摘要:

Heat of Dilution of Aqueous Solutions of Polyacrylates BY NONO ISE, U Z u E I MITA AND TSUNEO OKUBO Department of Polymer Chemistry, Kyoto University, Kyoto, Japan Received 10th July, 1972 The heats of dilution (AH) of aqueous solutions of polyacrylates having various gegenioas, i.e., sodium and tetra-alkylammonium (alkyl = methyl, ethyI, n-propyl and n-butyl), were measured. The (AH) value of the sodium polyacrylate was negative in the investigated concentration range (0.005-0.17 mol dm-3), whereas those of organic salts were positive. With the addition of urea, the endothermicity was lessened. These results are discussed in terms of hydrophobic interactions between solute and solvent. The observed (AH) values were found to be in agreement with the values cal- culated by the theory of Lifson and Katchalsky, when an appropriate value of d In a/d In Tis assumed (a is the ionic radius of rodlike macroion and T the temperature). However, the assumed value for this parameter was in disagreement with that obtained directly from the partial molal volume measurements (by a factor of -lo*).The heat of dilution is an important thermodynamic quantity of electrolyte solutions. For low molecular weight electrolytes, systematic study, experimental and theoretical? of this quantity has been conducted for dilute solutions. For high molecular weight electrolytes, however, there is very little information. The heat of dilution is related to the temperature dependence of the mean activity coefficient of polyelectrolytes, which has been intensively studied in this lab~ratory.~ Thus, in order to accumulate a complete set of information on the thermodynamic properties of polyelectrolyte solutions, we hoped to carry out further the AH measurements on various types of polyelectrolytes. We report results obtained for various salts of a poly(acrylic acid).The main purpose of this work is to investigate the specificity of gegenions on the heat of dilution. According to our measurements on the mean activity coefficients of polyelectrolytes, non-electrostatic interactions such as hydrophobic ones between solute and solvent molecules, as well as electrostatic interaction^,^ exert a great influence on the thermodynamic properties. Our aim in the present paper is to see how the non- electrostatic interactions affect the thermal properties.EXPERIMENTAL CALORIMETER The calorimeter used was of a twin-type, as shown in fig. 1. Two solution cells (A) are made of silver, 150 cm3 in volume, 4 cm in diameter and 12 cm in height. The two cells were arranged symmetrically. Solution A was poured into the silver cell and solution B was sealed in an ampoule (B), made of very thin glass and placed at the bottom of the cell. A stirrer (C, stainless steel), breaker @, stainless steel), and electric heater (E, covered with silver tube) were inserted into the cell from above. These two cells were placed in a vessel (F) made of stainless steel, 2000 em3 in volume. The synchronous motor and its accessories were mounted on the vessel, and heat transfer from the motor was prevented as completely as possibfe.The differences in temperature between the two solutions before and after mixing were monitored using 136 alumel-chromel thermocouples (G). Changes of 5 x 106N. ISE, K. MITA AND T. OKUBO 107 deg can be detected corresponding, when the calorimeter contains 150 cm3 of water, to ca. 0.04 J. The calorimeter assembly was submerged in a water bath, the temperature of which was regulated to within & 0.001"C. The Proportional-Integral-Differential (PID) circuit (Agne Research Center, Tokyo, Japan) was used in the inner bath and the current through the heater was continuously controlled. All measurements were made at 25°C. U U FIG. 1.-The calorimeter. PROCEDURE 60-100 cm3 of water was poured into the silver cell and 2-7 cm3 of sample solution was sealed in an ampoule.To maintain the symmetry of the two cells, almost same amounts of solution were poured into both silver cells. Thermal equilibrium between the bath and the cells was achieved after 8 h. After the steady state was attained the first ampoule in one of the cells (A,) was broken with the breaker (D). Mixing was complete within 5 min as shown by the temperature against time curve. When mixing was complete, another ampoule in the cell Az was broken and a second run carried out. Then, cell Al, followed by AZ, were heated by known amounts, the amount of heat being nearly equal to the heat of dilution, and the corresponding curves obtained. The heat of breaking an ampoule could be neglected, as ascertained by mixing water with itself. The precision of our calorimeter was tested by measuring the heat of dilution of MgCI, solutions.The results agree with the literature to within 5 %. Reproducibility was believed to be within f 5 %. The measurements were repeated several times and the heats of dilution reported in this paper are average values. MATERIALS Sodium polyacrylate (NaPAA) (degree of polymerization 640) was a gift from the Toa Gosei Chemical Co., Nagoya, Japan. An aqueous solution of NaPAA was passed through a column of cation- and anion-exchange resins in their acid forms. Completeness of conversion was checked by a flame test. Aqueous solutions of various salts of PAA108 HEAT OF DILUTION OF POLYACRYLATES Me4N, Et4N, Pr4N and Bu4N) were obtained by neutralizing the corresponding hydroxides. Urea used was for biochemicals (Merck).Deionized water obtained with cation- and anion- exchange resins was used. RESULTS AND DISCUSSION The heat of dilution in the present paper was obtained as an enthalpy change when a polymer solution is diluted from a concentration, nz, to a reference concentration, 0.005 mol ~ I r n - ~ . Results for R,NPAA are shown in fig. 2. -log m FIG. 2.-Heat of dilution of K4NPAA in the presence (0) and absence (0) of 3 M urea at 25°C. A, Me4NPAA ; B, Et4NPAA ; C, Pr4NPAA ; D, Bu4NPAA. The heat of dilution of NaPAA was slightly negative as is shown in fig. 3. This is in agreement with the results of Dolar et aL3 for Na, Zn, Li, K and Cs salts of a polystyrenesulphonic acid. The behaviour of the heat of dilution AH of NaPAA can be explained by electrostatic interactions between the macroions and their gegenions, and will be discussed below in detail.The heats of dilution of R,NPAA were positive, and the quantity of heat absorbed increased in the following order : Me,N < Et,N < Pr,N < Bu,N. Tetra-alkylammonium ions are known to be strong structure-formers (iceberg formation) as shown by measurements of activity coefficient, partial molal near infrared absorption,' solvation enthalpy,' etc. The structure-forming tendency becomes more and more distinct in the order (1). Lindenbaum?. lo reported the heat of dilution of low molecular weight tetra-alkylammonium halides at high concentrations. He explained his data in terms of the iceberg-enforcing effects of the organic ions. The concept of iceberg formation 6* l1 is also useful to explain the behaviour of the AH value of R,NPAA.It seems that a considerable amount of iceberg structure, formed by strong structure-making tetra-alkylammonium ions, is lost with dilution, and in extremely dilute solutions, the effect of hydrophobic bonds almost disappears and only the electrostatic intereactions are important. Because the energy of non-hydrogen bonded water is higher than that of water in an iceberg region, the dilution process of R,NPAA i s endothermic and the heat absorbed increases (1)N. ISE, K . MITA AND T. OKUBO 109 with the hydrophobicity of R4N+. This explanation is in accord with the concentra- tion dependence of the apparent molal heat capacity (ACp) of Bu,NBr in aqueous solutions ti : ACp increases with the salt concentration below 0.5 mol dm-3.In this concentration range, Bu,N ions will come closer together with increasing concentra- tion and increase the iceberg domains in a cooperative manner. Therefore, AC, increases with concentration. (Over 0.5 mol dm-3, the cations begin to link up and stabilize each other, and the number of water molecules forming the hydration shell around Bu,N+ decreases as discussed in detail below. Therefore, AC, decreases with the salt concentration at higher concentrations, as observed. 6, (fig. 2), becomes less positive with increasing concentration. The cation of Bu,NX (X = halogen) is known, from activity coefficient l2 and partial molal volume l 3 measure- ments, to aggregate at high concentrations. When aggregation occurs, some of the water molecules forming the iceberg structure around Bu4N+ are removed ; in other words, the overlapping of the " hydration iceberg shells " of the cations will cause a decrease in the number of hydrogen bonds per cation.When the solution is diluted, the aggregating ions separate and each Bu,N ion gains a complete hydration shell. Thus the magnitude of A H for Bu,NPAA in concentrated solutions is expected to be reduced. The heat of dilution of R,NPAA is believed to result froin three kinds of inter- action, i.e. (i) electrostatic interactiqns between charged solute molecules, (ii) solute- solvent interactions (iceberg formation or electrostriction), (iii) solute-solute inter- actions (aggregation). The first factor should give rise to a negative AH at very low concentrations, according to the Debye-Huckel theory. The second factor gives a positive AH as discussed above.The third is negative as argued above, and as confirmed by the fact that hydrophobic interactions between the side chains of protein molecules produce positive enthalpy changes.14 In the case of metal salts, contribution (i) is far greater than (ii) and (iii), and the dilution process is exothermic, whereas for R,NPAA (ii) and (iii) are greater than (i). The heat of dilution of R4NPAA was also measured in the presence of 3 M urea, which breaks hydrogen bonds between water mo1ecules.l 5-1 The results are given in fig. 2. On addition of urea AH became lower (more negative) and the magnitude of the shift was in the order : Me,N<Et,N<Pr,N<Bu,N. The order is quali- tatively accounted for by the fact that the stronger the hydrophobicity of the tetra- alkylammonium cation, the larger is the number of iceberg structural water molecules liberated by urea.This can be explained as follows : the hydrophobicity of Me4N+ is low so that the change in dielectric constant of the solution makes a larger contribution than the iceberg- breaking effect of urea. We can interpret the data of the heat of dilution of polyacrylates using the theory for paralIe1 oriented rodlike polyions proposed by Lifson and Katchalsky and Dolar et aZq3 According to the electrostatic enthalpy, He (the difference between the enthalpy of the real solution and that of the same solution in the hypo- thetical reference state in which all electrostatic interactions vanish) was given by eqn (2) The AHvalue of Bu4NPAA at concentrations greater than 0.03 mol For Me4N+, AH shifted slightly in the opposite direction.(1 +P2)+1n110 HEAT OF DILUTION OF POLYACRYLATES where z1 and z2 are the valencies of the macroions and gegenions, respectively. R is the charging parameter defined by D and b are the dielectric constant of solvent and the distance between the neighbour- ing charges of macroions, respectively. 103 y = 5 In- - 3 In rn 7ca2bN, (4) where NA is Avogadro’s number, a the radius of a macroion and tn the molar concen- tration. p is a constant given by 1 -p2 A = 1 +p coth B y ( 5 ) For (T/D)(dD/dT) and (T/V)(dV/dT) in eqn (2), Wyman’s l9 and Dorsey’s 2o data were used. In fig.3, the calculated values of AHfor NaPAA using eqn (2) (a = 5 A, b = 2.5 A) -mot 0 -log m are given on the figure for each line. FIG. 3.-Heat of dilution of NaPAA at 25°C : 0, observed ; -, calculated : values of d In a/d In T are compared with the experimental data. The agreement between calculated and observed values is not bad when d In a/d In T is 1.25. Similar results for the tetra- alkylammonium salts are shown in fig. 4. Calculated values show reasonable agree- ment with the observed ones when the parameter d In a/d In Tis chosen 5.0-7.5, 12.5, 16 and 20-25 for Me4N-, Et4N-, Pr4N- and Bu4N-PAA, respectively. It should be noted that the parameter for organic salts is much larger than for the sodium salt. The ionic radius, a, of polyions in eqn (2) implies the distance between the axis of the polyion and the surface which the gegenion approaches.R4N+ enhances water clusters around itself. Therefore, it is expected that a and its temperature dependence d In a/d In T will be influenced by organic cations.6* 1 3 9 21 It is also expected that the compact hydration shell around the organic cation is disrupted by increased temperature and that the alkyl chains of R,N+ will tend to expand. The d In a/d In T values for R4NPAA, therefore, should be much larger than those of nietal salts.N. ISE, K. MITA AND T. OKUBO 111 The water structural ionic volume can be estimated from the measurements of the partial molal volume.22 The d In a/d In T values estimated from the partial molal volume (7) are 0.06, 0.07, 0.08 and 0.11 for Me,N+, Et4N+, Pr4N+ and Bu4N+, res- pectively.On the other hand, those from the present measurements of AH are 6 , 12, 16 and 22 for Me4N+, Et,N+, Pr,N+ and Bu4N+, respectively. The values obtained from Vand AH are in clear disagreement by a factor of about 10'. This discrepancy must be taken into consideration, when the theory is revised. -log m FIG. 4.-Heat of dilution of R4NPAA at 25°C: 0, observed; -, calculated. A, Me,NPAA, d In uld In T = 5.0; B, Et,NPAA, d In a/d In T = 12.5 ; C, Pr4NPAA, d In a/d In T = 16.0 (upper line) and 15.0 (lower line) ; D, Bu~NPAA, d In a/d In T = 20.0. W. Schulze, 2. Elektrochem., 1954,58, 165. F. Vaslow, Abstracts, 145th Meeting of the American Chemical Society, New York, September 1963. T. Skerjanc, D. Dolar and D. Leskovsek, 2.phys. Chem., N. F., 1967,56,207,218 ; 1970,70,31. For a convenient review, see, N. Ise, Adv. Polymer Sci., 1971, 7, 536. E. Lange and A. L. Robinson, J. Amer. Chetn. SOC., 1931,53, 89. H. S. Frank and W. Y. Wen, Disc. Faraday SOC., 1957,24,133. K. W. B u d , J. Phys. Chem., 1967,71,1358. * C. V. Krishnan and H. L. Friedman, J. Phys. Chem., 1969,73, 3934. S . Lindenbaum, J. Phys. Chem., 1966,70,814. lo S . Lindenbawn, J. Phys. Chem., 1971,75,3733. G. Nkmethy and H. A. Scheraga, J. Chem. Phys., 1962,36, 3382, 3401. l2 S. Lindenbaum .and G. E. Boyd, J. Phys. Chem., 1968,72,911. l3 W. Y. Wen and S. Saito, J. Phys. Chem., 1964,68,2639. l4 G. Nkmethy and H. A. Scheraga, J. Phys. Chem., 1962, 66,1773. l5 C. Tanford, J. Amer. Chem. Soc., 1964,86,2050.112 HEAT OF DILUTlON OF POLYACRYLATES l6 G.C. Ktesheck and L. Benjamin, J. Phys. Chern., 1964,68,2476. l7 G. Nkmethy, Angew. Chem., 1967,79,260. l 8 S . Lifson and A. Katchalsky, J. Polymer Sci., 1954, 13,43. l9 J. Wyman, Phys. Rev., 1930,35, 623. 2o N. E. Dorsey, Properties of Ordinary Water-Substance, Reinhold, New York, 1940. 21 E. R. Nightingale, Jr., J. Phys. Chem., 1962,66, 894. 22 See, for example F. J. Millero, Chem. Rev., 1971,71,147.Heat of Dilution of Aqueous Solutions of PolyacrylatesBY NONO ISE, U Z u E I MITA AND TSUNEO OKUBODepartment of Polymer Chemistry, Kyoto University, Kyoto, JapanReceived 10th July, 1972The heats of dilution (AH) of aqueous solutions of polyacrylates having various gegenioas, i.e.,sodium and tetra-alkylammonium (alkyl = methyl, ethyI, n-propyl and n-butyl), were measured.The (AH) value of the sodium polyacrylate was negative in the investigated concentration range(0.005-0.17 mol dm-3), whereas those of organic salts were positive.With the addition of urea, theendothermicity was lessened. These results are discussed in terms of hydrophobic interactions betweensolute and solvent. The observed (AH) values were found to be in agreement with the values cal-culated by the theory of Lifson and Katchalsky, when an appropriate value of d In a/d In Tis assumed(a is the ionic radius of rodlike macroion and T the temperature). However, the assumed value forthis parameter was in disagreement with that obtained directly from the partial molal volumemeasurements (by a factor of -lo*).The heat of dilution is an important thermodynamic quantity of electrolytesolutions.For low molecular weight electrolytes, systematic study, experimental andtheoretical? of this quantity has been conducted for dilute solutions. For highmolecular weight electrolytes, however, there is very little information. The heatof dilution is related to the temperature dependence of the mean activity coefficient ofpolyelectrolytes, which has been intensively studied in this lab~ratory.~ Thus, inorder to accumulate a complete set of information on the thermodynamic propertiesof polyelectrolyte solutions, we hoped to carry out further the AH measurements onvarious types of polyelectrolytes. We report results obtained for various salts of apoly(acrylic acid).The main purpose of this work is to investigate the specificity of gegenions on theheat of dilution.According to our measurements on the mean activity coefficientsof polyelectrolytes, non-electrostatic interactions such as hydrophobic ones betweensolute and solvent molecules, as well as electrostatic interaction^,^ exert a great influenceon the thermodynamic properties. Our aim in the present paper is to see how the non-electrostatic interactions affect the thermal properties.EXPERIMENTALCALORIMETERThe calorimeter used was of a twin-type, as shown in fig. 1. Two solution cells (A) aremade of silver, 150 cm3 in volume, 4 cm in diameter and 12 cm in height. The two cellswere arranged symmetrically. Solution A was poured into the silver cell and solution B wassealed in an ampoule (B), made of very thin glass and placed at the bottom of the cell.Astirrer (C, stainless steel), breaker @, stainless steel), and electric heater (E, covered withsilver tube) were inserted into the cell from above. These two cells were placed in a vessel(F) made of stainless steel, 2000 em3 in volume. The synchronous motor and its accessorieswere mounted on the vessel, and heat transfer from the motor was prevented as completelyas possibfe. The differences in temperature between the two solutions before and aftermixing were monitored using 136 alumel-chromel thermocouples (G). Changes of 5 x10N. ISE, K. MITA AND T. OKUBO 107deg can be detected corresponding, when the calorimeter contains 150 cm3 of water, to ca.0.04 J.The calorimeter assembly was submerged in a water bath, the temperature of which wasregulated to within & 0.001"C.The Proportional-Integral-Differential (PID) circuit (AgneResearch Center, Tokyo, Japan) was used in the inner bath and the current through the heaterwas continuously controlled. All measurements were made at 25°C.U UFIG. 1.-The calorimeter.PROCEDURE60-100 cm3 of water was poured into the silver cell and 2-7 cm3 of sample solution wassealed in an ampoule. To maintain the symmetry of the two cells, almost same amounts ofsolution were poured into both silver cells. Thermal equilibrium between the bath and thecells was achieved after 8 h. After the steady state was attained the first ampoule in one ofthe cells (A,) was broken with the breaker (D).Mixing was complete within 5 min as shownby the temperature against time curve. When mixing was complete, another ampoule in thecell Az was broken and a second run carried out. Then, cell Al, followed by AZ, wereheated by known amounts, the amount of heat being nearly equal to the heat of dilution, andthe corresponding curves obtained.The heat of breaking an ampoule could be neglected, as ascertained by mixing water withitself.The precision of our calorimeter was tested by measuring the heat of dilution of MgCI,solutions. The results agree with the literature to within 5 %. Reproducibility wasbelieved to be within f 5 %. The measurements were repeated several times and the heatsof dilution reported in this paper are average values.MATERIALSSodium polyacrylate (NaPAA) (degree of polymerization 640) was a gift from the ToaGosei Chemical Co., Nagoya, Japan. An aqueous solution of NaPAA was passed througha column of cation- and anion-exchange resins in their acid forms.Completeness ofconversion was checked by a flame test. Aqueous solutions of various salts of PA108 HEAT OF DILUTION OF POLYACRYLATESMe4N, Et4N, Pr4N and Bu4N) were obtained by neutralizing the corresponding hydroxides.Urea used was for biochemicals (Merck). Deionized water obtained with cation- and anion-exchange resins was used.RESULTS AND DISCUSSIONThe heat of dilution in the present paper was obtained as an enthalpy change whena polymer solution is diluted from a concentration, nz, to a reference concentration,0.005 mol ~ I r n - ~ .Results for R,NPAA are shown in fig. 2.-log mFIG. 2.-Heat of dilution of K4NPAA in the presence (0) and absence (0) of 3 M urea at 25°C.A, Me4NPAA ; B, Et4NPAA ; C, Pr4NPAA ; D, Bu4NPAA.The heat of dilution of NaPAA was slightly negative as is shown in fig. 3. Thisis in agreement with the results of Dolar et aL3 for Na, Zn, Li, K and Cs salts of apolystyrenesulphonic acid. The behaviour of the heat of dilution AH of NaPAAcan be explained by electrostatic interactions between the macroions and theirgegenions, and will be discussed below in detail.The heats of dilution of R,NPAA were positive, and the quantity of heat absorbedincreased in the following order :Me,N < Et,N < Pr,N < Bu,N.Tetra-alkylammonium ions are known to be strong structure-formers (icebergformation) as shown by measurements of activity coefficient, partial molalnear infrared absorption,' solvation enthalpy,' etc.The structure-forming tendencybecomes more and more distinct in the order (1). Lindenbaum?. lo reported theheat of dilution of low molecular weight tetra-alkylammonium halides at highconcentrations. He explained his data in terms of the iceberg-enforcing effects of theorganic ions. The concept of iceberg formation 6* l1 is also useful to explain thebehaviour of the AH value of R,NPAA. It seems that a considerable amount oficeberg structure, formed by strong structure-making tetra-alkylammonium ions, islost with dilution, and in extremely dilute solutions, the effect of hydrophobic bondsalmost disappears and only the electrostatic intereactions are important.Becausethe energy of non-hydrogen bonded water is higher than that of water in an icebergregion, the dilution process of R,NPAA i s endothermic and the heat absorbed increases(1N. ISE, K . MITA AND T. OKUBO 109with the hydrophobicity of R4N+. This explanation is in accord with the concentra-tion dependence of the apparent molal heat capacity (ACp) of Bu,NBr in aqueoussolutions ti : ACp increases with the salt concentration below 0.5 mol dm-3. In thisconcentration range, Bu,N ions will come closer together with increasing concentra-tion and increase the iceberg domains in a cooperative manner. Therefore, AC,increases with concentration.(Over 0.5 mol dm-3, the cations begin to link up andstabilize each other, and the number of water molecules forming the hydration shellaround Bu,N+ decreases as discussed in detail below. Therefore, AC, decreases withthe salt concentration at higher concentrations, as observed. 6,(fig. 2),becomes less positive with increasing concentration. The cation of Bu,NX (X =halogen) is known, from activity coefficient l2 and partial molal volume l 3 measure-ments, to aggregate at high concentrations. When aggregation occurs, some of thewater molecules forming the iceberg structure around Bu4N+ are removed ; in otherwords, the overlapping of the " hydration iceberg shells " of the cations will cause adecrease in the number of hydrogen bonds per cation.When the solution is diluted,the aggregating ions separate and each Bu,N ion gains a complete hydration shell.Thus the magnitude of A H for Bu,NPAA in concentrated solutions is expected tobe reduced.The heat of dilution of R,NPAA is believed to result froin three kinds of inter-action, i.e. (i) electrostatic interactiqns between charged solute molecules, (ii) solute-solvent interactions (iceberg formation or electrostriction), (iii) solute-solute inter-actions (aggregation). The first factor should give rise to a negative AH at very lowconcentrations, according to the Debye-Huckel theory. The second factor gives apositive AH as discussed above. The third is negative as argued above, and asconfirmed by the fact that hydrophobic interactions between the side chains ofprotein molecules produce positive enthalpy changes.14 In the case of metal salts,contribution (i) is far greater than (ii) and (iii), and the dilution process is exothermic,whereas for R,NPAA (ii) and (iii) are greater than (i).The heat of dilution of R4NPAA was also measured in the presence of 3 M urea,which breaks hydrogen bonds between water mo1ecules.l 5-1 The results are givenin fig.2. On addition of urea AH became lower (more negative) and the magnitudeof the shift was in the order : Me,N<Et,N<Pr,N<Bu,N. The order is quali-tatively accounted for by the fact that the stronger the hydrophobicity of the tetra-alkylammonium cation, the larger is the number of iceberg structural water moleculesliberated by urea.Thiscan be explained as follows : the hydrophobicity of Me4N+ is low so that the changein dielectric constant of the solution makes a larger contribution than the iceberg-breaking effect of urea.We can interpret the data of the heat of dilution of polyacrylates using the theoryfor paralIe1 oriented rodlike polyions proposed by Lifson and Katchalsky andDolar et aZq3 According to the electrostatic enthalpy, He (the differencebetween the enthalpy of the real solution and that of the same solution in the hypo-thetical reference state in which all electrostatic interactions vanish) was given byeqn (2)The AHvalue of Bu4NPAA at concentrations greater than 0.03 molFor Me4N+, AH shifted slightly in the opposite direction.(1 +P2)+1110 HEAT OF DILUTION OF POLYACRYLATESwhere z1 and z2 are the valencies of the macroions and gegenions, respectively.Ris the charging parameter defined byD and b are the dielectric constant of solvent and the distance between the neighbour-ing charges of macroions, respectively.103y = 5 In- - 3 In rn7ca2bN, (4)where NA is Avogadro’s number, a the radius of a macroion and tn the molar concen-tration. p is a constant given by1 -p2A = 1 +p coth B y ( 5 )For (T/D)(dD/dT) and (T/V)(dV/dT) in eqn (2), Wyman’s l9 and Dorsey’s 2o datawere used.In fig. 3, the calculated values of AHfor NaPAA using eqn (2) (a = 5 A, b = 2.5 A)-mot 0-log mare given on the figure for each line.FIG. 3.-Heat of dilution of NaPAA at 25°C : 0, observed ; -, calculated : values of d In a/d In Tare compared with the experimental data.The agreement between calculated andobserved values is not bad when d In a/d In T is 1.25. Similar results for the tetra-alkylammonium salts are shown in fig. 4. Calculated values show reasonable agree-ment with the observed ones when the parameter d In a/d In Tis chosen 5.0-7.5, 12.5,16 and 20-25 for Me4N-, Et4N-, Pr4N- and Bu4N-PAA, respectively. It should benoted that the parameter for organic salts is much larger than for the sodium salt.The ionic radius, a, of polyions in eqn (2) implies the distance between the axis ofthe polyion and the surface which the gegenion approaches. R4N+ enhances waterclusters around itself. Therefore, it is expected that a and its temperature dependenced In a/d In T will be influenced by organic cations.6* 1 3 9 21 It is also expected thatthe compact hydration shell around the organic cation is disrupted by increasedtemperature and that the alkyl chains of R,N+ will tend to expand.The d In a/d In Tvalues for R4NPAA, therefore, should be much larger than those of nietal saltsN. ISE, K. MITA AND T. OKUBO 111The water structural ionic volume can be estimated from the measurements of thepartial molal volume.22 The d In a/d In T values estimated from the partial molalvolume (7) are 0.06, 0.07, 0.08 and 0.11 for Me,N+, Et4N+, Pr4N+ and Bu4N+, res-pectively. On the other hand, those from the present measurements of AH are 6 , 12,16 and 22 for Me4N+, Et,N+, Pr,N+ and Bu4N+, respectively. The values obtainedfrom Vand AH are in clear disagreement by a factor of about 10'. This discrepancymust be taken into consideration, when the theory is revised.-log mFIG. 4.-Heat of dilution of R4NPAA at 25°C: 0, observed; -, calculated. A, Me,NPAA,d In uld In T = 5.0; B, Et,NPAA, d In a/d In T = 12.5 ; C, Pr4NPAA, d In a/d In T = 16.0 (upperline) and 15.0 (lower line) ; D, Bu~NPAA, d In a/d In T = 20.0.W. Schulze, 2. Elektrochem., 1954,58, 165.F. Vaslow, Abstracts, 145th Meeting of the American Chemical Society, New York, September1963.T. Skerjanc, D. Dolar and D. Leskovsek, 2. phys. Chem., N. F., 1967,56,207,218 ; 1970,70,31.For a convenient review, see, N. Ise, Adv. Polymer Sci., 1971, 7, 536.E. Lange and A. L. Robinson, J. Amer. Chetn. SOC., 1931,53, 89.H. S. Frank and W. Y. Wen, Disc. Faraday SOC., 1957,24,133.K. W. B u d , J. Phys. Chem., 1967,71,1358.* C. V. Krishnan and H. L. Friedman, J. Phys. Chem., 1969,73, 3934.S . Lindenbaum, J. Phys. Chem., 1966,70,814.lo S . Lindenbawn, J. Phys. Chem., 1971,75,3733.G. Nkmethy and H. A. Scheraga, J. Chem. Phys., 1962,36, 3382, 3401.l2 S. Lindenbaum .and G. E. Boyd, J. Phys. Chem., 1968,72,911.l3 W. Y. Wen and S. Saito, J. Phys. Chem., 1964,68,2639.l4 G. Nkmethy and H. A. Scheraga, J. Phys. Chem., 1962, 66,1773.l5 C. Tanford, J. Amer. Chem. Soc., 1964,86,2050112 HEAT OF DILUTlON OF POLYACRYLATESl6 G. C. Ktesheck and L. Benjamin, J. Phys. Chern., 1964,68,2476.l7 G. Nkmethy, Angew. Chem., 1967,79,260.l 8 S . Lifson and A. Katchalsky, J. Polymer Sci., 1954, 13,43.l9 J. Wyman, Phys. Rev., 1930,35, 623.2o N. E. Dorsey, Properties of Ordinary Water-Substance, Reinhold, New York, 1940.21 E. R. Nightingale, Jr., J. Phys. Chem., 1962,66, 894.22 See, for example F. J. Millero, Chem. Rev., 1971,71,147

