J. Phys. Chem. B 1997, 101, 4111-4119
4111
Liquid Water-Acetonitrile Mixtures at 25 °C: The Hydrogen-Bonded Structure Studied through Infrared Absolute Integrated Absorption Intensities John E. Bertie* and Zhida Lan Department of Chemistry, UniVersity of Alberta, Edmonton, Alberta, Canada, T6G 2G2 ReceiVed: December 3, 1996; In Final Form: February 18, 1997X
The real and imaginary refractive index spectra of mixtures of water and acetonitrile over the full composition range at 25 °C were determined between 8000 and 700 cm-1 by calibrated multiple attenuated total reflection spectroscopy. Under the assumption of the Lorentz local field, the corresponding molar polarizability spectra, Rˆ m(ν˜ ) ) R′m(ν˜ ) + iR′′m(ν˜ ), were calculated and used to investigate the structure of the mixtures. The concentrations of water-bonded, acetonitrile-bonded, and non-hydrogen-bonded O-H groups, and of waterbonded and non-hydrogen-bonded acetonitrile molecules, were obtained from the integrated intensities COH and CCN, the areas under the O-H and CtN stretching bands in the ν˜ R′′m spectra. The results indicate that no enhancement of the water structure (OsH---O bonding) results from the addition of acetonitrile. In contrast, a monotonic decrease in the fraction of O-H groups that are bonded to oxygen is observed with increasing CH3CN content. At low acetonitrile concentration, xCH3CN e 0.05, where x is the mole fraction, the total fraction of OH groups that are hydrogen bonded increases slightly with increasing CH3CN content because the formation of OH---N bonds slightly exceeds the destruction of OsH---O bonds. The present results are consistent with the existence of microheterogeneity at compositions near 30-50 mol % of acetonitrile. However the fraction of OH groups that are hydrogen bonded to water is 0.50 at 50 mol % CH3CN and decreases to 0.35 at 70 mol % CH3CN. Both of these fractions are too small to support water clusters more complex than linear chains or hexagons
Introduction Binary solutions in which water is one of the components have been investigated intensively1-10 due to their importance in many branches of chemistry. Mixtures of water with acetonitrile are particularly favorable for study by vibrational spectroscopy because acetonitrile is a small molecule and the CtN group of the acetonitrile and the OH group of the water are good sensors of their environment. The strong absorption by water is often considered to make study by infrared spectroscopy excessively difficult, but this is not the case in attenuated total reflection spectroscopy.11,12 Different studies of CH3CN-H2O mixtures have given different results about their microstructure. Both molecular dynamic simulation1 and dynamic properties2 have suggested that when the acetonitrile content is sufficiently high the structure is microheterogeneous, i.e., that it contains clusters of molecules of one type, either CH3CN or H2O, and that the order in the water structure increases at low acetonitrile content. Raman spectroscopy3 and thermodynamic studies4,5 have shown no such increase in the water structure at low acetonitrile content but have supported the formation of clusters at intermediate compositions. The infrared study by Gorbunov and Naberukhin6 failed to indicate the existence of clusters at any concentration. Recent work has also not left a clear picture. Marcus and Migron9 made a detailed experimental and theoretical study and summarized the literature. They concluded that CH3CN is solvated by water at low mole fractions of CH3CN, xCH3CN, that microheterogeneity sets in above xCH3CN ) 0.22 ( 0.11, and that discrete complexes of H2O with one or two CH3CN molecules exist at high CH3CN concentrations. At xCH3CN g 0.95 the mixture inclines toward the structure of neat CH3CN and contains 1:1 H2O-CH3CN complexes as solute. Zhang et al.10 used infrared-visible sum frequency generation and X
Abstract published in AdVance ACS Abstracts, April 15, 1997.
