The infrared and Raman spectra of crystalline squaric acid

Speo-m

Acts, Vol. 26A,pp.2298to2QQ4.Pe*ysmon Pm

1970. Printed in NorthernIreland

F. G. BACZIN and CHARLESB. ROSE Department of Chemistry, University of Nevade, Rena, Nevada 89507 (Rsc&ed 4 O&&Y 1969) Ah&&-The vibrational spectrum of cry&&line squaric acid (1,2dihydroxycy&butenedione) has been recorded at and below ambient temperature. The infrared spectrum has been taken between 200 and 4000 cm-l, and the Raman spectrum hss been taken between 85 and 2000 cm-r. Eighteen of the twenty inframd active vibrations and twenty of the twenty-four Raman active vibrations have been observed. Factor group splitting has been detected in both the infrared end Raman spectra. The spectrum of the deuterium compound has also been recorded. A correlationbetween the apparent degreesof associationof several cgclcbutenedionesystems and their carbonyl stretching frequencies is given. The 0 - * - 0 distance in squaric acid (2.49 & 0.05 A) hss been estimated by a modifkation of the proceduregiven by Pimentel and McClellan. The results of our investigation indicate that squaric acid possesses an exceptionally strong hydrogen bond which is best describedby a very asymmetric double minimum potential function.

1. INTRODUCTIOW

the synthesis of 1,ZdihydroxycycZobutenedione (hereafter denoted as aquario acid) was reported several years ago [l], no structural or dynamic studies hrtve been reported on this molecule. The lack of such information was not, however, our sole reason for studying this substance. Squaric acid is the smallest system [2] (atoms per molecule) in a unique series of four member ring oompounds tha.t may be described as possessing spa hybridization about each of the carbon atoms. Thus, the electronic conf$uration is formally taken to be identical to the hybridization found about the carbon atoms of cyclobutadiene, a oompound whose preparation has eluded chemists for many years [3]. Moreover, the structure of squaric acid is, in several respects, as unique as that of c~clobutadiene. The HO-C==H (I) moiety of squaric acid is extremely unusual and is not found in any characterized aliphatic molecular systems other than the unique H&&O, series, of which squaric acid is the parent. The remainder of the molecule, containing the O=C-C=O (II) grouping, acts as an electron sink and through interactions with the (I) structure, in the same molecule and in other molecules, it gives rise to some unusual solution and solid state properties. The formation of the squrtrrtteions (C,0,2-) is strongly favored [4] once the molecule is in solution end &Ba result the pR, and pK, values a,re very small (1.2 and 3.48 respectively) [5]. In the solid state, however, the intermolecular interactions (lattice energy) are strong enough to limit squario acid’s solubility to about 2% in water at 296’K. COHEN et al. [1] noted the latter property and in conjunction with the compound’s high melting point (>573%) proposed ALTHOUGH

[l] S. COHEN, J. R. LAOEER and J. D. PAW, J. Am. Chsm. Sm. 81,3480 (1969). [Z] See for example dimethylcyclobutenedione(Ref. [Zl]) and diammocycZobutenedione(S. COHEN and S. G. COHEN, J. Am. Chem. Sot. 88, 1633 (lQSS)]. [S] See R. WILLST~~~, Aua Mehaem Leben, p. 99. Verlag Chemie (1949). [4] R. WEST and D. L. POWELL, J. Am. Chem. Soo. 85, 2677 (1963). 151 D. J. MACDONUD, J. Org. Ohem. 88,4559 (1968). 2293

2294

F. G. BAGIJN end C. B. Rosa

that the magnitude of the hydrogen bonding in this compound should be quite large. Prior to this current study the attention of previous investigators had been largely directed toward the properties of the squarate ion [4, 6-11-j rather than the acid itself. Thus, we were interested in recording the vibrational spectrum, in approximately describing the normal modes of vibration of this system, and in learning more about the nature and magnitude of the intermoleoular attraction and its effect on the intramolecular vibrational spectrum. 2. EXPERIMENTAL The squaric acid used in this study was obtained from Henley Chemioal Company, Inc. and was used without further purification. Elemental analysis gave the following result : c Calc.

