ASSOCIATION OF TRIFLUOROACETIC ACID
2039
Association of Trifluoroacetic Acid in Vapor and in Organic Solvents by Sherril D. Christian* and Thomas L. Stevens University of Oklahoma, Norman, Oklahoma
75069
(Received September 15, 1971)
Publication costs borne completely by T h e Journal of Physical Chemistry
Thermodynamic and infrared spectral results are reported for the monomer and dimer of trifluoroacetic acid (TFA) in vapor and in the solvents cyclohexane, carbon tetrachloride, benzene, and 1,2-dichloroethane. The more reactive solvents interact strongly with the TFA monomer, greatly reducing the energy and free energy of this species with respect to vapor and decreasing the carbonyl- and hydroxyl-stretching frequencies. The dimer, which is thought to be cyclic, is considerably less affected by change of medium; it is less than half as effectively solvated as two separated monomer molecules. A new spectral-partition method is introduced which can be used to study association equilibria in media which are opaque to radiation in spectral regions of interest. There have been numerous investigations of the association of trifluoroacetic acid (TFA)1-4 and other fluorinated carboxylic a ~ i d s .Most ~ ~ ~investigators ~ ~ have concluded that TFA associates primarily to the cyclic dimer, at least in the vapor phase and as a solute in nonpolar solvent^.^ However, infrared spectral results for solutions of the perfluorinated aliphatic acids in slightly basic solvents have been interpreted as evidence for the existence of linear dimers and/or higher polymers.5r7 Recently, Kirszenbaum, et u Z . , ~ have reexamined the infrared spectra of TFA in various nonpolar and slightly polar solvents. Their spectra (in the v(0H) and v(C=O) vibration regions) show no effects which are inconsistent with an assumed monomer-cyclic dimer equilibrium. Although the carbonyl- and hydroxylstretching vibrations of the TFA monomer shift progressively toward lower frequencies in the more reactive media, the dimer carbonyl-stretching frequency appears to be practically independent of the solvent. This paper summarizes spectral and thermodynamic constants obtained for TFA in a range of media and includes a discussion of the effects of nonpolar and slightly polar solvents on physical properties of the TFA monomer and dimer. Spectral results from this laboratory for TFA in vapor, CC14, benzene, and 1,2dichloroethane (DCE) are in essential agreement with those reported by Kirszenbaum, et uL4 Our thermodynamic results provide additional evidence that the more reactive solvents interact strongly with the TFA monomer. However, the observed medium effects on values of energies, free energies, and spectral characteristics of the TFA dimer support the view that the dimer is primarily cyclic in all the media investigated.
Experimental Section Trifluoroacetic acid (Matheson Coleman and Bell reagent grade) was distilled through a 30-plate bubblecap column at, a reflux ratio of 10: 1. Only the middle
portion (bp 72", corrected to 760 Torr) mas collected; it was stored in a desiccator over CaS04. Solvents were purified by distillation, where necessary, and stored over Cas04 or Pz05. All transfers of liquids were performed using closed all-glass systems fitted with Teflon-bore stopcocks; stock solutions of TFA in solvents were kept in vapor contact with CaS04. Great care was taken to avoid introducing water into spectral cells, either in the process of adding chemical components or during measurements. Complete details of procedures used in storing and transferring components are given in ref 3b. Infrared spectral measurements were made using a Beckman DK-1A extended-range spectrometer (in the 3500-cm-' region) and a Perkin-Elmer 12C singlebeam instrument equipped with a CaFz prism (in the lSOO-cm-' region). Three types of spectral experiments, requiring different apparatus and procedures, were performed: (1) vapor spectral studies, in which TFA samples at known pressures in the range 0-60 Torr were introduced into a gas cell thermostated at 25.0 k 0.2" ; (2) conventional condensed-phase experiments, in which solutions of TFA at known concentrations in various solvents were transferred into thermostated solution cells; and (3) spectral condensedphase-vapor-phase partition studies, requiring new techniques and apparatus. Since experiments of types (1) N. Fuson, M. L. Josien, E. A. Jones, and J. R. Lawson, J . Chem. Phys., 20, 1627 (1952). (2) R. E. Kagarise, ibid., 27,519 (1957) ; Naval Research Laboratory
Report No. 4955, Aug 8, 1958. (3) (a) S. D. Christian, Ph.D. Dissertation, Iowa State University, Ames, Iowa, 1956; (b) T. L. Stevens, Ph.D. Dissertation, University of Oklahoma, Norman, Okla., 1968. (4) M. Kirszenbaum, J. Corset, and M .L. Josien, J . Phys. Chem., 75, 1327 (1971), and references cited therein. ( 5 ) T.S.S. R. Murty, ibid., 75, 1330 (1971). (6) C. C. Costain and G. P. Srivast)ava,J . Chem. Phys., 35, 1903 (1961). (7) T. S. S. R. Murty and K. S. Pitzer, J . Phys. Chem., 73, 1426 (1969).