S1089-5647(96)03951-X CCC: $14.00
observed a transition in the hydrogen-bonded state of CH3CN molecules in the gas-liquid surface of CH3CN-H2O mixtures from hydrogen-bonded when xCH3CN < 0.07 to not hydrogen bonded at higher CH3CN content. The connection between these findings and the bulk properties is not clear. Huang and Wu7 studied CH3CN-H2O mixtures by measuring their nearinfrared third-harmonic susceptibility, χ(3), which sensitively reflects the microstructure of the liquid. Their results showed the existence of microheterogeneity when the mole fraction of CH3CN is >0.3 but gave no information at low acetonitrile content. The most recent study by infrared spectroscopy is that of Jamroz, Stangret, and Lindgren.8 They used transmission spectroscopy to study the CN stretching band of CD3CN-H2O mixtures and the OD stretching band of dilute HOD in CH3CN-H2O mixtures, all at 20 °C. They fitted the CN stretching band with a single Lorentzian band due to the hydrogen-bonded CtN and a triplet of Lorentzian bands due to the free CtN. From the fits for the different mixtures they calculated the percentage of CH3CN that is hydrogen bonded at each concentration and found that it is far smaller than they predicted from a close-packing model. They concluded that the molecular arrangement in the mixture is remarkably nonrandom, and strong preferential solvation occurs. With regard to the OD stretching band of dilute HDO in CH3CN-H2O mixtures, Jamroz et al. assigned a peak at 2631 cm-1 to HDO deuterium-bonded to CH3CN and a broad band at 2540 cm-1 to HDO-H2O deuterium bonding. They concluded that two types of water molecule and two types of CH3CN molecule exist over a wide composition range. One type consists of molecules that are “in close contact with molecules of the same kind, and the interactions among them are very similar to those in the pure solvent. Molecules of the other type interact strongly through hydrogen bonds with molecules of the other component”. Further, they concluded that the shift with concentration © 1997 American Chemical Society
4112 J. Phys. Chem. B, Vol. 101, No. 20, 1997 of the OD stretching band due to water-water bonding shows that the water molecules form dimers, trimers, and other oligomers, because “the formation of spherical clusters of water molecules would result in concentration-independent positions of the CN and OD stretching bands”. In the present work, calibrated infrared attenuated total reflection (ATR) spectra of CH3CN-H2O mixtures were measured13 and converted to accurate real, n, and imaginary, k, refractive index spectra.13-15 The n and k spectra were converted to complex molar polarizability spectra Rˆ m(ν˜ ) ) R′m(ν˜ ) + iR′′m(ν˜ ), through the use of the Lorentz local field.16,17 The absolute infrared integrated absorption intensities were measured for each composition as Cj, the areas under bands in the ν˜ R′′m spectrum.16,17 These intensities are estimated to be accurate to ∼3%. The dependence of the absolute intensities on composition was analyzed for the different bands to obtain quantitative information about the structure of the mixtures. The composition is described throughout this paper by the mole fraction of acetonitrile, xCH3CN. Experimental Section The spectra were recorded with the Bruker IFS 113V FT-IR spectrometer. A DTGS detector was used to keep the phase correction small. A globar source, 10 mm aperture, automatic gain selection, and an optical retardation velocity of 0.396 cm s-1 were used with both Ge-on-KBr and Si-on-CaF2 beam splitters. Each spectrum resulted from the average of 512 interferograms which was Fourier transformed with trapezoidal apodization (flat out to 0.85 XMAX) and one level of zero-filling. The nominal resolution was 2 cm-1 for water-rich mixtures with xCH3CN less than 0.20, and 1 cm-1 for the remainder. The CIRCLE multiple ATR cell13,14 contained an ATR rod made of ZnSe. For xCH3CN < 0.4, the spectra were measured in a short cell with effective number of reflections, NRF, ∼3.3 because the O-H stretching band is very strong. For xCH3CN > 0.4, a long cell with NRF ∼6 was used. Both cells were used for xCH3CN ) 0.4. Each ATR spectrum was measured as the ratio of the spectrum of the CIRCLE cell full of the liquid under study to the spectrum of the cell full of dry nitrogen gas. The negative decadic logarithm of the attenuated total reflectance, -log(ATR), gave the pATR spectrum. The IUPAC standard optical constants of benzene18 were used to calibrate NRF. Spectra were obtained between 8000 and 700 cm-1 by merging three spectra in the following way: the pATR values above 6500 cm-1 were set to zero; the spectrum recorded with the Si-on-CaF2 beam splitter was used from 6500 to 4500 cm-1; the average of the pATR spectra recorded with the two beam splitters was used between 4500 and 1200 cm-1; and the spectrum recorded with the Ge-on-KBr beam splitter was used from 1200 to 700 cm-1. The water was distilled water passed through a Millipore water system consisting of a SUPER-C carbon filter, two IONEX ion-exchange cartridges, and an ORGANEX-Q carbon filter. The acetonitrile was analytical reagent grade of 99.5% stated purity. The mixtures were made volumetrically (Table 1). Each mixture was kept in a closed glass bottle. Intensity Quantities The programs used to convert pATR spectra to real and imaginary refractive index spectra have been described.13-15 The values of the real refractive index at 8000 cm-1 at 25 °C were found to be 1.325 ( 0.003 for water and 1.325 ( 0.01 for acetonitrile, by fitting the values reported19,20 for several visible wavelengths to n2 ) A + Bν2 + Cν4 and then extrapolating to 8000 cm-1. Values of the real refractive index at several visible
Bertie and Lan TABLE 1: Volumetric Preparation of CH3CN-H2O Mixtures at 22.5 °Ca xCH3CN
V(CH3CN)/mL
V(H2O)/mL
0.00 0.05 0.10 0.15 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00
0.000 6.864 12.951 18.384 23.264 31.671 38.656 44.551 49.593 53.955 57.765 61.122 64.103
50.118 44.751 39.992 35.745 31.930 25.356 19.895 15.286 11.344 7.934 4.955 2.330 0.00
a xCH3CN is the mole fraction of acetonitrile. The liquid densities, d, and molar weights, M, used for the calculation are dCH3CN ) 0.7800 g/cm3 at 22.5 °C; MCH3CN ) 41.05 g/mol; dH2O ) 0.997 65 g/cm3 at 22.5 °C; MH2O ) 18.0153 g/mol.