Obs.

42.12 42.00

H

1.77 1.98

It should be noted, in relation to this analysis, that no evidence of a hydrate wss found. This type of “contamination” is found in oxalic acid, a compound to which squaric acid has been likened in the past [5, lo]. Thus, the possible spectral interference from lattice water is eliminated. The thin 6lms of squaric acid were grown on cesium iodide plates in a sublimation cell at 6 x 1O-5 ton: pressure. In order to hasten the growth process the cell was heated to ~373°K. The estimated film thickness was 5 ,u or less. The plates were placed in standard low temperature infrared cells and spectra were recorded at room temperature (296”K), dry-ice acetone temperatures (195’K) and at liquid nitrogen temperatures (77°K). All temperatures referred to are cryostat temperatures. The sample plate temperatures are usually 10-25’ higher depending upon the vacuum in the cell and heating by the infrared beam. Squaric acid-d, was prepared by refluxing the light compound in 4 ml aliquots of 99.8% D,O at 373°K for 24 hr and the solution then lyophilized. The procedure was repeated four times. Finally, metallic sodium was added to 2 ml of a solution of D,O and partially deuterated squaric acid. The system was then refluxed for 27 hr after which it was neutralized with 38% DCl in D,O. The resulting solution was then lyophilized. Unfortunately, it was not possible to follow the deuteration by observing the infrared spectrum ; hence the need for the extensive deuteration procedure. This problem will be examined further in the discussion section. The films of heavy squaric acid were grown and data recorded by the prooedure described for the light system. The infrared data were obtained on a Beckman Model IR-12 spectrophotometer (optical frequency range 4000-200 cm-l). The Raman data were recorded on a Cary Model 81 Raman Spectrophotometer equipped with Toronto Arc source and a [6] M. ITO and R. WEST, J. Am. Chem. Sot. 86,268O (1963). [7] R. WEST md H. Y. NIU, ibid 86, 2686 (1963) [S] [9] [lo] [ll]

R. WEST md H. Y. NIU, ibid. 86,2589 (1963). M. V. WERKEEXA, Disc. Abe. 26,639s (1966). P. H. TEDESCOmd H. F. WALTON, Inorg. G-m. 8, 932 (1969). D. T. IRELAND and H. F. WALTON, J. Php. Chem. 71, 761 (1967).

The in&wed and Rama

spectra of orystalline squaric acid

2296

filtering systam con&sting of a circulating saturated solution of sodium nitrite in water and a Brigham No. 8 gelatin filter. This filtering procedure was particularly effective in isolating the 4358 A Hg line, but did not prove as satisfactory in the higher frequency region when powdered crystal samples were used. A sample cell of the Keller-Busey design [12] was used to record the Raman spectra of polyorystalline squaric acid. Because of the relatively large amounts (l-2 g) of squaric acid necessary to fill the Keller-Busey cell, we were not able to record the Raman speotrum of the deuterated material. The average precision of the Raman data obtained from the solid cell was estimated to be f4 cm-l. The precision of the infrared data was estimated to be f6 cm-1 from 200-1000 cm-l, 3 10 cm-l between 1000 and 2000 cm-i, and f 15 cm-l between 2000-3000 cm-l. The principal o&useof this reproducibility problem arose from the inherent band shapes as can be seen in the following section. 3. EXPERWENTALRESULTS Figures 1 and 2 show the infrared spectra of the light compound as recorded at 296 and 77°K respectively. Figure 3 shows the infrared spectrum of the heavy compound at 296°K. From these spectra the reproducibility problem referred to in the previous section may be clearly observed, particularly in the 1000-1500 cm-r region. Furthermore, Figs. 1 and 2 show that, in contrast to most spectral temperature studies, the spectrum recorded at 77°K does not appear to be as well resolved as that taken at room temperature within the 900-1500 cm-l region. Because of these findings, all frequencies shown in Table 1 were taken from spectra recorded at 296°K (Fig. 1). In the same manner we have included only the spectrum of deutero-squaric acid recorded at 296°K. Furthermore, since there were no relevant changes observed at 195OK(dry-ice acetone slush), spectra recorded at these conditions are not included here. Figures 4 and 5 show the Raman spectra of polycrystalline samples of squaric acid and of a saturated solution of squaric acid in dimethylsulfoxide (DMSO) respectively. The designation e in Fig. 4 refers to mercury emission lines, A band (not shown in Fig. 4) was observed at 85 cm- l. A MgCO, blank was examined in the frequency region below 150 cm-l for comparison. The 85 cm-l band (Table 1) was clearly more intense than a very weak background feature seen at about 90 cm-l in the blank. The fluorescence seen in Fig. 5 was apparently concentration dependent and may indicate the formation of a DMSO:squaric acid complex. This hypothesis is strengthened by the development of a very light pink color in the solution as the concentration of squaric acid was increased. Our tentative conclusion concerning the species present in DMSO is that the squaric acid may lose one of its protons to DMSO to form the bisquarate ion. As we will see in the next section, the spectrum in Fig. 5 is not that of the squarate ion itself. Figure 6 is the infrared spectrum of dimethylcy&butenedione (DMCBD) [13]. It illustrates a model “non-associated” system with which we can compare the spectrum of squaric acid. [12] [13]