ISSN:0300-9599    

DOI:10.1039/F19736900106

出版商:RSC    

年代:1973

数据来源: RSC

摘要:

Metal-ion Catalyzed Decarboxylation of Oxaloacetic Acid BY -00 ITO, HIROSHI KOBAYASHI* AND mJI NOMIYA Dept. of Chemistry, Tokyo Institute of Technology, Oh-okayama, Meguro-ku, Tokyo, Japan Received 3rd July, 1972 Decarboxylation of oxaloacetic acid is accelerated by hydrated metal ions. The thermodynamic data of activation AG*, AH* and AS* for the ratedetermining process were determined. The main species involved in autodecarboxylation was identified as univaleat hydrogen oxdoacetate anion. Plots ofAH* against TAS* (T = 298 K) for the decarboxy€ations in the presence of various hydrated divalent and tervalent metal ions and also in the absence of hydrated metal ion show excellent linearity. A compensation effect was observed between the activation enthalpy and entropy terms.A barrier in the autodecarboxylation reaction is a highly entropy-decreasing process. The decarboxyla- tion of a metal chelate complex, however, is facilitated by decrease in A@ arising from an increase in T A P much larger than the increase in AH*. The metahion accelerating effect in this particular case is essentially an entropy phenomenon. The decarboxylation of oxaloacetic acid to pyruvic acid in aqueous solution is accelerated by metal The metal-ion promoted reaction is described in terms of a Michaelis-Menten reaction mechanism: fast metal complex formation (1) followed by a rate-determining reaction of the coordinating ligand (2). CH2 / \ c=o oxaloacetate dianion + Mn+ + O==C--C I II *+ 0- 0 ... .. . * . .. I 0- 0 (I) --+ pyruvate anion + C 0 2 + Mn+ (2) Metal ions play double r8les in accelerating the reaction.The metal ions cer- tainly increase the velocity of the rate-determining process, since the coordinating ligand has a higher reactivity due to conformational rearrangement on complex formation and an inductive and/or a mesomeric charge-transfer polarization effect arising from coordination to the metal cation. Furthermore, the metal ions with a higher tendency for complex formation may increase the concentration of the complex which keeps the coordinating ligand in a highly reactive state. The overall reaction rate constant koverall is given by a product of the equilibrium constant of the complex formation (1) Kand the rate constant of the rate determining process (2) k : koverall = kK. So far, the metal-ion promotion effect has been shown as due to an increase in koverall.The metal-ion enhanced reactivity of the coordinating ligand should be discussed in terms of an increase of k rather than koverall. In the present work, we determined the rate constant k for the decarboxylation of oxaloacetate catalyzed by various hydrated metal cations. From the temperature dependence of the rate 113114 DECARBOXYLATION OF OXALOACETIC ACID constant, AG*, AH* and AS* for process (2) were obtained. A mechanism will be proposed on the basis of the thermodynamic data of activation obtained. $ 5 E 3 1 4 1 EXPERIMENTAL MATERIALS Lanthanum, gadolinium, yttrium and lutetium sulphates were prepared from sulphuric acid solutions of the oxides, which were obtained by ignition of the oxalates (Shinetsu Chem.Ind. Ltd.). Other salts were commercially available and were repeatedly recrystallized. Oxaloacetic acid (B.D.H. Ltd.) was used without further purification. 0 E APPARATUS A N D METHOD Kinetic studies were carried out using a manometric apparatus. The reaction was initiated by dissolving the accurately weighed powdered sample of oxaloacetic acid in an aqueous solution containing the metal cation. The reaction flask was shaken in a water bath themostatted at 20,25,30 or 37°C. The reaction rate was determined by measuring the amount of carbon dioxide evolved at various stages of reaction. The pH and ionic strength of aqueous solutions were adjusted by adding sodium hydroxide sodium chloride and potassium nitrate. MEASUREMENTS OF N.M.R.SPECTRA N.m.r. spectra were recorded at room temperature on a 100 MHz Japan Electron Optics Laboratory spectrometer model JNM 4H-100. Sodium 2,2-dimethyl-2-silapentane-5- sulphonate @SS) was used as internal reference. RESULTS AND DISCUSSION AUTODECARBOXYLATION Oxaloacetic acid in aqueous solution spontaneously decomposes to pyruvic acid and carbon dioxide even in the absence of catalyst. KINETICS IN AQUEOUS soLuTIoNs.-Kinetic studies were carried out in aqueous solutions prepared according to Pedersen 2* from NaOH, NaCl and KN03. No pH change was observed upon decarboxylation of oxaloacetic acid. To initiate the reaction, the powdered sample of oxaloacetic acid was dissolved in an aqueous solution. The reaction may be heterogeneous for a few minutes after mixing.As shown in fig. 1, the amount of carbon dioxide evolved increased linearly with respect to time after about 7 to 9 min. Since it takes more than 20 h for 0.02 M 1 pH 3.4 0 2 pH 2.1 pH 1.8 - -5 I0 15 20 timelmin FIG. 1 .-Tile initial reaction in the autodecarboxylation of oxaloacetic acid in aqueous solutions containing NaOH and NaCl at 25°C ( I = 0.10 M).HIROO ITO, HIROSHI KOBAYASHI AND KENJI NOMIYA 115 oxaloacetic acid to be completely decarboxylated, the rate constant can be determined from the slope of the straight line obtained for the first 20min. The rate constant of the first order reaction is given by k = l / t 1n(pco/’pco -p), wherep is the manometer reading. The rate constants obtained by two different methods were coincident within experimental error.The rate constants of the autodecarboxylation of oxalo- acetic acid were determined at a variety of pH (table 1). TABLE TH THE RATE CONSTANTS OF THE AUTODECARBOXYLATION OF OXALOACETIC ACID IN AQUEOUS SOLUTION (I = 0.10) AT 25°C PH 1.8 2.1 2.8 3.4 3.7 k x 105/s-l 1.9 2.0 4.7 4.3 2.9 THE SPECIES INVOLVED IN THE AUTODECARFJOXYLATION.-~~ aqueous Solution, an equilibrium is present between the species SH2 (oxaloacetic acid), SH- and S2- (oxaloacetate dianion). K1 SH2 + SH-+H+ (3) SH- + S2-+H+ (4) K2 The concentrations of the species SH2, SH- and S2- are evaluated using Pedersen’s values of Kl and K2. Fig. 2 shows the concentrations of the existing species in solu- tion calculated as a function of pH. PH FIG. 2.-pH dependence of the existing species in aqueous solutions of oxaloacetic acid : -, [SH-] ; -.- , [S-1; -.. - , [SH21. The reaction rate is given as a function of hydrogen ion concentration [H+] and the autodecarboxylation rate constants of SH2, SH- and S2- denoted by ko, k , and k2, respectively : v = k[S] = k[Sl* = k,[SH2] + k,[SH-] + k2[S2-] = (ako + bkl + ck2)[Sl0, (5)116 DECARBOXYLATION OF OXALOACETIC ACID 1 [H+12/K,K, + [H+]/K2 + 1 ' c = Since a, b and c can be evaluated from the hydrogen ion concentration, the rate constants, ko, k , and k2 are calculated from the rates obtained for a variety of hydro- gen ion concentrations. This indicates that the species directly involved in the auto- carboxylation is SH- but not SH2 and S2-. Since ko and k2 are negligibly small, eqn (5) is approximately In fact, plots of k against b gave a straight line passing through the origin, within experimental error (fig.3). From the slope of the line, the rate constant kl is ZJ = k[S], = bkl[SIo. (6) 5.8 x 10-5 s-1. b x 103 FIG. 3.-Plots of the first-order rate constant of the autodecarboxylation of oxaloacetic acid against concentration of hydrogen oxaloacetate anion in units of the initial concentration of oxaloacetic acid, [SH-] = b[S],. N.M.R. SPECTRA OF OXALOACETIC ACID IN AQUEOUS SOLUTION.-N.m.r. spectra Of oxaloacetic acid in aqueous solutions, which were pH adjusted by adding solutions of HC1O4 and NaOH, were measured using DSS as internal reference. The -CH2- proton resonance of oxaloacetic acid is observed at 3.71 p.p.m. The methyl proton resonance of the decarboxylation product pyruvic acid appears at 2.35p.p.m.and the methyl proton resonance of the hydrate of pyruvic acid, 2,2- dihydroxypropionic acid, at 1.46 p.p.m. Oxaloacetic acid is present as the univalent anion SH- in aqueous solution at pH N 3. Two n.m.r. signals of comparable intensity were observed at 3.71 and 2.35 p.p.m. : a signal was observed at 1.46 p.p.m. about 2 h after dissolution. When the autodecarboxylation of oxaloacetic acid was almost completed, keeping the solu- tion at about 50°C for more than 24 h, an extremely intense signal characteristic of pyruvic acid was observed at 2.35 p.p.in., while only weak signals were observed at 3.71 and 1.46 p.p.m. In an aqueous solution at pH 6-7, oxaloacetic acid is present mainly as the divalent anion S2-. An intense signal was observed at 3.71 p.p.m., while only a very weak signal appeared at 2.35 p.p.m.The ratio of intensities of these two signals was not changed many hours after dissolution. No appreciable production of pyruvic acid was detected. In an aqueous solution of pH < 1, oxaloacetic acid is mainly present as the undissociated acid SH,. An intense signal characteristic of theHIROO ITO, HIROSHI KOBAYASHI AND KENJI NOMIYA 117 undissociated acid was observed at 3.00 p.p.m., while the signal at 3.71 p.p.m. was just a trace. Another signal at 1.70 p.p.m. seems to be characteristic of the solution of pH < 1, though it has not been well resolved from noise. No signal appeared at 1.46 or 2.35 p.p.m. for many hours after dissolution. Thus, no pyruvic acid was produced.These observations on the n.m.r. spectra support the conclusion from mano- metric studies that the main species involved in the autodecarboxylation is SH- but not SH2 and S2-. A pH-dependent equilibrium between pyruvic acid and its hydrate 2,2-dihydroxy- propionic acid in aqueous solution has been established by physical meas~rernents.~* N.m.r. studies also revealed that the hydration of the carbonyl group in pyruvic acid occurs only when the carboxyl group is protonated. Pyruvate anion is not hydrated. The details of the n.m.r. studies will be presented elsewhere. In the autodecarboxyla- tion of oxaloacetic acid, however, one r81e of the carboxyl proton in the active species SH- is in the hydration process prior to decarboxylation. When a proton attached to the second carboxyl group (the group furthest from the carbonyl) is transferred to the carbonyl oxygen, the coordinating water at the carbonyl carbon must be permanently deformed.This gives rise to proton attack by this water on the -CH2- group followed by decarboxylation. In an aqueous solution at pH 3, dehydration of the 2,2-dihydroxypropionate anion to pyruvate is very fast. In the initial stage of the reaction, the n.m.r. signal of the -CH2- protons was observed at 3.71 p.p.m. When the reaction was almost complete, only a weak signal of the CH3- protons in 2,2-dihydroxypropionate was observed at 1.46 p.p.m. whereas an intense signal of pyruvate anion was detected at 2.35 p.p.m. There is no great difference between CH,- proton resonance in pyruvate anion and undissociated pyruvic acid and thus n.m.r.gives no further information on the predominant form of pyruvic acid in solution. Acid dissociation measurements, however, suggests that the pyruvate anion is predominant in pH 3 solution.2 In the case of an aqueous solution of pH < 1, both carboxyl groups are protonated. Hydrogen-bonding of the y-carboxyl proton with the carbonyl oxygen is strong and thus a hydrogen-bonded chelate is formed as in undissociated pyruvic acid, giving rise to hydration. The water coordinated to the carbonyl is again deformed and then two successive proton transfers give a,a-dihydroxysuccinic acid ; an intense signal around 3.00 p.p.m. is assigned to its -CH2- protons. Since the a-hydroxy groups thus formed can accept no more protons, the second carboxyl group even if protonated cannot be decarboxylated.In an aqueous solution of pH 6-7, due to the divalent negative charge of the main species S2- an orientation effect of solvent water around the charged species arises. However, no further hydration of the central carbonyl is promoted. In fact, no decarboxylation was observed in the solution. Esterification also inhibits decarboxylation. DECARBOXYLATION IN THE PRESENCE OF HYDRATED METAL IONS In aqueous solution, the decarboxylation of oxaloacetic acid is accelerated by hydrated metal ions. Prior to decarboxylation, a metal chelate with the oxaloacetate divalent anion is formed.6* ' The structure of the metal chelate involved in the metal-ion promoted reaction has been established I f as (I) (eqn (1)).The free car- boxyl group is not protonated under the experimental conditions employed. Metal chelate formation is so fast that it could not be detected without employing a fast technique such as stopped flow. A rate for chelate formation lo3 times higher than the rate of decarboxylation of the chelating oxaloacetate was estimated.118 DECARBOXYLATION OF OXALOACETIC ACID In terms of the Michaelis-Menten reaction mechanism, the decarboxylatien of oxaloacetate S into pyruvate P in the presence of hydrated metal ion E is described as follows : k+ I k- 1 ES + E+P+C02. SH-+E f ES+H+ k Provided there is a constant concentration of intermediate metal chelate, the reaction velocity is given by or If k+,[SH-] 9 k-,[H+]+k, v = kp], and the rate constant k is determined as a ratio of the initial velocity and the initial concentration [El,.The condition k+,[SH-] 9 k-,[H+] + k is experimentally satisfied when the initial concentration of oxaloacetic acid [Sl0 i s much greater (approximately 100 times) than that of hydrated metal ion [Elo. The decarboxylation of oxaloacetic acid in an aqueous solution (PH 3.7 and ionic strength 0.1 M) containing hydrated aluminium ions shows excellent linearity between manometer readings and reaction time as shown in fig. 4. This indicates that the mean veIocity is constant in the initial stage of the reaction. When the first term of the right-band-side of eqn (10) is negligible, the mean velocity can be kept I n t - 5 10 15 timelmin FIG. 4.-The initial reaction in the decarboxylation of oxaloacetic acid in the presence of hydrated aluminium ion at various temperatures.[S], = 2 x mol l.-I, [El, = 6 x loe4 mol I.", pH = 3.7 and I = 0.10.HIROO ITO, HIROSHI KOBAYASHI AND KENJI NOMIYA 119 constant for a period of time even if the concentration [SH-] is decreasing. Fig. 4 also shows the temperature dependence of the rate. Plots of the initial rate against the initial concentration of metal ion [El, give a straight line passing through the origin. The slope of the straight line D = k[EIo gives the real rate constant k. The rate constants of the decarboxylation of oxaloacetic acid for various hydrated metal ions present in aqueous solution of pH 3.7 and ionic strength 0.1 M are summarized in table 2. TABLE Z-RATE CONSTANTS OF THE METAL-ION CATALYZED DECARBOXYLATIONS OF OXALO- ACETIC ACID (S-') 20°C 25°C 30°C auto.La3+ Gd3+ Y3+ Lu3+ CU2+ Ni2+ Zn2+ co2+ ~ 1 3 + 2 . 6 ~ 10-5 1.6 x 10-3 1 . 4 ~ 10-3 8 . 3 ~ 10-4 5 . 9 ~ 10-4 7 . o ~ 10-3 2 . 8 ~ 10-3 1 . 6 ~ 10-3 1 . 6 ~ 10-3 - 2 . 9 ~ 10-5 2 . 4 ~ 10-3 2 . 3 ~ 10-3 1.4 x 10-3 9.1 x 10-4 3 . 7 ~ 10-3 3 . o ~ 10-3 2 . 2 ~ 10-3 1.3 x 1.3 x 3 . o ~ 10-5 4.1 x 10-3 4 . 2 ~ 10-3 2.5 x 10-3 1.5 x 10-3 7 . 3 ~ 10-3 5 . 5 ~ 10-3 4 . 4 ~ 10-3 2 . 6 ~ 2.2x TEMPERATURE DEPENDENCE OF THE RATE CONSTANTS.-PlOtS O f log k against 1/T gave a straight line confirming an Arrhenius relationship k = A exp (-E/RT). The thermodynamic data of activation were evaluated from the temperature depend- ence of the rate constants assuming K = 1 in k = rc(kT/h) exp (AS* /It) exp (- AH* /RT).The results are summarized in table 3. Plots of AH* against TAS* (T = 298 K) for decarboxylation not only in the presence of various hydrated ions but also in their absence show excellent linearity. A compensation effect was observed between the activation enthalpy and entropy terms as shown in fig. 5. TABLE 3.-THERMODYNAMIC DATA OF ACTIVATIONS A G * / @ ~ I mol-1) at 25°C AH*/@c~~ mol-1) AS*/(C~~ mol-1 K-1) auto. La3+ Gd3+ Y3+ Lu3+ CuZ+ Niz+ Zn2+ coz+ ~ 1 3 + 23.6 20.0 21.0 21 .o 21.3 21.6 20.1 20.7 20.8 21.0 1.9 24.0 16.0 18.7 18.8 15.8 19.6 16.3 21.1 17.2 - 72.9 - 16.9 - 7.6 - 8.4 - 19.3 - 1.6 - 14.8 0.9 - 12.7 13.4 THE EFFECTS OF METAL IoNs.-The decarboxylation of oxaloacetic acid is acceler- ated when oxaloacetic acid chelates with metal ions.The activation energy is reduced upon metal chelate formation. The activation entbalpy and entropy of auto- decarboxylation were obtained AH* = 1.9 kcal mol-1 and AS* = -72.9 cal mol-l K-l. A barrier to the autodecarboxylation reaction is the high entropy decrease involved. Decarboxylation of a metal chelate complex, however, is accelerated since it is accompanied by some entropy-increasing processes. The enhancement of120 DECARBOXYLATION OF OXALOACBTIC ACID the rate due to chelation is due to a decrease in AG* arising from an increase in TASd* much larger than an increase in AH* up to 15-24 kcal mol-l. The metal-ion accelerating effect in this particular reaction is essentially an entropy phenomenon. The compensation relationship shown by the plots of AH* against TAS* (T = 298 K) holds not only for the metal-ion promoted decarboxylation regardless the charge on the metal ion but also for autodecarboxylation, hence the reaction mechanisms must be similar.The observed activation entropy of autodecarboxyla- tion -72.9 cal mol-l K-l indicates that a number of solvent water molecules are captured or orientated by the activated state. However, another entropy increasing effect is involved in the metal-ion promoted reaction. A tridentate chelate structure with both carboxyl groups and the carbonyl oxygen coordinated to the metal ion is assumed as the activated state. For this to be achieved, some of the coordinated and oriented water molecules must be set free. An increase in entropy arising from the formation of a tridentate activated complex compensates for an activation entropy decrease for the decarboxylation.The enthalpy increase necessary for the bond- breaking and the release of orientated solvent water can be compensated for by a much larger entropy increase. FIG. 5.-Plots of AH* against TAS* (T = 298 K ) . log K FIG. 6.