J. Phys. Chem. 1995, 99, 2256-2261
2256
ARTICLES Concerted Tunneling in a Mixed Ammonium (Co, Ni) Tutton Salt Suli Fei and Herbert L. Straws* Department of Chemistry, University of California, Berkeley, California 94720-1460 Received: October 26. 1994@
A series of mixed ammonium cobalt-nickel Tutton salts (NH&[CoxNil-,](H20)6(SO4)2 have been doped with a few percent of deuterium. Irradiation of the N-D stretch band I (-2260 cm-') rotates the ammonium group. In the dark this group relaxes back to its original orientational distribution. The HDO molecules also relax, probably by a 180" rotation of a HDO molecule. The samples are at 7 K, and the dark relaxation processes occur by tunneling. The rotation of the ammonium ions must be accompanied by the rotation of the water molecules.
I. Introduction The ammonium Tutton salts have long been studied as a series of simple isomorphous compounds. The general formula is (NH&M(H20)6(X04)2, where M is a divalent metal and another univalent ion can replace the ammonium ion. M can be any one of a large number of metals and, in particular, can be any of the first-row transition metals, and X can be any of a number of elements such as sulfur. The usefulness of the Tutton salts for studies of structure and spectra takes advantage of the possibility of making dilute crystals with one compound in another (for EPR studies, for example) and on the possibility of making a series of mixtures over the entire range of composition (for X-ray structural studies). The Tutton salts have a well-defined structure, and much is known about the subtle variation in this structure occasioned by a change in composition. We have been studying NH3D+-containing crystals at low temperature and have shown that the ammonium ions rotate upon infrared irradiation of any of the N-D stretching bands. We have previously investigated a number of Tutton The NH3D+ ions sit in an asymmetric site with four distinct orientations. The irradiation produces a nonequilibrium distribution of orientations, and in the dark, the NH3Df rotate back to equilibrium by tunneling. The changing orientational distribution can be followed in detail by monitoring the changes of the four N-D stretching bands. We have also studied first neat ammonium s ~ l f a t e ~ and , ~then , ~ disordered sulfates containing potassium and rubidium ions in addition to ammonium ion^.^^^ The mixed sulfates have a statistical distribution of ammonium sites and, instead of the discrete N-D bands, show a broad N-D band. The laser can bum spectral holes into this band. In contrast, mixed crystals of the Tutton salts of cobalt and nickel still show the discrete four N-D bands. The structures of the cobalt and nickel salts are very close, and so the mixed crystal has comparatively little strain. When deuterium is introduced into the crystal to produce NH3D+ ions, some of the deuterium also goes into the water to produce HDO. The 0 - D stretch bands are just to the highfrequency side of the N-D bands. These bands arise from the
* To whom correspondence should be addressed. @
Abstract published in Advance ACS Abstracts, February 1. 1995.
six crystallographically distinct 0 - D position in the crystal. These 0-D bands do not change during the ammonium reorientation in the neat compounds. There are, as might be expected, subtle changes in the spectra of the mixed cobalt-nickel salts as compared to those of the pure compounds. However, the overall hole buming and recovery process is very similar. The new result comes from careful observation of the 0 - D bands during the recovery of the ammonium ions toward their equilibrium distribution. Two of the 0 - D change their relative intensity, probably by a 180" rotation of an HDO molecule. This happens only in the mixed crystals and only during the recovery of the NH3D+ orientation by tunneling. The overall process is therefore a concerted tunneling process-most easily visualized as a geared motion of the various molecules.
