J . Chem. SOC., Faraday Trans. I , 1988, 84(2), 561-574 Solvent Properties of Polyaromatic Hydrocarbons Grazyna Geblewicz Chemistry Department, University of Southampton, Southampton SO9 5NH David J. Schiffrin" Wolfson Centre for Electrochemical Science, University of Southampton, Southampton SO9 5NH The solubility of inorganic and organic salts at 150 "C in polyaromatic hydrocarbons has been studied, and the results analysed using an electrostatic model. It is shown that only significant solubilities can be obtained with quaternary ammonium salts with C, or longer carbon chains. The n* solvatochromic parameter has been calculated for the different solvents from the absorption spectra of o-nitroaniline. It has been found that the Hildebrand solubility parameter and the work of cavity formation calculated from the scaled particle theory cannot account for the large solubility differences observed in the various solvents studied. It is proposed that the Gibbs energy of solvation is determined by the local polarisability, as shown by its linear dependence on n*.It is shown that these results rationalise previous observations on the solubilisation of transition-metal ions in aromatic solvents using hydrophobic ligands. The use of organic solvents in electrochemistry has been the subject of extensive research and there has been a particular interest in finding systems suitable for electrodeposition processe~.~-~ Most of the solvents which give good solubility and conductivity of electrolytes are themselves strong ligands and can be easily incorporated in the coordination sphere of transition-metal ions, which displaces the metal reduction potentials to values where the decomposition of the solvent interferes with elec- trodeposition.It is clear, therefore, that in the choice of aprotic solvents for metal electrodeposition, there are two conflicting requirements. A high dielectric constant appears to be necessary for achieving a significant solubility and conductivity of solutions of electrolytes, but this limits the choice to solvents with a high donor number, which are precisely those that produce significant redox shifts by coordination to the transition-metal ions. One approach that can avoid this problem is to ensure that the solvent is not involved in the coordination environment around the ions by the use of solvents of very low donor number, for example, the aromatic hydrocarbons, in which solubilisation of the transition-metal ions can be achieved by the use of hydrophobic ligands, such as the aliphatic amines.* In spite of their low dielectric constants, aromatic hydrocarbons can give conducting solutions with organic electrolytes at sufficiently high temperatures, as shown recently by Campbell et aL5 For instance naphthalene, 1 -methylnaphthalene and 1 -chloronaphthalene give conducting solutions at 150 "C with quaternary ammonium salts, enabling simple electrochemical processes to be carried out.The use of some aromatic hydrocarbons for electroplating has been described in the 1iteratu1-e.~. For example, toluene has been used for electroplating aluminium on steel in a bath consisting of the complex formed between triethylaluminium and sodium fluoride and operating at 80-95 0C.6 The advantages of the aromatic solvents include their high chemical stability, the very large potential polarisation window available, their high boiling point and low cost.The 56 1562 Solvent Properties of Polyaromatic Hydrocarbons useful potential window of an aromatic solvent will be determined to a first approximation, by the difference between the energy of the lowest unoccupied and highest occupied electronic energy levels. These differences are very large and in many cases of the order of 4 eV. In agreement with this, for instance, the difference between the oxidation and reduction potentials of naphthalene and biphenyl are 3.91 and 4.28 V, respectively.' Furthermore, the very low donor properties of these solvents make them ideal as media for reactions where the control of the coordination chemistry of the metal ion without interference by the solvent is required. The purpose of the present work was to understand the conditions required to obtain good solubilities and conductivities of salts in the aromatic solvents. The solubilities measured have been analysed using electrostatic solvation models and in terms of short-range solute-solvent interactions.The nature of these interactions has been further studied by measuring the solvatochromic shifts of indicator solutes and establishing relationships between the Gibbs energy of solvation and the polarisability scales.Experimental The solvents studied and their liquid range are shown in table 1. The thermal stability and very high boiling points of some of these compounds make them suitable candidates to bridge the temperature range between non-aqueous and molten salt electrochemistry. 1 -Chloronaphthalene (Aldrich) was distilled under vacuum and then dried over CaH,. 