The Journal of Physical Chemistry, Vol. 76,N o . 14, 1978
SHERRILD. CHRISTIAN AND THOMAS L. STEVENS
2040 1 and 2 resemble those commonly employed in infrared studies, the reader is referred to ref 3b for complete details. However, the spectral-partition method has several novel features which justify our describing the main features of the method here. Partition methods have been used with success in numerous studies of the activities and association reactions of solute speciesS8 In applying partition methods to studies of complex formation, it is necessary to establish equilibrium between an indicator phase (in which the component of interest is maintained at a known activity) and a second phase (the reaction phase) in which the association reaction is presumed to occur. From measurements of the concentration (or some related property) of thc associating component in the rcaction phase as a function of its activity in the indicator phase, thermodynaniic properties of complexes can be inferred. The spectral-partition method as applied here involves measurement of the spectra of TFA in the vapor phase above the reaction phase, which in the present experiments consists of TFA dissolved in organic solvent. Initially, base-line spectra are recorded with the thermostated, evacuated 10-cm cell (Figure 1) mounted in the DK-1A spectrometer; then, TFA vapors from an external reservoir are transferred into the cell under vacuum to give a monomer absorbance (at 3589 cm-l) in the required range (usually 0.2-0.7). After equilibration at 25.0 f 0.2", the spectra in the 3600-cm-' region are recorded. Next, a sample of dry solvent is delivered quantitatively through the sintered-glass disk, using a 0.2-ml Roger Gilmont Industries microburet (No. S-1200 A). Equilibrium is reestablished after 10 or 15 min, whereupon the vapor spectra are again recorded. Successive samples are added and spectra measured at intervals of 10-15 min, until a maximum reaction phase volume of not more than 4-5 ml has been reached. (In the 10-cm cylindrical cell, at volumes less than 4 ml, the liquid level lies safely below the beam, and there is little tendency for droplets to accumulate on the cell windows. Apparently, the energy absorbed from the beam maintains a sufficient vertical temperature gradient to prevent condensation of the volatile components on the windows.) From the vapor spectrum the absorbance of the TFA monomer
is obtained. Then, by utilizing results of experiments on TFA vapor at known pressures (vide infra) it is possible to infer the total molar amount of TFA in the vapor phase. The concentration of TFA in the condensed phase can be calculated from knowledge of the total amount of TFA in the cell, the volume of the cell, the volume of solvent added, and the vapor pressure of the solvent at 25.0'. Slight corrections are made for the reduction in vapor pressure of the solvent caused by the dissolved TFA.
Calculations and Results ( 1 ) Vapor-PhaseResults. Aleasuremcnts were made of the absorbance A of TFA vapor samples a t measured total pressures P in a 10.0-cm cell at 25.0'. Assuming the presence of only TFA monomers and dimers, individually obeying the ideal gas law, the total pressure may be expressed as
p = P M-k PD= P M-t K 8 x 2
(1)
where PMand PDare the monomer and dimer partial pressures and K 2 is the association constant for dimerization in reciprocal pressure units. If the absorbance a t 3589 cm-l is attributable solely to the monomer, which is assumed to obey Beer's law, PMmay be replaced by A,RT/eMb, where A M is the peak absorbance of the monomer band, enf is the absorptivity of the monomer, and b is the path length of the cell. Thus, eq 1 becomes
or
K,(RT)' +AX eMb
_P_ --- RT
AN
E
M
~
~
W)
~
Figure 2 displays absorbance measurements at 3589 cm-1 plotted as P I A Mus. A M ; from least-squares
I
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
AM
Figure 2. Vapor-phase spectral results for trifluoroacetic acid a t 25.0". Absorbance measured a t 3589 cm-1, using a 10-cm cell.