11. Structure and Spectra of the Tutton Salts The pure Tutton salts have been well-characterized, most recently in an extensive X-ray study of the ammonium salts of the first transition series.8 The metal ion is surrounded by a distorted octahedron of water molecules, which are in tum hydrogen bonded to sulfate ions. Ammonium ions separate the various sulfate ions, and the ammonium ions and the water molecules are not directly hydrogen bonded to one another. The six water molecules are related to each other by a center of inversion at the metal atom yielding three distinct pairs. Although the salts of the first transition series are isomorphous, subtle differences are observable and many are interpretable. For example, the variation of the M-0 distances has been systematically explained on the basis of M-0 bonding and the Jahn-Teller effect.8 A number of metal ions are expected to show Jahn-Teller distortions. Two ions, Cr2' and Cuz+,have orbital degeneracies involving the e, metal orbitals, which 0 bond to the ligands, and substantial Jahn-Teller distortions. Two, Fez' and Co2+, have a T, orbital degeneracy, which leads to relatively weak n bonds to the ligands and very small Jahn-Teller effects. This picture is well illustrated by explicit orbital calculations, which agree well with the results of polarized neutron diffra~tion,~ and by the distribution of M-0 distances across the transition series8 Since the Tutton salts have a well-characterized structure, they have been used to investigate both HDO and NH3D
0022-3654/95/2099-2256$09.00/0 0 1995 American Chemical Society
Concerted Tunneling in a Tutton Salt frequency-distance correlations.1° Both the Co and Ni N&+ salts were investigated and assigned. We will consider the HDO assignments further in connection with our spectra. At a finer level of detail, the Tutton salts-especially those with orbitally degenerate states-show the variation of the orbital mixing and the Jahn-Teller distortion with temperature, pressure, and composition. For example, single crystals of (ND4)2CU(H20)6(S04)2 show separate domains at low temperature. The two types of domain differ from one another only by small changes in the Jahn-Teller-active displacement coordinates.11 Equally remarkable is a change with pressure at low temperature for the same compound.12 There are a very large number of magnetic studies available, and attempts have been made to interpret these consistent with a bonding scheme and the structure,13 but this effort is still a very difficult 0ne.14 The dilution of a magnetic ion in a diamagnetic host is standard for magnetic studies. More recently, the Jahn-Teller distortions have been measured in a Cr/Zn mixed crystal as a function of composition.l5 Remarkably, the Jahn-Teller distortion of the water ligands about the Cr2+ switch direction on addition of as little as x = 0.07 mole fraction of Zn2+and remain in the new direction at all larger values of x. The X-ray data can be interpreted as showing static and/or dynamic disorder and are for room temperature. Especially in view of the temperature dependence seen for the Cu Tutton salt," we need to be cautious in extrapolating the Cr/Zn result to low temperatures. It is, however, clear that even the crystals containing the "strong" Jahn-Teller ions Cr2+and Cu2+show a remarkable variety of detailed structures that can interchange with small changes in the conditions.
In. Experimental Section Crystals of (NH4)2CO,Nil-,(H20>6(SO4)2containing about 5% D were formed by cooling a mixture of cobalt, nickel, and ammonium sulfates from a solution containing some D20. The crystallization of mixed Tutton salts has been systematically investigated.16 The composition of crystals shows a monotonic dependence on the composition of the solution in all cases. Studies using X-ray analysis show the same pattem.15 The small mixed crystals were mulled with Fluorolube (a fluorochlorocarbon grease) and sandwiched between two CaF2 windows. The windows were then mounted in a CTI closedcycle helium refrigerator. The refrigerator was held on a lathe tool carriage so that the sample was positioned in the beam of an infrared spectrometer. The spectrometer was a Nicolet 550 fitted with a homemade "auxiliary experimental module". The module contained a manually-controlled flip minor to switch between the spectrometer beam and the laser. After passing through the sample the infrared radiation, either from the laser or from the spectrometer, was monitored by a cooled InSb detector. The spectrometer was purged by either dry air or dry nitrogen, as necessary. The sample was masked using aluminum foil to leave a 1 mm2 aperture. The laser was a Laser Photonics SP5150 lead-salt laser with a few milliwatts of single-mode power and rather limited tuning range. The laser beam arose from different points on the leadsalt laser chip as the frequency was changed. This made it difficult to repeat experimentswith exactly the same laser power, as the laser had to be laboriously refocused after changes. Spectral holes were burned for about 30 min, and then the mirror was flipped to allow interferograms to be taken over time. We have previously shown that the interferometer light is sufficient to slowly pump all the N-D bands and so to alter the apparent recovery time constants.2 By doing both dark experiments, that is, those that were kept dark in between scans, and light
J. Phys. Chem., Vol. 99, No. 8, 1995 2257 NO bands
0.32
- o.Oo0
l
~
22M)
"
'
l
"
2250
"
l
"
"
l
2300
"
"
l
"
2350
"
l
~
"
~
2400
l
2450
2500
Frequency(cm")
Figure 1. Infrared spectrum of the 0-D and N-D stretching regions of ammonium cobalt sulfate hydrate (ACS) or cobalt Tutton salt with about 5% deuterium. The lower spectrum is of the original sample at 7 K, the upper spectrum is the difference spectrum between the sample after 30 min of laser irradiation and the original. The original spectrum is referred to the left-handabsorbance scale, and the difference spectrum is referred to the right-hand scale. The bands are numbered. The laser frequency was set to the peak position of N-D band I.