R. H. BUSEY and 0. L. KELLER, JR., J. C7wm. phys. 15,268 (1964). Professor A. T. Blomquist of Cornell University graciously furnished the spectrum of DMCBD from a thesis submitted by R. A. VIEBLMu, Cornell University (1962).

F. G. BAQLIN

4&O

3600

3200 I

2.300.

2400

2000 I

and C. B. ROSE

1700

1400

1100

‘800 1

300’

2

1

Wavenumbers

Fig. 1. The infked

40001700

I

I

3600

I

3200

2000

I

qmctrum of squaric acid recordedat 290°K.

I

2400

I

I

1

1100

2000

1 800

I

0

5

Wavenumbers

Fig. 2. The infrared spctmm of squtic acid recorded at 77°K.

4000

. .

3600

I

1

3200

2800

2400

I

I

2000

1700

1400

1100

1

*

800

500

Wdvenumbers

Fig. 3. The infiwed spectrum of squario acid-d, recorded at 296°K.

10

2297

The inkwed and Raman speaka of cryetallinesquaric acid

Wavenumbers

Fig. 4. The Raman spectrum of squaric acid recorded at 3OD’K.

L

ib3 Fig. 5. The Raman spectrum of a saturated solution of equarioacid in DMSO.

Emily, Table 1 gives our tentative spectral assignments for squaric acid, These will be discussed in the next se&ion, however one pertinent comment should be made here. There are extensive two phonon processes and broadening effects oocurring in this system which are clearly seen in both the in&red and Raman spectra. Since these effects do not allow an unambiguous analysis of certain spectral features, the approximate normal mode desoription is necessarily tentative.

2298

F. G.

BAGLIN

and C. B. Rosn~

Table 1. Assignment of fumhmentsl and prominent anhtumoniovibrations in aquarioa&d* Iufrered frequenoies (am-‘) Deuterium

-

Protinium

-

Reman frequenoiee (cm-i) Protinium 86 163

230

225

230(?)

aileut meeked 306 340 380 silent 613 636 127 862 760 760 1 1050 1060 1168 1270 1360 1090 1100 1620

silent 268 308 350 379 silent 626 635 731 866

241 274 308 &Lo. 383 445 626(?) 639 736 868

1660 1688 1818 2218 1680 to > 1610 2340

91.5 *+ 928 1 1056 1070 1166 1246 1318 1360 1380 1513 1643 1822 2216 2236 to 2.276 > 2360

2440

2440

2680 2866 2925

2600 28.50 2920

Aeeigmnente lattice mode: proposed origin. treneletion lattice mode: proposed origin, libretion l&ice mode: propoeed origin librstion. {intremol. bending modes tYl-S?r(CC’C’w,. (r(CW,l(y,) t ye-&CO)B,, [libretion] y*-n(CO)B, 268 + 86 = 363

Y,-Kw% t Y,-T(C’O’)-+, brn(CW(v3 Y,-n(C’O’m.

