544
R. W. LOVEJOY AND E. L. WAGNER
Infrared Spectra of Hypophosphorous Acid and Its Salts'
by R. W. Lovejoyz and E. L. Wagner Department of Chemistry, Washington State University, Pullman, Washington
(Received September 07, 1963)
Infrared spectra of polycrystalline films of H3P02 and D3P02 have been obtained a t liquid air temperature and were assigned on the basis of a C1 molecular model with a pentavalent phosphorus atom in a highly hydrogen-bonded structure. Polycrystalline spectra of KH2P0z, NaHZPO2, NH4(HZP02),Ca(HzP02)2,and the corresponding deuterated salts were also obtained a t low temperature and the anions assigned on the basis of Czvsymmetry, except for the calcium salt where evidence indicated a hypophosphite ion of C1 symmetry.
Introduction At various times in the past the structure of hypophosphorous acid, H3PO2, has been accounted for using three separate and distinct models, each employing a different number of hydroxyl groups. The first of these structures proposed on the basis of magnetic susceptibility measurements of several organic and inorganic phosphorus compounds, utilized two hydroxyl groups and a trivalent phosphorus atom, i.e., HP(OH)2. A second structure, containing a proton and a hypophosphite ion, H +(H2P02)-, was advanced on the basis of a Raman study of several oxygen-containing phosphorus compound^.^ A third structure, more consistent with the known chemical features of this acid than the above two, has assumed one hydroxyl group and a pentavalent phosphorus atom, H*P=O(OH). Experimental evidence supporting this last model has been obtained from n.m.r. spin-spin splittings of aqueous H3P02 so1utions.j While no direct structural investigation of HJ?OZhas been reported, the structures of two hypophosphite salts pertinent to this investigation have been deternined. Using X-ray methods, crystalline KHd(H2POz) was found6 to possess symmetry of the orthorhombic space group Cmma (DZhz1)with four molecules per unit cell, while neutron diffraction and X-ray diffraction showed that Ca(H2P0& has a monoclinic unit cell' of the space group C2/c (C2h6),also with four molecules per unit cell. The H2P02-ions of both these compounds were reported to possess CZv symmetry, having the shape of a distorted tetrahedron with oxygen atoms a t two of the corners, hydrogen atoms at the other two corners, and the phosphorus atom a t the tetrahedral T h e Journal of Physical Chemistry
center. However, as will be seen later, for Ca(H2P02)2 site symmetry considerations show this description of HzP02- ion is not strictly correct since the symmetry of the anion must be C1 and not Cev. While the hypophosphite salts have been included in several Raman s t u d i e ~ ~ , ~and - ~ ' infrared investigat i o n ~ ~ ~ -the ' ~ , only published vibrational data on itself is the Raman work of Simon and Feher.4 As a result of the uncertainty in the structural features of hypophosphorous acid, infrared spectra of crystalline H3P02 and D3P02have been obtained a t liquid air temperature in an effort to clarify this ambiguity. To aid in the vibrational assignment of the acid spectra, infrared spectra of the salts KHzPOZ,KDZPO~, NaH2P02, 2, and NaD2P02,XH4 (H2P02),NH4 (D2P02), Ca (H~POZ) Ca(DZPOzJ2 also have been obtained a t room tempera~~
~
~~~
~
~~
(1) Based on a thesis submitted by Roland W. Lovejoy in partial fulfillment of the requirements for the Ph.D. degree, Washington State University, 1960. (2) Department
of Chemistry, Lehigh University, Bethlehem, Pennsylvania. (3) P. Pascal, Compt. rend., 174, 457 (1922). (4) 8.Simon and F. Feller. 2 . anorg. allgem. Chem., 230, 289 (1936). (5) H. S. Gutowsky and D. W. McCall, J . Chem. Phys., 22, 162 (1954). (6) W. H. Zachariasen and R. C. L. Mooney, ibid., 2, 34 (1934). (7) B. 0. LoopBtra, J E N E R (Joint Estab. Nucl. Energy Res.) Rept. Publ., No. 15 (1958). (8) J. P. Mathieu and J. Jacques, Compt. rend., 215, 346 (1942). (9) T. J. Hanwick and P. 0. Hoffman, J. Chem. Phys., 19, 708 (1951). (10) M. Tsuboi, J . Am. Chem. Soe., 79, 1351 (1957). (11) J. S. Ziomek, J. R. Ferraro, and D. F. Peppard, J . Mol. Spectry., 8, 212 (1962). (12) D. E. C. Gorbridge and E. J. Lowe, J . Chem. Soc., 493 (1954). (13) C. Duval and J. Lecomte, Compt. rend., 240, 66 (1955).