experiments, in which the mirror was left in the spectrometer position after the initial laser burning period, we checked that what we report is due to the dark process. Most of the dark processes are not altered by the presence of the spectrometer light.
IV. Results and Preliminary Discussion A. Spectra. Five different mixed Tutton salts (N&)2(co,Nil-,)(H20)6(S04)2 were studied: x = 0 ( A N S ) , l/4 (ACNS II), l/2 (ACNS I), 3/4 (ACNS In), and 1 (ACS). Our previous studies on the neat confirmed by this study, show that irradiating N-D band I (the lowest frequency band) pumps intensity into the 11, 111, and IV bands. Figure 1 shows the spectrum of the x = 1 sample before irradiation and the difference spectrum [spectrum (bumed) - spectrum (before bum)], which make this clear. Figure 2 repeats this for the x = 0 sample, with the same result. The only evident differences in the x = 0 and x = 1 spectra are small shifts in the positions of the band (Table 1). When the irradiating laser beam is blocked, both of the neat samples relax back to their original orientational distribution. We have characterized the kinetics of the dark tunneling process by three rate constants connecting two sets of orientational sites.3 The sites are labeled, as in the spectra, I, 11, 111, and IV. Sites I, 111, and IV comprise one set and interchange with a rate constant Ka (Kmund). Site I1 is unique and interchanges with the other sites at the considerably slower rates Ki and KO (Kh and KOut).The names come from the schematic picture of the rate ~ c h e m e .The ~ kinetics of each band relaxing toward equilibrium after either a temperaturejump or a laser "bum" is characterized by combinations of these rates, and band II is by far the slowest to retum to equilibrium. At the low temperature at which we did most of the experiments (T= 6.5 K),band 11 was not always at its equilibrium intensity when we started a new experiment. The spectra of the mixed samples are very similar to those of the neat samples; the most obvious difference is that the spectra of the mixed Tutton salts show broader bands and small
2258 J. Phys. Chem., Vol. 99, No. 8, 1995
Fei and Strauss
0.40 I
0 900
0 036 OD bands
N D bands
I
OD bands ND bands
1
0.018
0.000
-0.018
ll8llllVV
I
VI
0.675
8C
e 0 450 9 9 0.225
-0.036
2200
2250
2300
2350
2400
2450
2200
2500
2250
Figure 2. Spectrum of the nickel Tutton salt (ANS) under similar conditions to those of Figure 1.
TABLE 1: Stretching Frequencies of N-D in NH3D+ Ion and 0 - D in HDO with 5% D Ammonium Tutton Salts (T = 6.5 K) N-D Bands freauencv (cm-’) compd” ACSb ANSb
ANCS(1)‘ ACNS(I1)‘ ACNS(II1)’
2300
2350
2400
2450
2500
Frequency (cm-’)
Frequency (cm.’)
