Langmuir 1994,10, 756-760
756
An Infrared and 31PMagic Angle Spinning Nuclear Magnetic Resonance Study of the Adsorption of Pc13 and OPC13on Silica B. A. Morrow* and S. J. Lang Department of Chemistry, University of Ottawa, Ottawa, Ontario, Canada K I N 6N5
Ian D. Gay* Department of Chemistry, Simon Fraser University, Burnaby, British Columbia, Canada V5A IS6 Received October 12,1993. In Final Form: December 17,199P Phosphorus trichloride initially physically adsorbs on silica at room temperature via an interaction with surface silanolgroups. Followingprolonged contact over several hours, it dissociatesto yield the chemisorbed SiOP-containing species SiOPCl2,SiOP=O(H,OH),and (Si0)2P=O(H). The last two of these species can also be produced by impregnation of silicawith phosphorous acid, H303, from aqueous solution. Phosphoryl chloride, O=PC13, also physically adsorbs on silanols, but the interaction is stronger than that which occurs with PCl3, and no further reaction occurs at room temperature. Some anomalies previously reported in the literature from earlier studies of PC13 adsorption on silica have been shown to be due to oPc13 which can be present as a very low level (1-2%) impurity in commerially available PC13.
Introduction The infrared spectrum of PCl3 adsorbed on silica has been studied by Bogatyrev and Chuiko,' who observed the appearance of two broad IR bands due to perturbed isolated surface silanolsnear 3650 and 3500cm-'. Although no specific attribution was specified for either band, they assumed that two different structures could result from the interaction between physically adsorbed Pcl3 and SOH. An additional IR band was reported at 605 cm-l which was attributed to the antisymmetric PCl stretching mode of PC13 coordinated to SiOH. However, for pure Pc13 the antisymmetric and symmetric PCl3 stretching modes2lie between 515 and 487 cm-l, and such a large high-wavenumber shift would not be expected for any molecular complex.3
Because 31P NMR spectroscopycan be additionally used to study the adsorption and reactions of this compound on silica, and because of the anomaly reported above, we have undertaken a reinvestigation of this system using both IR and solid-state NMR spectroscopy. In order to unravel some of the reaction paths of Pc13, we have also studied the adsorption of O=PCla, HsP03, and on silica.
Experimental Section The chemicals used for adsorption studies, PClS, OPCls, H3POa, and H8O4, were obtained from standard commercial sources. The silica used was Cab-0-Si1HS-5having a BrunauerEmmett-Teller (BET) (Nz) surface area of 325 m2/g and has been previously characterized by us using IR48 and NMR6 ~~
* Abstractpublishedin Advance ACSAbstracts, February 1,1994. (1)Bogatyrev, V.M.; Chuiko, A. A. Sou. R o g . Chem. (Engl. Transl.)
1984,50,50.
(2)(a) Nakamoto, K.Infrared and Ramun Spectra ojznorganic and Coordination Compounds, 4th ed.; John WileyL Sons: New York, 1986. (b)Frankiss, S. G.; Miller, F. A. Spectrochim. Acta 196621, 1235. (c) Nyquist, R. A. Appl. Spectrosc. 1987,41,272. (3)Edwards, H.G. M.; Woodward,L. A. Spectrochim. Acta 1970,26A, 1077. (4)Morrow, B. A.; McFarlan, A. J. J. Phys. Chem. 1992,96, 1395. (5)Morrow, B. A,; McFarlan, A. J. Langmuir 1991, 7, 1695. (6) Morrow, B. A.; Gay, I. D. J. Phys. Chem. 1988,92, 5569.