-Plots of log K (stability) against log k (reactivity).HIROO ITO, HIROSHI KOBAYASHI A N D KENJI NOMIYA 121 Higher polarization of the coordinating ligand is expected for tervalent than divalent metal cations. However, the rates for the tervalent metal chelates are smaller than those for the divalent metal chelates as seen in table 2. This might provide further evidence that the metal-ion accelerating effect is essentially an entropy phenomenon. Plots of log K at 298 K for metal chelate formations obtained by Gelles and Nancollas * against log k at 298 K for the metal-ion promoted reactions measured in the present work are linear as shown in fig.6. Such a linear free energy relationship obtained for a series of divalent metal ions is opposite in sign to a linear free energy relationship obtained for a series of tervalent metal ions. As is clearly shown by Nanc~llas,~ a metal chelate formed by oxygen donors such as in carboxy- lates is stabilized mainly by a large entropy term but destabilized by the enthalpy term. Since a common compensation relationship between AH* and TAS* was observed, different linear free energy relationships obtained for divalent transition metal ions and tervalent rare earth metal ions might arise from a difference in the relative contributions of entropy and enthalpy terms to the stability constants of the metal chelates.The authors thank Professors Kenji Tamaru and Iwao Yasumori for helpful discussions. (a) L. Krampits and C. Werkman, Boichem. J., 1941,35,595 ; (b) H. A. Krebs, Biochem. J., 1942,36, 303 ; (c) S. Ochoa, J. Biol. Chem., 1948,174, 115 ; ( d ) A. Kornberg, S. Ochoa and A. H. Mehler, J. Biol. Chem., 1948, 174, 159 ; (e) J. F. Speck, J. Biol. Chem., 1949, 178, 315 ; (f) R. Steinberger and F. H. Westheimer, J. Amer. Chem. SOC., 1951, 73, 429 ; (9) E. Gelles and J. P. Clayton, Trans. F'raday SOC., 1956, 52, 353; (h) J. V. Rund and R. A. Robert, J. Amer. Chem. SOC., 1964, 86, 367; (i) J. V. Rund and K. G. Claus, J. Amer. Chem. Soc., 1967,89,2256 ; (j) J.V. Rund and K. G. Claus, Inorg. Chem., 1968,7,860 ; (k) M. Munakata, M. Matsui, M. Tabushi and T. Shigematsu, Bull. Chem. SOC. Japan, 1970,43, 114. K. J. Pedersen, Acta Chem. Scand., 1952,6,243. K. J. Pedersen, Acta Chem. Scand., 1952,6,285. H. Strehlow, 2. Elektrochem., 1962,66,392. M. Becker, Ber. Bunsenges. phys. Chem., 1964, 68,669. E. Gelles and R. W. Hay, J. Chem. SOC., 1958,3673. ' E. Gelles and A. Salama, J. Chem. SOC., 1958,3683. E. Gelles and G. H. Nancollas, Tram. Farahy Soc., 1956, 52,98. G. H. NancolIas, Interactions in EZectrolyte Solutions (Elsevier, Amsterdam, 1966), p. 214.Metal-ion Catalyzed Decarboxylation of Oxaloacetic AcidBY -00 ITO, HIROSHI KOBAYASHI* AND mJI NOMIYADept. of Chemistry, Tokyo Institute of Technology, Oh-okayama,Meguro-ku, Tokyo, JapanReceived 3rd July, 1972Decarboxylation of oxaloacetic acid is accelerated by hydrated metal ions.The thermodynamicdata of activation AG*, AH* and AS* for the ratedetermining process were determined. The mainspecies involved in autodecarboxylation was identified as univaleat hydrogen oxdoacetate anion.Plots ofAH* against TAS* (T = 298 K) for the decarboxy€ations in the presence of various hydrateddivalent and tervalent metal ions and also in the absence of hydrated metal ion show excellentlinearity.A compensation effect was observed between the activation enthalpy and entropy terms. Abarrier in the autodecarboxylation reaction is a highly entropy-decreasing process. The decarboxyla-tion of a metal chelate complex, however, is facilitated by decrease in A@ arising from an increasein T A P much larger than the increase in AH*.The metahion accelerating effect in this particularcase is essentially an entropy phenomenon.The decarboxylation of oxaloacetic acid to pyruvic acid in aqueous solution isaccelerated by metal The metal-ion promoted reaction is described in termsof a Michaelis-Menten reaction mechanism: fast metal complex formation (1)followed by a rate-determining reaction of the coordinating ligand (2).CH2/ \ c=o oxaloacetate dianion + Mn+ + O==C--CI II*+ 0- 0 ... .. . * . ..I0-0(I) --+ pyruvate anion + C 0 2 + Mn+ (2)Metal ions play double r8les in accelerating the reaction. The metal ions cer-tainly increase the velocity of the rate-determining process, since the coordinatingligand has a higher reactivity due to conformational rearrangement on complexformation and an inductive and/or a mesomeric charge-transfer polarization effectarising from coordination to the metal cation. Furthermore, the metal ions with ahigher tendency for complex formation may increase the concentration of the complexwhich keeps the coordinating ligand in a highly reactive state.The overall reactionrate constant koverall is given by a product of the equilibrium constant of the complexformation (1) Kand the rate constant of the rate determining process (2) k : koverall =kK. So far, the metal-ion promotion effect has been shown as due to an increase inkoverall. The metal-ion enhanced reactivity of the coordinating ligand should bediscussed in terms of an increase of k rather than koverall.In the present work, wedetermined the rate constant k for the decarboxylation of oxaloacetate catalyzed byvarious hydrated metal cations. From the temperature dependence of the rate11114 DECARBOXYLATION OF OXALOACETIC ACIDconstant, AG*, AH* and AS* for process (2) were obtained. A mechanism will beproposed on the basis of the thermodynamic data of activation obtained.$ 5E 3 1 4 1EXPERIMENTALMATERIALSLanthanum, gadolinium, yttrium and lutetium sulphates were prepared from sulphuricacid solutions of the oxides, which were obtained by ignition of the oxalates (Shinetsu Chem.Ind. Ltd.). Other salts were commercially available and were repeatedly recrystallized.Oxaloacetic acid (B.D.H. Ltd.) was used without further purification.0EAPPARATUS A N D METHODKinetic studies were carried out using a manometric apparatus.The reaction wasinitiated by dissolving the accurately weighed powdered sample of oxaloacetic acid in anaqueous solution containing the metal cation. The reaction flask was shaken in a waterbath themostatted at 20,25,30 or 37°C. The reaction rate was determined by measuringthe amount of carbon dioxide evolved at various stages of reaction.The pH and ionic strength of aqueous solutions were adjusted by adding sodium hydroxidesodium chloride and potassium nitrate.MEASUREMENTS OF N.M.R. SPECTRAN.m.r. spectra were recorded at room temperature on a 100 MHz Japan Electron OpticsLaboratory spectrometer model JNM 4H-100. Sodium 2,2-dimethyl-2-silapentane-5-sulphonate @SS) was used as internal reference.RESULTS AND DISCUSSIONAUTODECARBOXYLATIONOxaloacetic acid in aqueous solution spontaneously decomposes to pyruvic acidand carbon dioxide even in the absence of catalyst.KINETICS IN AQUEOUS soLuTIoNs.-Kinetic studies were carried out in aqueoussolutions prepared according to Pedersen 2* from NaOH, NaCl and KN03.NopH change was observed upon decarboxylation of oxaloacetic acid.To initiate the reaction, the powdered sample of oxaloacetic acid was dissolvedin an aqueous solution. The reaction may be heterogeneous for a few minutes aftermixing. As shown in fig. 1, the amount of carbon dioxide evolved increased linearlywith respect to time after about 7 to 9 min.Since it takes more than 20 h for 0.02 M1pH 3.40 2 pH 2.1pH 1.8--5 I0 15 20timelminFIG. 1 .-Tile initial reaction in the autodecarboxylation of oxaloacetic acid in aqueous solutionscontaining NaOH and NaCl at 25°C ( I = 0.10 M)HIROO ITO, HIROSHI KOBAYASHI AND KENJI NOMIYA 115oxaloacetic acid to be completely decarboxylated, the rate constant can be determinedfrom the slope of the straight line obtained for the first 20min. The rate constantof the first order reaction is given by k = l / t 1n(pco/’pco -p), wherep is the manometerreading. The rate constants obtained by two different methods were coincidentwithin experimental error. The rate constants of the autodecarboxylation of oxalo-acetic acid were determined at a variety of pH (table 1).TABLE TH THE RATE CONSTANTS OF THE AUTODECARBOXYLATION OF OXALOACETIC ACID INAQUEOUS SOLUTION (I = 0.10) AT 25°CPH 1.8 2.1 2.8 3.4 3.7k x 105/s-l 1.9 2.0 4.7 4.3 2.9THE SPECIES INVOLVED IN THE AUTODECARFJOXYLATION.-~~ aqueous Solution, anequilibrium is present between the species SH2 (oxaloacetic acid), SH- and S2-(oxaloacetate dianion).K1SH2 + SH-+H+ (3)SH- + S2-+H+ (4)K2The concentrations of the species SH2, SH- and S2- are evaluated using Pedersen’svalues of Kl and K2.Fig. 2 shows the concentrations of the existing species in solu-tion calculated as a function of pH.PHFIG. 2.-pH dependence of the existing species in aqueous solutions of oxaloacetic acid : -, [SH-] ; -.- , [S-1; -.. - , [SH21.The reaction rate is given as a function of hydrogen ion concentration [H+] andthe autodecarboxylation rate constants of SH2, SH- and S2- denoted by ko, k ,and k2, respectively :v = k[S]= k[Sl*= k,[SH2] + k,[SH-] + k2[S2-]= (ako + bkl + ck2)[Sl0, (5116 DECARBOXYLATION OF OXALOACETIC ACID1[H+12/K,K, + [H+]/K2 + 1 'c =Since a, b and c can be evaluated from the hydrogen ion concentration, the rateconstants, ko, k , and k2 are calculated from the rates obtained for a variety of hydro-gen ion concentrations. This indicates that the species directly involved in the auto-carboxylation is SH- but not SH2 and S2-. Since ko and k2 are negligibly small,eqn (5) is approximatelyIn fact, plots of k against b gave a straight line passing through the origin, withinexperimental error (fig.3). From the slope of the line, the rate constant kl isZJ = k[S], = bkl[SIo. (6)5.8 x 10-5 s-1.b x 103FIG. 3.-Plots of the first-order rate constant of the autodecarboxylation of oxaloacetic acid againstconcentration of hydrogen oxaloacetate anion in units of the initial concentration of oxaloaceticacid, [SH-] = b[S],.N.M.R. SPECTRA OF OXALOACETIC ACID IN AQUEOUS SOLUTION.-N.m.r. spectra Ofoxaloacetic acid in aqueous solutions, which were pH adjusted by adding solutionsof HC1O4 and NaOH, were measured using DSS as internal reference.The -CH2- proton resonance of oxaloacetic acid is observed at 3.71 p.p.m.The methyl proton resonance of the decarboxylation product pyruvic acid appears at2.35p.p.m.and the methyl proton resonance of the hydrate of pyruvic acid, 2,2-dihydroxypropionic acid, at 1.46 p.p.m.Oxaloacetic acid is present as the univalent anion SH- in aqueous solution atpH N 3. Two n.m.r. signals of comparable intensity were observed at 3.71 and2.35 p.p.m. : a signal was observed at 1.46 p.p.m. about 2 h after dissolution. Whenthe autodecarboxylation of oxaloacetic acid was almost completed, keeping the solu-tion at about 50°C for more than 24 h, an extremely intense signal characteristic ofpyruvic acid was observed at 2.35 p.p.in., while only weak signals were observed at3.71 and 1.46 p.p.m.In an aqueous solution at pH 6-7, oxaloacetic acid is present mainly as thedivalent anion S2-. An intense signal was observed at 3.71 p.p.m., while only avery weak signal appeared at 2.35 p.p.m.The ratio of intensities of these two signalswas not changed many hours after dissolution. No appreciable production ofpyruvic acid was detected. In an aqueous solution of pH < 1, oxaloacetic acid ismainly present as the undissociated acid SH,. An intense signal characteristic of thHIROO ITO, HIROSHI KOBAYASHI AND KENJI NOMIYA 117undissociated acid was observed at 3.00 p.p.m., while the signal at 3.71 p.p.m. wasjust a trace. Another signal at 1.70 p.p.m. seems to be characteristic of the solutionof pH < 1, though it has not been well resolved from noise. No signal appeared at1.46 or 2.35 p.p.m. for many hours after dissolution. Thus, no pyruvic acid wasproduced.These observations on the n.m.r.spectra support the conclusion from mano-metric studies that the main species involved in the autodecarboxylation is SH- but notSH2 and S2-.A pH-dependent equilibrium between pyruvic acid and its hydrate 2,2-dihydroxy-propionic acid in aqueous solution has been established by physical meas~rernents.~*N.m.r. studies also revealed that the hydration of the carbonyl group in pyruvic acidoccurs only when the carboxyl group is protonated. Pyruvate anion is not hydrated.The details of the n.m.r. studies will be presented elsewhere. In the autodecarboxyla-tion of oxaloacetic acid, however, one r81e of the carboxyl proton in the activespecies SH- is in the hydration process prior to decarboxylation. When a protonattached to the second carboxyl group (the group furthest from the carbonyl) istransferred to the carbonyl oxygen, the coordinating water at the carbonyl carbonmust be permanently deformed.This gives rise to proton attack by this water onthe -CH2- group followed by decarboxylation. In an aqueous solution at pH 3,dehydration of the 2,2-dihydroxypropionate anion to pyruvate is very fast. In theinitial stage of the reaction, the n.m.r. signal of the -CH2- protons was observedat 3.71 p.p.m. When the reaction was almost complete, only a weak signal of theCH3- protons in 2,2-dihydroxypropionate was observed at 1.46 p.p.m. whereas anintense signal of pyruvate anion was detected at 2.35 p.p.m. There is no greatdifference between CH,- proton resonance in pyruvate anion and undissociatedpyruvic acid and thus n.m.r.gives no further information on the predominant formof pyruvic acid in solution. Acid dissociation measurements, however, suggeststhat the pyruvate anion is predominant in pH 3 solution.2In the case of an aqueous solution of pH < 1, both carboxyl groups are protonated.Hydrogen-bonding of the y-carboxyl proton with the carbonyl oxygen is strong andthus a hydrogen-bonded chelate is formed as in undissociated pyruvic acid, givingrise to hydration. The water coordinated to the carbonyl is again deformed andthen two successive proton transfers give a,a-dihydroxysuccinic acid ; an intensesignal around 3.00 p.p.m. is assigned to its -CH2- protons. Since the a-hydroxygroups thus formed can accept no more protons, the second carboxyl group even ifprotonated cannot be decarboxylated.In an aqueous solution of pH 6-7, due to the divalent negative charge of the mainspecies S2- an orientation effect of solvent water around the charged species arises.However, no further hydration of the central carbonyl is promoted.In fact, nodecarboxylation was observed in the solution.Esterification also inhibits decarboxylation.DECARBOXYLATION IN THE PRESENCE OF HYDRATED METAL IONSIn aqueous solution, the decarboxylation of oxaloacetic acid is accelerated byhydrated metal ions. Prior to decarboxylation, a metal chelate with the oxaloacetatedivalent anion is formed.6* ' The structure of the metal chelate involved in themetal-ion promoted reaction has been established I f as (I) (eqn (1)).The free car-boxyl group is not protonated under the experimental conditions employed. Metalchelate formation is so fast that it could not be detected without employing a fasttechnique such as stopped flow. A rate for chelate formation lo3 times higher thanthe rate of decarboxylation of the chelating oxaloacetate was estimated118 DECARBOXYLATION OF OXALOACETIC ACIDIn terms of the Michaelis-Menten reaction mechanism, the decarboxylatien ofoxaloacetate S into pyruvate P in the presence of hydrated metal ion E is describedas follows :k+ Ik- 1ES + E+P+C02.SH-+E f ES+H+kProvided there is a constant concentration of intermediate metal chelate, the reactionvelocity is given byorIf k+,[SH-] 9 k-,[H+]+k, v = kp], and the rate constant k is determined as a ratioof the initial velocity and the initial concentration [El,.The condition k+,[SH-] 9k-,[H+] + k is experimentally satisfied when the initial concentration of oxaloaceticacid [Sl0 i s much greater (approximately 100 times) than that of hydrated metal ion[Elo. The decarboxylation of oxaloacetic acid in an aqueous solution (PH 3.7 andionic strength 0.1 M) containing hydrated aluminium ions shows excellent linearitybetween manometer readings and reaction time as shown in fig. 4. This indicatesthat the mean veIocity is constant in the initial stage of the reaction. When the firstterm of the right-band-side of eqn (10) is negligible, the mean velocity can be keptIn t - 5 10 15timelminFIG.4.-The initial reaction in the decarboxylation of oxaloacetic acid in the presence of hydratedaluminium ion at various temperatures. [S], = 2 x mol l.-I, [El, = 6 x loe4 mol I.", pH =3.7 and I = 0.10HIROO ITO, HIROSHI KOBAYASHI AND KENJI NOMIYA 119constant for a period of time even if the concentration [SH-] is decreasing. Fig. 4also shows the temperature dependence of the rate. Plots of the initial rate againstthe initial concentration of metal ion [El, give a straight line passing through theorigin. The slope of the straight line D = k[EIo gives the real rate constant k. Therate constants of the decarboxylation of oxaloacetic acid for various hydrated metalions present in aqueous solution of pH 3.7 and ionic strength 0.1 M are summarizedin table 2.TABLE Z-RATE CONSTANTS OF THE METAL-ION CATALYZED DECARBOXYLATIONS OF OXALO-ACETIC ACID (S-')20°C 25°C 30°Cauto.La3+Gd3+Y3+Lu3+CU2+Ni2+Zn2+co2+~ 1 3 +2 .6 ~ 10-51.6 x 10-31 . 4 ~ 10-38 . 3 ~ 10-45 . 9 ~ 10-47 . o ~ 10-32 . 8 ~ 10-31 . 6 ~ 10-31 . 6 ~ 10-3-2 . 9 ~ 10-52 . 4 ~ 10-32 . 3 ~ 10-31.4 x 10-39.1 x 10-43 . 7 ~ 10-33 . o ~ 10-32 . 2 ~ 10-31.3 x1.3 x3 . o ~ 10-54.1 x 10-34 . 2 ~ 10-32.5 x 10-31.5 x 10-37 . 3 ~ 10-35 . 5 ~ 10-34 . 4 ~ 10-32 . 6 ~2.2xTEMPERATURE DEPENDENCE OF THE RATE CONSTANTS.-PlOtS O f log k against 1/Tgave a straight line confirming an Arrhenius relationship k = A exp (-E/RT).The thermodynamic data of activation were evaluated from the temperature depend-ence of the rate constants assuming K = 1 in k = rc(kT/h) exp (AS* /It) exp (- AH* /RT).The results are summarized in table 3.Plots of AH* against TAS* (T = 298 K)for decarboxylation not only in the presence of various hydrated ions but also in theirabsence show excellent linearity. A compensation effect was observed between theactivation enthalpy and entropy terms as shown in fig. 5.TABLE 3.-THERMODYNAMIC DATA OF ACTIVATIONSA G * / @ ~ I mol-1)at 25°C AH*/@c~~ mol-1) AS*/(C~~ mol-1 K-1)auto.La3+Gd3+Y3+Lu3+CuZ+Niz+Zn2+coz+~ 1 3 +23.620.021.021 .o21.321.620.120.720.821.01.924.016.018.718.815.819.616.321.117.2- 72.9- 16.9- 7.6- 8.4- 19.3- 1.6- 14.80.9- 12.713.4THE EFFECTS OF METAL IoNs.-The decarboxylation of oxaloacetic acid is acceler-ated when oxaloacetic acid chelates with metal ions. The activation energy is reducedupon metal chelate formation.The activation entbalpy and entropy of auto-decarboxylation were obtained AH* = 1.9 kcal mol-1 and AS* = -72.9 cal mol-lK-l. A barrier to the autodecarboxylation reaction is the high entropy decreaseinvolved. Decarboxylation of a metal chelate complex, however, is acceleratedsince it is accompanied by some entropy-increasing processes. The enhancement o120 DECARBOXYLATION OF OXALOACBTIC ACIDthe rate due to chelation is due to a decrease in AG* arising from an increase inTASd* much larger than an increase in AH* up to 15-24 kcal mol-l.The metal-ionaccelerating effect in this particular reaction is essentially an entropy phenomenon.The compensation relationship shown by the plots of AH* against TAS*(T = 298 K) holds not only for the metal-ion promoted decarboxylation regardlessthe charge on the metal ion but also for autodecarboxylation, hence the reactionmechanisms must be similar. The observed activation entropy of autodecarboxyla-tion -72.9 cal mol-l K-l indicates that a number of solvent water molecules arecaptured or orientated by the activated state. However, another entropy increasingeffect is involved in the metal-ion promoted reaction. A tridentate chelate structurewith both carboxyl groups and the carbonyl oxygen coordinated to the metal ion isassumed as the activated state.For this to be achieved, some of the coordinated andoriented water molecules must be set free. An increase in entropy arising from theformation of a tridentate activated complex compensates for an activation entropydecrease for the decarboxylation. The enthalpy increase necessary for the bond-breaking and the release of orientated solvent water can be compensated for by amuch larger entropy increase.FIG. 5.-Plots of AH* against TAS* (T = 298 K ) .log KFIG. 6.-Plots of log K (stability) against log k (reactivity)HIROO ITO, HIROSHI KOBAYASHI A N D KENJI NOMIYA 121Higher polarization of the coordinating ligand is expected for tervalent thandivalent metal cations. However, the rates for the tervalent metal chelates aresmaller than those for the divalent metal chelates as seen in table 2.This mightprovide further evidence that the metal-ion accelerating effect is essentially an entropyphenomenon. Plots of log K at 298 K for metal chelate formations obtained byGelles and Nancollas * against log k at 298 K for the metal-ion promoted reactionsmeasured in the present work are linear as shown in fig. 6. Such a linear free energyrelationship obtained for a series of divalent metal ions is opposite in sign to a linearfree energy relationship obtained for a series of tervalent metal ions. As is clearlyshown by Nanc~llas,~ a metal chelate formed by oxygen donors such as in carboxy-lates is stabilized mainly by a large entropy term but destabilized by the enthalpyterm. Since a common compensation relationship between AH* and TAS* wasobserved, different linear free energy relationships obtained for divalent transitionmetal ions and tervalent rare earth metal ions might arise from a difference in therelative contributions of entropy and enthalpy terms to the stability constants of themetal chelates.The authors thank Professors Kenji Tamaru and Iwao Yasumori for helpfuldiscussions.(a) L. Krampits and C. Werkman, Boichem. J., 1941,35,595 ; (b) H. A. Krebs, Biochem. J.,1942,36, 303 ; (c) S. Ochoa, J. Biol. Chem., 1948,174, 115 ; ( d ) A. Kornberg, S. Ochoa andA. H. Mehler, J. Biol. Chem., 1948, 174, 159 ; (e) J. F. Speck, J. Biol. Chem., 1949, 178, 315 ;(f) R. Steinberger and F. H. Westheimer, J. Amer. Chem. SOC., 1951, 73, 429 ; (9) E. Gellesand J. P. Clayton, Trans. F'raday SOC., 1956, 52, 353; (h) J. V. Rund and R. A. Robert,J. Amer. Chem. SOC., 1964, 86, 367; (i) J. V. Rund and K. G. Claus, J. Amer. Chem. Soc.,1967,89,2256 ; (j) J. V. Rund and K. G. Claus, Inorg. Chem., 1968,7,860 ; (k) M. Munakata,M. Matsui, M. Tabushi and T. Shigematsu, Bull. Chem. SOC. Japan, 1970,43, 114.K. J. Pedersen, Acta Chem. Scand., 1952,6,243.K. J. Pedersen, Acta Chem. Scand., 1952,6,285.H. Strehlow, 2. Elektrochem., 1962,66,392.M. Becker, Ber. Bunsenges. phys. Chem., 1964, 68,669.E. Gelles and R. W. Hay, J. Chem. SOC., 1958,3673. ' E. Gelles and A. Salama, J. Chem. SOC., 1958,3683.E. Gelles and G. H. Nancollas, Tram. Farahy Soc., 1956, 52,98.G. H. NancolIas, Interactions in EZectrolyte Solutions (Elsevier, Amsterdam, 1966), p. 214