1 -Methylnaphthalene (Lancaster Synthesis Ltd) was dried, distilled under vacuum and stored over CaH,. Fluoranthene (Aldrich) was twice recrystallised from acetone ; fluorene (BDH) was twice recrystallised from CCl, ; naphthalene (BDH) was three times recrystallised from ethanol ; m-terphenyl (Aldrich) was recrystallised from ethanol several times ; biphenyl (BDH) was twice recrystallised from ethanol. KC1 and KBr (BDH, spectroscopic grade) were recrystallised from water and dried for several hours under vacuum at 400 "C.Tetrabutylammonium tetrafluoroborate (TBABF,) (Aldrich) was recrystallised from water ; tetramethylammonium bromide (TMABr, Fluka), tetraethylammonium bromide (TEABr, Fluka), tetrapropylam- monium bromide (TPABr, Fluka) and tetrabutylammonium bromide (TBABr, Aldrich) were dissolved in hot ethanol and precipitated with ether which had been previously stored over CaH,. All these chemicals were dried at 80 "C under vacuum for several hours prior to use. Solubility measurements were made in a glass cell thermostatted in an oil bath. The mixture of molten solvent and excess salt was stirred in the cell with a PTFE-covered stirrer. The cell lid had ground joints for introducing nitrogen, for measuring the temperature and for sample removal.A dried nitrogen atmosphere was kept over the solutions throughout the experiments. After the addition of an excess quantity of the salt under study to the molten solvent, the mixture was heated and stirred for 8 - 10 h. After this, a sample was taken out by means of a heated pippete and put into a heated cell containing a fine frit. After filtering, the solution was placed in several previously weighed volumetric flasks. After cooling, the flasks were weighed again and the contents were dissolved in toluene, then mixed with triply distilled water and shaken for several hours using a mechanical shaker. The aqueous and organic phases were separated and the extraction procedure was repeated twice.The aqueous solutions were analysed using atomic adsorption spectrophotometry for potassium ions or by potentiometric titration with silver nitrate for the bromides. The conductivity of the solutions was measured using an AC Wienn bridge constructed from Sullivan components (0.05 YO accuracy). The conductometric cell had a constant of 0.727 cm-l. The measurements were carried out in the frequency range 0.5-5 kHz and in this range no significant frequency dispersion of the measured conductivity was observed.G . Geblewicz and D. J . Schiflrin 563 Table 1. Liquid range of the polyaromatic hydrocarbons studied boiling melting point/"C point/"C naphthalene 218 80.55 1 -methylnaphthalene 244.64 - 30.5 1 -chloronaphthalene 258.8" - 2.3 biphenyl 255.9 71 fluorene 293.5 116.7 fluoranthene 375 111 rn-terphenyl 365 89 At 753 mm.The high-temperature spectrophotometric measurements were carried out with a modified Pye-Unicam SP 700 spectrophotometer, and for the measurements at room temperature, a Pye-Unicam 8800 spectrophotometer was employed. o-Nitroaniline and m-nitroaniline were used as the indicator solutes to measure the shift of the absorption maximum caused by the solvent.8 Results and Discussion Solubility and Conductivity Results The solubility of KCl and KBr was studied at 150 "C. The values obtained were very low, of the order of 0.01 mmol dmV3. The results are given in table 2. An increase of temperature did not significantly affect the solubilities of inorganic salts and fig. 1 shows the temperature dependence of the solubility of KBr in naphthalene in the temperature range 100-200 "C.None of the solutions of inorganic salts showed any significant electrical conductivity. Organic electrolytes, such as tetramethyl- and tetraethyl-ammonium bromides had solubilities in the range 10-4-10-3 mol dm-3 as shown in table 3, but the conductance of these solutions at 150 "C was also very low, being less than 0.1 R-l cm-' for all of the solvents studied. The solubilities of all the organic electrolytes are compared in table 3. TBABr and TBABF, could be mixed with the solvents in all proportions. Significant conductivities were obtained for TPABr and even higher values were observed for TBABF, solutions. Fig. 2 shows the concentration dependence of the equivalent conductance of these salts in 1-chloronaphthalene at 150 "C.Analysis of Solubilities Since the solutions of salts of small ions like KCl, KBr, TMABr and TEABr were not conducting, it was reasonable to assume that extensive ion-pairing (or even the formation of higher neutral aggregates) occurred. In order to rationalise the different solubilities, a simple electrostatic association model has been used. As a first approximation, it has been considered that the solutions are totally ion-paired. This idea was verified by calculating the distances between ions forming the contact ion pairs using the Born cycle for the dissociation scheme of an ion pair, as already discussed by Denison and Ram~ey,~ and the results compared with the known ionic radii. The cycle employed is shown in fig.3. M i and X i represent the free solvated ions, (M+X-), the ion pair in the solvent, M,+ and Xi the free ions in vacuum. AGZOlv is the564 Solvent Properties of Polyaromatic Hydrocarbons Table 2. Solubilities (units loA5 mol dm-3) of inorganic salts at 150 "C ~ solvent KC1 KBr naphthalene 5.1 10.8 1 -met hy hap h t halene 3.5 2.3 1 -chloronaphthalene 7.3 3.2 biphenyl 4.2 2.5 fluorene 2.4 7.0 fluoran thene 2.6 5.2 rn-terphenyl 5.4 3.5 4 .: 4. w 0" 4 . ( - 3 .' 