&S@mc~cjq Y~-&c’o’)A,

Y,,-S(C’O’)B, Ye,? (see text) t vl,-vPw,, cV(cc)I t YwYwc’)% CVWH

11.0.

YI~S=W,,

1063 1062 1174 1261 1307 1362 1386 1614 ** 16321 1630 1677 1829

t vl~-v(C’C’)A,, [Y(C’O’)] t Y,,-VWP,, bKw1 t ~1,-~Ww% bvwl t *I,-&OH)-% bW)l t ~w&OHPp WW Yso-YW)-% [Y(CC’)I v*&C’O’)A,

LO. Zl.0.

ILO.

LO. ll.0. LO.

1380 +

Ym-Y

308 =

1688

(C’O’)B,

1063 + 1174 = 2227 yes-y(OD)Be, v(OH)Be likely involved in Fermi reeonence; Y.~? (eee text) 2(1174) = 2348 -1063 + 1307 = 2360 1368 + 1063 = 2438 1251+ 1307 = 2668 2(1307) = 2614 1614 + 1318 = 2868 1630 + 1307 = 2937

l Iufrered data reported in this table were recorded et 296’K, whereas the Reman data were recorded et 309’K. All eeeignmeuta aeeume C,, eymmetry (see text). t A pair of vibratioue whioh have been shown to be mechenically ooupled in similar aystame. The braokete refer to the proposed minor oomponent of the coupled pair. $ The prime-eeeeu throughout the eeeigmnente (C’) refer to the carbon end oxygen atoms of the oerbonyl group. 8 The deeiguetion no. refers to vibrations that were not observed. * * Aeaigued to factor group (Devydov) splitting.

4. Dnxx~ss~o~ 01~ RESULTS Introduction To our knowledge no X-ray studies have been performed on this molecular crystal. This laok of structure data has two important consequences. Fir&, if we observe splitting of intramolecular fundamental vibrations we oannot tell whether the number of molecules in the unit cell is directly indictated by the splitting or whether we are observing only the infrared active factor group oomponents. In addition, we have no knowledge of the actual site symmetry within the unit cell, e.g. if it were C,, then any assumptions about the presence of a moleoular plane would be 4.1

The infrared and Reman qmctra of crystalline sqmric mid

Wavelength

in

2299

microns

Fig. 6. The infrered spectrum of neat dim&hyl&obutenedionc

(see Ref. [13]).

incorrect. Second, it has recently been shown by other workers [14, 151 on similar intermolecular hydrogen bonded systems that a realistic normal coordinate analysis must include certain external modes of motion in the force field. Without unit cell dimensions and moleculrtr orientations, such a c&ulation oan easily be very misleading with regard to the descriptions of the low frequency and strongly coupled higher frequency normal modes of motion. We have assumed C,, symmetry for this molecule since we can show, on the basis of molecular models, that it is a reasonable assumption that squaric acid forms its hydrogen bonds in the form of itite planar chains. On the other hand, a sheet structure will force the hydrogen atoms out of the molecular plane and hence lower the molecular symmetry. The genuine irreducible representation of the 24 internal fundaamentalvibrations is given below followed by the irreducible representations for bond stretching coordinates, in-plane angle bending coordinates, and out-of plltne angle bending coordinates (prime denotes this irreducible representation). I? = 9A, + 4A, + 3B, + 8B,

(1)

r, = GA, + 4B,

(2)

IYE= 3A, + 4B, I’;

=4A,

+3B,

(3) (4)

The A, species is forbidden from the infrared spectrum ; all other species are infrared active whereas all species are Ramaa active. Thus, we expect 20 fundamental vibrations in the infrared spectrum and all 24 fundamental vibrations in the Ramitn spectrum. A majority of spectral features assigned to fundamental vibrations in squaric eoid were f&lit&d by the results of previous investigators [6, 14-191. [14] [lti] [16] [17] [IS] [lQ]

