J. Phys. Chem. 1994, 98, 10935-10939
10935
A 2HNMR Study of the Phases and Motions in Solid Oxonium Perchlorate, D3O+C104- t C. I. Ratcliffe Molecular Structures and Dynamics Group, Steacie Institute for Molecular Sciences, National Research Council Canada, Ottawa, Ontario KIA OR6, Canada Received: February 16, 1994; In Final Form: July 11, 1994@
2H NMR has been used to reexamine and clarify the phase behavior and molecular motions in solid D@+C104-. In the low-temperature phase reorientation about the molecular 3-fold axis of the D30+ ion is confirmed. The NMR results above 200 K, together with calorimetric data, show some dramatic effects due to eutectic behavior arising from small quantities of water in excess of the monohydrate stoichiometry. The true phases of D@'C104- have now been identified; only one solid-solid phase transition occurs, with hysteresis, at 245 K on cooling and 254 K on warming. Motion in the room temperature phase is not isotropic. Isotropic lines reported earlier presumably were due to the molten eutectic mixture.
Introduction Perchloric acid monohydrate, H30+C104-, is the classic example of an oxonium salt.' It has been studied more frequently than any other oxonium system. The literature is extensive, and the interested reader is referred to a recent review which includes most of the references concerning this salt;2 only those references which are most relevant to the current work will be included here. H@+C104- has two well-established crystalline phases: an orthorhombic phase, space group Pnma, stable at room temperature3s4and a monoclinic phase, space group P21/n, stable at low temperature^.^ Room temperature neutron diffraction studies6*' indicate orientational disorder of the oxonium ions. The presence of two phases for H@+C104- has been confirmed by adiabatic calorimetry,8 but a third intermediate phase occumng over a very short temperature range was suggested by the results for the deuteriated material D30+C104-.9 Curiously, powder diffraction studies on this intermediate phase showed very little difference between this and the lowtemperature phase.7 'H NMR studies have appeared frequently (see ref 2). Early 'H NMR line shape studies of perchloric acid monohydrate were the first experiments to unequivocally establish the existence of an oxonium ion H30+ as a stable entity,lo,l1and the H---H distance was determined to be 1.70 A. 'H line shapes, second moments (Mz), and spin-lattice relaxation times TI and Tle as a function of temperature12 showed the onset of reorientation about the 3-fold axis of the ion in the low-temperature phase, with an activation energy of Ea = 20.2 kJ/mol. This study also showed the presence of hysteresis at the phase transition. Results for the room temperature phase were all interpreted as indicative of isotropic motion. Results of a more recent lH T1 study gave Ea = 20.3 Wmol for the 3-fold motion.13 Inelastic neutron scattering measurements of the torsional mode frequency for the three-site potential in the low-temperature phase gave a barrier height of 19.24 k J / m ~ l . ' ~ lH NMR experiments in which the effects of proton-proton dipolar couplings have been removed have given information on the chemical shielding: The average chemical shift of the residual protons in the room temperature phase of D@+C104is 10.7 ppm with respect to TMS (6 scale).15 At 77 K in the low-temperature phase this value shifts to 13.5 ppm16 due to t @
NRCC No. 37277. Abstract published in Advance ACS Abstracts, September 15, 1994.
increased hydrogen-bond strengths, and the chemical shift anisotropy is 33.6 ppm. 2H NMR spectra have been obtained previously using continuous wave techniques.12 While these results gave some indication of the gross changes in line shape accompanying motion of the D30+ ion, there was clearly much scope for pulse Fourier transform 2H NMR to provide much improved spectra. It was hoped that a new study might also provide more details concerning the motions both in the low- and roomtemperature phases and perhaps cast light on the reported intermediate phase of D@+C104-. Preliminary 2H NMR spectra in the region above 200 K quickly showed some rather complicated behavior, which prompted a much more detailed study of line shapes in the phase transition region and some new calorimetric measurements.