ISSN:0300-9599    

DOI:10.1039/F19736900113

出版商:RSC    

年代:1973

数据来源: RSC

摘要:

Studies in Ion Solvation in Non-aqueous Solvents and Their Aqueous Mixtures Part 15.l-Free Energies of Transfer of Cadmium Chloride from Water to 10-40 % (wlw) Methanol-Water Mixtures at 25°C; Dissociation of CdCl+; Use of Dilute Cadmium Amalgam Electrodes. The Group I1 Cations. BY D. FEAKINS,* ALAN s. WILLMO~T AND ANN R. WILLMOTT Dept. of Chemistry, Birkbeck College, Malet Street, London WCl Received 19th July, 1972 The cell Cd(Hg) I CdC12 I AgC1-Ag has been used to obtain the standard molar free energies of transfer, AGt", of cadmium chloride from water to 10 %, 20 % and 40 % (w/w) methanol-water mixtures. Cadmium chloride is a weak electrolyte. This was allowed for in coniputing the standard e.m.f. values of the cell; dissociation constants of CdClf were also determined. The behaviour of the Group I1 cations, CdZ+, Zn2+, Sr2+ and Ba2+, in the methanol-water system is discussed in terms of the " acid-base " theory and their " hardness " or " softness '' as Lewis acids, and they are briefly compared with the univalent cations of Group I.In this part we describe the determination of AG:, the molar free energy of transfer, from water to 10, 20 and 40 % (w/w) methanol-water mixtures, at 25"C, for Cd2+-2C1-.$ AG," for Zn2+-2C1-, up to 32.2 % (w/w) of methanol, is obtainable from Corsaro and Stephens's data.2 A previous part gave AGg-values for Sr2+-2Cl- and Ba2+-2C1-. Experimental difficulties involved in getting information on the remaining Group I1 cations have yet to be overcome, but limited comparisons of the do ions from the A sub-group with the d10 ions from the B sub-group and of the Group I1 with the Group I cations are now possible .A saturated two-phase cadmium amalgam is well established as an electrode reversible to cadmium ions ; it is particularly valuable as a reference electrode because the activity of the metal in the amalgam is uniquely defined at constant temperature and pressure. For example, Harned and Fitzgerald used it in their study of the CdC1,-water system. A saturated amalgam is not essential for the present purpose. For suppose that we have a stock of a dilute amalgam of a particular composition, which need not be known exactly. Using some of it, we can measure the e.m.f. values of cell (I), E, Cd(Hg) I CdCl2 I AgC1-Ag (1) with water as solvent, at various molalities.By a suitable extrapolation we can obtain the standard molal e.m.f. appropriate to this amalgam, "EA, in water. We * present address : Dept. of Chemistry, University College, Belfield, Dublin 4. f present address : University Chemical Laboratory, South Parks Road, Oxford OX1 3QR. +c This nomenclature indicates that the transfer is of the independent ion constituents ; to speak of the transfer of CdC12 could imply transfer of the neutral species, which can exist in solution. As in the early papers in this Series we shall now use this nomenclature even where no ambiguity is likely. 122D. FEAKINS, A . S . WILLMOTT AND A . R . WILLMOTT 123 can then take more of the same amalgam, repeat the procedure using a niethanol- water mixture as solvent, and so obtain the standard molal e.m.f.sEg for this solvent. Whilst each Eg depends on the composition of the amalgam, the difference between them, AEE = wEg-sEg, does not, and it is AE: that we want. The great merit of the dilute amalgams, from our point of view, was that we already had a cell which had been used with alkaline-earth amalgamsY3 and which could be used without further modification for the dilute cadmium amalgams. In any event it is probably easier to handle dilute rather than saturated amalgams. The e.m.f. measurements were obtainable by a rapid continuous method and were comparable in accuracy with those obtainable with saturated amalgams. Cadmium amalgam does not react with water or aqueous methanol and can be used as an electrode in the form of a stationary pool.As a check we also determined the cell e.m.f. values for flowing electrodes; the two sorts of electrode gave results which agreed to within 50.05 mV of each other. The experimental cell was suitable for measurements either on the single cell (I) or a double cell (II) (11) In principle AEg could have been obtained directly from the e.m.f. values of cell (11), AE say, and it would not then have been necessary to keep the composition of the amalgam constant over a sequence of measurements. Cadmium chloride, however, is a weak electrolyte, and a complex iterative procedure (see below) is required to get from e.m.f. measurements at real concentrations to standard-state values. Not only would this be more complex for AE than for E but also measure- ment of AE rather than E would halve the amount of information available to deter- mine the optimum values of the various parameters in the analysis without halving their number.Nevertheless it was convenient experimentally to set up the double cell, with m, N m,. The e.m.f values of each single-cell component were measured as nearly simultaneously as possible, and as a check values of A23 were also measured directly. The dilute amalgam was stored in an atmosphere of dry nitrogen. One batch of amalgam (A) was enough for measurements on two double systems, water-10 % methanol and water-40 % methanol. We thus had two sets of e.m.f. values at various molalities of cadmium chloride in water as solvent; these were separated by about one month. The differences in e.m.f. between the sets were randomly distributed with respect to sign, and their mean was 0.08 mV.This suggests that any variation in the composition of the amalgam over the few days required to study a complete double system would be completely negligible. A second batch of amalgam (B) was used for the double system water-20 % methanol ; its composition differed slightly from that of A. Ag-AgC1 I CdC12(ms),S I Cd(Hg) I CdClz(mw),W 1 AgC1-Ag EXPERIMENTAL Cells for alkali-metal amalgams were described in Part 10 ; the modifications necessary for alkaline earth amalgams were given, with a diagram of the cell, in Part 13.3 This is the cell used in the present work. Each single cell (I) contained a silver-silver chloride electrode and an amalgam dropper of bore diameter either 0.15 or 0.2 mm, the tip of which was seated well down inside a cup 6 mm in diameter.These dimensions were not critical in the present work. The droppers to each single cell were fed from a common reservoir of amalgam. The cell and ancillary gas lines were evacuated at the beginning of the experiment then filled with nitrogen. The droppers and cups were filled with cadmium amalgam from the124 STUDIES I N ION SOLVATION I N NON-AQUEOUS SOLVENTS reservoir, the solutions were admitted to each single cell, and a slow stream of nitrogen was passed through each. At this stage the amalgam in each cup constituted a stationary electrode. The amalgam in each dropper made electrical connection with each pool electrode and a platinum wire sealed through the side of the dropper made the external connection to the voltmeter. The e.m.f.values of each cell were read within a few seconds of each other every 3 min. Equilibrium was considered to have been reached, usually in about 1 h, when the e.m.f. of the cell was constant to k0.02 mV for 30 min. The equilibrium e.m.f. could be maintained for several hours but normally we proceeded with the next stage of the experiment immediately. The same value of the e.m.f. was attained at once if the amalgam surface was renewed ; this suggested that the equilibration time was required solely for the silver-silver chloride electrodes. The cell e.m.f. values were then measured with flowing electrodes. Whilst for the experiments with the stationary pool electrodes nitrogen was bubbled through the solutions from the base of the cell, the gas was now by-passed over the surface of the solutions to avoid disturbances in the solution which could cause amalgam to be thrown onto the silver- silver chloride electrodes. As in the experiments with calcium and barium electrode^,^ amalgam was then run continuously into the cups.It spilled over from them via lips which directed the overspill away from the silver-silver chloride electrodes. Difficulties of electrical connection now meant that readings on each cell had to be taken in separate sequences. They were taken every 5 f for several minutes; equilibration was normally immediate, and the single cell values agreed to better than +0.05mV with those taken using the stationary electrodes. As a final check, the e.m.f.of the double cell, AE, was measured using the flowing electrodes. It always agreed to within kO.02 mV with the value expected from the observations on the single cells. In a series of measurements at different molalities the most concentrated solution was always studied first, and the remainder in order of decreasing molality. When all measure- ments had been taken on a particular solution, the cell was washed in situ, as before s with the next solution. In contrast to the previous pra~tice,~ however, the amalgam was not replaced by mercury during this operation, although fresh amalgam was admitted to the droppers and cups before e.m.f. readings were taken on the new solution. Up to six different solutions could be studied in a continuous operation in one day, but the cells and ancillary apparatus were washed out with water before the next day’s measurements.The digital voltmeter described previously was used for the e.rn.f. measurements. A concentrated cadmium amalgam (40 gm Cd/100 ml Hg) was prepared by the addition of cadmium (Johnson Matthey’s spectroscopically pure grade) to mercury (Scientific Supplies ; triple distilled ; 99.999 % pure) under nitrogen. This was diluted to approximately 0.003 m. Anhydrous cadmium chloride (Johnson Matthey’s spectroscopically pure) was dissolved in water to give a stock solution of known weight composition. This solution was slightly opalescent, owing to the hydrolysis of the cadmium ion. Sufficient hydrochloric acid of known weight composition was then added to give a clear solution. The stock solution was used to prepare all the cell solutions, in which therefore the stoichiometric ratio of Cd2+ to H+ concentration was constant.Silver-silver chloride electrode^,^ conductivity water and methanol were as before. RESULTS AND CALCULATIONS The e.m.f. of cell I is shown as a function of molality in table 1. Polarographic work has shown that at the ionic strengths and solvent compo- sitions used here, the only associated species present in solution in significant amount is CdCl+ ; CdC12, CdCl,, CdClz’ and CdC13- are found at higher ionic strengths and methanolic concentrations. Harned and Fitzgerald’s formalism is therefore useful : m mol of CdC12 are dissociated “ initially ” into m g-ion each of CdCl+ and C1- ; this is “ followed ” by partial dissociation of CdCl+ into Cd2+ and C1-.CdCl+fCd2+ + Cl-. (1)D. FEAKINS, A . S . WILLMOTT AND A . R. WILLMOTT 125 In our experiments, the stoichiometric concentration of added HC1 (see above) is Am; iz is constant for all m. If the degree of dissociation of CdCl+ is a then [Cd2+] = am ; [CdCI+] = (1 -a)m ; [Cl-] = (I +a+A)m and [Ht-] = Am. The molal dissociation constant of CdCI+, &, is given by eqn (2) a(l + a + d)mF(y) Kd = 1-C% where F(?> = YCdYCI/?CdCI* TABLE 1.-E.m.f. VALUES OF CELL (I), E/V AT VARIOUS MOLALITIES OF CADMIUM CHLORIDE rn/mol kg-’. set I amalgam A aqueous solutions 10 ”/, MeOH-H2O solutions rnXv E rns E 0.020 014 0.015 005 0.009 999 2 0.006 999 6 0.005 000 4 0.002 999 5 0.002 000 2 0.000 999 92 0.000 699 85 O.OO0 500 06 0.678 88 0.686 38 0.697 30 0.707 34 0.717 13 0.732 87 0.745 67 0.768 69 0.780 99 0.792 87 0.020 019 0.015 009 0.010 007 0.007 000 0 0.005 000 4 0.003 001 2 0.001 999 1 0.000 995 18 O.OO0 699 82 0.000 499 82 0.670 15 0.677 35 0.688 06 0.697 66 0.707 22 0.722 68 0.735 20 0.757 94 0.769 97 0.781 72 set I1 amalgam B aqueous solutions rnw E 0.040 035 0.662 97 0.019 951 0.680 15 0.010 004 0.698 51 0.007 003 9 0.708 55 0.004 998 3 0.718 31 0.003 000 0 0.733 94 0.002 000 0 0.746 76 O.OO0 999 91 0.769 79 0.OOO 700 27 0.782 13 0.000 500 01 0.794 14 20 % MeOH-H20 solutions ma E 0.040 025 0.646 59 0.020 001 0.662 69 0.010 008 0.679 87 0.007 008 9 0.689 50 0.005 006 3 0.698 61 0.003 000 0 0.713 31 0.002 000 1 0.725 49 0.001 000 8 0.747 41 0,000 700 10 0.759 31 O.OO0 500 00 0.770 86 set I11 amalgam A aqueous solutions 40 % MeOH-H20 solutions rnw E ma E 0.020 009 0.015 005 0.009 999 2 0.006 999 6 0.005 OOO 4 0.002 999 5 0.002 000 2 0.OOO 999 92 0.000 699 85 O.OO0 500 06 0.678 98 0.686 45 0.697 36 0.707 42 0,717 19 0.732 64 0.745 57 0.768 62 0.780 99 0.792 57 0.020 059 0.014 994 0.009 999 7 0.007003 1 0.004 999 9 0.003 005 3 0.002 002 8 0.001 001 7 0.000 700 94 0.000 499 95 0.645 73 0.652 17 0.661 71 0.670 50 0.679 00 0.692 44 0.703 47 0.723 87 0.734 96 0.745 75 All solutions contained HCI in concentration Am where 1 = 0.0022.126 STUDIES IN ION SOLVATION IN NON-AQUEOUS SOLVENTS deb ye-Hiickel theory gives eqn (3) for the mole fractional activity coefficient, fi, of an ion of valency zI : log fi = -Z;AJII(I + d ~ J I ) (3) where A and B are the molal Debye-Hiickel constants, d is the ion-size parameter and I the ionic strength, = m(l+2a+A). For F(y), application of eqn (3) gives -4AJI log F(Y) = 1 + 6~ JI -log (1 + 0.001MCrn). (4) M is the molecular weight or mean molecular weight of the solvent and Cm is the total molality of the ionic species.We may add an empirical term b,I to eqn (4) ; the second term of (4) will be very small for the present concentration range and may be absorbed, to a good approximation, in blI, giving Eqn (6) gives the e.m.f. of cell (I) E = Ek - (k/2) log acda& ; E = E& - (k/2)(10g cc + 2 log( 1 + cc +A) + 3 log 1.1 - (3k/2)10g 7 * (6) E: and a are respectively the molal standard e.m.f. and activity; k = (log, 10) RT/F. Expanding eqn (6) gives (7) Following the arguments given above we may write for y k , the mean molal activity coefficient of CdC12 Combination of eqn (7) with eqn (8) gives (9) E"' = Eg-2kbzI.(10) 3kAJI E+(k/2)(10ga+2 log ( l + a + 4 + 3 log m)+l+dBJI = E;-$kb21 or, say, We decided to proceed by seeking values of & and b, (eqn (2) and (5)) that would give values of a, and hence I, such that the function E"' showed the least deviation from linearity in I, if h, in eqn (5) and (8) was kept constant at 5.0, the value suggested by Harned and Fit~gerald.~ It was thought adequate to vary bl between -2.0 and +2.0 kg inol-l for all systems. An appropriate preliminary range of values of Kd was selected ; for water, for example, with Kd known to be about 0.01 mol kg-', this was 0.005-0.015 mol kg-l. Using eqn (2) and ( 5 ) we could, by an obvious iterative procedure, find a and hence also ?, at the various rn for a particular combination of b, and &.These values were substituted in eqn (9); EA and b, were found by the method of least squares, and the fit to eqn (10) assessed by noting the residual sum of squares (RSS). The analysis was carried out first in a coarse and then in a fine range. The University of London computer was used, with a programme written in EXCHLF. For any bl value within the range -2.0 to +2.0 kg mol-I, there is a value of 1yd for which the RSS is a minimum and close to that expected from the experimental error estimated as k0.07 mV. These minimum RSS values for each b, were notD. FEAKINS, A . S . WILLMOTT AND A . R. WILLMOTT 127 significantly different from each other ; thus the result is a range of Kd values each with its associated b l , which fit the data equally well.Any variation in E& across this range is insignificant. Kd is known more accurately the weaker the electrolyte, i.e., the higher the concentration of methanol. This is the expected behaviour. Kd cannot be quoted more precisely than in table 2, without some apriori assumption about the values of 6, and b2. Table 2 shows combinations of b2, & and Ez values which fit the data for bl = -2.0 and +2.0, together with the mean E i . Association constants Kl( = Kc are also listed. Three sets of values for aqueous solutions are quoted; each represents measurements made concurrently with a set for one of the methanol-water mixtures.Sets I and I11 were made with the same amalgam (A); the difference in E: is within experimental error ; so too are the slight variations in the range of Kd in the three cases. Mean values of AE;, from which AGp values are readily calculated, are also in table 2. Harned and Fitzgerald reported agreement to only 0.8 mV between water 2.0 - 2.0 10 ”/, methanol 2.0 - 2.0 water 2.0 - 2.0 20 % methanol 2.0 - 2.0 water 2.0 - 2.0 40 ”/, methanol 2.0 - 2.0 TABLE 2a9b b2 I Ka I kg niol-1 mol kg-1 set I (amalgam A) 0.252 0.011 29 -0.921 0.012 24 0.032 0.009 35 - 1.135 0.010 08 set I1 (amalgam B) 0.227 0.011 33 - 0.929 0.012 44 0.002 0.006 29 - 1.169 0.006 57 set 111 (amalgam A) 0.184 0.011 42 -0.993 0.012 57 -0.579 0.002 70 - 1.780 0.002 74 88.6 0.512 81 81.6 0.512 83 0.512 82 107.0 0.501 06 99.2 0.501 07 0.501 07 AEZ = 0.011 75 88.3 0.513 97 80.4 0.513 98 0.513 98 159.0 0.489 03 152.2 0.489 02 0.489 03 AEA = 0.024 95 87.6 0.512 77 79.6 0.512 79 0.512 78 370 0.460 33 365 0.460 35 0.460 34 AEZ = 0.052 44 Harned and Fitzgerald (saturated amalgam) water 2.0 0.633 0.010 13 98.7 0.573 80 - 2.0 -0.622 0.011 19 89.3 0.573 83 0.573 82 mean mean mean mean mean mean mean a Acciiiracy of AEh is 20.15 mV or better.6 P. J. Reilly and R. H. Stokes, Austral. J. Chem., 1970, 23, 1397. give Kl = 85+ 1 for aqueous solution from measurements on cells withadded NaCl.128 STUDIES IN ION SOLVATION IN NON-AQUEOUS SOLVENTS experimental and calculated e.m.f. values in the CdC1,-water system. This seemed unexpectedly poor considering the high quality of the work known to emanate from Prof.Harned's school. The fit to eqn (10) was also only k0.8 mV when our pro- gramme was tried on their data, using, as they did, the whole range 0.0005 to 1.0 m ; but when we used the same range, namely 0.0005-0.02m, as in our work, the fit improved to kO.07 mV. This suggests that the formation of the higher cadmium- chloride complexes cannot be neglected at the higher electrolyte concentrations used by Harned and Fitzgerald. Although the range of Kd values (see table 2) for the two investigations of the aqueous system do not overlap, the difference between the two investigations is very small. In table 3 we compare e.m.f. values, at various molalities, for cell (I) with a saturated amalgam electrode, with those using unsaturated amalgam A as electrode.The difference between the e.m.f. values of the two cells at any molality should be constant. The mean deviation from the average difference is only 0.16 mV, approxi- mately the sum of the experimental errors on the two determinations. TABLE 3.