0743-7463/94/2410-0756$04.50/0
spectroscopies. The silica was pressed into 2bmm-diameterdisks (10 mg/cm2)in a die at about lo7Pa of pressure. For IR studies, the intact disk was mounted in a previously described' 300-mLvolume IR cell and samples were activated in vacuum for 1h at 450 OC and then cooled to room temperature (22 f 1"C) prior to adsorption of PCl3 or OPCb. For NMR studies the disks were broken into small (about 1mm) pieces and were placed in 6-mm0.d. glass tubes which were then sealed to a vacuum line for subsequent treatment. Infrared spectrawere recorded using a Bomem Michelson MB100instrument at a resolution of 4 cm-'. The NMR spectra were recorded at either 1.4 or 3.5 T, giving 3lP resonance frequencies of 24.3 and 60.5 MHz, respectively. The magic angle spinner has been described previously,Bandall spectrawere proton decoupled. The spectra of PCb-treated Si02 were recorded at 24.3 MHz using either 90" pulse excitationor cross-polarizationwith a 2-ms contact time, with 2.5-kHz magic angle spinning (MAS). The spectra of HaPOd-treated Si02 were also obtained at 24.3 MHz, with 3-kHz MAS. For higher activation temperatures extensive dehydroxylation of the HPO4-treated SiOz samples made the acquisition of cross-polarization spectra impossible. Furthermore, for HaPO&reated Si02 samples activated at 400 "C, the 3lP TIwas too long to permit 90" pulse spectra to be obtained without the addition of about 300 Torr of 02 to the sample tube. Addition of dry 02 to these samples had no effect that could be observed by infrared spectroscopy. For lower activation temperatures spectra were acquired using 9 0 O pulses, but 02 was not added. It is doubtful that any of the HaPO&reated Si02 spectra are fully relaxed, and these spectra cannot be interpreted quantitatively. Since these spectra are only used for qualitative purposes this limitation is not important. The NMR spectra of HaPO3-treated Si02 were obtained at 60.5 MHz and about 2.5kHz magic angle samplespinning (MASS).These samples could not be activated above 200 O C without extensive decomposition, as shown by the loss of intensity of V ~ - Hin the infrared. The spectra presented here were acquired wing cross-polarization. The extensive sidebands present at this higher magnetic field were suppressedby using atotal suppressionof sidebands (TOSS) sequence. Chemical shifts were referenced to 85% HsPO4. In the descriptionsbelow, we will write SiOP=O(X,Y) for the structure (7) Morrow, B. A.; Ramamurthy, P. J. Phys. Chem. 1973, 77, 3052. (8)Gay, I. D. J. Magn. Reson. 1984, 58, 413.
0 1994 American Chemical Society
Adsorption of PCls and OPCls on Silica
Langmuir, Vol. 10, No. 3, 1994 767 .1
'1 W
0 Z
8LI:
W
0 .5-
0
z
v)
m
6
Q
E0 o In
m
6
3000 cm-l
2000
1000
Figure 1. Background infrared spectrum,at room temperature, of a silica disk after 1h of activation at 450 O C . The quantity of Si02 used waa 4 mg/cm2. Spectra shown in Figures 2 and 3
are difference spectra obtained after subtraction of this background spectrum. X
I I
SiP=O Y
3600
3800
-1
cm
3400
Figure 2. (A) IR spectrum after adsorption of an equilibrium pressure of 4.5 Torr of 'as-received" PCL on silica. (B) Ae for (A) but with purified PCL (see text). (C) As for (A)but with 1 Torr of OPCb. The absorbance scale applies to curves A and B; curve C is 5 times more intense (the absorbance scale should be multiplied by 5 to give the correct intensities).