Experimental Section D30+C104- was prepared from the two starting materials, 70% DC104 in D20 and 90% D2S04 in D20 (both from M.S.D. Isotopes). The preparation essentially followed the method devised by Smith for the preparation of anhydrous perchloric acidI7 with the exception that the anhydrous acid produced was condensed onto 70% DC104 in a 10 mm 0.d. Pyrex tube held in a dry ice/acetone bath. The cold bath was occasionally removed and the sample tube warmed to allow mixing. The reaction was stopped when the melting point of the product reached about 50 "C, which is the known melting point of the H@+C104- c o m p o ~ n d . ~The J ~ product (about 1 cm3) was then sealed in a short portion of the tube for use in the experiments. Calorimetric measurements were performed on this same sealed sample using a Tian-Calvet heat-flow calorimeter (Setaram, Model BT). The bulk of the sample melted in the 320332 K range, with the peak of the exotherm at 329 K. 2H NMR powder line shapes as a function of temperature were obtained at 27.63 MHz on a Bruker CXP-180 pulse spectrometer with a 4.24 T cryomagnet, using a variabletemperature N2 gas-flow probe with a Bruker B-VT-1000 controller. A phase alternated quadrupole echo pulse sequence'* was used with a delay time of 35 p between X and Y pulses of 2.6-3.0 p.
*HNMR Theoretical Background 2H NMR studies can provide very specific information about the motion of molecular bonds incorporating D atoms. All the
0022-3654/94/2098-10935$04.50/0Published 1994 by the American Chemical Society
10936 J. Phys. Chem., Vol. 98, No. 42, 1994
Ratcliffe
necessary theory of 2H NMR and the effects of molecular motion have been described in detail elsewhere,19-" and it will suffice to give a brief summary here: 2H ( I = 1) NMR in solids is dominated by a perturbation of the Zeeman interaction by the quadrupole coupling tensor which arises from the interaction of the nuclear quadrupole moment with the electric field gradient (efg) tensor. The 2H powder line shape, which comes from the summation of resonance line pairs for a distribution over all crystallite orientations, has either two or three pairs of characteristic features (edges, shoulders, and peaks) separated by frequencies
Av, = 3x12 200 kHz
x
where is the quadrupole coupling constant equal to e2qQ/hin hertz and the asymmetry parameter 7 = (AvP - Avu.)/Av,. The Avii are proportional to the principal axis components of the effective quadrupole coupling tensor, but information regarding signs of these is lost. Molecular motion at sufficiently high rates can cause changes in the 2H NMR line shape. Differing degrees of line shape averaging are observed depending on the rate of reorientation and the orientations of the principal axes of the efg tensor relative to the rotation axis. In the fast motion limit (reorientation rates about lo7 jumps/s or faster) analytical expressions for the line shapes are relatively easy to ~ b t a i n : ~The ~-~~ orientation of the tensor must first be described in a final reference frame in terms of the Euler angles
I
Figure 1. 2H N M R line shapes of D30+C104- as a function of temperature showing dynamic averaging by reorientation about the molecular 3-fold axis. TABLE 1: Observed Quadrupole Coupling Constants k) and Asymmetry Parameters ( q ) for D30+ in Various States of Motion in D30+C104-
x (kw rl 110/120 static 169.4" 0.146" 74.67 0.0621 222 3-site reor 259 complex anisotropic 21.77 0.636 a Average values (the line shape consists of a superimposition of three similar but unequal line shapes). phase I1 I1 I
motion
T(K)
L 222K
where R is the Euler angle rotation matrix describing the coordinate transformation and Vpasis the tensor in its principal axis system (a diagonal matrix). The fast motion line shape is then given by an effective tensor which is the populationweighted average of the tensor components in the final frame for all the orientations sampled during the motion.