3.1 I 1 1 I I I 1 _I 2.2 2.3 2.4 2.5 2.6 2.7 2.8 103 KIT Fig. 1. Temperature dependence of the solubility of KBr in naphthalene. Table 3. Solubilities (mol dm-3) of organic salts at 150 "C solvent TMAB TEABr TPABr naphthalene 5.7 x 10-5 2.8 x 10-3 1.907 1 -methylnaphthalene 7.1 x 10-5 3.5 x 10-3 0.127 biphenyl 3.7 x 10-5 9.9 x 10-4 0.011 fluorene 4.3 x 10-5 4.2 x 10-3 0.047 1 -chloronaphthalene 1.6 x 6.3 x 0.89 fluoran thene 1.8 x 7.8 x 1.89 rn-terphenyl 1.82 x 4.5 x 0.295G.Geblewicz and D. J . Schifrin 0.5 0.0 < 8 d -2.0 -2.5 565 - - rd r' - - t - 1.3 1.0 0.5 0.0 log c Fig. 2. Dependence of the equivalent conductance on the concentration of solutions of tetrapropylammonium bromide (A) and tetrabutylammonium tetrafluoroborate (a) in 1-chloronaphthalene at 150 "C. A ~ o S , i p Fig. 3. Born cycle representing the solvation of an ion pair in the solvents (AG,,,) followed by dissociation into the free ions in the solvent (AGionis). AGsolv is the Gibbs energy of solvation of the free ions. Gibbs energy of solvation of the ions from the gas phase, which was calculated from Abraham's solvation mode1.l0' l1 AGYonis is the Gibbs energy of ionisation of the ion pair in the solvent calculated assuming as a first approximation a uniform dielectric constant.(1) AGYOnis is given by: where 2 is the charge of ions, e is the charge of the electron, N is Avogadro's number, AG&is = Z2e2N[4z(r1 + r,) EE,]-'566 Solvent Properties of Polyaromatic Hydrocarbons E is the dielectric constant, E, is the permittivity of free space, and rl + r2 is the interionic distance in the ion pair. In these calculations, the dielectric constants and molar volumes were calculated at 150 "C from literature data.12* l5 AGZ,ip is the Gibbs energy of formation of an ion pair in the solvent from vacuum, which is given by This Gibbs energy can be calculated also from (2) AG,",,, = AG,O,,, - AG:onis.AG;,,, = AG: - AG;. (3) AGZ is the Gibbs energy of solution of the ion pair MX from vacuum, and AG; is the Gibbs energy of sublimation from the solid lattice to free ions in vacuum. From the above equations the distance of the ions in the ion pair can be calculated from (rl + r2) = (Z2e2N/4z~~o)[AG~o,v - AG," + AGZ1-l. (4) The Gibbs energies of solution, AG,", were calculated from the solubility data, assuming that the activity coefficient of the ion pairs was equal to unity, i.e. ( 5 ) where Csat is the molar concentration of the saturated solution. The Gibbs energy of the solid salt MX was calculated from the corresponding enthalpy (AH",) and entropy (A$) of lattice formation from the free ions in vacuum :"* l7 (6) where U,, is the lattice energy.Although Uo is calculated at 0 K, this value changes very little with temperature,'* and lattice energies were used directly in the following calculations. For KCl and KBr, A% was calculated using the Sackur-Tetrode equation for the gaseous ions; the entropies of the solids and other relevant data were taken from the literature." The values of the lattice energies for KCl, KBr, TMABr and TEABr used were 704,679, 549 and 514 kJ mol-l, respectively."* 18* 2o The entropy calculations for the quaternary ammonium ions is complicated by possible rotational and vibrational contributions. The entropy data at 25 "C given by Johnson21 were used, and these values were corrected to 150 "C using the heat capacity data for the corresponding gaseous alkanes given by Benson.22 The entropy of solid TMABr has been measured at different temperatures by Chang and W e s t r ~ m ~ ~ up to 350 K; no data are available at higher temperatures.However, since the heat capacity is almost linearly dependent on temperature up to 350 K (the linear regression coefficient in the range 200-350 K was 0.9999), it was assumed that this behaviour was valid up to 423 K and was used in the calculation of the entropy at this temperature. No entropy data are available for TEABr; therefore the approximation proposed by Johnson'' was employed. This makes use of the additivity rules for the entropy of the individual elements for analogous compounds. The entropy data for TEA1 were AG: = - R T In Csat AGZ = AH", - APLT = Uo + 2RT- TAPL in this calculation, i.e.