Results (a) Short AdsorptionTimes. The infrared spectrum of Si02 after vacuum activation is shown in Figure 1.The sharp peak at 3747 cm-l is due to noninteracting surface silanol groups [SiOH or Si(OH)zl,and the broader features below 2000 cm-l are due to fundamentals or combinations and overtones of Si02 modes of the ~ u b s t r a t e . ~PC1, *~ stretching m o d e ~ ~ 9lie ~ Jbetween ~ 700 and 400 cm-l (however, silica is totally absorbing below 550 cm-9, and P=O stretching vibrations are expected to lie between 1320 and 1100 cm-l. In the latter spectral range, only higher wavenumber P=O vibrations would be expected to be observed in view of the steeply rising background absorption of silica, the practical limit of IR transmission being near 1275 cm-l. The IR spectra shown in Figures 2 and 3 are presented as difference spectra after subtraction of the silica background following adsorption of a given compound. Figure 2A shows the spectral changes observed in the OH stretching region followingadsorption of PCb on silica; there was a decrease in the intensity of the isolated silanol peak at 3747 cm-l (the sharp band pointing down) accompanied by the appearance of two broader bands to lower wavenumber at 3650 and 3500 cm-'(peaks pointing upward). Accompanying these changes, additional bands were observed at 606 and 1290 cm-l (Figure 3A). The spectral features in the OH region, and at 605 cm-l, were similar to those previously reported by Bogatyrev and Chuiko.' Although the 606-cm-l band could be seen in the window of partial transparency without spectral subtraction, that at 1290 cm-l could not be observed without subtraction because of the rapidly rising silica background in this region. Bogatyrev and Chuiko presumably missed this band because they did not have the ability to carry out background substraction. (9) (a) Thomas, L. C. Interpretation of the Infrared Spectra of Organophphorua Compouncb; Heydon: London, 1974. (b) Corbridge, D.E.C. Top. Phosphonur Chem. 1969,6,235. (10) Nyquirt, R. A,; Puehl,C . W.Appl. Spectrosc. 1992,46,1552.
i 1320
1300 -1
cm
700
650
600
550
c"
Figure 3. (A) Low-wavenumber IR spectra of aa-receivedPCL adsorbedon silica. (B)Similar spectrafor pure 0-PCb adsorbed on silica. The intensityscaleappliesto curve A; the red intensities for curve B are 5 times greater.
A slPNMR analysis of the commercially available PCls solution indicated, in addition to a peak at 219 ppm due to PCb, a second peak at 3 ppm corresponding to about 1.696 of the total phosphorus. The 3 ppm peak is due to phosphorylchloride, OPCh. Table 1lists the vibrational frequencies of the Stretching modes of PCb and OPCls. The IR spectrum of the gaseous as-received PCla also revealed, in addition to the strong blended peak near 610 due to the two PCh stretching modes of PCb, additional weak bands at 595 and 1323 cm-l, clearly demonstrating that our sample was contaminated with OPCb. Phosphoryl chloride can be removed from PCh by briefly contacting the mixture with dry AlCb. The IR bands due
Morrow et al.
158 Langmuir, Vol. 10, No.3, 1994 Table 1. Infrared Frequencies (cm-l)of PCla and OPCla OPCls OPC& PCla OPCls PC&h (gas)' (soln)lO (SiOZ)' (SiOZ)'
v(m) u,(PC&)
u,(PClS) 0
1323 594 482
515 505
1300-1310 586-594 482-481
1290 605 n.0.
n.o.b n.0.
This work. b n.0. = not observed.
n
I 250 0
I
I
I70 0
I
I
I
90 0
1 10
I
I
-70 0
I
I -150 0
ppm
Figure 4. Cross-polarization alp MAS NMR spectrum of pure PCla after 18 h of adsorption on silica.