-E.m.f. VALUES OF CELL (I) EIV 0.0005 0.853 9 0.001 0.829 97 0.002 0.807 01 0.005 0.778 51 0.007 0.768 62 0.01 0.758 46 0.02 0.739 76 m saturated amalgam 4 amalgam A a 0.792 9 0.768 71 0.745 67 0.717 13 0.707 34 0.697 30 0.678 89 difference 0.061 0 0.061 26 0.061 34 0.061 38 0.061 26 0.061 16 0.060 87 mean difference 0.061 18 Vk0.16 mV difference in EA 0.061 00 V ES 0.573 82 0.512 82 a set I. DISCUSSION In comparing univalent with bivalent metal chlorides we note that AG,"(M2+ 2C1-) = AG,"(M2+) + 2AG,"(Cl-) and AG,"(M'C1-) = AG:(M+)+AG:(Cl-)* By comparing +AG,"(M2+-2C1-) with AG,"(M+Cl-) we bring the data to a common baseline, namely AG,"(Cl-), and compare AG," per unit charge in the case of the cations.Table 4 thus shows $AG," for Cd2+-2C1-, from the present measurements, for Sr2+ 2Cl- and Ba2+- 2Cl-, from a previous part,3 and for Zn2+ 2C1-, interpolated from Corsaro and Stephens's data. Values for AG,O for the alkali-metal chlorides and for Ag+Cl- lo are included for comparison. Fig. 1 shows AGp(M+Cl-) or +AG;(M2+-2C1-) plotted against the reciprocal of cationic radius (rL1) for the 20 % mixture. Previous discussion in the Series l1 has led to the idea that all molecules in a methanol-water mixture, at least of relatively high methanol content, are more basic than water molecules in pure liquid water.It has been postulated that the formal charge on a solvent oxygen atom is greater in the mixtures than in water, and alsoD. FEAKINS, A. S . WILLMOTT AND A. R . WILLMOTT 129 that such an atom is " softer '' in the mixtures than in water.' The cation-oxygen interaction thus leads to a negative contribution to AG,". Along the series of alkali-metal ions, AGt(Li+) cAG,"(Na+) for transfers from water to all methanol-water mixtures; this is to be expected if for these small, relatively " hard " ions the cation-oxygen interaction is dominated by electrostatic forces. By contrast, however, AG,"(Cs+) < AG,"(Rb+) for all transfers, and this suggests that, even for ions having the do configuration, '' soft-soft " interactions are important for large cationic radii, r,.TABLE 4.-FREE ENERGIES OF TRANSFER (MOLAR SCALE) FROM WATER TO METHANOL-WATER MIXTURES AT 25°C 3AGP (M2+ 2Cl-)/cal wt.% MeOH Srz+ * 2C1- Baz+ 2CI- Znzf * 2C1- Cd2+ - 2C1- 10 - 392 (307)c 302 20 773 784 (635)c 634 40 - 1597 - 1330 AG," (M+C1-)/cal.d Ag+CI- CsfC1- Li+CI- Na+CI- K+CI- Rb+CI- 10 313 437 428 428 395 353 20 623 883 876 861 803 (695) 40 1268 1785 1804 1753 1651 (1440) a values & 3 cal or better except b & 12 cal C & 15 cal. d all values L- 3 cal or better. Bracketed figures are interpolated. 1 cal = 4.184 J. 800 - - c8 O r u 700- 3 a 6 0 0 - 1 : 1 I 0.5 1.0 1.5 rc- 1 1A-l FIG. 1.-Plot of AG;)(M+Cl-) or +AG,"(M2+ 2C1-) against r; l . Our first comparison will be of the alkaline-earth ions with the alkali-metal ions ; both types have noble gas configurations.K+(1.33A) and Ba2+(1.35A) haveapproxi- mately the same radii, and +AG,"(Ba2+2Cl-) may usefully be compared with AG;(K+CI-). Table 4 shows that AG,"(K+CI-) > +AG,"Ofa2+-2Cl-) for all transfers studied. Both Sr2+ and Ba2+ are expected to behave as " hard " acids,12 and their inter- actions with oxygen atoms of coordinated solvent molecules should be dominated by electrostatic forces. Thus the dominant term in AG,"(M2+) will be approximately of the form -2en2Aq/rc, where n2 is the coordination number of the ion, e is the protonic charge, the ionic charge being 2e, and -Aq is the change in the charge on the oxygen atoms of the coordinated solvent molecules in pure water and in the 1-5130 STUDIES IN ION SOLVATION I N NON-AQUEOUS SOLVENTS mixture respectively.The corresponding expression for a " hard " univalent ion is -enlAq/rc. Thus for a bivalent and a univalent ion of the same radius, if (i) the above terms dominate AG," and (ii) n1 = n2, we would expect AG;(M+Cl-) N +AG,"(M2+-2C1-) on the simple " acid-base " theory. We first note that K+ is not a typically hard ion, so that not only is there a relative stabilization of Ba2+ per unit charge compared with K+ in the mixtures, but also that this stabilization is probably less than would be observed relative to a " hard " univalent ion of the same radius. The relative stabilization of Ba2+ could arise from two causes, among others. (i) The coordination number of the bivalent ion, n2, is expected to be larger than that of the univalent ion, nl, even if r, is the same.This is a common observation in coordination chemistry and arises from the Pauling electroneutrality principle. (ii) Polarization of the coordinated solvent molecules in the ionic field would enhance their basicity i.e. raise I Aq 1 . Such polarization should be the greater in the higher ionic field of the bivalent ion. Let us now intercompare the four bivalent ions Sr2+, Ba2+, Zn2+ and Cd2f. Only at 20 % methanol can we compare Sr2+ and Ba2+. Because of the relatively low accuracy in AG,"(Sr2+2C1-) and the cIoseness of the radii of the two do ions, it is not possible to establish a meaningful trend in AG,O as a function of r,, although the results are not inconsistent with what would be expected for " hard " cations, i.e.AGi(Ba2+) > AG,0(Sr2+). Zn2+ is regarded by Pearson l 2 as being on the borderline between a " hard " and a " soft " acid; there is however, a marked increase in the " softness " of the cations as we pass down the IIB sub-group from Zn2+ to Hg2+. Although mercury (11) complexes are in general more stable than those of zinc or cadmium, whatever the " hardness " or " softness " of the ligand, there is less differentiation between the stabilities of zinc and cadmium complexes unless the ligand is very " soft " (e.g. I-).14 This suggests a tendency for the decrease in electrostatic interaction energy with increasing r, to be approximately matched by an increase in the " soft-soft " interaction energy between ion and ligand as we pass from Zn2+ to Cd2+.These factors appears also to be determining AG,"(M2+), since the difference between AG,O(Cd2+*2Cl-) and AG,"(Zn2+-2C1-) both at 10 % methanol and at 20 methanol is small, and we would expect the increase in " softness " of the solvating molecules to be relatively small on passing from water to the mixed solvents. Fig. 1 shows what happens if, at 20 % methanol for example, we class the relatively hard Zn2+ with Sr2+ and Ba2+ and assume a straight line relationship between $AG,"(M2+a2Cl-) and r;l for these three ions. The point for Cd2+*2C1- lies well below this line, and this is analogous to the corresponding stabilization of AgtCI- in methanol-water mixtures with respect to a1 kali-metal chlorides of similar rildiusy6 which has its origins in the " softer " character of the dlo ion.We thank the University of London and the Clothworkers' Company for grants (to A. R. W.). Part 14, D. Feakins and P. J. Voice, J.C.S. Faruday I, 1972,68, 1390. G. Corsaro and H. L. Stephens, J. Electrochem. Soc., 1957, 104,512. D. Feakins and Ann R. Willmott, J. Chem. SOC. A, 1970, 3121. H. S . Harned and M. E. Fitzgerald, J. Amer. Chem. Soc., 1936,58, 2624. H. P. Bennetto, D. Feakins and K. G. Lawrence, J. Chem. SOC. A, 1968, 1493. D. Feakins, K. G. Lawrence and R. P. 1'. Tomkins, J. Chem. Soc. A, 1967, 753. ' D. Feakins, J. Chem. Soc., 1961, 5308. * D. Feakins and P. Watson, J . Chent. Soc., 1963, 4636.D. FEAKINS, A . S . WILLMOTT AND A . R . WILLMOTT 'OD. Feakins, K. G. Lawrence, P. J. Voiceand A. R.Willmott, J. Chem. Soc. A , 1970, 837. l2 R. G. Pearson, in Suruey of Progress in Chemistry, ed. A. F. Scott, Vol. 5 (Academic Press, l3 R. S. Nyholm, Proc. Chem. SOC., 1961, 273. 131 I. Tur'yan and B. P. Zhantalai, Russian J. Inorg. Chem., 1960, 5, 848. D. Feakins and P. Watson, J. Chem. Soc., 1963,4734. New York, 1969), p. 1. L. G. Sillen and A. E. Martell, Stability Constants of Metal-ion Coiizplexcs (The Chemical Society, London, 1964).Studies in Ion Solvation in Non-aqueous Solvents and TheirAqueous MixturesPart 15.l-Free Energies of Transfer of Cadmium Chloride from Water to 10-40 %(wlw) Methanol-Water Mixtures at 25°C; Dissociation of CdCl+; Use of DiluteCadmium Amalgam Electrodes. The Group I1 Cations.BY D. FEAKINS,* ALAN s. WILLMO~T AND ANN R.WILLMOTTDept. of Chemistry, Birkbeck College, Malet Street, London WClReceived 19th July, 1972The cell Cd(Hg) I CdC12 I AgC1-Ag has been used to obtain the standard molar free energiesof transfer, AGt", of cadmium chloride from water to 10 %, 20 % and 40 % (w/w) methanol-watermixtures. Cadmium chloride is a weak electrolyte. This was allowed for in coniputing the standarde.m.f. values of the cell; dissociation constants of CdClf were also determined. The behaviourof the Group I1 cations, CdZ+, Zn2+, Sr2+ and Ba2+, in the methanol-water system is discussed interms of the " acid-base " theory and their " hardness " or " softness '' as Lewis acids, and theyare briefly compared with the univalent cations of Group I.In this part we describe the determination of AG:, the molar free energy oftransfer, from water to 10, 20 and 40 % (w/w) methanol-water mixtures, at 25"C,for Cd2+-2C1-.$ AG," for Zn2+-2C1-, up to 32.2 % (w/w) of methanol, is obtainablefrom Corsaro and Stephens's data.2A previous part gave AGg-values for Sr2+-2Cl- and Ba2+-2C1-.Experimentaldifficulties involved in getting information on the remaining Group I1 cations haveyet to be overcome, but limited comparisons of the do ions from the A sub-group withthe d10 ions from the B sub-group and of the Group I1 with the Group I cations arenow possible .A saturated two-phase cadmium amalgam is well established as an electrodereversible to cadmium ions ; it is particularly valuable as a reference electrode becausethe activity of the metal in the amalgam is uniquely defined at constant temperatureand pressure. For example, Harned and Fitzgerald used it in their study of theCdC1,-water system.A saturated amalgam is not essential for the present purpose.For suppose thatwe have a stock of a dilute amalgam of a particular composition, which need not beknown exactly. Using some of it, we can measure the e.m.f. values of cell (I), E,Cd(Hg) I CdCl2 I AgC1-Ag (1)with water as solvent, at various molalities. By a suitable extrapolation we canobtain the standard molal e.m.f. appropriate to this amalgam, "EA, in water. We* present address : Dept. of Chemistry, University College, Belfield, Dublin 4.f present address : University Chemical Laboratory, South Parks Road, Oxford OX1 3QR.+c This nomenclature indicates that the transfer is of the independent ion constituents ; to speakof the transfer of CdC12 could imply transfer of the neutral species, which can exist in solution.Asin the early papers in this Series we shall now use this nomenclature even where no ambiguity is likely.12D. FEAKINS, A . S . WILLMOTT AND A . R . WILLMOTT 123can then take more of the same amalgam, repeat the procedure using a niethanol-water mixture as solvent, and so obtain the standard molal e.m.f. sEg for this solvent.Whilst each Eg depends on the composition of the amalgam, the difference betweenthem, AEE = wEg-sEg, does not, and it is AE: that we want.The great merit of the dilute amalgams, from our point of view, was that wealready had a cell which had been used with alkaline-earth amalgamsY3 and whichcould be used without further modification for the dilute cadmium amalgams.Inany event it is probably easier to handle dilute rather than saturated amalgams.The e.m.f. measurements were obtainable by a rapid continuous method and werecomparable in accuracy with those obtainable with saturated amalgams.Cadmium amalgam does not react with water or aqueous methanol and can beused as an electrode in the form of a stationary pool. As a check we also determinedthe cell e.m.f. values for flowing electrodes; the two sorts of electrode gave resultswhich agreed to within 50.05 mV of each other.The experimental cell was suitable for measurements either on the single cell (I)or a double cell (II)(11)In principle AEg could have been obtained directly from the e.m.f.values ofcell (11), AE say, and it would not then have been necessary to keep the compositionof the amalgam constant over a sequence of measurements. Cadmium chloride,however, is a weak electrolyte, and a complex iterative procedure (see below) isrequired to get from e.m.f. measurements at real concentrations to standard-statevalues. Not only would this be more complex for AE than for E but also measure-ment of AE rather than E would halve the amount of information available to deter-mine the optimum values of the various parameters in the analysis without halvingtheir number.Nevertheless it was convenient experimentally to set up the double cell, withm, N m,.The e.m.f values of each single-cell component were measured as nearlysimultaneously as possible, and as a check values of A23 were also measured directly.The dilute amalgam was stored in an atmosphere of dry nitrogen. One batchof amalgam (A) was enough for measurements on two double systems, water-10 %methanol and water-40 % methanol. We thus had two sets of e.m.f. values atvarious molalities of cadmium chloride in water as solvent; these were separatedby about one month. The differences in e.m.f. between the sets were randomlydistributed with respect to sign, and their mean was 0.08 mV. This suggests thatany variation in the composition of the amalgam over the few days required to studya complete double system would be completely negligible.A second batch of amalgam (B) was used for the double system water-20 %methanol ; its composition differed slightly from that of A.Ag-AgC1 I CdC12(ms),S I Cd(Hg) I CdClz(mw),W 1 AgC1-AgEXPERIMENTALCells for alkali-metal amalgams were described in Part 10 ; the modifications necessaryfor alkaline earth amalgams were given, with a diagram of the cell, in Part 13.3 This isthe cell used in the present work.Each single cell (I) contained a silver-silver chloride electrode and an amalgam dropperof bore diameter either 0.15 or 0.2 mm, the tip of which was seated well down inside a cup6 mm in diameter.These dimensions were not critical in the present work. The droppersto each single cell were fed from a common reservoir of amalgam.The cell and ancillary gas lines were evacuated at the beginning of the experiment thenfilled with nitrogen.The droppers and cups were filled with cadmium amalgam from th124 STUDIES I N ION SOLVATION I N NON-AQUEOUS SOLVENTSreservoir, the solutions were admitted to each single cell, and a slow stream of nitrogenwas passed through each. At this stage the amalgam in each cup constituted a stationaryelectrode. The amalgam in each dropper made electrical connection with each pool electrodeand a platinum wire sealed through the side of the dropper made the external connectionto the voltmeter. The e.m.f. values of each cell were read within a few seconds of eachother every 3 min. Equilibrium was considered to have been reached, usually in about 1 h,when the e.m.f.of the cell was constant to k0.02 mV for 30 min. The equilibrium e.m.f.could be maintained for several hours but normally we proceeded with the next stage ofthe experiment immediately. The same value of the e.m.f. was attained at once if theamalgam surface was renewed ; this suggested that the equilibration time was required solelyfor the silver-silver chloride electrodes.The cell e.m.f. values were then measured with flowing electrodes. Whilst for theexperiments with the stationary pool electrodes nitrogen was bubbled through the solutionsfrom the base of the cell, the gas was now by-passed over the surface of the solutions toavoid disturbances in the solution which could cause amalgam to be thrown onto the silver-silver chloride electrodes.As in the experiments with calcium and barium electrode^,^amalgam was then run continuously into the cups. It spilled over from them via lipswhich directed the overspill away from the silver-silver chloride electrodes. Difficultiesof electrical connection now meant that readings on each cell had to be taken in separatesequences. They were taken every 5 f for several minutes; equilibration was normallyimmediate, and the single cell values agreed to better than +0.05mV with those takenusing the stationary electrodes. As a final check, the e.m.f. of the double cell, AE, wasmeasured using the flowing electrodes. It always agreed to within kO.02 mV with thevalue expected from the observations on the single cells.In a series of measurements at different molalities the most concentrated solution wasalways studied first, and the remainder in order of decreasing molality.When all measure-ments had been taken on a particular solution, the cell was washed in situ, as before s withthe next solution. In contrast to the previous pra~tice,~ however, the amalgam was notreplaced by mercury during this operation, although fresh amalgam was admitted to thedroppers and cups before e.m.f. readings were taken on the new solution. Up to six differentsolutions could be studied in a continuous operation in one day, but the cells and ancillaryapparatus were washed out with water before the next day’s measurements.The digital voltmeter described previously was used for the e.rn.f. measurements.A concentrated cadmium amalgam (40 gm Cd/100 ml Hg) was prepared by the additionof cadmium (Johnson Matthey’s spectroscopically pure grade) to mercury (Scientific Supplies ;triple distilled ; 99.999 % pure) under nitrogen.This was diluted to approximately 0.003 m.Anhydrous cadmium chloride (Johnson Matthey’s spectroscopically pure) was dissolvedin water to give a stock solution of known weight composition. This solution was slightlyopalescent, owing to the hydrolysis of the cadmium ion. Sufficient hydrochloric acid ofknown weight composition was then added to give a clear solution. The stock solutionwas used to prepare all the cell solutions, in which therefore the stoichiometric ratio ofCd2+ to H+ concentration was constant.Silver-silver chloride electrode^,^ conductivity water and methanol were as before.RESULTS AND CALCULATIONSThe e.m.f. of cell I is shown as a function of molality in table 1.Polarographic work has shown that at the ionic strengths and solvent compo-sitions used here, the only associated species present in solution in significant amountis CdCl+ ; CdC12, CdCl,, CdClz’ and CdC13- are found at higher ionic strengthsand methanolic concentrations.Harned and Fitzgerald’s formalism is therefore useful : m mol of CdC12 aredissociated “ initially ” into m g-ion each of CdCl+ and C1- ; this is “ followed ”by partial dissociation of CdCl+ into Cd2+ and C1-.CdCl+fCd2+ + Cl-.