Y. MDUWA, J. W. BRUCE and R. J. JAKOBSEN,J. Mol. &w&y 24,314 (1967). M. S~~KI and T. SEIBEAN~TJ~EI, J. Mol. Spay 28,394 (1968). IL NAKAMOTO,Y. A. SARBEA and G. T. BEENKE, J. Chem. Phye. 42, 1662 (1966). DE.HAIJBD and A. NOVAK, Spectmchim.Actu 21,1217 (1966). P. MIRONE and P. CELORBOLI,ibid. l&l426 (1962). R. C. LORD and D. G. &A, J. Am. Chma. Sot. 7@, 2401 (1967).

2300

F. CL BAGILINend

C. B. Ram

4.2 The qmtral region from 86 to 860 cm-l In this region, 10 of the 24 internal crystal lattice vibrations occur and a large number of the external lattice vibrational features should also be found. These latter effects were measured mainly by Raman scattering techniques and only three were identified. The translational type lattice modes usually give rise to nonobservable or very weak bands in the Raman spectrum, whereas the librational modes usually give rise to prominent bands (especially if they arise from a totally symmetric lattice mode). Unfortunately, no such general rule is applicable to the infrared intensities. On this basis and the fact that none of these bands are observed in solution spectra (see Fig. 5), we have assigned the three lowest frequency features of Table 1 to external lattice vibrations and qualitatively described their origin. The remaining assignments (Y~-Y& were made primarily from the experimental and theoretical work on the squarate ion [0], secondarily from the group theory demands placed on these bands by others found at higher frequencies, and finally by the qualitative depolarization ratio measurements on saturated squaric acid-DMSO solution spectra. The designated order of the sources of assignment given above indicates their contribution to the final results shown in Table 1. We were unable to make better use of depolarization ratios because of the large number of coincidences between the solvent and the solute bands. The bands at 274,639 and 736 cm-l were found to be dp, dp and p respectively. This result is consistent with the other criteria for spectral assignments given above. One band, Y@(Y(CO)AJ, was not assigned to a definite band in the region. However, it is felt that this normal coordinate very likely contributes to both y1 and vs. We have assigned y1 to the band at 241 cm-l, due to the absence of a coincident infrared band, due to the band’s intensity and finally by analogy to previous work on maleic anhydride [18]. The assignments of ygand or,,to the 735 and 858 cm-l bands were done on the basis of the assumed chain geometry of the system, the very strong hydrogen bond present in this system and the fact that ring breathing modes (involving two oxygen and the four carbon atoms) may also contribute to these normal coordinates. As a result, the carbonyl rock and deformation modes are found at these unusually high frequencies. Indeed, a general and final comment about this low frequency region is that the simplified designations given in Table 1 may well involve six atoms, rather than just four as indicated by the mode descriptions. 4.3 The spectral region from860 to 1400 cm-l In this region of the vibrational spectrum more is known about normal mode descriptions and the frequencies at which they are likely to appear, than in the previous region. However, as has been found in potassium acid maleate and croconic acid [ZO] (a five member ring of similar structure to squaric acid but containing an additional carbonyl group) the identitlcation of the individual normal coordinates is at best very difficult. The shape of the band, seen in Figs. 1 and 2, between 860 and 1400 cm-l well illustrates the difllculty. The order of the assignments and the approximate description of the normal modes is after %KAMoTO and co-workers [16]. Another source of supporting evidence comes from the work on succinic acid pc] K. YAMADA, N. MI~JN~ and Y. H~ATA, Bull. Chem. Sot. Japan 81,643 (1955).