to OPC13 were removed, and an IR spectrum observed followingadsorption of purified PCl3on silica did not show the 605- and 1290-cm-l peaks. Further, the perturbed silanol band at 3500 cm-l was not observed; only that at 3650 cm-1 remained (Figure 2B). Conversely,adsorption of pure OPC13on the same silica sample gave IR bands at 3500 (Figure 2C), 1290, and 605 cm-l (Figure 3B). The assignments for both compounds adsorbed on Si02 are shown in Table 1. Although it might seem surprising that OPC13as such a low-level impurity in PCl3 could give IR bands of significant intensity, the phosphoryl groups confers a significant basicity to this compound1' and it would be selectively adsorbed by silica. Accordingly, the OPCl3-perturbed silanol band is at a lower wavenumber than that which is perturbed by PCl3. Finally the shift of u ( P 4 )to lower wavenumber and of u,(PCl3) to higher wavenumber for adsorbed OPC13 relative to gaseous OPC13 is consistent with a P=O-*HO hydrogen-bonded interaction.10 Following evacuation at room temperature, all of the spectral features noted above slowly decreased, and disappeared after about 1 h of evacuation. Therefore, all changes can be attributed to physical adsorption of PCl3 or OPCls on silica. Finally, NMR spectra which were obtained after PCl3 had been in contact with Si02 for up to a few hours showed no additional peaks other than that due to physically adsorbed PCl3 (see below regarding the formation of chemisorbed species after longer contact times). (b) Long Adsorption Times. If pure PCl3 was left in contact with Si02 for 18 h, new IR bands at 1290 and a doublet near 2490 cm-l were observed, the latter spectral region being associatedwith PH stretching modes. These bands could not be removed by prolonged evacuation at room temperature, or even at 150 "C. The NMR spectrum after 18 h of adsorption showed new weak peaks at 184, 24, -5, and -16 ppm (Figure 4). By virtue of its chemical shift alone the 184 ppm peak can be assigned to SiOPC12. A NMR peak at 186 ppm was attributed to this species by Bogatyrev et al. following reaction of PCl3 with Si02 at 320 OC;12 they may also have observed a peak at 156 ppm attributable to (Si0)2PCl. (11) Corbridge, D. E. C. Studieain InorganicChemistry. Phouphorous, an Outline of ita Chemistry, Biochemistry and Technology; Elsevier: Amsterdam, 19m Vol. 6. (12) Bogatyrev, V. M.; Brei, V. V. Chuiko, A. A. Theor. Exp. Chem. (Engl. Trcrml.) 1988,24,603.
These assignmentsare reasonable given the followingdata for similar model compounds:13
MeOPClz (Me0)aPCl
NMR
El
181 169
508/445cm-lfor P-Clz 496 cm-1 for PCl
The above vibrational frequencies"1ie in a region of strong Si02 absorption, and being at the low-wavenumber limit of our spectrometer, the infrared bands due to SiOPC12 or (SiO)2PC1would be virtually impossible to observe. Insight into the nature of the lower frequency NMR signals can be obtainedwith reference to the NMR spectra of some known Si-containingmodel compounds, and with reference to the NMR and IR spectra of H3PO3 and HaPO4adsorbed on silica. Although the IR spectrum showed a doublet at 2490 cm-l, the ratio of the intensities of the two components varied from experiment to experiment, suggesting that two P-H-containing species were generated. Finally, the chemical shift of H3POl is 0 ppm; that of H3P03 is 5 ppm in solution,16and we found that this gave two lines at 8 and 13.2 ppm for the solid. As will be discussed below, the -5 ppm signal is most probably due to SiOP=O(H,OH), and that at -16 ppm is due to the diadsorbed species (SiO)ZP=O(H). Assuming that some of the species generated are chemically bonded to the surface via a SiOP linkage, we have examined the NMR spectra of a series of model compounds which contain varying numbers of MesSiOgroups bonded to phosphorus. These are listed below:
NMR shift
compound
Me&3iOP=O(H,OH)
-116
(MeaSiO)#--O(H)
-14.9:' -14.3,'8 -13.3l6 -28.3lS
(MesSiO)P4
The value of the chemical shift need not concern us, but the trend is worth noting, namely, an approximately 1214 ppm shift to lower frequency with each additional Si0 attachment to P. A similar trend can be found in the NMR spectrum of phosphoric acid impregnated silica. Silica was impregnated with phosphoric acid (1.2 P/nm2),dried in air at 110 OC, and placed in an NMR tube for vacuum treatment at room temperature or at 200 or 400 OC prior to sealing the tube. The 31PNMR spectrum of the room temperature evacuated sample gave three resonances at -10, -20, and -35 ppm, the -20 resonance being the most intense (Figure 5A). The spectrum of the 200 OC activated sample was similar to the room temperature sample, whereas the 400 OC activated sample had no resonance at -10 ppm and the -35 ppm resonance (now shifted to near -38 ppm) was more intense than that at -20 (Figure 5B). Mudrakovskii et a1.20obtained the 31PNMR spectra of H~PO4-impregnatedSi02 samples containing 1.8 or 5.9 P/nm2 after calcining in air at 100,250, or 500 OC. They (13) Mark, V.; Dungan, C. H.;Crutchfield, M. M.; Van Wazer, J. R. Top. Phosphorru, Chem. 1967,5, 227. (14) Fritzowsky, V. N.; Lentz, A.; Goubeau, J. 2. Anorg. Allg. Chem. 1971, 386, 203. (16)Berenstein,T.;Fink,P.; Mastikhin, V. M.;Shubin, A. A. J. Chem. SOC.,Faraday Tram. 1 1986,82,1879. (16) Racorded in CDCh solution as part of this study; the compounds were prepared by an adaptation of the m e t h d described in ref 18. (17) Livantsov,M. V.; Prkhchenko, A. A.; Lutoenko,1. F.J. Gen. Chem. USSR (Engl. Troml.) Zh. Obshch. Khim. 1986,55,1976. (18) Brazier, J. F.;Houalla, D.; Wolf, R. Bull. SOC. Chim. h..1970, 1089.