(3) where pi is the population factor of site i. Intermediate rate line shapes depend on the specific rate of motion and the experimental details (e.g., on the echo delay time) and have to be calculated numerically. Methods for doing this are ~ e l l - k n o w n . ~ ~ ~ ~ ~ - ~ ~
Results and Discussion For convenience the well-established low-and room-temperature phases will be labeled phase I1 and phase I, respectively. Low-Temperature Phase 11. The temperature variation of the 2H Nh4R line shapes for phase I1 shown in Figure 1 is entirely consistent with 3-fold reorientation of the D3O+ ion. Below 120 K the line shape corresponds to an essentially static ion. Crystallographically, the three D atoms are inequivalent (with 0-D--0 hydrogen bond lengths of 2.63, 2.64, and 2.71 A).5 Unfortunately, the three are not distinguishable in the static line shape due to overlap, but since the line shape has reasonably well-defined features, one must presume that the three quadrupole coupling tensors are quite similar. (This is not the case for all D30' salts.)2 According to the empirical correlation of Berglund et al.25 relating 0---0distances and 2H quadrupole
100 kHz 1
I
Figure 2. Details of the ZHNMR line shape at 222 K showing the presence of non-axially symmetric anisotropy after motional averaging. coupling constants k),the three D's would have x ' s of 179.6, 182.7, and 201.8 kHz (average = 188.0 kHz), somewhat larger than the observed average = 169.4 lcHz at 110 K (Table 1). The apparent discrepancy might be explained (a) as a result of librational averaging reducing the observed line width or (b) as a result of overestimating the x ' s because of deviation from the correlation or errors in the 0---0distances. Inequivalence of the three D's is also confirmed by the fact that the 3-fold averaged line shape in the fast motion limit is slightly nonaxial, x = 74.67 kHz and 7 = 0.06213 at T > 220 K (Figure 2 and Table 1). Threefold averaging of equivalent 2H gives axial line shapes (7= O).20 Intermediate rate line shapes occur over the range 140-200 K. The lines shapes do not provide enough information to carry out an exact analysis of the dynamics, on account of the lack of complete 3-fold symmetry. It is reasonable, however, to take 3-fold symmetry as a first approximation in order to extract further useful and valid information. The low-temperature line shape parameters may then be used as representative of the static tensor for all three D's and the 222 K line shape with the same x but with 7 reduced to zero as the fast motion limit C3 averaged tensor. Working through the mathematics using eq 2 and 3
x
J. Phys. Chem., Vol. 98, No. 42, 1994 10937
Phases and Motions in D3O+C1O4-
TABLE 2: Calculated Quadrupole Coupling Constants c;C) and Asymmetry Parameters (7) for Various Simple Motions of D30+ Based on the Observed Static Line Shape Parameters and B = 73.15'
x (MZ)
motion static (assumed from observed)
inversion or 180" reorientation C3 reorientation or (C3 inversion)
+
169.4 147.0 74.67
4 0.146
for 3-fold
3.2E4
,)&,,(
1.OEB
n/l
b,
Reorientation of 030' 1.OE8
0.016 0
leads to the following simple relationships between the static and C3 averaged tensor components:
+ Avo + Avz sin28) (4) Avz' = AvXxsin2 p + A v cos2 ~ p
Avn' = AvD' = 0.5 (Av, cos2 ,8
Calculations based on these equations then lead to a value of p = 73.15' for the angle between the 3-fold reorientation axis and each 0-D bond and show that the static tensor is oriented with its smallest principal component (n) perpendicular to the 3-fold axis. This angle /3 corresponds to a D-0-D bond angle of 112", in close agreement with values found from neutron diffraction studies of solids (108.6- 116.0') and from ab initio calculations on the isolated H30+ molecule (1 11.6114.4°).2 It is also of interest to calculate the averaged tensor expected for rapid inversion of the D30' ion, for which the barrier in the isolated ion is expected to be in the range 6.39.6 kJ/mol from ab initio calculations.2 It turns out that the result is exactly the same as for 180' reorientation about the C3 axis. The calculated line shape (Table 2) clearly is not observed experimentally. Further calculations demonstrate that the averaged line shape for rapid inversion and C3 reorientation is exactly the same as for C3 reorientation alone. In phase 11, however, inversion seems highly unlikely because of the hydrogen-bonding situation, but it is not implausible for phase I. Intermediate-rate echo line shapes were simulated for 3-fold reorientation based on the same C3 approximation and parameters derived above. The simulations were carried out using a substantially modified version of the program originally written by Vega and L U Z . ~The ~ results (Figure 3) show a reasonable similarity to the observed sequence of line shapes. The match is not perfect; however, the inequivalence of the three sites and the presence of a second component, discussed below, may well account for this. The motion produces an attenuation of the echo signal intensity which reaches a minimum at about 6 x lo4 jumps/s (at about 150-160 K in the observed spectra). In the region of this minimum the spectrum of the second component will therefore appear to be enhanced relative to the D30+ spectrum. An Arrhenius plot of log(rate) versus temperature, obtained by comparing the simulated and experimental line shapes, gave a rough estimate for the activation energy E, = 23.0 kJ/mol. Considering the problems discussed above, this compares very well with the values 20.2 and 20.3 kJ/mol obtained from lH T I measurement^.^^^^^ Phase I. Above the transition into phase I the line shapes characteristically showed the sharp line, as reported in the early 2H NMR continuous wave study,lZ which was interpreted as indicating isotropic motion of D3O+ in phase I. However, in the new spectra it was found that this sharp line was underlain by a second broad component which constitutes a major proportion of the intensity (Figure 4). In the first studies this broad line shape did not show a recognizable tensor powder pattern. In fact, various measurements of the spectrum after melting and refreezing the sample gave slightly different results each time. Sometimes numerous pairs of sharp lines were
3.2E3
I
200 kHz
1
Figure 3. Intermediate rate line shapes calculated for 120" jumps of a symmetric D30+ion about its C3 axis. x = 169.4 kHz, T,I = 0.146, and p = 73.15'.