TABLE 2: Physical Properties of H2O-CH3CN Mixtures22 xCH3CN
Vma at 25 °C (mL/mol)
xCH3CN
nDb at 20 °C
xCH3CN
Qmixc at 20 °C (cal/mol)
0.00 0.05 0.10 0.15 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00
18.0682 19.4649 21.2077 22.8642 24.5611 28.0215 31.4983 35.0462 38.5940 42.2345 45.8561 49.443 53.016
0.00 0.07 0.10 0.16 0.31 0.35 0.42 0.49 0.60 0.61 0.69 0.79 1.00
1.3330 1.3414 1.3430 1.3459 1.3478 1.3479 1.3479 1.3478 1.3473 1.3472 1.3470 1.3462 1.3444
0.00 0.019 0.033 0.056 0.148 0.198 0.390 0.450 0.533 0.618 0.745 0.907 1.00
0.00 -5.34 -3.94 +3.54 +67.60 +100.46 +169.83 +185.07 +198.63 +200.93 +197.4 +125.16 0.0000
a V is the molar volume of mixture. b n is the real refractive index m D at sodium D line. c Qmix is the molar heat of mixing.
wavelengths are not available for all of the mixtures so the fitting procedure could not be used for the mixtures. The real refractive indices for sodium D light of the different mixtures are listed in Table 2. They differ by 0.2, the value shown is the area between 2275 and 2210 cm-1 measured above a straight line through the ordinate values at 2590.8 and 2150.2 cm-1. The number in parentheses is the maximum deviation in the last digit. c The fraction of H-bonded CN groups determined by eqs 2 with the intensity parameters C′CN ) 0.1198 km/mol and C′CNH ) 0.206 km/ mol. d The fractions of water-bonded and free OH groups determined from eqs 3, x′CNH, and C′OH ) 0.41 km/mol, C′OHO ) 5.82 km/mol, and C′OHN ) 5.67 km/mol.
mol-1, was multiplied by the mole fraction of water in the mixture and subtracted from the total area under the spectrum of the mixture. The intensities CCN so obtained are shown for the mixtures by the crosses in Figure 6. The second method used a baseline drawn between the ordinate values at 2590.8 and 2150.2 cm-1, and the area between 2275 and 2210 cm-1 was measured above this baseline. The intensities CCN measured by the second method are shown by the open circles in Figure 6. They are about 3% larger than those from the first method for xCH3CN > 0.2 and are smaller for xCH3CN < 0.2. For the higher concentrations of acetonitrile, xCH3CN > 0.2, the ν˜ R′′m spectrum of water multiplied by the mole fraction of water is slightly higher than the baseline of the spectrum of the mixture, so the first method clearly overestimates the water absorption. For xCH3CN < 0.2, the second method gives a distorted spectrum after the baseline is subtracted, clearly because it overestimates the water absorption at these high water concentrations. Thus the intensities, CCN, that are used are those from the first method for xCH3CN e 0.2 and those from the second method for the remaining solutions. They are tabulated in Table 3 together with the maximum deviations from the mean. They are shown by the filled circles in Figure 6 with error bars that show the 95% confidence limits. These CCN values all lie well above the ideal solution line. Hydrogen Bonding of Acetonitrile. The nonlinear curves in Figures 5 and 6 indicate that the O-H and CtN groups are involved in the interaction between the water and acetonitrile molecules in the mixtures. The interaction is undoubtedly hydrogen bonding between lone pair electrons on the CtN group and the proton of the O-H group. This is expressed as CH3CN‚‚‚H-O-H, or CNH for short. Thus, the CN group can be taken to be either bonded or not. The integrated intensity under the CtN stretching band in one mole of mixture at concentration xCH3CN can be related to the extent of hydrogen bonding through
4116 J. Phys. Chem. B, Vol. 101, No. 20, 1997
Bertie and Lan
Figure 6. CCN, the integrated intensity under the CtN stretching band in ν˜ R′′m spectra of CH3CN-H2O mixtures at 25 °C plotted against mole fraction of acetonitrile. The integration limits were 2275 and 2210 cm-1: (4) indicates the total intensity; (×) indicates the intensity corrected for the water absorption by the first method described in the text; (O) indicates the intensity corrected for the water absorption by the second method described in the text; (b) indicates the final intensities used. All symbols show the average from g4 spectra. The error bars for the total intensity show the maximum deviation from the average while those for the final intensity show the 95% confidence limit. The straight line shows the behavior expected for an ideal solution.