(7) where S,' and SBr' are the individual entropic contributions of these elements. These values were calculated at 423 K from Latimer's and SOTEAI was calculated from ref. (23) using the same approach as described above for correcting the entropy data to 423 K. A summary of the values used in the calculation is given in tables 4 and 5 . The calculated interionic distances of KCl, KBr, TMABr and TEABr ion pairs in several solvents obtained from eqn (4) are compared with the corresponding crystal radii in table 6. The results obtained for KC1 and KBr are reasonable for contact ion pairs.G. Geblewicz and D. J. Schifrin 567 Table 4. Thermodynamic data used in the calculations of the Gibbs energy of lattice formation" s:/ AGOL/ kJ mo1-' s:/ solute J mol-' deg-' J mol-' deg-' ~~~ ~ KCl 263.3 100.0 639.0 KBr 251.2 1 14.4 620.1 TMABr 528.2 265.5 444.9 TEABr 71 1.8 406.3 391.8 " Si and Sz are the standard entropies of the solute in the gas and in the solid phase, respectively.Table 5. Gibbs energies of solution (AG,"/kJ mol-l) and solvation (AG,",,,/kJ mol-') of the solutes in different solvents (1 50 "C) naphthalene 1 -methylnaphthalene 1 -chloronaphthalene solute AG: AG,",,, AG,O AG:ol" AG: W O I " KCl 30.9 -425.5 36.1 -435.4 33.5 -471.8 KBr 32.1 -411.8 37.5 -421.4 36.4 - 457.1 TMABr 34.3 -311.1 33.6 -317.6 30.7 - 349.5 TEABr 20.7 -295.1 19.9 -301.2 17.8 -331.9 Table 6. Comparison of the values of calculated ionic radii of ion pairs in different solvents with crystallographic ionic radii (in nm) l-chloro- 1 -methyl- crystal naphthalene naphthalene naphthalene solute radii" ( E = 3.7) ( E = 2.50) (E = 2.61) KCI 0.318 0.280 0.331 0.290 KBr 0.334 0.296 0.344 0.302 TMABr 0.454 0.584 0.595 0.537 TEABr 0.506 0.890 0.784 0.699 a Crystal radii data from ref.(1 1) and (16). In the case of TMABr and TEABr the distance of closest approach is greater than the sum of the ionic radii, probably indicating a greater interaction with the solvent. From the temperature coefficient of the solubility [d In S/d(l/T)] shown in fig. 1, a value of the enthalpy of solution (AK) of -6.2 kJ mol-' was calculated for KBr. The small value of A% is a further indication of the extensive ion-pairing and perhaps of higher aggregation which occurs in these solutions, i.e.the electrostatic interactions in solution are similar to those on the lattice, and solvation makes a minor contribution to the energy of the aggregates. It can be concluded also that one of the main factors determining the solubility is the lattice energy of the salt, which does not depend very much on temperature. Salts formed from small ions like KC1, KBr, TMABr and TEABr dissolve in polyaromatic hydrocarbons in the form of neutral ion aggregates and the very low conductivities of the solutions confirm the strong ion-ion interactions in these cases.568 Solvent Properties of Polyaromat ic Hydrocarbons Analysis of the Conductivity Data In general the dependence of the equivalent conductance on concentration of a salt in a solvent of low dielectric constant presents a minimum followed by an inflection point on the conductivity curve at higher concentrations caused by the formation of quadrupoles.25-29 The results presented in fig. 2 are indicative of the presence of triple ions and probably of higher aggregate^.^, The TBABF, data were analysed using the Fuoss-Kraus theory,28 from which the equivalent conductance is given by where g(c) is a factor taking into account interionic interaction terms, A, and A: are the limiting equivalent conductances of the fully dissociated salt and the triple ions TBA(BF,), and (TBA),BF,+, Kp and kT are the ion pair and triple ion formation constants, respectively. In this theory, the two possible triple ions are treated as species with identical properties. g(c) was found to be very close to unity, mainly due to the very low permittivity of the medium ( E = 3.7).This term departs from unity only for much more polar solvent^.^^^ 31 No conductivity data are available for the polyaromatic solvents, and the values of A. had to be estimated. Owing to the low conductivity of the solutions studied, only an approximate value of A, needs to be used in eqn (8). The contribution to A. by the cation can be estimated quite accurately from the Walden product, which is 240 and 210 C2-l cm2 equiv-l Pa s for TPA+ and TBA+, respectively, for a wide range of For the inorganic ions, the value of the Walden product is more dependent on the nature of the solvent, and unfortunately no reliable data appear to be available for BF,. For this ion, the data for I- were used in view of the similarity of the ionic radii of BF, and I-.For Br- and I-, the Walden product was estimated by averaging data for solvents of similar viscosity as 1-chloronaphthalene at 150 "C (8.4 Pa g) and a value of 360 was This is a very approximate procedure, but as discussed before, large errors in A, do not affect significantly the calculations based on eqn (8). The values of A. were 71 and 68 C2-l cm2 equiv-l for TPABr and TBABF,, respectively. Fig. 4 shows the test of the Fuoss theory for TBABF,; the correlation coefficient was 0.985. Taking A:= values of Kp = 6.3 x lo4 dm3 mol-1 and k, = 30 dm3 mol-1 were obtained. These constants are similar to those obtained for other low dielectric constant systems.28 