The infrared and Raman spectra of cry&abe

squaric acid

2301

by SUZTJKX and S~~KAN~~~EI[15]. The general rule of placing the asymmetrio component of a partictular vibration at a higher fiequenay than the symmetrio component was followed for those assignments where no other criteria were available. The increase in frequency of ylBand vl, upon deuteration is equally well supported by other investigators [14, 16, 171. The OH torsional modes of species & and 8, were only partially characterized. This was due to the fact that neither of these vibrations could be observed in the Raman spectrum. The B, component, observed in the in&ared, appears to be split into two components (915 and 928 cm-l in the light compound and 750, 760 om-l in the heavy compound). We tentatively assign this phenomenon to factor group (Davydov) splitting. The very intense Raman band at 1174 cm-l and weaker band at 1053 cm-l were polarized in the DMSO solution work and are assigned to A, modes, principally involving carbon-carbon stretching coordinates. The remaining bands in this region gave rise to a poorly defined Raman spectrum. In the solution spectrum of Fig. 5, it can be seen that polarization measurements were not possible on the bands between 1250 and 1400 cm-l. 4.4 The spectral region above 1400 cm-l

Three fundamental vibrations are expected below 2000 cm-l. They are the C==C stretch and two C==O stretches. All Raman bands observed in this region were weak, broad and poorly defined; therefore, we relied primarily on the infrared results. The infrared spectrum of the solid (Fig. 1) shows a strong broad band at 1513 cm-l with a slight shoulder at 1540 cm- l. The Raman spectrum of the solid (Fig. 4) shows two weak features at 1514 and 1532 cm-l. Since the frequencies are reproducible to only &lo cm-l, the most reasonable explanation would seem to be to assign these observed Raman peaks to the symmetric component of a factor group splitting interaction for the A, C==Cstretch, v~,,. This symmetric splitting would not be expected to be as well defined in the infrared spectrum. This observation is verified by examining Fig. 1. In the spectrum of DMCBD [13, 211, Fig. 6, it is clear that the C=C stretch must be assigned to the 1580 om-l band. This higher value found for vaOin DMCBD is due to the lesser degree of delocalization of the electrons in DMCBD as compared to squaric acid. This assignment of the v,,, to the 1513 band in squaric acid is also consistent with the spectral results of monomethoxy-, dimethoxy-, diethoxy- and diamino- derivatives of squaric acid [22]. The assignment of the carbonyl vibrations in squaric aoid are made in a rather straightforward manner from infrared data, but are rendered somewhat less certain by the Raman data (Figs. 4 and 5) tm was mentioned above. Again, relying on the spectral data recorded for similar systems [22] and illustrated here by DMCBD [21], we have assigned val to the symmetric O=O stretch at 1643 om-l in the infrared and 1630 cm-l in the Raman (note that these frequencies are again subjeot to the &IO cm-l uncertainty). The antisymmetric component, vaa,is assigned to the 1822 cm-l band in the infrared spectrum and is assigned to the 1829 cm-l band in the Raman (this band is located between the two emission lines seen in Fig. 4). Upon 6rst inspection one might object to the large frequency of separation between the two [21] A. T. BUJMQUISTand R. A. VIEIGINca, TebrahedronLett. 665 (1961).

[22] See the second part of Ref. [2].

F. G. BAULIN and C. B. Rosn

2302

components val and vzz and to their unusually low and high values. However, in COHEN and COHENS’work [22] on several different derivatives of squaric acid mentioned above, the antisymmetric component was found above 1800 cm-i in four out of the five systems they prepared. Only in the diethoxy- and dimethyl- derivatives is the high frequency component below 1800 cm-l. On the other hand, whereas the antisymmetric stretching frequencies are rather insensitive to the nature of the groups at the 1 and 2 positions on the ring, the symmetric stretching frequenoies are quite sensitive to the nature of adjacent groups which allow increased intermolecular association. Consider the values of the antisymmetric and symmetrio carbonyl stretching frequencies given in Table 2. Notice that in the substance Table 2. Carbonyl frequencies found in various cyclobutenedione derivatives Molecule dimethyl-* diethoxydimethoxymonomethoxydiSAIlO-

dihydroxy-

Antisymmetric stretch (cm-r) 1786 1786 1802 1818 1818 1829

Symmetric stretoh (cm-l) 1740 1724 1724 1724 1668 1630

Melting pointst b.p. 506 b.p. 560 328 406 523 d 673 d

* DMCBD frequencies from Ref. [21], all others are taken from Ref. [22]. t &king points are reported exoept where substances are liquids et room temperature or where they decompose (d). All temperatures are in “K.