545
INFRARED SPECTRA OF HYPOPHOSPHOROUS ACID
io00
3500
3000
2500
2000
1500
1000
I
500
FREQUENCY IN WAVES PER C M
Fjgure 1. Infrared spectra of sodium and potassium hypophosphite.
Thin crystalline films of the hypophosphite salts, suitable for infrared study, were obtained by dissolving the compounds in anhydrous redistilled methanol and spraying these solutions onto a polished alkali halide backing plate with a modified chromatographic sprayer. After the methanol had evaporated, a thin crystalline film of the salt remained on the backing plate which was then mounted in a conventional glass low temperature infrared ce11.l' The calcium salt was not sufficiently soluble in methanol to permit satisfactory film formation, consequently spectra of this salt were limited to those obtained a t room temperature from pressed pellets of KBr or NaC1. Spectra of W3P02and D8PO2were obtained by the same technique as used for the salts. However, since the acids are solids with melting points near room temperature, they could be sprayed directly onto a backing plate by warming them slightly above their melting point. As soon as the acid film was formed, it was immediately frozen with liquid air to a clear colorless - ' I , . '
"
/ " " I '
"
'
I
' I " "
'
"
7
ture and liquid nitrogen t e m p e r a t ~ r e . ' ~A complete vibrational assignment was made for the hypophosphite ion in each of these salts that agreed satisfactorily with B normal coordinate treatment of this ion. Using these results the infrared spectra of crystalline H3P02 and D8P02were assigned on the basis of a CI symmetry model. The acid results were most satisfactorily explained by postulating a structure of molecular aggregates with strong hydrogen bonding between the aggregates.
Experimental Methods and Results Crystalline hypophosphorous acid was obtained from either 30 or 50% aqueous solutions of the acid (J. T. Baker Chemical Company) by removing the water through vacuum distillation and cooling the concentrated liquid acid, as described by Jenkins and Jones ' 6 Samples of the crystalline acid were analyzed volumetrically,16 and the acid assayed 100.1% H3P02. Deuteriohypophosphorous acid was prepared by repeabed exchange of R3POnwith D20,followed by vacuum sublimation of the frozen exchange water. Several of the hypophosphite salts are available conimercially (Fisher Scientific Company) and were used without further purification. The deuteriohypophosphite salts could not be prepared by a direct exchange with DZO; it was necessary to first synthesize and isolate Ba(D2P02)2and then allow this compound to react with a stoichiometric amount of a water-soluble metal sulfate.
L?%k--Tk--A-
' '*do0 FRERIENWm
wm
'
'
'.bo'
"
,
'
a
' *
1000
j
'
'
LOO
PER CY
Figure 2. Infrared spectra of ammonium and calcium hypophosphite. The dashed spectrum of Ca(H2P02)zwas obtained from a pellet containing 5 mg. of sample, the solid spectrum from a pellet containing 1 rng. (14) At the time this investigation was completed the work of Ziomek, Ferraro, and Peppard had not yet been published. The infrared frequencies of the hypophosphite salts obtained from this investigation are in substantial agreement with those reported by these workers. (15) VV. A. Jenkins and R. T. Jones, J . A m . Chem. Soc., 7 4 , 1353
(1952). (16) R. T. Jones and E. H. Swift, Anal. Chem., 2 5 , 1272 (1953). (17) E. L. Wagner and D. I?. Hornig, J . Chem. Phys., 18, 296 (1950).