x(Co)
band1
band11
band111
bandIV
1.0 0.0 0.5 0.25 0.75
2259.26 2261.86 2260.7 2261.9 2260.5
2290.30 2289.12 2290.5 2289.7 2290.5
2328.53 2331.05 2330.2 2330.8 2329.3
2365.91 2364.83 2365.6 2364.9 2366.4
Figure 3. Spectra of the mixed sample ACNS(III),
x = 0 75 (d), compared to the spectra of the neat samples ANS (a) and ACS (b) Spectrum c is the weighted sum of (a) and (b) Vertical lines are drawn to call attention to the small shifts of the bands For N-D I, spectra (a) and (b) are in different positions, and the weighted sum (c) lines up with the observed spectrum of the m x e d salt d. For N-D IV, neither (a) nor (b) nor their sum (c) lines up with the spectrum of the mxed salt For 0 - D band V, the mixed salt has the appearance of two separate bands-rather different from (a), (b), or (c) 0 900
ND bands
OD bands
I
0 675
0 - D Bands frequency (cm-I) compd” ACSb
~ ( C O ) band I
1.0 ANSb 0.0 0.5 ACNS(1)’ ACNS(I1)‘ 0.25 ACNS(II1)‘ 0.75
2380.93 2379.18 2380.4 2380.4 2380.4
8
band 11, I11 band IV band V band VI 2436.38 2439.30 2437.6 2439.0 2437.1
2448.64 2448.64 2448.6 2448.4 2448.7
2462.06 2465.56 2463.2 2465.8 2461.4
2491.25 2488.91 2490.1 2489.4 2490.9
C
m n
5
0450
0.225
ACS = ammonium cobalt sulfate, ANS = ammonium nickel sulfate, and ACNS = mixture of ACS and ANS: (N&)2CoXNil-,(H20)6(S04j2. Error range izO.05 cm-’. Error range hO.1 cm-l.
shifts in frequency. Figure 3 shows a typical spectrum, that of ACNS 111. The spectrum is compared to the spectra of the neat ACS and A N S , and vertical lines are drawn through the positions of the bands to emphasize the differences. Figure 3 also shows the spectrum calculated by adding the spectra of the neat compounds in the nominal ratio 3: 1. This matches the observed spectrum well except near HDO band V. Note the nominal ratio is that of the solution from which the ACNS sample was crystallized and is not likely to be the actual ratio in the small area of the sample that was in the beam. The spectra with their broadened bands are rather insensitive to the value of x , and we did not characterize the concentration more accurately. The spectra vary quite systematically with composition, and Figure 4 shows the spectra of ACNS I1 to illustrate this point. Figure 4 also shows the shift in the position of the N-D band IV from its expected position. This band shifts in the opposite direction in Figure 3 (ACNS 111) and is broad in both cases. The laser frequency for buming the neat samples was chosen to be at the center frequency of peak I. However, for the mixed salts, frequencies of 2257 and 2266 cm-’ were chosen. Figure
2200
2250
2300
2350
2400
2450
2500
Frequency (cm-’)
Figure 4. Spectrum of ACNS(II), x = 0.25, compared to the spectra of its constituents. Compare to Figure 3, which is similar, except here the shifts are less prominent. 5 shows the spectrum of ACNS (I), x = 0.5, on an expanded frequency scale. The 2257 cm-I irradiation will bum mainly the NH3D+ in “Co-like” sites, while the 2266 cm-’ irradiation will affect mainly those in the “Ni-like” sites. The effect of buming the mixed crystals is very similar to the effect of buming the neat crystals. A typical result is shown in Figure 6. The bum is at 2257 cm-’, and the hole is at 2259.5 cm-I. The antiholes at the position of band IV are shifted to lower wavenumbers from the apparent maximum of the band. This also is true if the bum is at 2266 cm-’. Deconvolution of the HDO I-NH3D+ IV band does not seem to account for this shift. Band I1 is decreasing in intensity, due either to complex relaxation paths or to residual thermal relaxation, and we
J. Phys. Chem., Vol. 99, No. 8, 1995 2259
Concerted Tunneling in a Tutton Salt
0.34
2200
2250
2300
2350
2400
Frequency(cm-l)
Figure 5. Spectrum of ACNS(I), x = 0.5, compared to those of its constituents. This spectrum is shown with an expanded horizontal scale to emphasize the differences in position for N-D I in the spectra of the neat samples (a and b). 0.30
,
1 ND bands
0.016
0.24
0.008
0.000
0.18
4.008 0.12
4.016
2200
2250
2300
2350
2400
Frequency(cm-')
Figure 6. Spectrum of ACNS(I), x = 0.5, and the difference spectrum after burning at 2257 cm-'. The vertical lines show the shifts of the holes as compared to the original bands.