CCN ) xCH3CN(x′CNC′CN + x′CNHC′CNH)
(2)
x′CN + x′CNH ) 1
(2a)
with
Here x′j (j ) CN or CNH) is the fraction of CtN groups that are free or hydrogen bonded, and C′j (j ) CN or CNH) is the integrated intensity per mole of free or hydrogen-bonded CtN groups. In order to obtain the fractions x′CNH and x′CN in each mixture, the intensities C′CN and C′CNH in eq 2 were assumed to be independent of composition. In pure acetonitrile, the CtN group is not hydrogen bonded. Consequently, its area, CCN ) 0.1198 km/mol (Table 3), was taken as C′CN. Jamroz et al.8 found that the CtN stretching band of free CD3CN appeared only as a shoulder when the mole fraction of CD3CN was less than 0.14, indicating that nearly all of the CD3CN is hydrogen bonded to water at low concentration. It was therefore assumed in the present work that all of the acetonitrile in the water is hydrogen bonded at infinitely low CH3CN concentration. The intensity per mole of the hydrogen-bonded CN stretching vibration, C′CNH, is then given by the slope of the intensity versus mole fraction curve at the limit xCH3CN f 0.0. This slope was found by fitting the intensities at xCH3CN ≈ 0.05, 0.10, and 0.15 to the quadratic equation CCN ) bxCH3CN + cx2CH3CN. It was found that b ) C′CNH ) 0.206 km/mol. It should be noted that these intensities C′CN and C′CNH are smaller than the theoretically significant intensities, because the integration range was limited by practical considerations and
Figure 7. Percentage of acetonitrile molecules in CH3CN-H2O mixtures that are free (O) or hydrogen bonded (b) plotted against composition.
could not be extended far enough to include all of intensity in the near-Lorentzian8 components of the band. The fraction of acetonitrile that is hydrogen bonded, x′CNH, was calculated through eqs 2 and 2a for each mixture and the values are listed in Table 3. The values of x′CNH and x′CN are plotted as percentages in Figure 7 against the mole fraction of acetonitrile. They are also plotted, in the lower box in Figure 7, against the number of water molecules per 100 acetonitrile molecules. This inconvenient abscissa quantity is included to allow visual comparison with Figure 3 of Jamroz et al.’s paper,8 which shows the same quantities calculated from fits of their absorbance spectra and from a close-packed model. Our values agree very well with their experimental values. Specifically, the two curves have the same general shape, the maximum value of x′CNH × 100% agrees at 79 and 78%, and the only significant differences are at xCH3CN ) 0.20 and 0.30, i.e., 400 and 230 molecules of water per 100 molecules of CH3CN, respectively, where our values are 56 and 50% compared with their 68 and 59%. Hydrogen Bonding of Water. The absorption by the mixtures in the O-H stretching region results from free O-H groups and two types of hydrogen-bonded O-H groups: OH bonded to CH3CN, abbreviated to OHN, and OH bonded to water, abbreviated to OHO. Accordingly, the intensity per mole of mixture, COH, is given by
COH ) 2xH2O(x′OHNC′OHN + x′OHOC′OHO + x′OHC′OH ) (3) where
x′OHN + x′OHO + x′OH ) 1
(3a)
Here x′j (j ) OHN, OHO, OH) is the fraction of OH groups
Liquid Water-Acetonitrile Mixtures at 25 °C that are of the type OHN, OHO, or free OH, C′j (j ) OHN, OHO, OH) is the integrated intensity of one mole of such groups, and there are 2 mol of OH group per mole of H2O. The molar intensities C′j (j ) OHN, OHO, OH) were taken to be independent of composition. Information from the literature had to be used with the data from this work in order to find their values. In pure liquid water at room temperature, not all of the O-H groups are hydrogen bonded (OHO). Luck et al.25 determined the number of OH---O hydrogen bonds as 1.73 mol per mole of water. Accordingly, the value of C′OHO is calculated from the COH value of pure water (Table 3) as 10.08/1.73 ) 5.82 km/mol of OHO groups. To find C′OH, it was noted that the integrated intensity of the O-H stretching vibrations in the gas phase is about 7% of that in the liquid.26 Thus, the value of C′OH was taken to be 0.07C′OHO ) 0.41 km/mol of OH groups. C′OHN was found in the following way