It can be concluded that in the case of TBABF, solutions, the major proportion of the electrolyte exists as ion pairs with only a minor proportion present as triple ions, and the formation of quadrupoles is less likely owing to the larger size of the ions.In the case of the TPABr solutions, eqn (8) was not applicable. The reason for this is not clear at present, but it should be noticed that the conductance is very low and it is very likely that higher aggregates are formed. A problem with this system is that the conductivity at concentrations lower than 0.05 mol dm-3 was very difficult to measure with the bridge employed (< lo-' R-l cm-l), and therefore the concentration region of the minimum that would be expected to be observed with this electrolyte in very dilute solutions, could not be studied. The influence of the ionic size can be clearly seen from the results in fig. 2; when the size of the ion is increased by changing the hydrocarbon chain from C , to C,, and Br- for BF,, the conductivity is increased by two orders of magnitude.When comparing the conductivities and solubilities of all the alkylammonium salts, it can be seen that an increase in the length of the carbon chain results both in smaller association constants due to purely electrostatic effects, and also in larger interactions of the salt with the solvent. These two effects give rise to significant solubilities andG. Geblewicz and D. J. Schiffrin 569 Fig. 4. Fuoss-Kraus plot for TBABF, in 1-chloronaphthalene at 150 "C. conductivities. Furthermore, an increase in the ionic radii decreases the lattice energy of the salt and these two factors determine a critical condition for total miscibility between the salt and the solvent.This appears to occur for a C , chain in the quaternary ammonium cation. An important conclusion of this analysis is that in order to dissolve salts of small ions, e.g. transition-metal ions, the usually very high lattice energy of the solid can be overcome by coordination of the cation with hydrophobic ligands that display sufficiently strong dispersion interactions with the polyaromatic solvents. Therefore, the same arguments discussed before for increasing the solubility of inorganic and quaternary ammonium salts in aromatic solvents can be applied to the dissolution of transition-metal ions in solvents of low dielectric constant. These principles have been used already for the electrodeposition of nickel and cobalt from molten naphthalene.' The value of K p for TBABF, obtained corresponds to a value of the parameter b in the Fuoss-Kraus theory of 10.However, the triple ion constant k, calculated with b = 10 is 140 dm3 mol-l, nearly five times greater than the experimental result. The discrepancy between theory and results can be due to the neglect of quadrupole formation. For instance, using b = 10, a quadrupole asociation constant of 3.1 dm3 mol-1 can be ca1c~lated.l~. 34 Unfortunately, the amount of experimental data did not justify a three-parameter fit, and for the same reason, attempts to calculate quadrupole constants for TPABr were unreliable.More work is required to understand the behaviour of these strongly associated solutions. Characterisation of the Solvation Ability of the Polyaromatic Solvents It is surprising that very large differences in solubilities were observed for the different solvents studied (table 3), if it is considered that the polyaromatic solvents all have very low and similar dielectric constants. Also, many of their other properties, such as surface tension, solubility parameters etc., are very similar. The energy of solution of a solute570 Solvent Properties of Polyaromatic Hydrocarbons can be considered as resulting from the creation of a suitably sized cavity in the bulk solvent, the reorganisation energy of the solvent round the cavity and the interaction energy of the solute with the reorganised and therefore solubilities will be affected by solvent-solvent and solute-solvent interactions. The solvent-solvent interactions which determine the cavity term can be taken into account by using either the Hildebrand solubility parameter, 8H,38i 39 which represents the energy of vaporisation of the solvent per unit molar volume, or calculated from theories of liquids.40 In the case of the solubility parameter, contributions from hydrogen bonding, 6,, polar, d9, and dispersion interactions, 6d, are usually c~nsidered.~~ However, for the polyaromatic solvents (probably with the exception of 1-chloronaphthalene), the hydrogen bonding and polar interactions are expected to be relatively small.compared with the dispersion interactions, and therefore SH was calculated taking into account only the later contributions.The solubility parameter was calculated from39 37 The solvent reorganisation energy is expected to be relatively 6, = - 4 . 