Figure 1. Spectral-partition cell. The Journal of Physical Chemistry, Vol. 76, No. 14,1976
(8) S. D. Christian, A. A. Taha, and B. W. Gash, Quart. Rev., Chem. SOC.,24, 1 (1970).
ASSOCIATION OF TRIFLUOROACETIC ACID
1
"
" ; & '
"
'
i o '
"
'
204 1 (3) Spectral-Partition Results. I n the spectralpartition studies, data are reduced t o sets of values of the monomer absorbance in the vapor phase, A M ,and the formal concentration of TFA in the condensed phase, f ~ .Since the TFA monomer in solution is assumed to obey Henry's law, its concentration may be equated to AxRT/(eMKHb), where K H is the Henry's law constant for the monomer in units of pressure/ concentration. Then, the formal pressure equation may be written
A '
%
AMRT
Figure 3. Spectral results for dilute solutions of trifluoroacetic acid in carbon tetrachloride. Absorbance measured in a 5-em cell a t 3508 cm-l; formal concentration in molar units.
analysis of these data in the form of eq 2, it is possible to calculate EM and K2. The dashed line in Figure 2 corresponds to the least-squares values = 143 i 5 cm-' 1. mol-' and K , = 0.30 =t0.02 Torr-l or 5660 370 1. mol-l. Previously, Ling, et ~ l .reported , ~ the value K 2 = 0.30 Torr-' for TFA vapor at 25", based on vapor density and infrared spectral measurements. ( 2 ) Dilute Solution Measurements for TFA in CC14Hydroxyl-Stretching Region. I n the condensed phase, if it is assumed that the monomer and dimer individually obey Henry's law, the total or formal concentration of dissolved TF-4 may be expressed as
*
f.4
= Cllr
+ 2K2C1f2
(3)
where CMis the monomer concentration and K1 is the dimerization constant in reciprocal concentration units. The band attributed to monomeric TFA in CC1, is a t 3508 cm-'; assuming that the absorbance at this frequency is linear in monomer concentration, one may write
=
m
+ 2K2 (-)AMRT ExKHb
(5a)
or
where AM and E M are properties of TFA in vapor, and and K2 pertain to the liquid phase. Figure 4 shows spectral-partition data plotted as fA/AM vs. A Mfor four TFA-solvent systems. The solid lines represent leastsquares fits of data in the form of eq 5; Table I1 summarizes constants inferred from the spectralpartition results. In the case of the benzene solutions, significant departure from linearity is observed at the highest TFA concentrations. Therefore, for this solvent system, least-squares constants were calculated from results for solutions a t molarities less than 0.21. It should be noted that the presence of linear as well as cyclic dimers in the condensed phase would not affect the validity of the K 2value inferred from spectral-
fA
(4) Figure 3 includes data for TFA in CCL at three temperatures, plotted in the form jJAM vs. A M . Leastsquares analysis of the data, using eq 4, yields the values of the constants shown in Table I. From the temperature dependence of the Kz values, the enthalpy change for the dimerization reaction is calculated t o be -9.0 f 0.4 lical/mol, in good agreement with that reported by Kagarise (-8.85 kcal/mol).2 A.
Table I : Dimerization Constants and Monomer Absorptivities T.OK
Kz,1. mol-1
288 298 313
205 f 8 128 f 7 58 f 4
t u , om-1
1. mol-1
138 i: 2 139 rt 2 137 i: 2
Figure 4. Spectral-partition data for trifluoroacetic acid in four solvents at 25.0'. Absorbance of vapors measured at 3589 em-f in a 10-cm cell; formal concentrations of trifluoroacetic acid in condensed phases in molar units. (9) C. Ling, 6. D. Christian, H. E. Affsprung, and R. W. Gray, J . Chem. SOC.A, 293 (1966); C. Ling and H. E. Affsprung, unpublished work. The Journal of Phwical Chemistry, Vol. 76,No.