from the OH stretching intensities at high CH3CN concentrations. The slope of the graph of COH against xCH3CN (Figure 5) as xCH3CN f 1.0 was determined from a quadratic function fitted to the values at xCH3CN ) 0.70, 0.80, and 0.90, as described earlier for CCNH. Its value is 5.55 ( 0.14 km/mol of water. To translate this into the intensity per mole of bonds, the fraction of the OH bonds in a water molecule that forms OH---N bonds at infinite dilution is required. This was obtained from the previous determination of x′CNH, the fraction of the acetonitrile that is hydrogen bonded. The number of H-bonded CH3CN molecules per mole of mixture is given by xCH3CNx′CNH. Thus, the number of H-bonded CH3CN molecules per mole of water molecules is given by xCH3CNx′CNH ÷ xH2O, which is the slope of the graph of xCH3CNx′CNH versus mole fraction of water. This is the same as the number of OHN hydrogen bonds per mole of water molecules, the number that is sought for infinite dilution. To find this quantity at infinite dilution the product quantity xCH3CNx′CNH was calculated for xCH3CN ) 0.70, 0.80, and 0.90, i.e., for xH2O ) 0.30, 0.20, and 0.10, and fitted to a quadratic function in xH2O which passes through the origin, as described previously for C′CNH. The slope was 0.90 at xH2O ) 0.0. Consideration of the difficulty of separating the H2O and CH3CN absorptions, and the fact that the fit included data for water mole fractions as high as 0.30, yielded an estimated error in this slope of (0.2. Thus, the number of OHN bonds per mole of water molecules was found to be 0.90 ( 0.2. Half of this value, 0.45 ( 0.1 is x′OHN at xH2O f 0, the fraction of OH groups that is hydrogen bonded to acetonitrile when the water is infinitely dilute in acetonitrile. It may seem anomalous that about half of the OH groups of dilute water in acetonitrile are not hydrogen bonded. This idea is not new, although the evidence that has been cited in the past is not convincing. Bonner and Choi27 studied the water overtone band, ν2 + ν3 near 5250 cm-1 for various wateracetonitrile mixtures. The band they assigned to the free bonds provided 28% of the total area at 5 mol % of water. However, the detailed significance for this work of their results and conclusions is very much in doubt. At very low water concentration, there is little doubt that water molecules are isolated and no OH---O hydrogen bonding occurs in the solution. Therefore, the derivative of the area COH at xH2O ) 0.0 was derived from eq 3 as
dCOH/dxH2O ) 2(x′OHNC′OHN + x′OHOC′OHO + x′OHC′OH) (4) with dCOH/dxH2O ) 5.55 km/mol, x′OHN ) 0.45, x′OHO ) 0.0, C′OHO ) 5.82 km/mol, x′OH ) 0.55, and C′OH ) 0.41 km/mol. Accordingly, the value of C′OHN is 5.67 km/mol. The possible
J. Phys. Chem. B, Vol. 101, No. 20, 1997 4117
Figure 8. Percentage of OH groups in CH3CN-H2O mixtures that are free (O), hydrogen bonded (b), OsH---O bonded (2), or OsH---N bonded (4), plotted against composition.
error in x′OHN ) 0.45 ( 0.1 allows C′OHN to be as large as 7.3 km/mol and as small as 4.63 km/mol. Equations 3 were then used with these values of the parameters C′OH, C′OHO, and C′OHN to calculate x′OH and x′OHO, the fractions of nonbonded and OHO-bonded OH bonds, at the different concentrations. The fraction of OHN bonds, x′OHN, is needed for this at each composition and was calculated from the fraction of acetonitrile molecules that is hydrogen bonded, x′CNH, which was calculated in the previous subsection from the CN stretching intensity. Specifically the number of hydrogen-bonded CH3CN molecules per mole of solution is x′CNHxCH3CN. This equals the number of OHN bonds per mole of solution, which is also given by 2x′OHNxH2O. Hence,
x′OHN ) xCNHxCH3CN/2xH2O
(5)