5 8 + 1 0 8 ~ - 1 1 9 ~ ~ + 4 5 ~ ~ (9) where x = (n2- l)/(nz + 2 ) and n is the refractive index (6, units in MPa'j2). The solubility parameters for the different solvents were compared with the energy required to create a cavity in the solvent calculated using the scaled particle theory of fluids, in which the reversible work required to form a cavity of radius r in a fluid, We, is given hv40 " J -=In(l-Y)+ wc kT where Y = 7&/6 is the reduced number density, p = N / V is the number density, V is the molar volume, a, and az are the hard-sphere diameters of the solvent and solute molecules, respectively, such that the cavity radius is (a, + a,)/2, R = a,/a,.The effective hard-sphere diameter of the solvent can be calculated from the solubility of gases4, However, in the absence of solubility data for the solvents studied, the relationship between the van der Waals volumes and a, observed by de Ligny and van der Veen4' was used. These authors found that for a large number of solvents, the relationship between a, and Vw was (1 1) where Vw is the van der Waals volume in cm3 mo1-l. Eqn (1 1) has been found to be applicable to a wide range of compounds with geometries far removed from spherical and for this reason, it has been used in the present work. Values of Vw were calculated from the corresponding group contributions given by B ~ n d i .~ ~ Since a, plays a central role in the rigid-sphere fluid model, it was desirable to ascertain the suitability of the approach of de Ligny and van der Veen by calculating these values using another method. This was done using the relationship between surface tension and o1 for a hard- sphere liquid discussed by Mayer :44 ~ X N ~ T ; = - 10+ 1.13 Vw o,RT(2+ Y) ' = 4V(1- P) where y is the surface tension. This value was calculated at 150 "C from literature data.12-15 The two sets of values are shown in table 7 and, as can be seen, there is a nearly constant difference of 0.5 nm between the two methods of calculating a,. This difference is not surprising considering that the group additivity rules given by Bondi had been calculated at 25 "C.In spite of this difficulty, it can be said that the ordering of the values obtained for the cavity term for the different solvents is correct. Values of WJRT are also given in table 7. The similarity in the values of 6, for the different solvents indicates that the energy of formation of a cavity cannot account for the different solubilities observed.G. Geblewicz and D. J. Schiffrin 57 1 Table 7. Hard-sphere diameters, work of cavity formation, solubility parameters and solvatochromic parameters for polyaromatic solvents at 150 "C solvent ui/nm ui/nm W C W 4I/ (from V,) (from 12) (from V,) MPak Z* naphthalene 0.617 0.566 9.7 19.9 0.69 1 -chloronaphthalene 0.642 0.597 10.6 20.8 0.89 1 -methylnaphthalene 0.647 0.596 10.1 19.9 0.82 biphenyl 0.662 0.621 9.7 19.9 0.54 fluorene 0.675 11.9 fiuoranthene 0.709 13.7 Furthermore, the differences observed in WJkTdo not account either for the differences in solubilities.For instance, the cavity term for fluoranthene is greater than that of biphenyl, but the solubility of TPABr is nearly 200 times greater in the former solvent. It must be concluded that the differences in solubilities are related to short-range solute-solvent interactions, where not only the average molecular polarizability is important, but also the degree of packing of the solvent round the ions is relevant. In order to test these ideas, it was desirable to have a probe for the local solvent polarizability, through which the solvation behaviour of the same family of solvents could be tested. The technique employed was based on the measurement of shifts in the absorption spectra of molecules which have significantly different dipole moments in the ground and excited states. These solvatochromic shifts have been extensively used in the past for the characterization of organic solvents.** 37 In the ground state, solvent molecules are best oriented to the indicator solute.In the excited state, the solute has a different dipole moment (usually different in magnitude but not in orientation), and during the electronic transition, solvent reorientation is ruled out by the Frank-Condon principle. To stabilize the new charge distribution in the solute, the redistribution of the charge in the environment is only possible through electronic reorganization of the solvent ;35 therefore solvatochromic shifts are very convenient for distinguishing between solvents of similar chemical structure.The shift in the adsorption maximum can be related to the local polarizability of the solvent, and hence to a variety of thermodynamic and kinetic propertie~.