attribute no particular significance to the decrease. The hole in band IV also showed the 2 cm-I shift in ACNS 11(x = 0.25) but not in ACNS III (0.75), which suggests that the shift is a property of the Ni-like site. B. Kinetics. As already mentioned, the kinetics of the relaxation of the orientational distribution is described by at least three rate constants, and this results in complicated kinetic^.^ To start with, we fit the relaxation data to singleexponential functions, with the results shown in Table 2. We first consider the N-D bands. The data are all for recovery of the bands, that is, the dark process. After a burn of band I, the bands of the ACS recover with a time constant of about 8 h. (The various bands show time constants of 6- 11 h.) This may be compared to the overall relaxation after a T-jump of about 11 h.2 The rate measured for A N S is considerably faster, averaging about 5 h. For the mixed salts we have different results for burning at 2257 cm-' (Co-like sites) and 2266 cm-' (Ni-like sites). All the mixed salt relaxation times are shorter
than those of the neat salts, and the relaxation of the putative Ni-like sites is faster than that of the Co-like sites. C. 0 - D Spectra and Kinetics. For the neat salts, nothing happens to the 0-D bands as either a function of time or laser irradiation at the N-D stretch frequencies. For the mixed salts, the results are remarkably different. Nothing happens on laser burning of the N-D I band, but some of the 0-D bands exchange intensity during the relaxation process. This is illustrated for ACNS (I) (x = 0.5) in Figure 7a, which shows the 0-D stretch region of the spectrum immediately after the burn at 2266 cm-' and 72 h later. The region around 2460 cm-' (band V in Table 1) decreases in intensity, while the bands at 2437 cm-I (bands I1 and 111) increase. Figure 7 also show the results for ACNS (11) (x = 0.25) and ACNS (In) (x = 0.75). The changes are larger for the samples which contain more Ni and so are associated with the Ni-like sites. The irradiation of the ND I band was done at each of our two wavenumbers, 2257 cm-' emphasizing the Co-like sites and 2266 cm-' emphasizing the Ni-like sites. The change in the 0-D bands was greater for the Ni-like burn in each case (Table 2). The half-time for the 0-D stretches is considerably longer than those for the N-D stretches, but quite comparable to the longer time constants observed in the neat crystals. For example, the three elementary constants for neat ACS at 7 K are 18, 28, and 63 h.3 For the NJ&D+ bands, the intensities change as a function of temperature. Two types of change are observed. As the temperature is raised from low temperature, the relative intensities of the four bands change as the thermal energy kT becomes equal to AE,the difference in ground state energies among the four possible orientations.2 The NH3D+ bands change a great deal with temperature, and so, as noted above, some (band 11) take a long time to come to equilibrium as the temperature is rapidly lowered. As the temperature is raised toward room temperature, the NH3D+ bands broaden and tend to merge.2 In our previous studies of the OD bands of the neat Tutton salts, no change, other than the broadening at high temperature, was noted. For the mixed salts ACNS (I) (Figure S), a marked change in relative intensity occurs between 70 and 200 K. Band V goes up in relative intensity, while band MI1 goes down.
V. Discussion We start by considering the assignment of the 0-D bands. There are six of these, but they are not clearly resolved. The frequency of the ODX bands goes down as the strength of the hydrogen bond between the D and X atoms increases. The strength of the hydrogen bond is usually assumed to increase monotonically as the 0-X distance decreases as measured, for example, by X-ray diffraction. Previous workers have simply assigned the bands for a series of Tutton salts as a monotonic function of the known distances.1° Substantially, the same arguments have been used for many of the Tutton salts, for example, to determine the distance in the hexaaqua complexes in solution.'' The assignments for our Tutton salts on the same basis are shown in Table 3. Comparing the values for ACS and ANS in Tables 1 and 3 shows the small shifts in both frequency and distance that occur in these compounds. The absolute values of the distances vary, as shown by the differences between the neutron diffraction'* and the X-ray diffraction values for A N S in Table 3. However, the order of the distances remains the same, and so taking the distances from different sources does not change the assignment. It is important to notice that the distance values quoted in Table 3 are all for room temperature, and thus the assignment of closely spaced 0-D bands may change at the low temperatures used for the vibrational spectra. We suggest an alternate assignment below.