Volume 68, Number 3
March, le64
546
R. W. LOVEJOY AND E. L. WAGNER
Table I : Infrared Wave Numbers for the Hypophosphite Salts a t -------
2345 2320 1988
1694 1630
1194 1172 1082 1054
Peak wave numbers (cm. )-'-NaDnPOz Ca(HzP0z)z' Ca(ThP0z)P NHd(HzP0,) NNI(DZPO~) 7
KDzPOZ
KHzPO4
1194 846 798 1062
NaNzPOz
2361 2318 1982 1633
1187 1167 1090 1073
812
673
474
468
1685 1610
1182 838 802 1052
2383
1719
1184 1147 1098 1063 1038 926
1183 828 812 1056
823 812 793 472
673
809
664
467
460
Assignment
NHa+ stretch N H 4 + combination ul(a1)PHz symmetric stretch u,(bl)PHz antisymmetric stretch Y? u7 combination 2V l NHa+ bend
2971 2881
2966 2874
2360 1970
1706 1960
1466 1405 1194 1179 1142 1028
1461 1400 1189 825 815 1030
796
668
V7(bl)PH2rock
431
432
vp(
680
708
706
a
- 190"
+
v8(b2)P02 antisymmetric stretch v2(a1)PH2scissor vdbz)PH?wag us(al)POz symmetric stretch ? vb(a2)PHztwist P H D deformation
al)POz scissor
These wave numbers were observed a t room temperature.
Table 11: Infrared Wave Numbers for Hypophosphorous Acid a t - 180" Peak wave numbers (am. -1) HaPOz DaPOz
2700 2200 2388 2319 1610 1360 1259 1184 1124 1055 950 803 700 428
0-H stretch 1796 1768
PH2 antisymmetric stretch
1140
P-0-H (in-plane) bend
Figure 3.
3500
FREWENCY
IN WPVESzoo0 PER
W
500
Infrared spectra of hypophosphorous acid.
glass in the evacuated low temperature cell. The film was then allowed to warm up slowly until crystallization took place; after the film had crystallized completely it was recooled to liquid air temperature and the infrared spectrum obtained. The spectra were scanned over the 4000-400 cm.-l region using a Perkin-Elmer Model 21 spectrometer equipped with either NaCl or KHr optics. Spectra obtained from representative filnis of the hypophosphite salts are shown in Pig. 1 and 2, while spectra of hypoThe Journal of Physical Chemistry
PH2 symmetric stretch
P=O stretch 812 775 97 7 640
E E d
4ooo
Assignment
416
PH2 scissor PH2 wag P-0 stretch PH? rock P-0-H (out-of-plane) torsion HO-P=O scissor
phosphorous acid are shown in Fig. 3. The band center wave numbers of the salts are given in Table I and those of the acid are presented in Table 11.