~~-~' This solvent polarity scale correlates solvatochromic shifts with n -+ n* and p -+ n* electronic spectral transitions and the derived solvatochromic parameter n* is defined by8T 35 where vo is the wavenumber of the absorption maximum for an indicator solute in cyclohexane for which n* is assumed to be zero (in this solvatochromic scale, n* is taken as one for dimethyl sulphoxide); v, is the absorption maximum in the solvent under study and s is a measure of the indicator solute sensitivity to interactions with the cybotactic environment.o-Nitroaniline (v, = 26550 cm-', s = 1536 cm-') was used as the indicator solute for the determination of n* (table 7) and the results for 1-chloronaphthalene were confirmed using rn-nitroaniline (v, = 28870 cm-', s = 1664 cm-'), in order to establish any possible indicator dependence of the results, The two indicators gave the same value of n* within experimental accuracy. Although the spectrophotometric measurements have been carried out at 150 "C, there is a problem in the definition of the solvatochromic scale from these data. The absorption maximum depends on the local polarizability, and therefore the spectra obtained should be temperature-dependent. This dependence was measured for572 Solvent Properties of Polyaromatic Hydrocarbons 40 r( I Z 30 E 24 O', u Q c, 20 0.4 0.5 0.6 0.7 0.8 0.9 n* Fig.5. Dependence of the Gibbs energy of solution of TMABr (m) and TEABr (0) on the solvatochromic parameter n*. naphthalene and 1-chloronaphthalene for which dv/dT are 3.0 and 1.5 cm-l K-l, respectively. No data are available for the temperature dependence of the absorption spectra of the indicator for the reference solvents used in the definition of the n* scale, and in consequence the results given in the present work are referred to a scale at 25 "C. This will not alter the conclusions regarding the comparison of the relative local polarizability of the different solvents. The relationship between the solvatochromic parameter and the local polarisability contribution to the solution energy is given by37 (14) AG," = (AG,") , + sn* + h6, where (AG,"), is a Gibbs energy of solution referred to a reference solvent and h is a constant.The calculated values of 6, and n* are compared in table 7; it is evident that solvent-solvent interactions have no significant effect on the solubilities and the differences observed result from solute-solvent interactions which are reflected in changes of the n* values, i.e. in the local polarizability of the medium. Fig. 5 shows the dependence of the free energy of solution of TMABr and TEABr calculated from the solubilities of these salts as a function of the solvatochromic parameter. The linear character of this dependency (except for fluorene) confirms that the main origin of the differences in the solubilities observed is the local polarizability of the solvent.Furthermore, the values of s calculated from the slopes dAG,/dn* are - 1.42 and - 1450 cm-l for TMABr and TEABr, respectively. These results are in excellent agreement with many other determinations of s for a wide range of solvents and obtained from the analysis of totally different properties.' This gives further support for the applicability of solvatochromic scales to the description of solution phenomena.35* 37G. Geblewicz and D. J. Schiflrin 573 In the previous discussion all the polyaromatic solvents in the molten state have been considered as homogeneous, isotropic liquids. For some of the solvents, this assumption may be quite incorrect. For example, from the measurements of the temperature coefficient of the viscosities it has been postulated that compounds like m-terphenyl associate in the melt into clusters or ‘cybotactic’ group^,^^^^^ i.e.into groups of molecules arranged side-by-side or end-to-end, giving rise to a microcrystalline structure in dynamic equilibrium with molecules in a random orientation. The high boiling point of fluorene compared with m-terphenyl, which has the same molecular weight, reflects the limited freedom of the fluorene molecules in the melt45 and indeed some of the hydrocarbons studied produce probably highly structured liquids. In consequence, the polarizability of the solvent may depend strongly on its orientation with respect to the solute. This high anisotropy in the polarizability of organic molecules may change the influence of the solvent on the electronic spectral transitions of the solute and could explain the departure of the measured value of n* for fluorene from the trend observed for the other solvents. Conclusions The polyaromatic compounds have been shown to be good solvent media at high temperatures, giving conducting solutions with hydrophobic electrolytes. The limiting requirement for solubilizing ions and obtaining conducting solutions in the case of the symmetric tetraalkylammonium salts is a C, chain. The use of C, compounds leads to total miscibility between the salt and the solvent.The conductivity measurements show the usually high degree of association of electrolyte solutions. The salts of small ions (KCl, KBr, TMABr, TEABr) are associated in neutral aggregates and their solutions are non-conducting. Increasing the size of the ions decreases the degree of association and the formation of charged aggregates occurs.This results in an increase of the equivalent conductance with concentration (fig. 2). Because there is no defined minimum on the curves but rather inflection points, it has