ll 20 kHz I Figure 4. *HN M R line shape of D30tC104-in phase I at 259 K. The central spike (truncated) is due to the liquid phase containing excess water. I
apparent. Such behavior is typical of samples where the crystallite orientations are not distributed completely at random. Eventually, it was found that by melting the sample, and cooling quickly in an ice bath with vigorous agitation, that occasionally an isotropic distribution of crystallite orientations was produced, as evidenced by a recognizable 2H NMR powder pattern, = 21.77 kI-Iz, TI, = 0.636 (Figure 4). This is narrower, by a factor of more than 3, than the 3-fold averaged line shape of phase 11. In every case the sharp central line was present. This twocomponent behavior was the first indication that something unusual was occurring. Unfortunately, it is not possible to produce a unique model to describe the dynamics of the D30+in phase I; the system is underdetermined. The problem is exacerbated by the absence of a definite orientation for the molecular axis (from the diffraction s t ~ d i e s ~and , ~ , from ~ , ~ rotational potential calculations).26 Furthermore, the effective static which cannot be measured for this phase, is probably higher than that in phase 11, due to the longer, and therefore weaker, hydrogen bonds. However, the reduction in line width suggests a further rapid motion in addition to rapid reorientation about the molecular 3-fold axis. One can exclude a simple two-site flip of the molecular 3-fold axis, since in a two-site flip the xx or y y component of the 3-fold averaged tensor would remain unchanged and lead to a broader line shape than that observed. The fact that the line shape, and hence the new motion, is
x
x,
10938 J. Phys. Chem., Vol. 98,No. 42, 1994
Ratcliffe
I
220
J I 100 kHz I Figure 5. ZHNMR line shapes of (a) the sample at 233 K with solid eutectic (warmed from 200 K) and (b) at 233 K with molten eutectic (cooled from 270 K). Note the central hump and weak broad wings in (a) which are replaced by the sharp liquid line in (b). The liquid line has been truncated in data collection and processing.