(1D. FEAKINS, A . S . WILLMOTT AND A . R. WILLMOTT 125In our experiments, the stoichiometric concentration of added HC1 (see above)is Am; iz is constant for all m.If the degree of dissociation of CdCl+ is a then[Cd2+] = am ; [CdCI+] = (1 -a)m ; [Cl-] = (I +a+A)m and [Ht-] = Am.The molal dissociation constant of CdCI+, &, is given by eqn (2)a(l + a + d)mF(y) Kd =1-C%where F(?> = YCdYCI/?CdCI*TABLE 1.-E.m.f. VALUES OF CELL (I), E/V AT VARIOUS MOLALITIES OF CADMIUM CHLORIDErn/mol kg-’.set I amalgam Aaqueous solutions 10 ”/, MeOH-H2O solutionsrnXv E rns E0.020 0140.015 0050.009 999 20.006 999 60.005 000 40.002 999 50.002 000 20.000 999 920.000 699 85O.OO0 500 060.678 880.686 380.697 300.707 340.717 130.732 870.745 670.768 690.780 990.792 870.020 0190.015 0090.010 0070.007 000 00.005 000 40.003 001 20.001 999 10.000 995 18O.OO0 699 820.000 499 820.670 150.677 350.688 060.697 660.707 220.722 680.735 200.757 940.769 970.781 72set I1 amalgam Baqueous solutionsrnw E0.040 035 0.662 970.019 951 0.680 150.010 004 0.698 510.007 003 9 0.708 550.004 998 3 0.718 310.003 000 0 0.733 940.002 000 0 0.746 76O.OO0 999 91 0.769 790.OOO 700 27 0.782 130.000 500 01 0.794 1420 % MeOH-H20 solutionsma E0.040 025 0.646 590.020 001 0.662 690.010 008 0.679 870.007 008 9 0.689 500.005 006 3 0.698 610.003 000 0 0.713 310.002 000 1 0.725 490.001 000 8 0.747 410,000 700 10 0.759 31O.OO0 500 00 0.770 86set I11 amalgam Aaqueous solutions 40 % MeOH-H20 solutionsrnw E ma E0.020 0090.015 0050.009 999 20.006 999 60.005 OOO 40.002 999 50.002 000 20.OOO 999 920.000 699 85O.OO0 500 060.678 980.686 450.697 360.707 420,717 190.732 640.745 570.768 620.780 990.792 570.020 0590.014 9940.009 999 70.007003 10.004 999 90.003 005 30.002 002 80.001 001 70.000 700 940.000 499 950.645 730.652 170.661 710.670 500.679 000.692 440.703 470.723 870.734 960.745 75All solutions contained HCI in concentration Am where 1 = 0.0022126 STUDIES IN ION SOLVATION IN NON-AQUEOUS SOLVENTSdeb ye-Hiickel theory gives eqn (3) for the mole fractional activity coefficient,fi, of an ion of valency zI :log fi = -Z;AJII(I + d ~ J I ) (3)where A and B are the molal Debye-Hiickel constants, d is the ion-size parameter andI the ionic strength, = m(l+2a+A). For F(y), application of eqn (3) gives-4AJIlog F(Y) = 1 + 6~ JI -log (1 + 0.001MCrn).(4)M is the molecular weight or mean molecular weight of the solvent and Cm is thetotal molality of the ionic species. We may add an empirical term b,I to eqn (4) ;the second term of (4) will be very small for the present concentration range and maybe absorbed, to a good approximation, in blI, givingEqn (6) gives the e.m.f. of cell (I)E = Ek - (k/2) log acda& ;E = E& - (k/2)(10g cc + 2 log( 1 + cc +A) + 3 log 1.1 - (3k/2)10g 7 *(6)E: and a are respectively the molal standard e.m.f. and activity; k = (log, 10) RT/F.Expanding eqn (6) gives(7)Following the arguments given above we may write for y k , the mean molal activitycoefficient of CdC12Combination of eqn (7) with eqn (8) gives(9)E"' = Eg-2kbzI.(10)3kAJIE+(k/2)(10ga+2 log ( l + a + 4 + 3 log m)+l+dBJI = E;-$kb21or, say,We decided to proceed by seeking values of & and b, (eqn (2) and (5)) that wouldgive values of a, and hence I, such that the function E"' showed the least deviationfrom linearity in I, if h, in eqn (5) and (8) was kept constant at 5.0, the value suggestedby Harned and Fit~gerald.~It was thought adequate to vary bl between -2.0 and +2.0 kg inol-l for allsystems. An appropriate preliminary range of values of Kd was selected ; for water,for example, with Kd known to be about 0.01 mol kg-', this was 0.005-0.015 mol kg-l.Using eqn (2) and ( 5 ) we could, by an obvious iterative procedure, find a andhence also ?, at the various rn for a particular combination of b, and &.Thesevalues were substituted in eqn (9); EA and b, were found by the method of leastsquares, and the fit to eqn (10) assessed by noting the residual sum of squares (RSS).The analysis was carried out first in a coarse and then in a fine range. TheUniversity of London computer was used, with a programme written in EXCHLF.For any bl value within the range -2.0 to +2.0 kg mol-I, there is a value of1yd for which the RSS is a minimum and close to that expected from the experimentalerror estimated as k0.07 mV. These minimum RSS values for each b, were noD. FEAKINS, A . S . WILLMOTT AND A . R. WILLMOTT 127significantly different from each other ; thus the result is a range of Kd values eachwith its associated b l , which fit the data equally well.Any variation in E& acrossthis range is insignificant. Kd is known more accurately the weaker the electrolyte,i.e., the higher the concentration of methanol. This is the expected behaviour.Kd cannot be quoted more precisely than in table 2, without some apriori assumptionabout the values of 6, and b2. Table 2 shows combinations of b2, & and Ez valueswhich fit the data for bl = -2.0 and +2.0, together with the mean E i . Associationconstants Kl( = Kc are also listed. Three sets of values for aqueous solutions arequoted; each represents measurements made concurrently with a set for one of themethanol-water mixtures. Sets I and I11 were made with the same amalgam (A);the difference in E: is within experimental error ; so too are the slight variations inthe range of Kd in the three cases.Mean values of AE;, from which AGp values are readily calculated, are also intable 2.Harned and Fitzgerald reported agreement to only 0.8 mV betweenwater 2.0- 2.010 ”/, methanol 2.0- 2.0water 2.0- 2.020 % methanol 2.0- 2.0water 2.0- 2.040 ”/, methanol 2.0- 2.0TABLE 2a9bb2 I Ka Ikg niol-1 mol kg-1set I (amalgam A)0.252 0.011 29-0.921 0.012 240.032 0.009 35- 1.135 0.010 08set I1 (amalgam B)0.227 0.011 33- 0.929 0.012 440.002 0.006 29- 1.169 0.006 57set 111 (amalgam A)0.184 0.011 42-0.993 0.012 57-0.579 0.002 70- 1.780 0.002 7488.6 0.512 8181.6 0.512 830.512 82107.0 0.501 0699.2 0.501 070.501 07AEZ = 0.011 7588.3 0.513 9780.4 0.513 980.513 98159.0 0.489 03152.2 0.489 020.489 03AEA = 0.024 9587.6 0.512 7779.6 0.512 790.512 78370 0.460 33365 0.460 350.460 34AEZ = 0.052 44Harned and Fitzgerald (saturated amalgam)water 2.0 0.633 0.010 13 98.7 0.573 80- 2.0 -0.622 0.011 19 89.3 0.573 830.573 82meanmeanmeanmeanmeanmeanmeana Acciiiracy of AEh is 20.15 mV or better.6 P. J. Reilly and R. H. Stokes, Austral. J. Chem.,1970, 23, 1397. give Kl = 85+ 1 for aqueous solution from measurements on cells withadded NaCl128 STUDIES IN ION SOLVATION IN NON-AQUEOUS SOLVENTSexperimental and calculated e.m.f. values in the CdC1,-water system. This seemedunexpectedly poor considering the high quality of the work known to emanate fromProf.Harned's school. The fit to eqn (10) was also only k0.8 mV when our pro-gramme was tried on their data, using, as they did, the whole range 0.0005 to 1.0 m ;but when we used the same range, namely 0.0005-0.02m, as in our work, the fitimproved to kO.07 mV. This suggests that the formation of the higher cadmium-chloride complexes cannot be neglected at the higher electrolyte concentrations usedby Harned and Fitzgerald.Although the range of Kd values (see table 2) for the two investigations of theaqueous system do not overlap, the difference between the two investigations isvery small. In table 3 we compare e.m.f. values, at various molalities, for cell (I) witha saturated amalgam electrode, with those using unsaturated amalgam A as electrode.The difference between the e.m.f.values of the two cells at any molality should beconstant. The mean deviation from the average difference is only 0.16 mV, approxi-mately the sum of the experimental errors on the two determinations.TABLE 3.-E.m.f. VALUES OF CELL (I)EIV0.0005 0.853 90.001 0.829 970.002 0.807 010.005 0.778 510.007 0.768 620.01 0.758 460.02 0.739 76m saturated amalgam 4 amalgam A a0.792 90.768 710.745 670.717 130.707 340.697 300.678 89difference0.061 00.061 260.061 340.061 380.061 260.061 160.060 87mean difference 0.061 18 Vk0.16 mVdifference in EA 0.061 00 VES 0.573 82 0.512 82a set I.DISCUSSIONIn comparing univalent with bivalent metal chlorides we note thatAG,"(M2+ 2C1-) = AG,"(M2+) + 2AG,"(Cl-)andAG,"(M'C1-) = AG:(M+)+AG:(Cl-)*By comparing +AG,"(M2+-2C1-) with AG,"(M+Cl-) we bring the data to a commonbaseline, namely AG,"(Cl-), and compare AG," per unit charge in the case of the cations.Table 4 thus shows $AG," for Cd2+-2C1-, from the present measurements, forSr2+ 2Cl- and Ba2+- 2Cl-, from a previous part,3 and for Zn2+ 2C1-, interpolatedfrom Corsaro and Stephens's data.Values for AG,O for the alkali-metal chloridesand for Ag+Cl- lo are included for comparison.Fig. 1 shows AGp(M+Cl-) or +AG;(M2+-2C1-) plotted against the reciprocalof cationic radius (rL1) for the 20 % mixture.Previous discussion in the Series l1 has led to the idea that all molecules in amethanol-water mixture, at least of relatively high methanol content, are more basicthan water molecules in pure liquid water.It has been postulated that the formalcharge on a solvent oxygen atom is greater in the mixtures than in water, and alsD. FEAKINS, A. S . WILLMOTT AND A. R . WILLMOTT 129that such an atom is " softer '' in the mixtures than in water.' The cation-oxygeninteraction thus leads to a negative contribution to AG,".Along the series of alkali-metal ions, AGt(Li+) cAG,"(Na+) for transfers from waterto all methanol-water mixtures; this is to be expected if for these small, relatively" hard " ions the cation-oxygen interaction is dominated by electrostatic forces.By contrast, however, AG,"(Cs+) < AG,"(Rb+) for all transfers, and this suggests that,even for ions having the do configuration, '' soft-soft " interactions are importantfor large cationic radii, r,.TABLE 4.-FREE ENERGIES OF TRANSFER (MOLAR SCALE) FROM WATER TO METHANOL-WATERMIXTURES AT 25°C3AGP (M2+ 2Cl-)/calwt.% MeOH Srz+ * 2C1- Baz+ 2CI- Znzf * 2C1- Cd2+ - 2C1-10 - 392 (307)c 30220 773 784 (635)c 63440 - 1597 - 1330AG," (M+C1-)/cal.dAg+CI- CsfC1- Li+CI- Na+CI- K+CI- Rb+CI-10 313 437 428 428 395 35320 623 883 876 861 803 (695)40 1268 1785 1804 1753 1651 (1440)a values & 3 cal or better except b & 12 cal C & 15 cal.d all values L- 3 cal or better. Bracketedfigures are interpolated. 1 cal = 4.184 J.800 - - c8O r u 700-3a6 0 0 -1 : 1 I0.5 1.0 1.5rc- 1 1A-lFIG.1.-Plot of AG;)(M+Cl-) or +AG,"(M2+ 2C1-) against r; l .Our first comparison will be of the alkaline-earth ions with the alkali-metal ions ;both types have noble gas configurations. K+(1.33A) and Ba2+(1.35A) haveapproxi-mately the same radii, and +AG,"(Ba2+2Cl-) may usefully be compared withAG;(K+CI-). Table 4 shows that AG,"(K+CI-) > +AG,"Ofa2+-2Cl-) for all transfersstudied.Both Sr2+ and Ba2+ are expected to behave as " hard " acids,12 and their inter-actions with oxygen atoms of coordinated solvent molecules should be dominated byelectrostatic forces. Thus the dominant term in AG,"(M2+) will be approximatelyof the form -2en2Aq/rc, where n2 is the coordination number of the ion, e is theprotonic charge, the ionic charge being 2e, and -Aq is the change in the charge onthe oxygen atoms of the coordinated solvent molecules in pure water and in the1-130 STUDIES IN ION SOLVATION I N NON-AQUEOUS SOLVENTSmixture respectively.The corresponding expression for a " hard " univalent ionis -enlAq/rc. Thus for a bivalent and a univalent ion of the same radius, if (i) theabove terms dominate AG," and (ii) n1 = n2, we would expect AG;(M+Cl-) N+AG,"(M2+-2C1-) on the simple " acid-base " theory.We first note that K+ is not a typically hard ion, so that not only is there a relativestabilization of Ba2+ per unit charge compared with K+ in the mixtures, but alsothat this stabilization is probably less than would be observed relative to a " hard "univalent ion of the same radius.The relative stabilization of Ba2+ could arise from two causes, among others.(i) The coordination number of the bivalent ion, n2, is expected to be larger thanthat of the univalent ion, nl, even if r, is the same.This is a common observation incoordination chemistry and arises from the Pauling electroneutrality principle.(ii) Polarization of the coordinated solvent molecules in the ionic field would enhancetheir basicity i.e. raise I Aq 1 . Such polarization should be the greater in the higherionic field of the bivalent ion.Let us now intercompare the four bivalent ions Sr2+, Ba2+, Zn2+ and Cd2f.Only at 20 % methanol can we compare Sr2+ and Ba2+. Because of the relativelylow accuracy in AG,"(Sr2+2C1-) and the cIoseness of the radii of the two do ions, it isnot possible to establish a meaningful trend in AG,O as a function of r,, although theresults are not inconsistent with what would be expected for " hard " cations, i.e.AGi(Ba2+) > AG,0(Sr2+).Zn2+ is regarded by Pearson l 2 as being on the borderline between a " hard "and a " soft " acid; there is however, a marked increase in the " softness " of thecations as we pass down the IIB sub-group from Zn2+ to Hg2+.Although mercury (11)complexes are in general more stable than those of zinc or cadmium, whatever the" hardness " or " softness " of the ligand, there is less differentiation between thestabilities of zinc and cadmium complexes unless the ligand is very " soft " (e.g.I-).14 This suggests a tendency for the decrease in electrostatic interaction energywith increasing r, to be approximately matched by an increase in the " soft-soft "interaction energy between ion and ligand as we pass from Zn2+ to Cd2+. Thesefactors appears also to be determining AG,"(M2+), since the difference betweenAG,O(Cd2+*2Cl-) and AG,"(Zn2+-2C1-) both at 10 % methanol and at 20 methanolis small, and we would expect the increase in " softness " of the solvating moleculesto be relatively small on passing from water to the mixed solvents.Fig. 1 shows what happens if, at 20 % methanol for example, we class the relativelyhard Zn2+ with Sr2+ and Ba2+ and assume a straight line relationship between$AG,"(M2+a2Cl-) and r;l for these three ions. The point for Cd2+*2C1- lies wellbelow this line, and this is analogous to the corresponding stabilization of AgtCI-in methanol-water mixtures with respect to a1 kali-metal chlorides of similar rildiusy6which has its origins in the " softer " character of the dlo ion.We thank the University of London and the Clothworkers' Company for grants(to A. R. W.).Part 14, D. Feakins and P. J. Voice, J.C.S. Faruday I, 1972,68, 1390.G. Corsaro and H. L. Stephens, J. Electrochem. Soc., 1957, 104,512.D. Feakins and Ann R. Willmott, J. Chem. SOC. A, 1970, 3121.H. S . Harned and M. E. Fitzgerald, J. Amer. Chem. Soc., 1936,58, 2624.H. P. Bennetto, D. Feakins and K. G. Lawrence, J. Chem. SOC. A, 1968, 1493.D. Feakins, K. G. Lawrence and R. P. 1'. Tomkins, J. Chem. Soc. A, 1967, 753. ' D. Feakins, J. Chem. Soc., 1961, 5308.* D. Feakins and P. Watson, J . Chent. Soc., 1963, 4636D. FEAKINS, A . S . WILLMOTT AND A . R . WILLMOTT'OD. Feakins, K. G. Lawrence, P. J. Voiceand A. R. Willmott, J. Chem. Soc. A , 1970, 837.l2 R. G. Pearson, in Suruey of Progress in Chemistry, ed. A. F. Scott, Vol. 5 (Academic Press,l3 R. S. Nyholm, Proc. Chem. SOC., 1961, 273.131I. Tur'yan and B. P. Zhantalai, Russian J. Inorg. Chem., 1960, 5, 848.D. Feakins and P. Watson, J. Chem. Soc., 1963,4734.New York, 1969), p. 1.L. G. Sillen and A. E. Martell, Stability Constants of Metal-ion Coiizplexcs (The ChemicalSociety, London, 1964)