(19) See ref 13, p 321. (20) Mudrakowkii, I. L.; Mastikhin, V. M.; Kotearenko, N. S.; Shmachkova, V. P. Kinet. Catal. (Engl. 7'raml.) 1988,29, 166.
Adsorption of PCls and OPCls on Silica
l 100 0
I
/ 60 0
l
/ 20 0
I
l -20 0
I
Langmuir, Vol. 10, No. 3, 1994 I69
I -60 0
I
l -100
wm Figure 5. slP MAS NMR spectra of H a 0 4 (curves A and B 90" pulses) andHa08 (curvesCandD; cross-polarization with TOSS) adsorbed via impregnationof silica (seetext for details). Curves A and C were observed after room temperature evacuation of the impregnated samples. Curve B was observed after evacuation of (A) at 400 "C, and curve D was observed after evaluation of (C)at 150 OC.
3500
cm-l3000
2500
Figure 6. Infrared spectra of HaO4-impregnated silica (800 pmoVg after evacuation at (A) room temperature, (B)260 O C , and (C) 400 O C .
convincingly argued that a NMR peak near -10 ppm was due to chemically attached SiOP=O(OH)2. A signal near -25 to -31 ppm which was not present after roasting the 1.8 P/nm2sampleat 100 O C appeared after roasting at 250 OC and was attributed to a bridging phosphate group. Finally, a peak at lower frequency (-42 to -46 ppm), which was only observed after roasting the 1.8 P/nm2sample at 500 "C, was attributed to a silicon pyrophosphate phase. Infrared spectra show that, with increasing evacuation temperatures (room temperature, 260 and 400 "Cstudied here, see Figure 6), there is a substantial decrease in the number of H-bonded SiOH/POH interactions with in-
creasing temperature (broad band near 3100 cm-l). Additionally,between 250and 400 "C, the number of isolated POH groups (sharpband, 3670cm-l) decreases. This can indicate that more surface SiOP bonds form, on average, form each adsorbed molecule, but does not of itself rule out the possibility that condensation to form P 4 P species also could occur. The weight of evidence from the IR and NMR experiments, when taken with the NMR data from the model compounds, would suggest that the two resonances for HsPOdSi02 at -10 and -20 ppm can be attributed to SiOP=O(OH)2 and (Si0)2P=O(OH), respectively. A similar trend (a chemical shift to lower frequency leading to a greater number of surface P-0 bonds when the temperature is increased) has been reported in a NMR study of phosphated Ti02 and Zr02.21 The third peak near -36 to -38 ppm may be attributable to a triply bound species, ( S i O ) W , or to a polyphoephate species. However, the latter attribution is unlikely given a Si density22 of 7-8 per nm2and a P density of 1.2 per nm2,and in view of the appearanceof the -35 ppm signaleven in the absence of thermal activation. Finally, further evidence of this trend (an increasing number of SiOP bonds as the temperature is increased) also comes from a study of the IR and NMR spectra of phosphorous acid, HsP03, on silica. As stated previously, the slP NMR spectrum of HsP03 in solution has a single peak at 5 ppm, and as a solid there are two lines at 13.8 and 8 ppm. The NMR spectrum of silica doped with HsPOs, dried in air, and evacuated at room temperature prior to sealing the tube has a single peak at about -6 ppm with a slight asymmetry to lower frequency, Figure 5C. If the sample was evacuated at 150 OC instead of room temperature, the peak appeared near -18 ppm, with asymmetry to higher frequency, Figure 5D. The IR spectrum in each case showed a characteristic P 4 vibration at 1290 cm-l, and a PH absorption near 2485 cm-l. As with the HsPOr-impregnated sample, heating would be expected to lead to an increasing number of surface