1978
SHERRIL D. CHRISTIAN AND THOMAS L. STEVENS
2042
for the dimerization reaction. The dimerization constant may be related to the dimer and monomer absorbances by the expression
Table 11: Association Constants and Henry's Law Constants for Trifluoroacetic Acid (TFA) in Several Solvents a t 25.0'
System
TFA-cyclohexane TFA-CCla TFA-benzene TFA-1,2-dichloroethane
K
Conoentration range, M
K t , 1. mol-1
0.01-0.07 0,007-0.23 0.04-0.21 0.02-0.3s
192 + 36 149 f 19 2.6 & .9 1 . 5 i .4
K H ,Torr 1. mol-1 592 f 110 275 rt 35 32.5 i 3 . 5 23.8 f 4 . 2
(AD/Ax*) ( E M ' / E D ) ~
(6) where E M and ED are absorptivities of monomer and dimer, respectively, and b is the cell length. Assuming that the ratio E A ~ ~ / E Dis constant with changing temperature, a plot of log A D / A R Ius. ~ 1 / T should yield a straight line with slope -AH/2.303R. Figure 6 shows plots of this type for the three solvent systems; values of AH obtained are -11.7 0.6 kcal/mol in cyclohexane, -9.2 f 0.4 kcal/mol in CC14, and - 7.0 zt 0.8 kcal/mol in dichloroethane. The AH value for CC14 agrees iTell with that determined from the dilute solution measurements (vide supra) and Tvith that reported by Kagarise., Summary of Thermodynamic Results. The spectral results described above provide the basis for deriving thermodynamic constants for the individual steps in the cycles
*
partition data. The calculated Kz would represent the sum of the association constants for the two types of dimers, whereas conventional dilute solution studies will yield correct values for Kz only if the monomer and dimer hydroxyl-stretching bands do not overlap significantly. (4) Condensed-Phase Studies of the Carbonyl-Stretching Region. Spectra in the 1800-cm-' region were obtained for solutions of TFA in cyclohexane, CCL, and 1,Zdichloroethane, using a Beckman VLT-2 thermostated cell with AgCl windows. Path lengths used were 0.2 mm for dichloroethane and cyclohexane and 1.0 mm for CC1,. A typical spectral curve is shown in Figure 5 , where absorbance is plotted against frequency for a solution of TFA in dichloroethane at 32.8". The peak at 1805 cm-1 is attributed t o monomer and that a t 1789 cm-l is ascribed to the dimer. The individual absorbance curves for the monomer and dimer have been inferred using a nonparametric graphical method developed by Stevens for resolving overlapping-band ~pectra.~bSpectral features of the monomer and dimer bands agree generally mith those reported by Kirszenbaum, et u Z . , ~ although the dimer frequency in 1,3dichloroethane was observed at 1789 cm-I, as compared with the value 1782 cm-l given earlier. However, there is more overlap between the monomer and dimer bands in dichloroethane than in the other two solvents, and therefore the dimer band maximum is more difficult to locate. Measurements of the temperature dependence of the dimer and monomer absorbance at the peak maxima of these species may be used to infer the enthalpy change
it
2TFA in solvent
Discussion Qualitatively, the thermodynamic results in Table I11 show the important influence of the solvent on the energetics of the monomer and the dimer of trifluoro-
dichloroethane
3.0
T
Figure 5 . Spectra of the trifluoroacetic acid monomer and dimer in the carbonyl-stretching region. The Journal of Phgisical Chemistrg, Vol. 76, No. 14, 1972
(TFA), in solvent
for each of the four solvents investigated. Table I11 lists the values of Gibbs free energy (AG) and internal energy ( A E ) changes for the solvation and reaction steps in these cycles.
I
31-
__ i f
2TFA in vapor I_ (TFA)2in vapor
+
OK-'
3.5
Figure 6. Absorbance ratio measurements for trifluoroacetic acid in several solvents as a function of temperature.