2260 J. Phys. Chem., Vol. 99, No. 8, 1995
Fei and Strauss
TABLE 2: Kinetics of the Orientation of NHD+ and HDO after Laser Irradiating temp
freq
sample
(K)
(cm-')
ACNS(1)
7 7 18 7 7 7 7 7 7
2257 2266 2266 2257 2266 2257 2266 2259 2262
ACNS(I1) ACNS(II1) ACS ANS
NW) t 1 / 2 (h)
1.7i0.5 0.7 k 0.1 0.6i0.1 2 f 0.5 1.3 k O S 2.5i0.5 0.8i 0.5 8 i1 2.7i O S
int (%)
60.0 50.9 42.8 58.8 45.0 78.0 30.7 75.0 108
ND(I1) int (%)
ND(II1)
fIl2 (h)
1.2iO S -3.2 712 -3.2 30 f 2 -3.3 1.8 k0.5 -21 1.5 k 0.5 -3.9 2.7i0.5 -5.2 2.3i 0.5 -4.7
t l (h) ~
int (%)
1.5f 0.2 1.5f 0.2 1.7f 0.5 8 i2 2 k 0.5 1.8f 0.8 2.5f 0.5 11 i1 5.5 k 1
-7.5 -6.9 -7.7 -5.4 -11.1 -16.1 -5.5 -15.0 -17.6
ND(IV) tln
(h)
1.7f 0.2 0.8k 0.2 1.2k 0.5 5.5i 1 2 f0.5 3.5k 0.8 2 f0.5 6kl 7f1
OD(I1,III)
int (%)
t , (h) ~
int (%)
-4.8 -5.5 -12.0 -4.0 -14.0 -14.0 -11.3 -25.0 -9.0
19 f 1 20 f 5 10 f 1 25 f 5 10 f 1 6f1 6 i1
1.5 2.3 1.5 1.5 5.2 0.8 1.2
(h)
int (5%)
19f 1 15f 1 10 i 1 28 i 6 15 1.2 5k1 5 i1
-11.1 -12.5 -6.5 -7.8 -20.0 -3.2 -5.2
Intensity variation = [int(after buming) - int(before buming)] x 100%/int(beforebuming). tl,z = half-life.
2250
23W
I
OD bands
ND bands
iiaiiiiv v
I
2350
2400
2450
2500
Frequency (cm")
2375
2405
2435
2485
2495
Frequency (cm")
Figure 7. Spectra of the mixed Tutton salts in the 0-D region. The samples were burned at 2266 cm-I for '12 h (solid lines) and then left to recover in the dark for 72 h (dotted lines). Spectrum (a) is ACNS(I), x = 0.5; (b) is ACNS(II),x = 0.25; and (c) is ACNS(III),x = 0.75. The relative order of the distances switches in going from ANS to ACS, but these switches are among p s r s whose distances vary from one another by less than 0.01 A. Table 3 gives an indication of the small differences in the crystal structure between ANS and ACS, with distances that vary on the order of 1%. The mixed crystals show some strain, enough to broaden the various infrared bands, but not enough to make them merge together. The mixed crystals presumably contain a random arrangement of Co and Ni ions. The water molecules are, of course, attached to one of these, while the ammonium ions are surrounded by a number of nearest-neighbor metal ions. The ammonium sites can therefore be intermediate between Co-like and Ni-like sites, and the hole buming can emphasize the role of one or another. We do not attempt to interpret in detail the small shifts in position that we observe in the mixed crystals and the further shifts observed in the antiholes. It is in the kinetics that we see the largest changes on forming the mixed crystals. The general decrease in the relaxation times (Table 2 ) must come from lowered barriers to reorientation, and this lowering must be due to the crystal strain. Since the ammonium ions are in a statistical distribution of environments, there is undoubtedly a statistical distribution of barriers. The
Figure 8. Spectra of ACNS(I),x = 0.5, as a function of temperature. Note the increase in 0-D V and the concomitant decrease in 0-D IIlIII between 70 and 100 K. Band IV stays about the same. By 200 K, the spectra have broadened severely, but band V is still relatively intense. TABLE 3: Assignment of OD Bands in ACS and ANS ANS ACS band" atomsb distance' distanced atomsb distanced I 8-4 2.697 2.721 9-3 2.707 I1 9-3 2.709 2.736 8-4 2.714 I11 8-6 2.758 2.776 9-5 2.763 IV' 9-5 2.765 2.781 8-6 2.769 V' 7-5 2.782 2.786 7-5 2.793 VI 1-6 2.828 2.853 7-6 2.831 Bands in order of increasing wavenumber; refer to Table 1, Figure 1, and Figure 2. The numbering of the atoms as in refs 10, 17, and 18. The first atom is a water 0, and the second is a sulfate 0. From neutron diffraction at room temperature (ref 18). Quoted in ref 10 from X-ray data at room temperature. e A possible altemate assignment is to switch IV and V; see text. mixed Tutton salt crystals have a rather different type of disorder than the mixed ammonium-alkali metal sulfates we examined previ~usly.~~~ The most interesting new process we found in the mixed salts is the relaxation of the HDO distribution. The bands that change in intensity are IIAII and V. The most likely process that produces the observed change in intensities is similar to that for the NH3D+ ions, that is, a change in orientation. Temperature studies show that the orientation that gives rise to band V is lower in energy than that of the IMII orientation and that the difference is about 70 K or 50 cm-'. The changes in the infrared spectra show only that the 0-D moves to a new crystallographic