expeoted to be the most non-associated, DMCBD, the frequencies are close together. The antisymmetrio stretch is found at a lower value and the symmetric stretch at a higher value than in the other molecules. Indeed, the relative degrees of expected association seems to be indicated by the positions of the carbonyl frequencies in these compounds. Thus, in order of increasing association of the cyclobutenedione derivatives we would have dimethyl- (DMCBD), diethoxy-, dimethoxy-, monomethoxy-, diamino-, and finally dihydroxy- (squaric acid). Melting and boiling points are included in Table 2 as an approximate cross check on the degree of association. Then this series really represents the ionicity of the protons on the groups (atoms) located at the 1 and 2 positions of the cyclobutenedione ring and their ability to form head to tail dimers with the carbonyl groups of nearest neighbor molecules. These dimers could be stabilized by intermolecular resonance forms in which one carbonyl group remains intact and the other captures a proton from the group (atom) at the 1 or 2 position. Thus, the resulting resonance structure’s relative importance as a contribution wave function to the overall intermolecular potential function would increase from DMCBD to squaric acid. Furthermore, as discussed by MIRONE and CHIORBOLI [18] on an intramolecular basis for maleic anhydride, the separation of the two carbonyl stretching frequencies in the oyclobutenedione derivatives would be attributed to the intermolecular eleotronic (resonance) argument given above. The last assignments to be made in squaric acid are the antisymmetric and symmetric O-H stretching frequencies vzSand yap. The latter has not been observed in similar systems [15, 161 and was not observed by us. The remainder of the discussion deals with the location of the antisymmetric stretching vibration, Y(OH)

The infrared end Ramen ape&a

of orpetallinerquario aoid

2303

B,. From Figs. l-3 it is clear that the OH vibration probably lies between 2000 and 2600 cm-l. The features at 2850 and 2920 am-l may be oonfldently assigned to anharmonio features as shown in Table 1 (see Figs. 1 and 3). The lack of any signifloant change in the intensity of this very broad absorption between 296 and 77’K is indiaative of the presence of a very asymntet& double minimum potential fimotion in the O-H ---0 system [23]. The fad that the intensity of this broad band has not appreciably changed in the spectrum of the squario aoid-d, (Fig. 3) may be due to the presence of several anharmonic features as well as Fermi resonance [14, 171. The three shoulders on this broad band oould be assigned to transitions of the asymmetric double minimum potential; however, we favor the assignment of these features to overtone and/or oombination bands. The fact that no really significant changes occur in these features in the spectrum of the heavy oompound suggests that Fermi resonance may not be important. However, sinoe the band is so very broad, we do not believe it is profitable to further disouss the origin of these shoulders. Figure 3 gives the only really concrete information we have on the antisymmetrio OH stretching frequency. The apparent shift in the 1650 cm-l band in Fig. 1 to 1600 cm-l in Fig. 3 is, on closer inspection, seen to be a new band whose maximum is located somewhere between 1680 cm-l and 1610 cm-l. Indeed the 1650 om-l band is still present and can be seen as a shoulder on this new band. We have assigned this new absorption to the r(OD)B, vibration. By assuming an isotopio shift parameter of 1.414, we can calculate that the Y(OH)B, must be in the region of, and most likely between, 2235 and 2275 cm-l. This value is to be compared to the values observed for the following crystals: acetic acid, 2875 cm-l; formic acid, 2900 cm-l; suooinic acid, 2990 cm-l; and potassium hydrogen maleate, 2008 cm-l (calculated, but not observed). The calculated hydrogen bond frequency of this latter substance arises from an intramolecular bond and has been shown by several structural studies [24-271 to exist in a single minimum environment. The O-H force constant in potassium hydrogen maleate was calculated to be 1.10 may-n/A in contrast to the O-H force constant value in monomeric acetic acid of 6.9 mdyn/A. Clearly, the O-H force constant of squario acid must be nearer the 1.10 mdyn/A value. A crude oaloulation of this force constant for squario acid yields a value of 3.0 mdyn/A. Since this calculation did not consider this to be a three body problem, 3.0 mdyn/A must be the upper limit of the true value. If the 153 cm-l band seen in Fig. 4 is assigned to Y, (PINENTEL and MCCLELLAN’S [28] notation for the interbody vibration), then the value of the force constant for this mode is 1.6 mdyn/A. The true values of both the intra and inter molecular hydrogen bond force constants very likely lie somewhere between these two values. In order to estimate the 0 -* 0 distance in squario acid we used the general procedure followed by Pimentel and MoCellan as modified by BELLAMY [29]. The l