anisotropic is in keeping with the phase I orthorhombic structure where the D30+ sits on a mirror plane. As mentioned earlier, rapid inversion could also be taking place but with no effect on the averaged line shape. Furthermore, other evidence suggests that this phase is a proton conductor, which would entail transfer of protons between H30+ via the Clod- ionsz7 Phase Transition Region. For this particular sample the following reproducible behavior in the phase transition region was established. On cooling from phase I, at 244.2 K the 2H NMR spectrum changed gradually over a period of minutes from the one observed in the room temperature region to one which showed the 3-fold averaged low-temperature line shape and a remnant sharp component. Down to 218.4 K these two components only were seen. Below 218.4 K the 3-fold averaged line shape remained, but the sharp component disappeared to be replaced by a much broader component. On warming from this state the 3-fold averaged and the broad component remained up to 243 K, where over a period of several minutes the broad component disappeared and the sharp central line returned (Figure 5). The 3-fold averaged component remained up to 254.3 K where it disappeared and was replaced by the anisotropic room temperature component. Two calorimetric scans on the sample used in the NMR experiments showed similar behavior (Figure 6): When the sample was precooled to about 210 K, only one endothermic transition was observed at 253.8 K on warming, but after it was precooled to 170 K an additional endothermic peak appeared at 243.6 K. It is quite clear that at all times two components were present, and since the crystal structures of both phases I and I1 each show only one crystallographic type of H30+ ion, the obvious conclusion is that one component is not solid D3O+C1O4-. The origin of this mysterious behavior is to be found in the HzO/HC104 composition d i a g r a n ~ , *which ~ , ~ ~ shows a eutectic point to the water-rich side of the H30+C104- composition. The eutectic point can be estimated at roughly 254 K and 76.5 wt % HC104 from ref 28. Everything can be explained if there is a small excess of DzO over the D@+C104- stoichiometry. Hence, although much of the material is already solid D30+C104at room temperature, more D@+C104- crystallizes as the sample is cooled further and the composition of the excess solution falls down the melting point curve toward the eutectic
1
I
260
I
I
300
1
1
340
TIK Figure 6. Calorimetry scans (thermopile output) of the D30+C104sample on warming: (a) after precooling to 210 K and (b) full scan after precooling to 170 K. The temperature scales are identical.
Figure 7. Difference spectrum (solid line) obtained by subtracting spectra with the eutectic mixture in molten and solid states (at about 240 K). Imperfect subtraction gives rise to the two sharp spikes and inversion of the central liquid line (truncated). The spectrum of D502+C104- at 237 K is superimposed (dashed line) for comparison. point. It seems that in the sample studied the residual eutectic mixture can supercool quite a long way, so that if the sample is not cooled far enough, this residue does not solidify and on warming only the phase I1 to I transition of D30+C104- is detected. On the other hand, if the sample is cooled far enough, one can observe two consecutive endothermic transitions, the first due to the melting of the eutectic mixture and the second due to the 11-1 transition. Note also that the long introductory tail to the melting endotherm of D30+C104- in the calorimetric scan (Figure 6) is characteristic of the presence of impurity, which in this case is the eutectic mixture. By subtracting spectra obtained from the sample with the eutectic mixture in its solid and molten states, it is possible to obtain a rough difference spectrum corresponding to the second solid component in the eutectic mixture (Figure 7). This difference spectrum is reminiscent of that observed for DsOz'CF3S03- in the region where the terminal DzO groups undergo rapid 180' flips.24 Furthermore, the eutectic point is close to the composition of D50z+C104-, and a preliminary spectrum of a sample of this compound at 237 K compares reasonably well with the difference spectrum (Figure 7). The spectra of these Ds02+ compounds actually consist of two superimposed lines: one, a broad component with 77 1 arising from 180" flipping of the terminal D20 units, and the other, a narrow component corresponding to the static D atom in the very strong hydrogen bond between the two D20 units.
-
Phases and Motions in D&+C104Thus, it is clear that there are only two solid phases for &O+C104-. The phase transition occurs with hysteresis at 245 K on cooling and 254 K on wanning and is accompanied by a sudden change in the dynamics of the D30+ ion, from 3-fold reorientation in phase I1 to 3-fold reorientation plus some other rapid anisotropic motion in phase I.
Conclusion The current work demonstrates that the previously reported intermediate phase of D3O+C1O4- is not a phase of this compound at all; the effects observed arise from eutectic behavior. This explains why the neutron diffraction patterns of phase 11and the supposed intermediatephase were so similar. Furthermore, the “isotropic” phase previously detected in NMR experiments is in fact a liquid phase coexisting with phase I. Activated reorientation about the molecular 3-fold axis and the inequivalence of the three 2H sites in phase 11 is c o n f i i e d . There is a further rapid reorientational motion in phase I which is anisotropic, in keeping with the site symmetry. Eutectic behavior may be a complicating factor to be considered in other hydrated proton or hydrated hydroxyl systems, particularly those with a rich acidwater or base/water phase diagram and especially if it is difficult to produce the exact stoichiometry. This suggests that great care should be exercised in obtaining conductivity measurements in solid phases of such systems, in order to be sure that the conduction is through the solid and not through a thin film of electrolytic liquid; i.e., very precise control of the stoichiometry must be ensured. In this respect the proton conductivity measurements of H3O+ClO4- previously determined as a function of temperat ~ r e appear * ~ to be valid, since the results show a marked jump at the 11-1 transition.