ISSN:0300-9599    

DOI:10.1039/F19736900122

出版商:RSC    

年代:1973

数据来源: RSC

摘要:

Mechanism of the Cathodic Reduction of Acetophenone inAcidic Aqueous-methanolic MediaB Y M. P. J. BRENNAN AND 0. R. BROWN*Electrochemistry Research Laboratories, Dept. of Physical Chemistry,University of Newcastle upon Tyne, Newcastle upon Tyne NEI 7RU, EnglandReceived 26th July, 1972The reduction, in acidic aqueous-methanolic media, of acetophenonc to its pinacol at severalhigh hydrogen overvoltage metal cathodes follows the simple classical mechanism of reversibleaddition of a proton and an electron followed by the rate-determining dimerisation of free radicals.Several points are raised concerning recent work which produced apparently quite different results,thereby prompting reconsideration of this system. It is shown that similar results can be obtainedif artefacts are not correctly compensated or if electrodes, deactivated with respect to the electrontransfer process, are used.The electrochemical reduction of acetophenone and other aromatic ketones hasbeen the subject of a vast number of investigations (e.g., ref.(l), (2)), mainly usingthe dropping mercury cathode to examine dilute solutions of the depolariser. Inacidic aqueous-organic solutions, preparative experiments have shown that theproduct on mercury cathodes is predominantly the hydrodimer, acetophenonepinacol. In such media the polarographic mechanism has been shown to be a rate-determining dimerisation of free radicals which are formed in Nernstian equilibriumat the cathode surface with the ketone and protons :C6H5COCH3 + H+ + e + C6H5c(OH)CH,2C6H,c(OH)CH3 -+ C6H5C(OH)CH, IC~HSC(OH)CH~.Recently Conway et al.proposed that, on a stationary mercury cathode inmore concentrated solutions of acetophenone, the dimerisation step takes placebetween adsorbed radicals on the electrode surface. This suggestion was based ontheir results obtained with sulphuric acid solutions in a solvent mixture (80mol "/omethanol+20 % water). In discussion one of the present authors questionedthe validity of those results which did not correspond with his own unpublishedmeasurements on a nominally identical system. Those kinetic measurements werebriefly described and were shown to be consistent with the simple classical schemeoutlined above. Subsequently Conway and Rudd published further results whichdid not fit the classical behaviour but were not fully explained by any single simplemechanism.In this paper, we present evidence in favour of the classical mechanism and proposepossible explanations for earlier discrepancies.EXPERIMENTALAll work was carried out in a solvent consisting of molar sulphuric acid in 80 moi %Methanol was A.R.(B.D.H.) and was distilled prior to use.Potentials are quoted relative to the Hg/Hg2S04 referencemethanol+20mol % water.Water was thrice-distilled.13M. P . J . BRENNAN AND 0. R. BROWN 133electrode in aqueous molar sulphuric acid, which was found to have a potential relative to ahydrogen electrode in the working electrolyte of + 0.615 V (including liquid junctionpotential).Preparative electrolyses were carried out in divided cells at electrodes formed from sheetsof tin, lead (Johnson-Matthey " specpure ") and cadmium (Koch-Light), and at a largepool electrode in the case of mercury, which was twice distilled at reduced pressure immedi-ately prior to use.The electrolyses were in all cases carried out with an acetophenone con-centration of 0.3 M, in order to duplicate as closely as possible the conditions under whichvoltammetric measurements were made. The electrode potential was - 1.32 V in the caseof mercury, lead, and cadmium, and - 1.44 V in the case of tin. The electrolyte was purgedcontinuously during a run with a stream of nitrogen.The electrolyte after reduction was neutralised with solid sodium carbonate A.R.(B.D.H.),filtered to remove the precipitate of sodium sulphate which formed, and made up to a knownvolume with methanol. The neutralised electrolyte was then analysed quantitatively byg.1.c. on an F.and M. 810 instrument. A TWEEN 80 column (6 mmx 2 m 20 % TWEEN80-DTATOPORT W (A.W.) 60-80) was used to analyse for acetophenone and phenylethanol,at a temperature of 150°C and a carrier gas flow rate (N2) of 50 ml min-'. Analysis foracetophenone pinacol was performed on a GESE-52 column (6 mmx 2 m 10 % GESE-52-DIATOPORT S 80-loo), at 200" and a carrier gas flow rate of 50 ml min-l. Both aceto-phenone and the pinacol gave well-shaped Gaussian peaks.The cell used in voltammetric studies was a divided cell of 150 ml capacity. Deoxygena-tion was achieved by flushing with a stream of nitrogen, and the cell was allowed to deoxy-genate for one hour before measurements were made.The metal rotating disc electrodesemployed (diam. 0.40 cm) were constructed from small cones of the respective metals,machined from Johnson Matthey " specpure" rods and held in a Teflon holder. Themercury electrode was a gold disc which was amalgamated prior to use by cathodic polar-isation in a solution of mercuric nitrate. The mercury surface could be renewed duringoperation by means of a mercury-filled glass syringe incorporated into the cell which wasused to direct a jet of mercury on to the electrode surface. The cadmium, lead and tinelectrodes were prepared by polishing on a fine emery paper and then a final polishing with aGamma grade alumina (Griffin and George). The lead electrode was briefly electro-p~lished.~ All the electrodes were thoroughly washed in a jet of thrice-distilled water afterpreparation.The construction and preparation of pyrolytic graphite electrodes has beendescribed previously.8Polarisation curves were obtained on electrodes rotating at 200-400 rad s-' by cycliclinear potential sweep (10 mV s-l) over the potential range studied, which was generally- 1150 to - 1600 mV. Before adding acetophenone to the cell, the background hydrogenevolution current was allowed to settle to a reproducible cyclic voltammogram. Aceto-phenone was then added with a g.1.c. syringe through a septum in the cell above the electro-lyte surface in aliquots corresponding to total concentrations of 0.001 M, 0.003 M, 0.01 M,0.03 M, 0.1 M, 0.3 M and 1 .O M.After each addition the polarisation curve was remeasured,using a complete cycle of the potential waveform. The effect of varying the rotation speedupon limiting currents was also studied. Prior to each measurement with the amalgamatedgold electrode the mercury fiIm on the surface was renewed. This was found to be necessaryto prevent a gradual decrease in hydrogen overvoltage over a period of several minutes,presumably arising from diffusion of gold atoms to the eIectrode solution interface.The potentiostat and function generator lo are described elsewhere. Polarisationcurves were recorded on a Bryans 21 000 X-Y recorder. Currents during preparativeelectrolyses were recorded on a TOA EPR-2T pen recorder.Thermostats were not usedbut the laboratory temperature varied by no more than 1°C from 22°C throughout the work.RESULTSOn cathodes of mercury, tin, lead and cadmium no reaction product other thanacetophenone pinacol was detected by g.1.c. In particular, phenylethanol and ethyl-benzene were absent. The consumption of acetophenone corresponded onImercury134 ELECTROCHEMICAL REDUCTION OF ACETOPHENONElead, and cadmium to 1.0rt0.05 electrons per molecule. However, in common withearlier work,ll difficulty was encountered in isolating a quantitative yield of thehydrodimer. The major source of this problem appeared to be the inclusion ofpinacol in the sodium sulphate precipitate. Repeated washing of that precipitateenabled yields in excess of 60 % to be isolated.We note that Conway et al.4 recordcomplete conversion to pinacol on a mercury cathode.Our kinetic results obtained on cathodes of mercury, lead, cadmium, pyrolyticgraphite planes and tin are presented in fig. 1-5 respectively. Hysteresis betweenI 0’I I-1.25 -1.3 -1.35EIVFIG. 1 .-Polarisation data on a mercury film cathode. Concentrations refer to acetophenone in 1 Msulphuric acid in 80 mol % MeOH+ 20 mol % water. Solid lines connect points uncorrected fordiffusion of reactant. Crosses are additional data after correction for diffusion.FIG. 2.-Polarisation data on a lead cathodeM. P. J . BRENNAN AND 0 . R . BROWN 135MII I I - 1.25 -1.3 -'I. 35EIVFIG. 3.-Polarisation data on a cadmium cathode.too 4i I I-1.25 -1.3 -I.35EIVFIG. rF.-Polarisation data on a pyrolytic graphite (nominal face) electrode. Tafel slope values aregivenI36 ELECTROCHEMICAL REDUCTION OF ACETOPHENONEanodic and cathodic scans was negligible. Before being plotted, the experimentaldata have been corrected for ohmic contributions to overpotential and backgroundcontributions to the total current. Moreover, mass transfer limitations on the sizeof the currents have been avoided by setting aside results in which currents exceeded10 % of the diffusion limited current. The limiting currents were shown to be masstransfer controlled by varying the rotation speed of the electrode; fig. 6 indicates1007 10EcE ---. *i- 1.3 -l.L - 1.5EiVFIG. 5.-Polarisation data on tin, uncorrected for background currents.- . - . -- , 0.03 M;, 0.1 M. .....5 0-LO-0.01, 0.02 003depolariser concentration/MFIG. 6.-Limiting current densities in 0.01 M FIG. 7.-Limiting current densities in aceto-acetophenone solution at a mercury film electrode. phenone solutions at a mercury film electrodeat w = 375rads-lM. P. J . BRENNAN AND 0. R. BROWN 137that the Levich relationship l 2 holds for currents at a mercury cathode with DK =1.3 x cm2 s-1 . Further evidence for the diffusion controlled nature of the limit-ing currents is provided by their first order behaviour (fig. 7). For the more con-centrated solutions it was necessary to calculate the limiting currents, using thediffusion constant obtained from dilute solution measurements.Ohmic compensation was made by displacing the electrode potential to morepositive values by an amount equal to the product of the current and the effectiveresistance between the interface and the Luggin probe.This latter quantity (of theorder 10 ohm) was obtained from the limiting slopes at high current densities of thepolarisation curve of the most concentrated (1.0 M) acetophenone solution. Thevalue so obtained was reproducibly accurate within 10 % ; data requiring correctionby more than 20mV were rejected. Thus electrode potential data are consideredaccurate within 2 mV.That background currents can not simply be subtracted from the measuredcurrents to obtain with complete accuracy the ketone reduction current is illustratedby results obtained on a tin cathode (fig.5). Addition of the ketone actually cancause a decrease in the total current at more anodic potentials. Thus the hydrogenevolution reaction on tin is inhibited by adsorption of the organic reactant. Thiseffect is also evident in the results of Conway et a2. (ref. (4), fig. 1) but noticeablyabsent when absolute methanol was employed as solvent (ref. (6), fig. 2) probablyowing to the absence of appreciable acetophenone adsorption in that medium. Toavoid errors in the subtraction of background currents, we have rejected data inwhich the total current was less than double the background current at the same(corrected) electrode potential. These strict criteria for data analysis precluded theexamination of all data obtained with tin cathodes because of the large hydrogenevolution currents on that metal.On mercury, lead and cadmium the essential features of the results are as follows.(1) Tafel slopes are (40+2) mV.(2) All reaction orders with respect to acetophenone are 1.5 f 0.1.(3) Within experimental error (electrode potential +2 mV, concentrations & 10 %)results are independent of the metal used as cathode material.(4) Currents depended on rotation speed only near the limiting current region.If allowance is made for a roughness factor of pyrolytic graphite faces of between2 and 3, then on pyrolytic graphite the polarisation curve in the most concentrated(1.0 M) solution coincides with those obtained on the metal cathodes.However,’as theconcentration of acetophenone is decreased the Tafel slopes for graphite cathodesgradually steepen : 1.0 M ; 42 mV, 0.3 M ; 45 mV, 0.1 M ; 49 mV, 0.03 M ; 51 mV,0.01 M ; 57 mV.In consequence, reaction orders are a function of current densityfalling from 1.9 at high current densities to 1.3 at lower values.DISCUSSIONThe classical mechanism gives rise to simple and characteristic kinetic behaviourat a planar electrode with constant surface area. The diffusion equation in the steadystate for the radical R in the schemek2K+H++e + R + R2isd2a,dx2D,- = 2k2a138 ELECTROCHEMICAL REDUCTION OF ACETOPHENONEwhich solves easily to(da,/du)L = 4k2a;/3 DR, (2)where x represents distance into solutioii from the surface and DR is the diffusioncoefficient of R.As Nernstian equilibrium is supposed in the electron transferstep, we can write= r&pRafi+ut exp 3(E, - E)F2RT (3)where uK is the activity of the ketone, K, at the electrode surface and Eo the standardpotential for the first step. When diffusion control by K is unimportant, it is evidentthat this model predicts reaction orders of 3 for the neutral ketone and the protonand Tafel slopes of 2.3 x 2RT/3F (equal to 40 niV at 25°C). Moreover, completeindependence of electrode material is expected as no heterogeneous process contri-butes to the determination of the overall rate. The complete agreement betweenour results and these predictions is clear.We have not, in this study, examined the reaction order in H+, as the seconddissociation of sulphuric acid is not complete and varies with concentration.How-ever, in an earlier unpublished study made in 40 % ethanol+water mixtures, hydro-chloric acid was used as acid and ionic strength was maintained constant with KCI.Then the reaction order in H+ was found to be 1 . 5 k O . l .The data, temporarily set aside because of diffusion limitations, were tested forcompliance with this model by making a plot which corrects for depletion of theketone at the cathode surface. At constant rotation speedaK(s) = aK(l - i/id) (4)where id is the diffusion limiting current and aK(s) the value of aK at the surface.Substituting (4) into (3) we obtain- i = ( 4k2D, T)+ufi+a( 1 - t - e x p ~ ( E o 2RT - E)F .FTherefore the remaining data were plotted in the form log i+ + log (id/[id - i]) against E.In this way, the slopes of the Tafel lines are seen to remain unchanged (dashed linesin fig.1, 2).The foregoing theory is sometimes associated with the names of Koutecky andHanus who tackled l3 the more difficult problem of solving the corresponding diffusionequations for the case of an expanding spherical (i.e., dropping mercury) electrode.Conway and Rudd drew attention to the limitations of that theory but were notspecific in their criticisms. To our knowledge every previous attempt to use thetheory of Koutecky and Hanus has treated results obtained at the dropping mercuryelectrode to which the simple exact treatment l4 correctly used by us is not strictlyapplicable.This highlights a feature of the dropping mercury electrode, but in noway reflects upon the correctness or otherwise of the classical mechanism. Indeedit has been shown to operate in several reduction systems ; for example N-alkylpyridin-ium salts,l tropylium ion,16 tridecahydrodecaborate( 1 -),I aromatic aldehydes,lbenzocy~lanones,~ and organomercury cations,20 in addition to aromatic ketones.2It is true that complicating features are sometimes found, particularly inhibition byinsufficiently soluble products which accumulate on the cathode surface. Bonnaterreand Cauquis 22 have recently treated the general steady state case of a mechanism inwhich electron transfer is followed by dimerisationM. P . J . BRENNAN AND 0. R . BROWN 139The bimolecular nature of the mechanism implies some interesting kinetic con-sequences.At sufficiently low current densities, efficient stirring should enable theradicals to escape from the diffusion layer before appreciable dimerisation hasoccurred, even if that reaction is diffusion controlled. Then one should observeTafel slopes of 2.3RT/F, currents dependent on rotation speed and reaction orders ofunity. The current density level at which the kinetics should change in this waycan be calculated if a value for the rate constant k2 is assumed. Suppose the dimer-isation is diffusion controlled (k2 = 1O1O 1. mol-1 s-l). Then the transition occurswhen6where 6 is the diffusion layer thickness, typically em. Then the current density isThe magnitude of the background hydrogen evolution current on all the metalsstudied exceeded this value by several orders of magnitude.As such free radicaldimerisation reactions normally approach diffusion control 2 3 it is unlikeIy that thispredicted change in kinetic behaviour can be experimentally verified.By contrast, at sufficiently high current densities the rate of the chemical stepmust ultimately prevent equilibrium being maintained in the charge transfer stepwhich consequently will become rate determining. This transition must occurwhen the dimerisation rate exceeds the rate of the charge transfer in the anodicdirection ; it will result in a resumption of first order kinetics and a change in Tafelslope to approximately 4.6RTIF. As the rate constant ko for the charge transfercannot exceed 24 N lo4 cm s-l we can calculate an upper limit to the current densitylevel at which that transition should occuri = 2 F ( T ) k2aiD, = Fk,a,.That isThis current density is also experimentally inaccessible.Clearly therefore there areno a priori objections to the classical mechanism being operative at the currentdensities used in our measurements.However, on graphite electrodes there is an indication that electron transfercontributes to the determination of the overall rate ; real current densities obtainedfor the more dilute acetophenone solutions fall below those observed when usingmetallic electrodes by factors as large as three. This is not surprising ; the anisotropyof graphite, particularly with regard to electrical conductance, is well known and thedistribution of energy levels from which electron transfer occurs will be different fromthat of metal cathodes.Although studies of slower electron transfer reactions havenot manifested any anisotropic behaviour in pyrolytic graphite 2 5 it is reasonable tosuppose that the very high values of the electron transfer rates needed to maintainequilibrium in the system described here may disturb the equilibrium distribution ofelectrons at the graphite surface.The absence of detectable non steady state currents in cyclic voltammetry whic140 ELECTROCHEMICAL REDUCTION OF ACETOPHENONEConway and Rudd interpreted as a sign of " complete irreversibility " arisesbecause the dimerisation reaction is extremely rapid so that the concentration ofradicals at the surface is low and the reaction layer very thin.This is shown bysimple calculation. At a current density of 10 mA cm-2, aR at the surface is given byeqn (2)aRCs) - 4 xagain assuming k2 = 1 0 ' O I . mol-1 s-l. Tli~is the reaction layer thickness 26 11 isinol C M - ~Clearly failure to observe oxidation currents does not preclude reversibility in theelectron transfer step.The proposed scheme must be able to explain the solvent effect on the kinetics ofthe reduction. We have verified the observation of Conway et aL49 that in absolutemethanol slopes of 4.6RT/F are observed. We interpret this phenomenon in termsof a decrease in the rate of the cathodic electron transfer step relative to its rate inpart-aqueous solutions.As already mentioned, Conway and Rudd's data indicatethat adsorption of the ketone from methanol solution is weak. As the adsorbedmolecules are probably most eligible for electron transfer, it follows that adsorptionof acetophenone facilitates the electron transfer equilibrium to be maintained inaqueous media.It is important to stress that, in proposing the classical mechanism, we do notnecessarily exclude the possibility that adsorbed ketone radicals are intermediates inthe course of the formation of free radicals. However, spontaneous desorption ofchemisorbed radicals is unlikely. In view of the adsorption of the analogous radicalsderived from a~etone,~' the behaviour of acetophenone appears strange. Thedifference in behaviour of the two ketones probably lies in the difference in theirbasicities. In an acid solution a substantial proportion of acetone molecules isprotonated,28 whereas the fraction of protonated acetophenone molecules is severalorders of magnitude less.29 It is likely, therefore, that whereas protonated acetonemolecules are reduced the predominant reactant in the acetophenone system is theneutral molecule.Consequently, the product of electron transfer must be a radicalanion which is immediately repelled from the negatively charged cathode and thenrapidly protonated to form a free radical in the acid medium. Although thissequence cannot be proved or disproved in the solvent system used in this work, it canbe distinguished by kinetic studies in absolute methanol. This we intend to carry out.Having shown our results to be entirely consistent with and characteristic of theclassical mechanism and that no objections to the possibility of the operation of thatmechanism appear to exist, we must now re-examine the papers of Conway et aL49to seek an explanation for the discrepancy in results.First we show (fig. 8) ourresults for a mercury electrode without compensation for potential error and back-ground currents. Their similarity to the results of Conway et al. in regard to reactionorders is remarkable. The account given in their papers to the details of the correc-tion procedures is brief, even when allowing for the elementary nature of this routineexercise. Although it is stated " that all points on Tafel lines were corrected forresidual currents " the obvious curvature of the lines at low current densities throwssome doubt on that operation.A similar difficulty exists regarding ohmic over-potential compensation in those results. The authors consider IR effects as analternative to mass transfer control when seeking to explain the tendency towardslimiting currents at cathodic potentials. Having shown that IR compensatioM. P. J . BRENNAN AND 0. R. BROWN 141alone does not linearise the limiting currents, they then appear to dismiss IR com-pensation and it is subsequently not mentioned. It is, however, well known that athigh current densities electrode kinetic measurements are affected both by IR errorsand mass transfer control.Conway and Rudd's final conclusion concerning thenature of the limiting currents is not clear. Initially they suggest the possibilitythat " the limiting currents are in part a kinetic phenomenon " because the currentsat negative potentials were less than first order. This is easily explained in terms ofohmic error. Moreover their limiting currents should be treated with caution asrotating electrodes were not employed in their study.FIG. 8.--Polarisation data on a mercury film cathode before correction for ohmic error and back-ground currents.A further disturbing feature of the earlier work concerns the measurements ofreaction orders. The two papers (ref. (4), fig. 3 and ref. (61, fig. 4(a)) indicate somedegree of inconsistency in the measurements of the order with respect to acetophenone.There is doubt also concerning the measurements of reaction order in protons.Asthe authors refer to " limiting proton migration ", and make no mention of a support-ing electrolyte to maintain the ionic strength, it appears that during those measure-ments the ionic strength and therefore the activity coefficient of the acid was allowedto vary. There is also a problem caused by incomplete dissociation of sulphuric acid.Finally we draw attention to the fact that on cathodes less active than clean metalsurfaces (in our case pyrolytic graphite planes), Tafel slopes with the values found byConway and Rudd can be recorded and attributed to mixed kinetic control betweenelectron transfer and subsequent dimerisation.We look forward to the publication of an examination of this system from a thirdlaboratory in order that these discrepancies can be resolved.V.N. Pavlov, Ya. M. Zolotovitskii, S. G. Mairanovskii and G. A. Tedoradze, Sou. Electro-chem., 1965, 1, 371.P. Zuman, D. Barnes and A. Ryvolova-Kejharova, Disc. Furuduy SOC., 1968,45,20142 ELECTROCHEMICAL REDUCTION OF ACETOPHENONER. Pasternak, Helv. Chim. Acta, 1948,31, 753.B. E. Conway, E. J. Rudd and L. G. M. Gordon, Disc. Faraday Sac., 1968,45, 87.0. R. Brown, Disc. Faraday Sac., 1968,45, 125. ' E. J. Rudd and B. E. Conway, Trans. Faraday SOC., 1971, 67,440. ' E. Jones and H. R. Thirsk, Nature, 1953, 171, B43.M. P. J. Brennan and 0. R. Brown, J. Appl. Electrochem., 1972,2, 43.A. Bewick and 0. R. Brown, J. Electroanal. Chem., 1967,15,129.0. R. Brown, Electrochim. Acta, 1968, 13, 317.l1 S. Swam and G. H. Nelson, Trans. Electrochern. Soc., 1935, 67, 201.l 2 V. G. Levich, Acta Physicochim. U.R.S.S., 1942, 17, 257.l 3 J. Koutecky and V. Hanus, Coll. Czech. Chem. Comm., 1955, 20, 124.l 4 V. Hanus, Chem. Zvesti, 1954, 8, 702.l 5 S. G. Mairanovskii, Doklady Akad. Naitk S.S.S.R., 1956, 110, 593.A. M. Khopin and S. I. Zhadnov, Sou. Electrochem., 1968,4,200.D. E. Smith, E. B. Rupp and D. F. Shriver, J. Amer. Chem. SOC., 1967, 89, 5568. *' E. Laviron, Coll. Czech. Chem. Comm., 1965,30,4219.l 9 M . Levy and V. Toure, Compt. rend., 1968,266CY 1390.2o K. P. Butin, I. P. Beletskaya, A. N. Ryabtsev and 0. A. Reutov, Sou. Electrochem., 1967, 3,2 2 R. Bonnaterre and G. Cauquis, J. Electroanal. Chem., 1971, 32, 199.23 E. T. Denisov, Russ. Chem. Rev., 1970,39, 31.24 R. A. Marcus, J. Chem. Phys., 1965, 43, 679.25 I. Morcos and E. Yeager, Electrochim. Acta, 1970, 15, 953.26 S, G. Mairanovskii, Catalytic and Kinetic Waues in Polarography (Plenum Press, New York,27 0. R. Brown and K. Lister, Disc. Faraday Suc., 1968, 45, 106.28 S. Nagakura, A. Minegishi and K. Stanfield, J. Amer. Chem. SOC., 1957,79, 1033.29 G. Culbertson and R. Pettit, J. Amer. Chem. SOC., 1963, 85, 741.1183.S. G. Mairanovskii and V. N. Pavlov, Russ. J. Phys. Chern., 1964,38,980.1968), p. 229