J. Phys. Chem., Vol. 99, No. 8, 1995 2261
Concerted Tunneling in a Tutton Salt position but does not show how it got there. The water molecules are arranged in a distorted octahedral arrangement about the metal center. The interaction of the metal with the water has been studied in detail for the deuterated vanadium and nickel Tutton salt.19 The water molecule is inclined about 40" to the VO4 plane as measured about an axis through the D atom, which is in the VO4 plane and perpendicular to the bisector of the DOD bond. This odd angle is neither the angle expected for sp3 lone-pair bonding (55") nor that for sp2bonding (0"). The conclusion is that the orbitals of the oxygen are best described by s p ~ *and p n orbitals, but a more complete description requires more complicated hybridization schemes. There is some evidence that the 0-H bond becomes weaker on complexation.20 Can the D atoms interchange with the H (D) on the neighboring water molecules? This seems unlikely. The rate constants at room temperature for proton exchange involving a hydronium ion and a water molecule and of a water molecule with a hydroxide ion are 8 x lo9 and 4 x lo9 W(mo1 s), respectively.21 The reactions have an activation energy of about 3 kcaVmol and so would have a negligible rate at the temperature of our experiment. The existence of both the metal-oxygen interaction and hydrogen bonding makes comparison of the structure and the kinetics between the crystals and solution particularly difficult. l7 However, theoretical studies of proton exchange reaction by tunneling in small water clusters also give barriers on the order of 3-4 kcdmol, and the activation energies may indeed be of this order in a variety of hydrogen-bonded structures.22 Our measurements show that deuterium moves from a position that contributes to band V (due to water 7) to one that contributes to band IYIII (waters 8 and 9). This implies either that the assignment of the 0 - D is incorrect or that the D actually exchanges from one water molecule to the next. We have already mentioned that the hydrogen-bonded distances as measured by room temperature diffraction methods are subject to change in forming the mixed crystals and in lowering the temperature. One such change, a switch in the assignment of band IV and V (Table 3), would rationalize the exchange process-which would be around oxygen 9 for the Ni salt and around oxygen 8 for the Co salt. The switch 9-8 in going from Ni to Co also provides a rationale for our observation that the HDO changes occur only in the Ni-like sites. The water changes are only observed in the mixed salts. They occur in the dark process after the end of the laser excitation of the N-D stretch. The time scale of the change is that of the slower of the NH3D+ reorientational time constants. The simplest interpretation would be of a concerted geared tunneling motion in which rotation of the ammonium ion moves the sulfate, which in turn rotates the water. The idea of coupled motion is common in kinetic studies in which the participation of the molecules in question can be demonstrated by the existence of secondary isotope effects. Demonstrating specific coupling in which one molecule tunnels together with another
one is much more difficult. Hole burning and similar spectroscopic studies provide good tools for this purpose. An example in which such phenomena are likely is seen in the relaxation of pentacane doped in benzoic after electronic absorption. The absorption process leads to changes in the pentacane sites, and these sites transform into one another in the dark.23 These studies differ from ours as they start by electronic excitation and depend on the rate of intersystem crossing. Considerably more work needs to be done on our system to decide the many questions raised. Future experiments include trying laser irradiation of the various HDO bands and of the other NH3D+ bands, as well as mixing other metal ions in the Tutton salts. The demonstration of the concerted tunneling suggests that this will occur in many other systems. These systems will provide a rich field of study for hole burning and other multicolor spectroscopies.