ASSOCIATION OF TRIFLUOROACETIC ACID
2043
Table I11 : Energy and Free Energy Changes for Association and Solvation of Trifluoroacetic Acid Reaction
Medium
2TFA = (TFA)z
TFA(vapor) = TFA(medium)
-AGO, koala
- - E o , koala
Vapor 5.12 f 0.04 (13.4)b Cyclohexane 3 . 1 1 & 0.12 11.7 4= 0 . 6 CC14 2.97 & 0.08 9 . 2 =f= 0 . 4 (2.88 f 0.03)c (9.0 zk 0.4)c Benzene 0.57 & 0.18 (7.4)d 7 . 0 3.8 1,2-Dichloro- 0.24 f 0.14 ethane Cyclohexane 2.04 f 0.11 2 . 3 I: 0.4
represent internal energy and Gibbs free energy changes for transferring the complex from the ideal vapor phase a t unit molarity into the unit molarity ideal dilute solution state in solvent f i j and ZAEOreactants S
'
(TFA)z(vapor) = (TFA)z(medium)
3 . 4 I: 0 . 6 5 . 2 j, 0 . 6 5 . 3 =k 0 . 6
2.07 f 0.23
2.9 =t 1 . 0
2.83 f 0.18 CC14 Benzene 2.97 =t0.16 1,2-Dichloro- 3.00 f 0.27 ethane
2 . 6 =t 1 . 4 4 . 4 =t 1 . 3 4 . 2 Z!Z 1 . 5
a Standard states are unit molarity, ideal dilute solution states for components in all phases. Internal energies are assumed to be equal to enthalpies for species in condensed phases and to differ from enthalpies by -RT in the gas phase. b C. Ling, S. D. Christian, H. E. Affsprung, and R. W. Gray, J . Chem. SOC.A , 293 (1966). Dilute-solution studies, this work. d Assumed to be the same as A f l for the reaction 2TFA = (TFA)z in diphenylmethane, reported by W. s. Higazy and A. A. Taha, J . Phys. Chem., 74, 1982 (1970).
ZAGOreactants SY '
are the sums of the corresponding energy and free energy changes for the monomers which unite to form the complex. In the case of the TFA dimerization reaction ZAE'reactants
2.51 f 0.07 3.76 & 0.07 3.94 f 0.12
CCla Benzene 1,2-Dichloroethane Cyclohexane
and
V
=
~AE'TFA and SY '
S V '
2AG"reactants
=
~AG'TFA V 8'
8V '
It has been noted that a and a' for a given electron donor-acceptor association reaction (of either the charge-transfer or hydrogen-bonded type) are generally nearly equal to each other and nearly invariant throughout a range of different media. (In several previous communications from this laboratory,ll the constant a has been used to represent both the energy and the free energy ratio, although the ratios Q and a' are not necessarily equivalent. It is probably desirable t o use different symbols to designate these two ratios, while recognizing that they are ordinarily nearly equal.) In terms of the parameters a and a' one may write the thermodynamic identities AGO, = AGO,
+ (1 -
a')ZAG'rnonorners
(7)
-7'8
and acetic acid. Even the nearly inert solvents considerably reduce the energy and free energy of the individual species. That the monomer is much more strongly solvated than the dimer (per mole of CF3COOH) is indicated by the marked decrease in the magnitude of the energy of the dimerization reaction in the condensed phases. Changes observed in AGO and AE" (and of course A x o ) for the reaction 2TFA S (TFA), parallel those noted by Allen, et al.,1° for the dimerization of benzoic acid in several media. The influence of solvents on the energies and free energies of formation of molecular complexes may be rationalized in terms of a solvation model developed in this laboratory.*~l' Application of the model requires introduction of the dimensionless parameters AE'aornplex YS '
a=
A E o V = AEOs
+ (1
- a)ZAE'rnonorners
(8)
SV '
where AGO, and AGO, are the standard free energy changes for the association reaction in the vapor and solvent phases, respectively, and AEoY and AE", are the corresponding internal energy changes in the two phases; unit molarity ideal dilute solution states are employed throughout. Equations 7 and 8 imply that if a and a' are nearly equal and constant for the TFA dimerization reaction in the media investigated here, plots of - AGO, us.
- ~AG'TFAand - AE", Y' 8
us.