Acknowledgment. The author thanks Dr. Y. P. Handa, Institute of Environmental Chemistry, NRCC, for performing the calorimetric measurements. References and Notes (1) Volmer, M. Ann. Chem. 1924,440, 200.
J. Phys. Chem., Vol. 98, No. 42, 1994 10939 (2) Ratcliffe, C. I.; Irish, D. E. In Water Science Reviews; Franks, F., Ed.; Cambridge University Press: Cambridge, 1986; Vol. 2, 1988; Vol. 3. (3) Lee, F. S.; Carpenter, G. B. J. Phys. Chem. 1959,63,279. (4) Truter, M. R. Acta Crystallogr. 1961,14, 318. (5) Nordman, C. E. Acta Crystallogr. 1962,15, 18. (6) Smith, H. G.; Levy, H. A. Abstracts, Come11 Meeting of the American Crystallography Association, Ithaca, NY, 1959; Abstract K-6, p 41. (7) Domoslawski, J.; Golab, M. Physica 1980,IOIB, 217. (8) Janik, J. M.; Rachwalska, M.; Janik, J. A. Physica 1974,72, 168. (9) Czamiecki, K.; Janik, J. A,; Janik, J. M.; Pytasz, G.; Rachwalska, M.; Waluga, T. Physica 1977,85B, 291. (10) Kakiuchi, Y.; Shono, H.; Komatsu, H.; Kigoshi, K. J . Chem. Phys. 1951,19, 1069. (11) Richards, R. E.; Smith, J. A. S. Trans. Faraday SOC. 1951,47, 1261. (12) O’Reilly, D. E.; Peterson, E. M.; Williams, J. M. J . Chem. Phys. 1971,54, 96. (13) Herzog-Cance, M. H.; Pham Thi, M.; Potier, A. J . Mol. Struct. 1989, 196, 291. In Solid State Protonic Conductors 111; Goodenough, J. B., Jensen, J., Potier, A., Eds.; Odense University Press: 1985; p 129. (14) Janik, J. M.; Pytasz, G.; Rachwalska, M.; Janik, J. A.; Natkaniec, I.; Nawrocik, W. Acta Phys. Pol. 1973,A43, 419. (15) Ratcliffe, C. I.; Ripmeester, J. A.; Tse, J. S. Chem. Phys. Lett. 1985, 120, 427. (16) Sears, R. E. J.; Kaliapemmal, R.; Ratcliffe, C. I. J . Chem. Phys. 1990,93,2959. (17) Smith, G. F. J . Am. Chem. SOC. 1953,75, 184. (18) Davies, J. H.; Jeffrey, K. R.; Bloom, M.; Valic, M. I.; Higgs, T. P. Chem. Phys. Lett. 1976,42, 390. (19) Spiess, H. W.; Sillescu, H. J . Magn. Reson. 1981,42, 381. (20) Barnes, R. G. Adv. Nucl. Quad. Reson. 1974,I , 335. (21) Vega, A. J.; Luz, J. J . Chem. Phys. 1987,86, 1803. (22) Wittebort, R. J.; Olejniczak, E. T.; Griffin, R. G. J . Chem. Phys. 1987,86,5411. (23) Greenfield, M. S.; Ronemus, A. D.; Vold, R. L.; Vold, R. R.; Ellis, P. D.; Raidy, T. E. J . Magn. Reson. 1987,72, 89. (24) Ratcliffe, C. I. J . Phys. Chem. 1987,91,6464. (25) Berglund, B.; Lindgren, J.; Tegenfeldt, J. J . Mol. Srruct. 1978,43, 179. (26) Czarniecki, K. Physica 1981,IOJB, 226. (27) Potier, A.; Rousselet, D. J . Chim. Phys. Physicochim. Biol. 1973, 70, 873. (28) van Wyk, H. J. Z. Anorg. Allg. Chem. 1906,48, 1. (29) Brickwedde, L. H. J . Res. Natl. Bur. Stand. 1949,42, 309.