Discussion of the Salt Spectra Assuming Cpv point group symmetry for the free H2POz- ion, the usual group theory treatment predicts nine normal modes of vibration: four a1 modes, one 8 2 , two bl, and two bz modes. I n the infrared spectrum all fundamentals should be active except a2, as well as all
INFRARED SPECTRA OF HYPOPHOSPHOROUS ACID
overtones except Aznfor odd n, and all binary combinations except A1 X Az and B, x Bz. A suitable set of symmetry coordinates for the hypophosphite ion has been given by Ziomek, Ferraro, and Peppard, l 1 which can be described as symmetric and antisymmetric stretching modes for both the PHZ and POz subgroups, scissor deformation modes for the PHz and POz subgroups, a PH, twist, a PH2rock, and a PHz wag. These results provide an adequate basis for interpreting the spectrum of the HzPOz- ion for the three crystalline salts, KHZPO2,NaHZPO2,and NH,(HzPOZ). There was little ambiguity about the symmetries of the cation and anion for the ammonium compound. From the space group symmetry and unit cell population of this salt, Halford's table'* predicts the HzP02- ion should occupy sites of Czv symmetry and that the N H a + ions should be a t D, sites. While space group information is not available for the other two salts, the spectra of crystalline KHZPO2and NaH2PO2were explained adequately on the basis of Czvsymmetry for the anion. It is known that P-H stretching modes give rise to absorptions in the 2300-2500 cm.-l region. The KHzPOzspectrum exhibited two bands in this region, a weak band a t 2345 cm.-l and a more intense band a t 2320 cm.-l. The PH2 symmetric and antisymmetric stretching modes were assigned to these two absorptions, in this case the symmetric mode vl, being assigned to the higher band and the antisymmetric mode V 6 , to the lower band. This assignment is just the reverse of that made by Ziomek, Ferraro, and Peppard." However, for crystalline KHZPO2the most satisfactory agreement between observed and calculated Teller-Redlich product ratios was obtained with the above assignment. This assignment also gave very good agreement between observed and calculated wave numbers from the normal coordinate analysis. Finally, the assignment was consistent with the Raman depolarization ratios observed for these two modes." The K H ~ P Obands Z a t 1194, 1054, and 474 cm.-' were associated with the three POz subgroup motions, since they were found a t approximately the same wave numbers in the spectrum of KD2PO2. The 1054 crn.-' band was assigned to the POn symmetric stretching mode v 3 . This was reasonable in light of the polarized Raman shift for the H2P02-ion reported at 1048 cm.-'. The band a t 1194 cm.-l then corresponded to the POz antisymmetric stretching mode Vg, while the absorption a t 474 cm.-l was assigned to the PO, scissor deformation v h . A fairly strong Raman shift reported near 924 cm.-l for the hypophosphite ion was completely absent in the infrared spectrum of crystyHine KH2POZ. This
547
wave number must then correspond to the PH2 twisting mode v6, which is predicted to be infrared inactive but Raman active. This leaves three absorptions in the KH2POZspectrum to be explained, a t 1172, 1082, and 812 cm.-l, which were associated with motions of the PH2 subgroup. The rather weak band a t 1172 crn.-', which was partially polarized in the Raman spectrum, was assigned to the PHZ scissor deformation v2. Assuming that the wagging mode occurs a t a lower frequency than the scissor deformation, the weak band a t 1082 cm.-l could reasonably be assigned to the PHz wagging mode v g . The PHz rocking mode v7, was then assigned to the remaining band a t 812 cm.-'. Since all nine fundamentals are thus assigned in the KH,P02 spectrum, the very weak absorption a t 1988 cm.-l must be either a combination or overtone band. In this instance it was assigned to the combination mode vz
+
v7.
The assignment of the NaHZPOzspectrum was similar to that of the potassium salt with minor wave number differences. The most pronounced difference between these two spectra concerned the PHz rocking mode a t 812 cm.-', which was split into a triplet in the sodium salt spectrum. An analogous band shape has been noticed by Corbridge and Lowe,12who reported a very strong single peak a t 817 cm.-l for NaHzPOz.HZ0and a triplet structure for the same peak in the anhydrous salt spectrum. The isotopic frequency ratios of the two bl modes for both potassium and sodium hypophosphite deserve some comment. It was found that on deuteration, the PH, antisymmetric stretching frequency for both salts shifted by greater amounts than expected. For this mode the ratio v(PDz)/v(PH2)was 0.702 for the potassium salt and 0.694 for the sodium salt. On the other hand, the isotopic ratio for the other bl mode, the PHz rock, shifted by smaller amounts than expected, being 0.829 and 0.832 for the potassium and sodium salts, respectively. I n the limiting case of a normal mode in which only hydrogen atoms are vibrating, deutetium substitution should yield a frequency that decreases by the factor 0.707 relative to the unsubstituted frequency. The fact that on deuteration the PHz antisymmetric stretch was shifted by a factor slightly less than 0.707, while the PHz rock was shifted by a factor considerably greater than 0.707 may be explained by proposing a strong mechanical coupling of the two b1 modes in these salts. This coupling could result in a mixing of the two modes that would differ more for the H2P02ion than for the DZPOZ- ion. Finally, a weak absorption was observed near 707 ~~
~
(18) R. S. Halford, J . Chem. Phys., 14, 8 (1946).
