R. Brundle, “Molecular Photoelectron Spectroscopy,” Wiley, London, 1970. (14) L. 1,. Lohr and M. B. Robin, J . Amer. Chem. Soc., 92, 7241 ( 1970).
I
1
1
I
I
8
IO
I
I
I
1
14
16
18
7 600
x .d
5
6
400
200
6
12
20
1PIeV)-
Figure 5. The 584-A excited photoelectron spectrum of triphenyl phosphine, obtained at 105 “C
changed to OCH) and 1,l-dichloroethylene (15). Therefore, bands due to the ethylenic x orbital, chlorine nonbonding orbitals and some of those discussed above for (CH30)3P can be anticipated. In this regard, the band corresponding to ionization from the ethylenic 7r orbital is distinct at 9.4 eV. Appropriate spectrum stripping techniques may enable us to further interpret the spectrum of dichlorvos on the lines mentioned in this paper. In the disulfoton molecule there are three sulfur atoms in different environments ; consequently, we might expect to find three PE spectral bands corresponding to ionization from the different sulfur lone-pair orbitals in the 8-10 eV range (16). The spectrum of disulfoton actually shows a composite band from 8.3 to 9.7 eV in which at least two peaks at 9.0 and 9.3 eV and a shoulder at 8.7 eV can be distinguished. The remainder of the spectrum shows few features which can be assigned, but again, reference to spectra of the appropriate simpler molecules making up the pesticide may aid in its interpretation. The PE spectrum of butonate is tantalizing. At present it would be foolhardy to assign peaks, but it should be possible when more information about ionization cross-sectional areas and the effect of adjacent groups on ionization potentials become available. There are obviously a number of peaks in the chlorine 3p lone pair region at about 12.5 eV and these are overlapped by oxygen 2p lone pair peaks and C-0 u . It is a characteristic spectrum despite its apparent shapelessness, which is overemphasized by the reduction in size. CONCLUSIONS
The spectra shown in this paper were hard to obtain because of the relative involatility of the samples and since, because of the deleterious effects of these compounds and their decomposition products, only photomultipliers approaching the end of their working life were used. Developments, in the design of energy analyzer and sampling systems, which have taken place since these experiments, and which are in hand, make it certain that much better spectra of these compounds will be obtained with much less labor. For example, the spectrum of triphenylphosphine (Figure 5 ) was obtained on the PerkinElmer PS18 within 2 hours. The finer details which will then be available should lead to more refined interpretations than we have attempted here. In particular it will be possible to examine the hypothesis that the UV-PE spectra may be used to identify the presence of (15) N. Jonathan, K. Ross, and V. Tomlinson, Znt. J . Mnss.Spectram. Zon. Phys., 4, 51 (1970). (16) D. Betteridge, “Molecular Spectroscopy 1971,” P. Hepple, Ed., Institute of Petroleum, London, 1972, p 82.
ANALYTICAL CHEMISTRY, VOL. 44, NO. 12, OCTOBER 1972
0
2009
atoms in different molecular environments in much the way that is common in X-ray PES (cf. the different sulfur-peaks in disulfoton and oxygen peaks in butonate). So far most of the compounds subject to UV-PES have had a high degree of symmetry, so this possibility has not been made evident. It is an exciting possibility, especially as the chemical shift in X-ray PES may be nullified if there are two opposing tendencies such as ligand to metal cr-donation and metal to ligand a-back donation (17). We also feel that the simple arguments used to explain the spectra could well be amplified (or vitiated) by a more detailed analysis of the spectra in which band areas are more carefully measured and correlated with ionization cross-sectional areas. A program of computer analysis of the spectra is in hand to carry out this work. As it stands, we feel this study shows that it may be possible to assign peaks in complex spectra by simple methods. Although the assignments may need revision or incompletely or inadequately reveal the true nature of the orbital from which ionization occurs, the methods proposed do allow the correct number of bands and their approximate order to be deduced. As before we note that the spectra are quite distinct and form the basis of identification, also that the peculiarly phos-
phorus characteristics of P lone-pair and the oxygen lone pair in the P=O moiety are clearly observable in the spectra. The possibilities for detecting trace impurities, such as water in PCI5, and of resolving the spectra of mixtures have been demonstrated. Quantitative application of the above findings will require further work on the inlet system so that the amount of sample entering the target chamber is known as well as better handling of the output data. Work on these matters is in progress, and preliminary results from a computer program fully confirm the hand calculations reported here (18).
(17) W. E. Morgan, W. J. Stec, R. G. Albridge, and J. R. Van Wazer, Inorg. Chem., 10, 926 (1971).
(18) D. Betteridge, M. A. Stevens, and M. Thompson, University College Swansea, unpublished work, 1971172.
ACKNOWLEDGMENT
Thanks are due to Albright and Wilson and to Shell for the provision of the various samples examined in this work. We are also grateful to Perkin-Elmer Ltd. (Beaconsfield) for the use of a model PS18 photoelectron spectrometer. RECEIVED for review February 11, 1972. Accepted June 5 , 1972. We gratefully acknowledge the support of the Agricultural Research Council, who provided the instrument and a Fellowship to A.D.B., and of the Science Research Council, who supplied a Fellowship to M.T. and a maintenance grant to N.R.K.
Microwave Spectroscopy Analysis of the Distribution of Deuterium in Propene-D, Obtained from Catalyzed Hydrogen-Deuterium Exchange Reactions LeRoy H. Scharpen and Roger F. Rauskolb Scientijc Instruments Division, Hewlett-Packard Company, 1601 California Auenue, Palo Alto, Calg. 94304
Chadwick A. Tolman Central Research Department, E . I . du Pont de Nemours & Company, Experimental Station, Wilmington, Del. 19898 Five subspecies of propene-d,, differing in the location of the deuterium atom, can be distinguished by microwave spectroscopy. Quantitative analysis of these subspecies by microwave spectroscopy was done for samples obtained by hydrogen-deuterium exchange between CHsOD and propene-do in the presence of homogeneous catalysts of Pt, Rh, and Ni. The relative amount of each subspecies in a sample varied considerably for the catalysts used. The relative standard deviation of the analytical result was about 2% of the values found for subspecies concentrations ranging from 1.5% to 17% of the total sample. Most spectrometer functions, including signal processing, were controlled by computer.
MICROWAVE SPECTROSCOPY is the study of the centimeter and millimeter wavelength absorption spectra of molecules in their vapor state at low pressure. Although this technique has been used to study a variety of chemical problems ( I ) , there has been a paucity of reported analytical applications (1) D. R. Lide, Jr., Surc. Progr. Chem., 5 , 9 5 (1969). 2010
(I, 2) and, as a result, a lack of knowledge among those interested in chemical analysis concerning the capabilities and methodology of the technique. Lide (I) recently listed the lack of reliable, easy to operate spectrometers as one reason for slow development of the analytical potential of the technique and problems associated with accurate measurement of absorption line intensities, required for quantitative analysis, as another. The availability of commercial spectrometers has obviated Lide’s first point. In this paper, we consider some aspects of the second question, the reliability and accuracy of quantitative data. To do this we treat our solution of a specific analytical problem-namely, the determination of the amounts of the five subspecies of propene-dl (C3HsD) present in samples of partially deuterated propene obtained in catalytic hydrogen-deuterium exchange reactions. Interpretation of the analytical data in terms of the mechanism
(2) L. H. Scharpen and V. W. Laurie, ANAL.CHEM., 44, 378R (1972).
ANALYTICAL CHEMISTRY, VOL. 44, NO. 12, OCTOBER 1972