Acknowledgment. We are pleased to acknowledge support from the National Science Foundation, Grant CHE-9220908, and thank Anthony Konashenok for his help. References and Notes (1) Trapani, A. P.; Straws, H. L. J . Chem. Phys. 1987,87, 18991900. (2) Trapani, A. P.; Gender, S. W.; Straws, H. L. J. Chem. Phys. 1987, 87,4456-64. (3) Trapani, A. P.; Strauss, H. L. J . Am. Chem. SOC.1989,I l l , 910917. (4) Burrows, W.; Straws, H. L. J. Chem. Phys. 1990,93, 7510-12. (5) Burrows, W.; Straws, H. L. J. Chem. Phys. 1993,99, 5668-76. (6) Cho, H.-G.; Strauss, H. L. J. Chem. Phys. 1993,98, 2774-82. (7) Cho, H.-G.; Strauss, H. L. J. Chem. Phys. 1993,99, 5661-68. (8) Cotton, F. A.; Daniels, L. M.; Murillo, C. A,; Quesada, J. E. Inorg. Chem. 1993,32, 4861 and references therein. (9) Chandler, C. S.; Christos, G. A.; Figgis, B. N.; Reynolds, P. A. J. Chem. SOC., Faraday Trans. 1992,88, 1961. (10) Oxton, I. A.; Knopp, 0. J . Mol. Struct. 1978,49, 309. (1 1) Figgis, B. N.; Reynolds, P. A.; Henson, I.C.; Mutikainen, I. Phys. Rev. B 1993,48, 13372. (12) Simmons, C. J.; Hitchman, M. A,; Stratemeier, H.; Schultz, A. J. J . Am. Chem. SOC. 1993,115, 11304. (13) Doerfler, R. J. Phys. C: Solid State Phys. 1987,20, 2533. (14) Figgis, B. N.; Reynolds, P. A.; Cable, J. W. Mol. Phys. 1993,80, 1377. (15) Araya, M. A.; Cotton, F. A.;Daniels, L. M.; Falvello, L. R.; Murillo, C. A. Inorg. Chem. 1993,32, 4853. (16) Hill, A. E.; Durham, G. S.; Rici, J. E. J. Am. Chem. SOC.1940,62, 2723. (17) Beagley, B.; Eriksson, A.; Lindgren, J.; Persson, I.; Pattersson, L. G. M.; Sandstrom, M.; Wahlgren, U.; White, E. W. J . Phys.: Condens. Matter 1984,I , 2395. (18) Maslen, E.; Ridout, S. C.; Watson, K. J.; Moore, F. H. Acta Crystallogr. 1988,C44, 412. (19) Duth, R. J.; Figgis, B. N.; Forsyth, J. B.; Kucharsk, E. S.;Reynolds, P. A. Proc. R . SOC.London 1989,A421, 153. (20) Beaumont, R. C. Inorg. Chem. 1969,8, 1805. (21) Glick, R. E.; Tewari, K. C. J. Chem. Phys. 1966.44, 546. (22) Luth, K.; Scheiner, S. Int. J. Quantum Chem. 1992,526, 817. (23) Astilean, S.; Corval, A.; Casalegno, R.; Trommsdorff, H. P. J. Lumin. 1994,58, 275. JP942883K