Volume 68, Number 3
March, 1964
548
R. W. LOVEJOY AXD E. L. WAGNER
cm.-I in the KD2PO2 and NaD2P02spectra whose intensity decreased as the degree of deuteration increased. This absorption was assigned to a P H D deformation mode resulting from incomplete deuteration of the salt. The assignment of the crystalline NHd(HzP02) spectrum was straightforward. Treating the NH, + ion first, it is reasonably certain that the strong absorption a t 2971 cm.-l corresponded to the tetrahedral NH4+ stretching mode v3, while the 2881 cm.-I band u4 combination mode of the was assigned to the u2 tetrahedral KH, + ion. The 1466-1405 cm.-l doublet then corresponded to the tetrahedral NH4+ bending mode v4. The assignment of the H2P02- bands was analogous to that of the sodium and potassium salts, with slight wave number differences. Unlike the other salt spectra, however, only a single absorption a t 2360 cm.-' was detected in the P-H stretching region which was assigned to the PH2 antisymmetric stretch. The spectrum of crystalline Ca(H2P02)zmas of interest since theoretical and experimental evidence indicated that the H2PO2- ions of this salt possessed C1 rather than CZvsymmetry. From an examination of the unit cell projection along the b-axis, it was evident that the four Ca+2 ions occupied the set of Cp sites, while the eight H2P02-ions were a t sites of C1 symmetry. There are then no selection rules for this compound and all nine fundamental modes should be infrared active. Experimentally, the crystalline Ca(H2P02), spectrum exhibited the usual HzP02- ion fundamentals (except the PH2 symmetric stretch, which again was not observed) plus a weak sharp absorption a t 926 crn.-I. This band, which was completely absent in the other salt spectra and forbidden on the basis of Czv selection rules, was assigned to the PH2 twist Vg. Other corroborating evidence supporting the vibrational assignments for KHzP02, P\'aHEP02,and "4(H2PO2)was obtained from the Teller-Redlich product rule and a riormal coordinate treatment for the C Z ymodel of the hypophosptiite ion. Using the equilibrium bond distance values determined by Loopstra' (1.39 A. for the P-H bond distance, 1.50 4. for the P - 0 bond distance) and assuming all angles to have the tetrahedral value, the harmonic product rule ratios were: 0.508 for the four a1 modes, 0.581 for the two b1 modes, and 0.727 for the two bz modes. From the KHzPO2and KDzPOa spectra the experimental product rule ratios for t/he a], b,, and b2 modes were 0.519, 0.582, and 0.738, respectively. These values deviated by less than 2.5 yofrom the harmonic values and in the expected direction. The normal coordinate calculation was carried out
+
Thc Journal of Physical Chemistry
using the Wilson FG matrix method with the following simpiified valence force potential function 2" = f h [ ( A h J 2 d2fhn[Ahlhz12
+ (W21+
j'oI(A01)~
+
+ (Aoz)'] +
d z S , u [ A ~ ~ ~ z 2jhh[(Ahi)(Ahz) ]2 J
+
+ 2dfP [(Aoi) (Ahlhz) + + 2dJP [(Aoi)(Ahioi) + ( A d (Ahaoi) + (bo,)(Ahioz) + ( W ( A h z o 2 ) I -I- d2j'ho[(Ahi0i)2 + (Ahi02)' + 2foo[(Aoi)(Aoz)I (Aoz)(Ahihz) ]
(Ahd
+ (Ahzoz)'1
where the Ah and Ao values are changes in the P-H and P-0 internal coordinates, respectively; the Ahh values, Aoo, and Aho values are changes in the angular internal coordinates of the H-P-H, 0-P-0, and H-P-0 groups, respectively; aiid d represents the root mean square average of the P-H and P-0 bond distances. Expressions for the symmetry factored G matrix elements have been given by Gerding, >faarsen, and Zijpls; the equilibrium P-H and P-0 bond distances used were 1.39 and 1.50 8., and the H-P-H and 0-P-0 angles mere taken as 105 and 117O, respectively.' The procedure used was to adjust the force constants of the symmetry factored secular equation until the calculated wave numbers for each block gave the best agreement with the observed values from both the H2P02- and DzP02- spectra. A comparison of the calculated and observed wave numbers for KH2P02given in Table 111 shows the maximum deviation between calculated and observed results was less than 1% for 32 of the 36 wave numbers. The force constants that yielded the set of calculated wave numbers were: f h = 3.07, f u = 8.14, f h n = 0.39, s o u = 0.68, j h o = 0.46, fohh = -0.46, j o h o = 0.24, f h h = 0.08, and f0o = 0.20 mdyne/k.