-2 A E 0 T ~ A S Y '
should be linear and have equal Figure 7 shows data from Table I11 plotted in these forms; the two solid lines are both drawn with slopes equal to -0.56, corresponding t o a = a' = 0.44. Thus, the
ZAEOreactants BV '
and
(10) G. Allen, J. G. Watkinson, and K. H. Webb, Spectrochim. Acta, (1966). (11) (a) S. D. Christian, J. R. Johnson, H. E. Affsprung, and P. J. Kilpatrick, J . Phys. Chem., 70, 3376 (1966); (b) J. Grundnes and S. D. Christian, J . Amer. Chem. Soc., 90, 2239 (1968); (c) 5 . D. Christian and J. Grundnes, Acta Chem. Scund., 2 2 , 1702 (1968); (d) J. R. Johnson, P. J. Kilpatrick, S. D. Christian, and H. E. Affsprung, J . Phys. Chem., 72, 3223 (1968); (e) S. D. Christian, J . Amer. Chem. Soc., 91, 6514 (1969); (f) S. D. Christian, Office of Saline Water Research and Development Progress Report No. 706, July 1971. 22, 807
AGOcornBlex
a' =
"'8
ZAGoreactants SV '
where AEOeornplex VS '
and
AGooornplex V S'
The Journal of Physical Chemistry, Vol. 76, N o . 14, 1978
SHERRIL D. CHRISTIAN AND THOMAS L. STEVENS
2044
o
z
L
a
6
IO
12
KCAL
Figure 7 . Dependence of the energy and free energy of dimerization of trifluoroacetic acid on transfer energies and free energies of the monomer; solvents cyclohexane (C), carbon tetrachloride (Ca), benzene (B), 1,Zdichloroethane ( D ) .
internal energy arid the free energy results conform reasonably well to a solvation model in which the dimer is somewhat less than half as effectively solvated as two separated monomer molecules. Stevens has developed a quasi-lattice model for predicting a from group interaction energy parameters, tabulated for specific atomic and submoleeular groups of molecules participating in complex formation reaction^.^^,^^' Assuming that (TFA), is cyclic, he estimates values of a equal to 0.44, 0.59, 0.56, and 0.42, respectively, for the solvents cyclohexane, cc14, benzene, and 1,2-dichloroethane. The experimental value of a (0.44) is one of the lowest which has been reported for complex formation reactions. This in itself suggests that the dimer is predominantly cyclic, and the reasonable agreement with values of a estimated from the lattice calculations strengthens the argument that the dimer is cyclic. The spectral results from this laboratory and those of
The Journal of"Physical Chemistry, Vol. 76, N o . 14, 1978
Kirszenbaum, et ~ l .do , ~not indicate that more than one self-associated form of TFA exists in the solvents investigated. Gradual changes do occur in values of the frequencies and half-widths of the dimer spectral bands and of the thermodynamic properties of the dimer as the medium is varied from vapor to nonpolar solvents to slightly polar solvents. However, these effects are of the sort to be expected even if only a single monomer-dimer association reaction is occurand there seems to be no reason to assume the presence of an open dimer. The spectral-partition results for TFA in benzene (see Figure 4) suggest that there may be partial polymerization to species larger than the dimer at TFA concentrations greater than 0.2 M . However, in CC1, at concentrations up to 0.2 M we can find no thermodynamic or spectral evidence for higher polymerization. Moreover, the value obtained for K z in CC14, using the conventional spectral method in the very dilute region ( f A < 3.5 X lop3 M ) , agrees well with that obtained with the spectral-partition method at much higher concentrations. This is also an indication that higher polymers and large activity coefficient effects are not important. The present results therefore do not support the observation by Murty and Pitzer' that there is a pronounced dilution effect on spectra in the carbonyl stretching region for TFA in CCl, in the concentration range 1.3 X 10-+3.3 X 10-2 M . Possibly the reported dilution effect may be attributed to the presence of small concentrations of water in the TFA solutions. Water is known to interact strongly with TFA to form hydrogen-bonded complexes in both vapor and solvents, and it is exceedingly difficult to remove the last traces of water from liquid TFA or its solutions in organic solvents.
Acknowledgment. This work was supported by Sational Science Foundation Grant No. GP-23278.