bonds, with a corresponding displacement of the chemical shift to lower frequency. The NMR peaks observed from HsPOs-impregnated Si02 are very close to those observed for overnight adsorption of PCla on silica, being at -5 and -16 ppm, and the PH stretching frequency is the same as that observed for the absorption of HsPOs on SiO2. The evidence suggests that these two surface species are SiOP-0(H,OH)(-6ppm) and ( S i 0 ) 2 P 4 ( H )(-16ppm), the small shift from those observed from the doping of Si02 with HzPOs presumably being due to the presence of other adsorbed species from PCls adsorption, as already discussed. Finally, the weak peak at 24ppm only appeared in crosspolarized spectra, and it cannot be assigned at present. However it disappears after heating the NMR tube at 150 O C for 2 h, and the peaks formerly at -5 and -16 ppm merged into a single broad peak near -10 ppm.
Discussion The reaction of PCls on silica has produced some unexpectedproducts, and this has necessitated a study of the reactions of phosphoryl chloride, phosphoroua acid, and phosphoric acid on the same surface in order to determine the nature of the surface species formed. (21) Randarevich, S. B.; Strelko, V. V.;Belyakov, V. N.;Korovin, V. Y.;Bortun, A. 1. Theor. Erp. Chem. (Engl. Tronel.) ISM,%, 607. (22) Berendsen, G.E.;de Galan, L.J. Liq. Chromatogr. 1978, I, 405.
760 Langmuir, Vol. 10, No. 3, 1994
It is apparent that Pels is not particularly reactive with silica at room temperature and that the results previously reported in the literature can be attributed to the effects of OPC13 as a minor contaminant in PCl3. OPC13 itself does not chemically adsorb on Si02 at room temperature although it relatively strongly H-bonds to surface SiOH groups. For short adsorption times, Pels weakly H-bonds to SiOH groups. For a longer adsorption time, one mode of the reaction of PCk is a simple one-step process involving surface SiOH: PCl,
+ SiOH
-
SiOPC1, + HC1
Although the above reaction is expected, we did not anticipate that all chlorine atoms could be stripped from PC13 to yield the surface species SiOP=O(H,OH) and (SiO)2P=O(H). It is difficult to devise a mechanism for their formation, but given the propensity of P(II1) to undergo oxidation to P(V) species, the first step of the reaction is probably
Morrow et al. We have also founds that a similar process is the first step in the reactions of PMe2Cl and PMeC12 with Si02. Although we have no IR or NMR evidence for the generation of the intermediate C12P=O(H), this is expected to be a reactive species and it presumably subsequently reacts further with one or two SiOH groups to generate SiOP=O(H,OH) or (SiO)ZP=O(H)and two or one HC1 molecules, respectively. Further speculation concerning the mechanism of reaction is unwarranted, particularly as one other unidentified species (+24 ppm NMR peak) was also present as a relatively minor product. Conclusions Phosphorus trichloride reacts with silica at room temperature to give surface SiOPCl2, SiOP-O(H,OH), and ( S i 0 ) 2 P 4 ( H ) species. The last two surface species are also produced from the adsorption of phosphorous acid, H3P03, on silica at room temperature. Phosphoryl chloride, O=PCk, does not chemically adsorb on silica at room temperature, but it strongly H-bonds with surface silanol groups.
Acknowledgment. We are grateful to the Natural Sciences and Engineering Research Council of Canada and Eeso Petroleum (Canada) for financial support. CgP =O(H)
(23)Morrow, B. A.; Laug,S.;Gay, I. D.To be published.