Discussion of the Acid Spectra The models initially considered for the interpretation of the hypophosphorous acid spectrum were the ionic model and the two covalent models with either pentavalent or trivalent phosphorus atoms. The ionic model would presumably consist of a proton and an anion of C,, symmetry and one might expect the crystalline acid spectrum to be similar to that, of the salts, except for the effect of the protons. It is clear that, whatever the effect may be of the protons on the acid spectrum, there should be no distinct OH vibrational modes as there would be for either covalent structure. While the gross features of the acid spectrum are consistent with these general considerations, it is a t -~ (19) H. Gerding, J. W. Yharsen, and D. H. Zijp, Roc. t r m . ehim., 77, 361 (1958).
549
INFRARED SPECTRA OF HYPOPHOSPHOROUS ACID
Table I11 : Calculated Wave Numbers for KH2P04and KDzPOz KDnPOz
KHoPO4
Symmetry species
a1
a2
bi bz
Obsd.
Calod.
Per cent deviation
2345 1172 1054 474 926a 2320 812 1194 1082
2344 1179 1058 474 934 2285 813 1193 1082
0.0 +0.6 +0.4 0.0 +0.9 -1.4 +o. 1 -0.1 0.0
Obsd.
Calod.
1694 1346 1062 468 680" 1630 673 11194 798
1691 839 1061 468 673 1645 671 1192 797
Per cent deviation
-0.2 -0.8 -0.1 0.0 -1.0 +1.0 -0.2 -0.2 -0.1
Assignnient
PHz symmetric stretch PHz scissor PO2 symmetric stretch PO2 scissor PHz twist PHZantisymmetric stretch PHa rock POZantisymmetric stretch PHz wag
These wave numbers were taken from the data for the calcium salts.
best difficult to conceive of a structure existing in the solid state with discrete protons and hypophosphite ions at lattice sites within the unit cell and one is led to conclude that this model is not very realistic. Furthermore, since there were observable and important differences between the acid and the salt spectra, the ionic model was abandoned and will not be considered furt her. Both covalent models have as their highest possible symmetry a single reflection plane and thus belong to the C, or C1 point groups. Either covalent structure would permit twelve fundamental modes to be active in the infrared and Raman effect, however, in principle the spectra of these two models should be distinguishable on the basis of the different types of normal modes involved. While a covalent model is chemically more reasonable than the ionic structure, the fact that the acid spectrum lacked typical OH absorptions in the 3000-3600 cm. -l region indicated that the simple molecular model is an oversimplification of the crystalline state environment. The presence of very bralad absorption bands near 2700, 2200, and 1600 cm.-', with the second band being somewhat obscured by the absorptions at 2388 and 2319 cm.-l, suggested that, a highly hydrogen-bonded covalent structure was a much more likely possibility. Furthermore, the frequencies of these three absorptions tended to favor the pentavalent over the trivalent phosphorus model. Evidence supporting this conclusion was found in the work of Braunholtz, et U L . , ~ who ~ correlated the presence of the 0
//
X subgroup (where X represents phosphorus, arsenic,
\
OH or sulfur) with the appearance of three very broad absorptions near 2600, 2200, and 1700 cm.-l. These