been concluded that triple ions mainly contribute to the conductance, and quadrupole formation reduces the conductivity at high electrolyte concentration. The effect of the solvent polarizability has been found to determine the solubility of the salts in the various polyaromatic hydrocarbons studied. The Gibbs energy of solution shows a linear dependence on the z* solvatochromic parameter and does not depend on the energy of cavity formation.The most important conclusion from this work is that the solubility of organic electrolytes is mainly determined by the local polarizability of the solvent and conditions have been found to obtain good solubilities and conductivities in polyaromatic solvents. This opens the possibility of using these organic non-polar solvents as very satisfactory inert media for the electrodeposition of metals and for other electrochemical reactions due to their extremely low donor numbers and very large potential windows. The authors gratefully acknowledge the financial support and encouragement given by the Commission of the European Communities to this project, under contract RNF 024- UK of the Recycling of Non-ferrous Metals Programme. One of us (G.G.) thanks the University of Warsaw for granting leave of absence during this work.Part of the equipment used was purchased with funds from the University Grants Committee Special Award to the Electrochemistry Group at Southampton University, 1985. Helpful discussions and comments by Dr R. J. Potter are acknowledged.574 Solvent Properties of Polyaromatic Hydrocarbons References 1 A. Brenner, A h . Electrochem. Electrochem. Eng., ed. C. W. Tobias (Interscience, New York, 1967), 2 T. Takei, Electrochim. Acta, 1978, 23, 1321. 3 A. Reger, E. Peled and E. Gileadi, J. Phys. Chem., 1979, 83, 873. 4 G. Geblewicz, R. J. Potter and D. J. Schiffrin, Trans IMF, 1986, 64, 134. 5 R. H. Campbell, G. A. Heath, G. T. Hefter and R. C. S. McQueen, J. Chem. SOC., Chem.Commun., 6 R. Suckentrunk and H. Tippman, Galvanotechnik D. 7968 Saulgau, 1982, 273,2. 7 G. J. Hoijtink, Reel. Trav. Chim., 1958, 77, 555. 8 M. J. Kamlet, J. L. Abboud and R. W. Taft, J. Am. Chem. SOC., 1977,99, 6027. 9 J. T. Denison and J. B. Ramsey, J. Am. Chem. SOC., 1955, 77, 2615. vol. 5. 1983, 1123. 10 M. H. Abraham and J. Liszi, J. Chem. SOC., Faraday Trans. I , 1978, 74, 1604. 1 1 M. H. Abraham and J. Liszi, J. Inorg. Chem., 1981, 43, 143. 12 International Critical Tables of Numerical Data (McGraw-Hill, New York, 1929). 13 G. B. Arrowsmith, G. M. Jeffery and A. I. Voge, J. Chem. SOC., 1965, 2072. 14 R. N. Rampolla and C. P. Smyth, J. Am. Chem. SOC., 1958,80, 1057. 15 Beilsteins Hanbuch der Organischen Chemie, 1979, Band 513 s, 1327-2134. 16 D. A. Johnson, Some Thermodynamic Aspects of Inorganic Chemistry (Cambridge University Press, 17 A.B. D. Lever, S. M. Nelson and T. M. Shepherd, Inorg. Chem., 1965, 4, 810. 18 R. Boyd, J. Chem. Phys., 1969, 51, 1470. 19 JANAF Thermochemical Tables, ed. D. R. Stull and M. Prophel (NSRDS-NBS37 Natl. Bur. Stand. 20 C. De Visser and G. Somsen, Reel. Trav. Chim. Pays-Bas, 1971,90, 1129. 21 D. A. Johnson and J. F. Martin, J. Chem. SOC., Dalton. Trans., 1973, 1585. 22 S. W. Benson, F. R. Cruickshank, D. M. Golden, G. R. Haughen, H. E. ONeal, A. S. Rodgers, 23 S-S. Chang and E. Westrum, J. Chem. Phys., 1962, 36, 2420. 24 W. Latimer, J. Am. Chem. SOC., 1921, 43, 818. 25 R. Fuoss and C. Kraus, J. Am. Chem. SOC., 1933, 55, 2386. 26 R. Fuoss, Chem. Rev., 1935, 17, 27. 27 R. Fuoss and C. Kraus, J. Am. Chem. SOC., 1935,57, 1. 28 C. Kraus and R. Fuoss, J. Am. Chem. SOC., 1933, 55, 21. 29 C. Kraus and R. Vingee, J. Am. Chem. SOC., 1934, 56, 51 1. 30 H. E. Maaser, M. Delsignore, M. Newstein and S. Petrucci, J. Phys. Chem., 1984, 88, 5100. 31 M. Delsignore, H. E. Maaser and S. Petrucci, J. Phys. Chem., 1984, 88,2405. 32 B. Krumgalz, J. Chem. SOC., Faraday Trans. I , 1983, 79, 571. 33 S. Boileau and P. Hemery, Electrochim. Acta, 1976, 21, 647. 34 R. M. Fuoss and C. Kraus, J. Am. Chem. Soc.,. 1933,55, 3614. 35 M. H. Abraham, M. J. Kamlet and R. W. Taft, J. Chem. SOC., Perkin Trans. 2, 1982, 923. 36 M. J. Kamlet, T. N. Hall, J. Boykin and R. W. Taft, J. Org. Chem., 1979, 44, 2600. 37 M. J. Kamlet, J. L. M. Abboud, M. H. Abraham and R. W. Taft, J. Org. Chem., 1983,48, 2877. 38 J. H. Hildebrand and J. H. Scott, The Solubility of Nonelectrolytes (Dover Publication Inc., New York, 39 A. F. M. Burton, Chem. Rev., 1975, 75, 731. 40 R. A. Pierotti, Chem. Rev., 1976,76, 717. 41 R. A. Pierotti, J. Phys. Chem., 1965, 69,281. 42 C. L. De Ligny and H. G. Van der Veen, Chem. Engl. Sci., 1972, 27, 391. 43 A. Bondi, J. Phys. Chem., 1964, 68,441. 44 S. W. Mayer, J. Phys. Chem., 1963, 67, 2160. 45 E. McLaughlin and A. R. Ubbelohde, Trans. Faraday SOC., 1957, 53, 628. 46 J. Andrews and A. R. Ubbelohde, Proc. R. SOC. London, Ser. A, 1955, 228,435. Cambridge, 1982). U.S.A., 2nd edn, 1971). R. Shaw and R. Walsh, Chem. Rev., 1969, 69, 279. 3rd edn, 1964.). Paper 71585 ; Received 1st April, 1987