With eqs 5 and 3a and the C′ values of the intensities per mole of bond, eq 3 can now be solved for the one unknown at each concentration. The resulting fractions x′OH, x′OHO, and x′OHN of OH, OHO, and OHN bonds are plotted against mole fraction of acetonitrile in Figure 8, together with the sum x′OHO + x′OHN. x′OH and x′OHO are tabulated in Table 3 together with the x′CNH from the previous subsection that gave the x′OHN through eq 5. The possible errors in C′OH and, particularly, C′OHN yield possible errors in x′OHO that range from (0.006 at xCH3CN ) 0.05 through (0.04 at xCH3CN ) 0.50 and (0.06 at xCH3CN ) 0.70 to (0.08 at xCH3CN ) 0.90. Clearly these possible errors do not change substantially the chemical information contained in Figure 8. Discussion The numerical results of this work can be summarized as follows. At very low mole fraction of acetonitrile, where acetonitrile and water mix exothermically, the CH3CN is all hydrogen bonded. The hydrogen-bonded fraction drops rapidly to 50%
4118 J. Phys. Chem. B, Vol. 101, No. 20, 1997 by xCH3CN ∼ 0.3 and then drops more slowly but uniformly to zero in pure acetonitrile. The fraction of OH groups that forms OH--N bonds is zero in pure water and rises uniformly to 0.45 in slightly wet CH3CN. At xCH3CN ) 0.3, 0.5, and 0.7 this fraction is ∼0.10, 0.18, and 0.30. The fraction of OH groups that forms OH--O bonds is ∼0.865 in pure water and decreases monotonically to zero in pure CH3CN. From xCH3CN ) 0.1 it decreases roughly linearly in mole fraction to xCH3CN ∼ 0.6 and then decreases more rapidly to pure CH3CN. At xCH3CN ) 0.3, 0.5, and 0.7 this fraction is 0.68, 0.52, and 0.35. The fraction of OH groups that forms hydrogen bonds of both types rises slightly from pure water to 10% CH3CN and then drops steadily to 0.45 in wet CH3CN. The fraction of OH groups that is not hydrogen-bonded decreases slightly from pure water to xCH3CN ∼ 0.10 and then climbs steadily to 0.55 in wet CH3CN. At xCH3CN ) 0.3, 0.5, and 0.7 this fraction is ∼0.22, 0.30, and 0.37. Thus, in the equimolar mixture, the fractions of OH groups that form OH--O bonds, OH--N bonds, and are not bonded, are 0.52, 0.18, and 0.30 respectively. These results must be compared with the large body of literature which reports studies of the CH3CN-H2O system by a variety of methods. Marcus and Migron9 made a detailed experimental and theoretical study of this system and their 1991 paper makes a convenient starting point for the discussion. They measured Taft’s28 π*, R, and β, which reflect the polarity, hydrogen-bond donor ability, and hydrogen-bond acceptor ability of mixtures of water and acetonitrile, and they analyzed thermodynamic data from the literature by the inverse Kirkwood-Buff (IKBI) integrals29,30 and the quasi-lattice quasichemical (QLQC) method.31 From their results and survey of the previous literature they presented the following picture of the structure of these solutions. At low CH3CN concentrations the CH3CN enters cavities in the water structure. Solvation by the water is generally agreed, but different methods disagree on the importance of water structure-making or structure-breaking to this process. Beyond xCH3CN ∼ 0.15 (0.10-0.33 by various methods) the CH3CN cannot be accommodated in the cavities. Beyond this limit microheterogeneity sets in, wherein the hydrogen-bonded structure in the water clusters is enhanced relative to that in neat water. More specifically, microheterogeneity means a preference for neighbors of the same kind that extends over several concentric shells around a given molecule. Thus the mole fraction of the water is greater near a water molecule than the bulk value, and the mole fraction of CH3CN is greater near a CH3CN molecule than the bulk value. The weight of evidence is in favor of microheterogeneity existing in the middle range of compositions. At xCH3CN > 0.7 the water clusters are so few and far apart that new kinds of interactions occur. Water-CH3CN interactions that could be discounted in the middle range are now important. Marcus and Migron viewed this region as containing discrete water-acetonitrile complexes surrounded by a rather inert acetonitrile solvent. The complexes were taken to be CH3CN--HOH and CH3CN--HOH--NCCH 3. At xCH3CN ) 0.95 another change is indicated by some methods. The 1:1 H-bonded structure may still exist, but the CH3CN “has a weakly manifested structure as in the neat liquid”.9 It can be noted again here that Huang and Wu7 obtained clear evidence for the existence of microheterogeneity when the mole fraction of CH3CN is >0.3.