[23] [ 241 [25] [26] [27] [28] [29]

R. L. SO~~ORJAI snd D. F. HORNIU,J. C?m% P&a. 36,198O (1962). 5. W. PETERSON and H. A. LEW, J. Chn. P&s. 29, 948 (1968). S. FORSEN,J. Chtxn. P?y.s 81, 862 (1969). 8. DARLOWand W. COCERAN, Aota Cryst. 14, 1260 (1961). R. BLINC, D. HADZI 8nd A. NOVAK, 2. Electroch. &Q,566 (1960). G. C. PEMENTEL end A. L. IkfOCLmLm, Ths HydrogenBond, p. 68. W. H. Freeman (1960). L. J. BELLA~ and A. J. OWEN, Spe&roc7&n.A&u $%A, 329 (1969).

2304

F. G. BAGIZN

and C. B. Roa~l

calculated distance is 2.49 f 0.06 8. Examination of the 0 - - - 0 diattmce values for m-ystalsfound in Pimentel and McClellan’s book shows that only cobaltous aoid, crystalline hydrogen fluoride, the bifluoride ions in various salts and, from Nakamoto’s work, potassium hydrogen maleate have shorter and hence stronger hydrogen bonds than squaric acid. If this distance is correct, within the error limits shown, the hydrogen bond of squaric acid should be of about the same strength as the hydrogen bond in pots&urn dihydrogen arsenate (2.628) [30].Thus, squario acid’s enthalpy of hydrogen bonding would appear to be stronger than that of any known carboxylic acid and, in fact, stronger than nearly every organic system studied to date. It will remain for further studies to bear out this assertion. 4.5 Anhamzonic features It is clear from Fig. 1 and Table 1 that several weak features in the infrared spectrum have not been assigned. However, this is caused by the selection rules being so lax that only those anharmonicities resulting in an A, species are infrared inactive. Because of the large number of plausible explanations for these features, it was not deemed reasonable to guess at their assignment. This decision was also reached on the basis of the uncertainties in the assignments of some of the fundamental frequencies. We have assigned the more prominent overtones and combinations bands or indicated where others might be interfering with fundamental modes (the former case is illustrated by the 1200-1400cm-l region and the latter by the 1500-1900cm-l region in Fig. 1). Acknozule&ement-The authors wish to thauk Dr. DAVID 5. MAODONALD for a generous gift of squaric acid, the Research Advisory Board of the University of Nevada for fhmncial assistance, and the helpful comments of Mr. Hnnn HABERof Cary Instruments. Note added in proof Examination of the squaric acid sample via lsser Raman spectroscopyhas revealed that the strong line at 163 cm-1 is indeed composed of three bands. The main peak is split into two components at 152 and 164 cm-l. In addition, the high frequency shoulder,which wss identitled in the text as a lamp artifact, is in fact a real feature. The maximum of this signal is located at 160 cm-l. All these bands show strong temperaturedependencywith respect to frequencyand intensity. This reinforcesour assignments of them to lattice vibrations. Special thanks are due to the following instrument companies for supplying this data: Jarrell-Ash, Spectra-Physics and Spex Industries. [30] Ref. 1281,p. 284.