authors also pointed out that in the spectra of compounds containing two OH groups bonded to X, the absorption near 1700 cm.-' appeared to be missing or ill defined in comparison with the 2600 and 2200 cm.-' absorptions. The frequencies and broad nature of the three hypophosphorous acid absorptions were thought to be in sufficient agreement with the other set to be taken as evidence for the fact that only one OH group is bonded to the phosphorus atom in this acid. Also, since the 1600 cm. absorption was always a prominent feature of the hypophosphorus acid spectrum, we believe the pentavalent phosphorus model more nearly accounts for the observed spectrum than does the trivalent configuration. For the pentavalent acid model the PH2 symmetry coordinates should be similar to those of the hypophosphite salts. One would thus expect to find PH2 symmetric and antisymmetric stretching modes, as well as a scissor deformation, a wag, a rock, arid a twisting mode. By analogy with the hypophosphite salts the pair of bands a t 2388 and 2319 cm.-l in the acid spectrum were assigned to the PH, antisymmetric and symmetric stretching modes, respectively. The scissor mode was assigned to the 1124 cm. -l band, the PH2 wag to the 1055 cm.-' band. and the PH, rock to the 803 cm.-' band. The PH1 twist, which should give rise to a band near 900 cm. -I, was not detected. The various phosphorus-oxygen modes of the pentavalent model are somewhat different from those of the salts and consist of a P=O stretch, a P-0 stretch, and a 0
// \
P scissor deformation. The latter mode was easily asOH
(20) J. T. Braunholtz, G. E. Hall, F. G . Mann, and N. Sheppard, J . Chem. Soc., 868 (1959).
Volume 68, Number 9 March, .19(74
R. W. LOVEJOY AND E. L. WAGNER
550
signed to the 428 cm.-' absorption by analogy with the salts. The P=O stretch was assigned to the 1184 cm.-' absorption, which is in agreement with the findings of Bellamy and BeecherZ1for other oxyphosphorus compounds. The P-0 stretching mode then corresponded to the 950 cm. band. This leaves three modes to be accounted for: the 0-H stretching vibration, the P-0-H (in-plane) bending mode, and the P-0-H (out-of-plane) torsion. The broad absorptions near 2700 and 2200 cm.-' were assigned to the hydrogen-bonded 0-H stretching mode, in keeping with the assignment consistently made by other authors. The remaining P-0-H modes were more difficult to assign uniquely since there is more uncertainty about the absorption features of these two bands. The in-plane bending mode should occur in the 1000-1700 cm. -l region and has been assigned to the absorption near 1030 cm. -l in the spectra of diphenyland dibenzylhydrogen phosphate. 21 We prefer to assign this mode to the rather broad absorption near 1600 cm.-' and the weaker band a t 1360 cm.-', although lacking further solid state information it is difficult to justify why there should be two stretching and two bending modes. One possible explanation niay be that strong hydrogen bonding results in exceptional
Th,e Journal of Physical Chemistry
intermolecular coupling. 2o The out-of-plane torsiona mode, which is known to occur in the region below 800 cm.-' in hydrogen-bonded liquids and solids and in hypophosphorous acid, was tentatively assigned to the broad absorption near 700 cm.-'. One further comment on the acid spectra should be made. I n Fig. 3 two slightly different spectra are shown for H3P02. Both types of spectra were observed in a random fashion from crystalline films formed as described from the same starting material. There appeared to be no tendency for one type of convert to the other on warming the film or allowing it to stand a t low temperature. The reason for this effect is not known although it may be that the acid molecules are oriented differently in the two cases. In summary, we believe that the observed hypophosphorous acid spectrum can be explained most reasonably in terms of a pentavalent phosphorus atom model, in which the individual acid molecules are held together in a crystalline structure through strong hydrogen bonding. I n all probability the site symmetry of the individual acid molecules is no higher than C1.
(21) L. J. Bellamy and
L. Beecher, J . Chem. Soc., 1701
(1952).