Bertie and Lan Jamroz et al.8 have studied infrared spectra of CH3CN-H2O mixtures as outlined in the Introduction. Of relevance here are the following points. The fraction of hydrogen-bonded CH3CN they found is in excellent agreement with that found in this work by a different method. Their comparison of these numbers with the much greater fractions they obtained from a random close-packing model led them to conclude that the arrangement in the mixture is remarkably nonrandom and strong preferential solvation of H2O by H2O and CH3CN by CH3CN occurs in the mixtures. Their study of dilute OD bonds in CH3CN-H2O mixtures led them to conclude that two types of water and two types of acetonitrile exist over a wide composition range, those in contact with like molecules and those in contact with molecules of the other type. The CN stretching band and the bands assigned to OD--O bonds shifted steadily with mole fraction of CH3CN, from which they concluded that the water molecules form dimers, trimers and higher oligomers, because “the formation of spherical clusters (globules) of water molecules would result in concentration-independent positions of the CN and OD stretching bands”.8 Mixing water and acetonitrile is an exothermic process below xCH3CN ) 0.04 and is endothermic elsewhere. The present results (Figure 8) show a slight increase in the total fraction of hydrogen-bonded OH bonds between xCH3CN ) 0.0 and 0.05, with the number returning to its value in pure water near xCH3CN ) 0.1. The increase is achieved by the formation of OH--N bonds instead of OH--O bonds. Thus, the acceptance of CH3CN into the lattice is not accompanied by enhancement of the water structure, the fraction of OH--O bonds decreasing steadily with increasing CH3CN content. It can be noted from considerations of molecular size that there is not enough water to fully surround an acetonitrile molecule when the mole fraction of acetonitrile is significantly larger than ∼0.08, i.e., 1 CH3CN to 12 H2O. Thus full solvation of acetonitrile can not occur for xCH3CN much above 0.08. The rapid decrease in the fraction of hydrogen-bonded CH3CN molecules with increasing CH3CN content ends near 50% near xCH3CN ) 0.3. This rapid decrease is consistent with the establishment of microheterogeneity with the formation of clusters of acetonitrile molecules, and the wide composition range reported for the onset may reflect the sensitivity of the different methods to the phenomenon. There is general agreement that microheterogeneity exists at xCH3CN ) 0.33, which is where the fraction of hydrogen-bonded CH3CN molecules is ∼0.5 and starts to decrease more slowly with increasing CH3CN content. At this point, the fractions of OH bonds that form OH--N bonds, OH--O bonds, and are not bonded are about 0.1, 0.7, and 0.22. Thus, the present results do not support the proposal noted above that the water structure is enhanced (above that in pure water) by the microheterogeneity. The fraction of OH bonds that are OH--O bonded shows that the region of microheterogeneity cannot extend much beyond xCH3CN ) 0.50. This can be seen by considering the fraction of OH bonds that must be hydrogen bonded in various water clusters. In H2O(H2O)4, i.e., a water molecule solvated by four others, the fraction of bonded OH bonds is 0.40. It increases to 0.47 in H2O(H2O)14, 0.50 in H2O chains and ice-like hexagonal rings, 0.67 in isolated dodecahedra, and 0.74 in a 39-molecule piece of ice Ih. Thus, the type of cluster that can form is severely limited when the fraction of the OH bonds that form OH--O bonds is 0.5 or less. This fraction, x′OHO, is 0.5 in the equimolar mixture. At xCH3CN ) 0.7 it is only 0.35, the same as the fraction of free OH bonds and nearly the same as the fraction that forms OH--N bonds, 0.30. These considerations lead to broad agreement with Jamroz
Liquid Water-Acetonitrile Mixtures at 25 °C et al. with respect to the acetonitrile-rich mixtures. We conclude that the structure must change near xCH3CN ) 0.5 from a microheterogeneous one, in which large clusters of acetonitrile molecules and large clusters of hydrogen-bonded water molecules can exist, to one in which water is increasingly hydrogen bonded to CH3CN rather than H2O with increasing xCH3CN, and in which only short water chains and rings can exist as well as some isolated water molecules. Above xCH3CN > ∼0.85, the structure must change again because there are enough CH3CN molecules to surround each water molecule, and the water molecules increasingly form just a single hydrogen bond to one of the surrounding CH3CN molecules. Conclusions The results of this study are generally supportive of the current picture of the structure of acetonitrile-water mixtures and provide the following specific details for the first time. The fraction of hydrogen-bonded OH groups increases slightly in the region below xCH3CN ∼ 0.05, where the mixing process is exothermic. This arises because more OH---N bonds are formed than OH---O bonds are broken. The fraction of hydrogenbonded CH3CN molecules decreases rapidly as xCH3CN increases to 0.3, consistent with the establishment of microheterogeneity as xCH3CN approaches 0.3. The water structure is not, however, enhanced by the microheterogeneity since the fraction of OH bonds that are OH---O bonded decreases monotonically with increasing CH3CN content. The microheterogeneity cannot exist far beyond the equimolar mixture, because there are insufficient OH---O bonds to support larger units than short chains or small rings of water molecules when xCH3CN > 0.5. At very high acetonitrile mole fractions, 90% of the water molecules bond to one CH3CN molecule and are presumably solvated by other CH3CN molecules. Supporting Information Available: The Compact Tables A-I to A-VIII are available as Supporting Information in digital form at the ACS Web site. See any current masthead page for access information. The complete n and k spectra of acetonitrile, water, and all of the 11 mixtures studied are available in digital form from J.E.B.’s Web site, http://www.ualberta.ca/∼jbertie/ jebhome.htm. References and Notes (1) Kovacs, H.; Laaksonen, A. J. Am. Chem. Soc. 1991, 113, 5596. (2) Goldammer, E.; Herz, H. G. J. Phys. Chem. 1970, 74, 3734.
J. Phys. Chem. B, Vol. 101, No. 20, 1997 4119 (3) Kabisch, G. Z. Phys. Chem., Leipizig 1982, 263,48. (4) Moreau, C.; Douheret, G. Thermochim. Acta 1975, 13, 385. (5) Armitage, D. A.; Blandamer, M. J.; Foster, M. J.; Hidden, N. J.; Morcom, K. W.; Symons, M. C. R.; Wootten, M. J. Trans. Faraday Soc. 1968, 64, 1193. (6) Gorbunov, B. Z.; Naberukhin, Y. I. Zh. Strukt. Khim. 1975, 16, 816. (7) Huang, J. Y.; Wu, M. H. Phys. ReV. E 1994, 50, 3737. (8) Jamroz, D.; Stangret J.; Lindgren, J. J. Am. Chem. Soc. 1993, 115, 6165. (9) Marcus, Y.; Migron, Y. J. Phys. Chem. 1991, 95, 400. (10) Zhang, D.; Gutow, J. H.; Eisenthal, K. B.; Heinz, T. F. J. Chem. Phys. 1993, 98, 5099. (11) Bertie, J. E.; Lan, Z. Appl. Spectrosc. 1996, 50, 1047. (12) Bertie, J. E.; Zhang, S. L. J. Chem. Phys. 1994, 101, 8364. (13) Bertie, J. E.; Eysel, H. H. Appl. Spectrosc. 1985, 39, 392. (14) Bertie, J. E.; Harke, H.; Ahmed, M. K.; Eysel, H. H. Croat. Chem. Acta 1988, 61, 391. (15) Bertie, J. E.; Zhang, S. L.; Manji, R. Appl. Spectrosc. 1992, 46, 1660. (16) Bertie, J. E.; Zhang, S. L.; Keefe, C. D. J. Mol. Struct. 1994, 324, 157. (17) Bertie, J. E.; Zhang, S. L.; Eysel, H. H.; Baluja, S.; Ahmed, M. K. Appl. Spectrosc. 1993, 47, 1100. (18) Bertie, J. E.; Keefe, C. D.; Jones, R. N. Tables of Intensities for the Calibration of Infrared Spectroscopic Measurements in the Liquid Phase; International Union of Pure and Applied Chemistry; Blackwell Science Ltd.: Oxford, U.K., 1995. (19) Tiltona, L. W.; Taylor, J. K. J. Res. Natl. Bur. Stand. 1938, 20, 419. (20) Timmermans, J. Physico-Chemical Constants of Pure Organic Compounds; Elsevier: Amsterdam, 1965; Vol. 2, p 529. (21) Bertie, J. E.; Jones, R. N.; Apelblat, Y. Appl. Spectrosc. 1993, 47, 1989. (22) Timmermans, J. Physico-Chemical Constants of binary systems; Interscience: New York, 1961; Vol. 4, p 65. (23) Narvor, A. L.; Gentric, E. G.; Saumagne, P. Can. J. Chem. 1971, 49, 1933. (24) Bertie, J. E.; Lan, Z., unpublished work. (25) Luck, W. A. P.; Borgholte, H.; Habermehl, T. J. Mol. Struct. 1988, 177, 523. Kretzer, L. E.; Fritzsche, M.; Luck, W. A. P. J. Mol. Struct. 1988, 175, 277. (26) Bertie, J. E.; Ahmed, M. K.; Eysel, H. H. J. Phys. Chem. 1989, 93, 2210. (27) Bonner, O. D.; Choi, Y. S. J. Phys. Chem. 1974, 78, 1723. (28) Kamlet, M. J.; Abboud, J.-L. M.; Abraham, M. H.; Taft, R. W. J. Org. Chem. 1983, 48, 2877. (29) Matteoli, E.; Lepori, L. J. Chem. Phys. 1984, 80, 2856. (30) Blandamer, M. J.; Blundell, N. J.; Burgess, J.; Cowless, H. J.; Horn, I. M. J. Chem. Soc., Faraday Trans. 1990, 86, 277. (31) Marcus, Y. J. Chem. Soc., Faraday Trans. 1 1989, 85, 381.