ISSN:0300-9599    

DOI:10.1039/F19736900132

出版商:RSC    

年代:1973

数据来源: RSC

摘要:

Charge Transfer Complexes of Iodine with Mercaptans and Related Sulphur Compounds BY G. REICHENBACH, S. SANTINI AND U. MAZZUCATO lstituto di Chiinica Fisica, Universith di Perugia, 06100 Perugia, Italy Received 26th June, 1972 Charge transfer complexes of iodine (u acceptor) with ethyl and phenyl mercaptan, and diphenyl sulphide and disulphide (n-donors) have been studied in cyclohexane at 25°C. Spectral data, stability constants, enthalpies and entropies of complexation are reported. As comparison with literature data for related compounds, the stability of the complexes is correlated with the donor structure and discussed in terms of electronic and steric factors. The effect of substituents on Kct of thiophenols has been investigated leading to an Hammett reaction constant of - 1.1 1.In the course of a study of the iodine-catalysed isotopic exchange between diphenyl disulphide (Ph,S,) and thiophenol (PhSH),' it seemed interesting to investigate the possible role played by the charge transfer (c.t.) complexes formed between the catalyst and the sulphur derivatives. The c.t. complexes between iodine (a acceptor) and the n-donors dialkyl sulphides (R2S) and disulphides (R2S2) are well known,2 as are those of iodine+alkyl aryl sulphides (PhSR) (thioani~oles).~ However, data on aryl mercaptans (thiophenols) are absent and the study of the equilibrium iodine+ethyl mercaptan (RSH) has not been completed because of its low stability.2c The study of the complexes of iodine + thiophenols and other sulphur-derivatives was thus undertaken along the same lines as a series of iodine complexes of similar w+o* nature which used more basic pyridine donor^.^ Stability constants and thermo- dynamic data are reported for PhSH, EtSH and Ph,S and the results discussed by comparison with literature data for related complexes.EXPERIMENTAL Twice sublimed iodine (Carlo Erba RP) was resublimed with potasium ioaide and kept in a dessicator. Cyclohexane (Carlo Erba RP) was purified following Vogel,s dried and stored under nitrogen. All the donors were commercial products crystallized before use from absolute ethanol or distilled under nitrogen. Nitrogen (99.9 %) was purified by bubbling through a Fieser solution,6 dried and then saturated with cyclohexane. The stability constants were calculated from spectrophotometric measurements on the assumption of 1 : 1 complexation (see Discussion).A single beam Unicam SP500/2 spectrophotometer (with 1 cm path quartz cells and a thermostatted jacket) was used. The same donor-solvent system was used as a blank when necessary, to compensate for the free donor absorption ([D]>[I,], weak complex). Prelim- inary spectra were run with a double beam Optica CF4-DR spectrophotometer. The Van de Stolpe equation was applied in the form : 1 1 1 where E~ is the apparent molar extinction coefficient (observed absorbance/total iodine concentration), Ef and cc are the molar extinction coefficients for the free iodine and the 143144 SULPHUK CHARGE TRANSFER COMPLEXES complex, and [D] is the donor molar concentration. Five different donor concentrations were generally used in each experiment.The initial concentrations were 5x M for iodine, 10-1-10-2 M for thiophenols, 5 x 10-'-1.7 x For EtSH, concen- trations of the donor were 2~ 10-l-4.3~ lop3 M in the ultraviolet and 0.4-0.1 M in the visible experiments. Equilibrium constants and molar extinction coefficients for thiophenols were calculated at different wavelengths in the range 320-360nm, at the tail of the charge transfer band. In the case of EtSH and Ph2S, Kct was calculated also at the c.t. peak, where the donor does not overlap. A good linearity of the experimental plots was found in every case (correlation coefficients were in the range 0.9969 to 0.9999). The reported Kct values are averages of 3-5 values calculated at different A (mean deviations are of the order of 1 %).When the perturbed iodine band was sufficiently blue-shifted with respect to the free iodine, measurements were also performed in the visible range, producing good agreement with the ultraviolet data. Under these conditions all the systems were stable over a period of several hours, as shown by the constancy of the spectrum at the experimental temperatures. M for Ph2S. All the measurenients were done under a nitrogen atmosphere. RESULTS AND DISCUSSION The n-donor S-derivatives display a basic character intermediate between the corresponding 0- and N-derivatives. In order to investigate the donor properties of the sulphur atom towards iodine and to compare them with those of other hetero- atoms, ethyl and phenyl mercaptan and diphenyl sulphide were first investigated. Table 1 shows the equilibrium constants at different temperatures and the thermo- dynamic parameters for these compounds.Table 2 collects similar data reported in the literature for related donors together with the donor ionization potentials, for comparison purposes. For all the three donors of table 1 the perturbed iodine band shifts towards the blue by complexation and a new c.t. absorption appears in the U.V. region, as shown in fig. 1 for EtSH. TABLE 1 .-TEMPERATURE DEPENDENCE AND KELATED THERMODYNAMIC PARAMETERS OF THE C.T. EQUILIBRIA OF IODINE WITH ETHYL AND PHENYL MERCAPTANS AND DIPHENYL SULPHIDE IN CY CLOHEX ANE donor tl"C KctlM-1 -AH/kcal mol-1 -AS/cal mol-1 K-1 EtSH 17 28.7 4.6 25 23.2 40 15.7 PhSH 16 14.7 25 11.6 34.6 9.3 40 7.9 PhzS 17.5 4.9 25 3.9 35 2.8 4.3 5.7 9.0 9.6 16.6 ETHYL A N D PHENYL MERCAPTAN Introduction of an ethyl group in H,S (Kct = 1.3 M-I at 25°C) l2 considerably increases the base strength of sulphur.In the aliphatic series EtNH,, EtSH and EtOH (no. 4, 1, 3 in table 2), the S-derivative gives weaker c.t. complexes than the amine but stronger than the alcohol, as foreseen from the blue-shift of the iodine peak 2c and expected from the relative basicity of the donor atoms and from their ionization potentials. The combination of the inductive and mesomeric interactionsG. REICHENBACH, S . SANTINI AND U. MAZZUCATO 145 between the lone pair and the n-electrons of the aromatic ring, which have a strong electron-releasing influence on the base strength of the nitrogen atom (Kct decreases about fifty times on going from ethyl- to phenyl-amine 13), have, however, a small overall effect in this case, the S-phenyl-derivative displaying almost the same stability as EtSH.The smaller effect in PhSH compared to PhNHz could be due both to the greater dimensions of the sulphur centre (the orbital overlap with iodine is less hindered by the phenyl group) and to a return of charge from the ring to the 3d- orbitals of the sulphur, which enhances its donating ability. TABLE 2.-cHARGE TRANSFER EQUILIBRIA OF IODINE WITH ETHYL AND PHENYL MERCAPTANS AND OTHER ALIPHATIC AND AROMATIC RELATED COMPOUNDS IN INERT SOLVENT AT 25°C donor (1) EtSH (2) MeOH (3) EtOH (5) EtOEt8 (6) EtSEt 2b (7) EtSSEt 26 (8) PhSH (4) EtNHz’ (9) PhSCH3 (10) PhzS (12) S(C2H4)2S (1 1) MeSeMe KctIM-1 23.2 0.44 0.39 0.84 560 - 170 - 5 11.6 12.1 3.9 47 1 77 -AH/ -AS/ kcal moi-1 cal mol-1 K-1 4.6 9.0 1.9 3.3 2.1 4.1 7.4 12.3 4.3 9.9 7.8 15.9 4.6 12.3 4.3 9.6 5.5 13 5.7 16.6 8.6 16.4 6.2 12.1 W n m 289 242 243 246 249 302 310 (310/sh) 333 3 50 433 N 300 solvent cyclohexane CC14 CC14 n-heptane n-hept ane n-heptane cyclohexane cyclohexane cyclohexane CC14 CC14 cc14 I .P.* lev 9.3 10.8 10.5 8.9 9.5 8.4 8.3 8.3 8.9 8.2 8.7 8.5 *Ionization potentials are taken from K.Watanabe, T. Nakayama and J. Mottl, J. Quant. Specfr. Radiative Transfer, 1962, 62, 369, except for PhSCH3 (B. G. Gowenlock, J. Kay and J. R. Mayer, Trans. Faraday SOC., 1963,59,2463), for Ph2S (G. Innorta, personal communication) and for S(C2H4)2S (R. W. Kiser and E.J. Gallegos, J. Phys. Chem., 1962,66,947). 0.4 0.3 3.2 3.1 FIG. 1.-Absorption spectra of the system ethyl mercaptan+iodine in cyclohexane at 25°C: A, 0.1 M ethyl mercaptan ; B, c.t. band (0.025 M EtSH+ 5 x M 12) ; C, 4.5 x M iodine; D, perturbed iodine (4.5 x M I, + 0.3 M EtSH). Cell path 1 cm.146 SULPHUR CHARGE TRANSFER COMPLEXES Recent data from photoelectron spectroscopy l4 seem to agree with this behaviour, as the first I.P. of PhSH (assigned to a lone pair electron with considerable amount of benzene ring character) is only at 8.3 eV. The situation is different for other aromatic compounds such as PhOH and PhNH2, where the lowest I.P. values are assigned to the n-system while n-I.P. values are at rather higher levels (1 1.3 and 10.6 eV for PhOH and PhNHz, respectively).15 These values could explain why phenol acts as a n-donor and aniline acts as both an 12- and n-donor, depending on the type of acceptor.2a The decrease of n-I.P.from PhOH to PhSH (already observed in table 2 from EtOH to EtSH) and the consequent interchange of the n and n orbital energy levels seems thus to be responsible for the change of the c.t. interaction from 71 to n. SULPHIDES Replacement of the hydrogen atom of mercaptans by alkyl- or aryl-groups has different effects. The complex Et,S (no. 6) displays the highest K,, observed for these S-derivatives, due to the electron-donating effect of the alkyl groups. However PhSCH3 (no. 9) has a stability similar to PhSH. This may indicate that the increase of charge density on the sulphur atom due to the CH,-group is matched by a partial charge redistribution on the ring and by a small steric effect.Steric hindrance becomes even more important in the diphenyl-derivative where K,, is reduced to less than 4 M-l. The close approach to the reaction center of an acceptor molecule as large as iodine is, in fact, prevented by two rings and the freedom of phenyl rotation is decreased in the complex. This should reflect a greater negative complexation entropy, as experimentally found (- 16.6 cal mol-l K-l). NATURE OF THE COMPLEXES All complexes were assumed to be of 1 : 1 stoichiometry. Arguments in favour of this assumption are : (1) the good linearity of the experimental plots (correlation coefficients in the range 0.9969 to 0.9999) in a rather wide concentration interval ; (2) relatively constant K,, values with analytical wavelength (see experimental) ; (3) linearity of the Van de Stolpe plots was further checked by Scatchard plots l6 for EtSH and Ph2S, and satisfactory correlation coefficients (R = 0.9961 and 0.9982 respectively) for 10 measurements in a wide range of saturation fractions (s from 0.09 to 0.82 for EtSH and from 0.065 to 0.67 for PhzS) were obtained.For PhSH it was difficult to extend the donor concentration range because of its strong absorption in the analytical spectral region. Other minor points in favour of 1 : 1 complexation come from the analogy with the PhSCH, + Iz complex (for which analysis indicates 1 : 1 a~sociation),~ and with the (PhCH,),S + I, complex (for which structural data indicate bimolecular associa- tion in the solid state)." The presence of termolecular species cannot be completely ruled out l9 but, in the light of the above arguments, we think that the observed K,, values closely represent the correct values for 1 : 1 association.A linear dependence of the energies of the intermolecular c.t. bands on the I.P. values of the donors, frequently observed for weak interactions,2a is only approxi- mately followed by the compounds investigated. This is not unexpected however, since aliphatic and aromatic donors are involved. Moreover, complex formation may be controlled not only by electronic factors but also by different steric require- ments of the S-derivatives which may influence the position of the c.t. absorption. In this respect, the isoequilibrium relationship of fig.2 between AH and AS, usually shown by series of related equilibria involving moderate changes in structure,2u is also indicative. The plot, which refers to various compounds of table 2, shows some8 - I L 6 - 2 - (d Y % 4 - 1 . \ 4 2 - L------ 10 (5 2 0 -AS/cal mol-' K-I FIG. 2.-Relationship between enthalpies and entropies of formation for c.t. complexes of a series of n-donors with iodine as the common acceptor. Numbers refer to the donors in table 2. 1 The irregularities observed in the AH against A S plot of fig. 2, as well as in other relationships predicted by Mulliken's theory, could hence be due both to a primary steric effect and to a contribution of the mystem in the c.t. interaction of the sulphur atom of the aromatic derivatives. Excluding complexes having aromatic donors in the plot of fig.2, an isoequilibrium temperature of 499 K can be evaluated for these iodine complexes, similar to that reported for the iodine +aliphatic amines complexes (545 K)9 but rather smaller than that estimated by the data for iodine+alkylbenzenes (680 In the latter case the n+tP nature of the complexes leads to a stronger dependence of the complex stability on the energetic, rather than on the entropic, parameter. As to the nature of the donor orbital involved in the S-complexes, we believe that the relatively high stability (benzenes have, comparatively, rather lower K,, values) points to donation from the sulphur centre in both aliphatic and aromatic derivatives. The influence of temperature on the extent of complex formation is also consider- able, leading to enthalpy values characteristic of n-donor complexes.It is known from X-ray studies on complexes of iodine with dibenzyl sulphide l8 and 1,4-dithiane 21 that the S-1-1 atoms all lie in a straight line, which is probably true also for the mercaptan complexes. For the phenyl-derivatives, the spatial distribution may be148 SULPHUR CHARtiE TRANSFER COMPLEXES different from the alkyl-ones, at least as far as the angle between the I-1-S axis and the plane containing the C-S-C atoms is concerned. That S-derivatives act as n- donors is also suggested by the primary steric effect of substituents adjacent to the sulphur, as well as by the position of the halogen band in the far-infrared.22 The possibility of two simultaneous complexes involving both n- and n-electron binding has been proposed in other cases,23 but seems improbable here, as a-acceptors give high intermolecular overlap and so high degree of association with n-donors.THIOPHENOLS Table 3 shows the stability constants at 25°C for somep- and m-substituted thio- phenols. The substituent effect, which follows the Hammett equation (fig. 3), is transmitted rather efficiently to the reaction centre with an increase of the donor properties of the sulphur atom for thiophenols with elect ron-donating substituents ( p = - 1.1 1, linear correlation coefficient R = 0.997). The fact that the two m- derivatives investigated fit the Hammett plot supports a n-a* nature also for the aromatic S-complexes.The reaction constant is very similar to that found in the same solvent for thioanisoles (- 1.3) but rather smaller than that for pyridines ( -2.34),4 as can be expected from the reduced charge localization. A comparison with the acid-base equilibrium constants (K, is the dissociation constant of the donor) shows a linear relationship between pK,, and pK,, as expected from the fact that both protonation and complexation processes obey the Hammett equation and from the assumption that the reaction centre is the same in the two cases. TABLE 3 .-SUBSTITUENT EFFECT ON CHARGE TRANSFER EQUILIBRIA OF SOME SUBSTITUTED THIOPHENOLS WITH IODINE IN CYCLOHEXANE AT 25°C COMPARED WITH THEIR DISSOCIATION CONSTANTS subst ituent KctlM-' PKi ' none 11.6 7.76 P-CH3 18.2 8.03 p-OCHj 24.4 7.99 p-c1 6.4 6.96 P-F 9.1 m-CH, 14.1 7.96 m-C1 4.7 6.74 t, - 0 .4 - 0 . 2 0 . 2 0.4 0 FIG. 3.-Hammett plot for the stability constants of the charge transfer complexes between thio- phenols and iodine in cyclohexane at 25°C.G. REICHENBACH, S. SANTINI AND U. MAZZUCATO 149 DIPHENYL DISULPHIDE Diphenyl disulphide, which is the other partner of the iodine-catalysed isotopic exchange reaction,' was also investigated but complex formation was not observed (maximum concentrations : [I2] = 5 x M, [Ph,S2] = 0.1 M). A similar nega- tive result was also obtained with the 4,4'-dichloro-derivative, while for the 4,4'- dimethoxy-derivative complexation was detected by slight spectral modifications. Their extent, however, did not allow reliable equilibrium measurements to be carried out.A strong decrease of K,, has been reported also for Et2S2 (no. 7) and explained as being due to a smaller donor-acceptor overlap integral for the disulphide than for the monosulphide, owing to the different geometry (1-1 molecule pointing per- pendicularly to the middle of the S-S group in the first case).2c While the reason for the practically undetectable complex formation in the case of Ph2S2 is not com- pletely understood, explanations similar to those advanced for Et,S2 are probably operative.2 The possibility of two stability constants for this complex, corresponding to the complexation with each S-atom, can be discounted for steric reasons. Also in other cases, only 1 : 1 complexes are formed when iodine concentration is not too high, as found e.g.by McCullough et al. with 1,4-dithiane and diselenane." CONCLUSIONS The donor properties of the S-derivatives investigated depend to a small extent on their structure, the stability constants being in the range 4-23 M-l at room temper- ature for the donors of tables 1 and 3. The possibility of direct n-donation for aromatic derivatives seems to be ruled out by the present results for c.t. equilibria and by recent photoelectron spectroscopy data. It should, however, be recalled that the characterization of the complex is only approximate. Like anilines and pyridine~,~ inercaptans and sulphides cannot be classified rigorously into n- or n-donors because the unshared electrons of the nitrogen or sulphur atoms are more or less extensively conjugated with the n-orbitals of the ring.One can thus infer only that the sulphur atom is the donor site in the c.t. interaction with iodine, with a conjugative contribu- tion from the n-system in the aromatic derivatives. We thank Dr. G. Distefano for helpful discussions. work in progress ; see also A. Fava, G. Reichenbach and U. Peron, J. Amer. Chem. SOC., 1967, 89, 6696. (a) see for instance R. Foster, Organic Charge Transfer Complexes (Academic Press, London, 1969) ; (b) H. Tsubomura and R. P. Lang, J . Amer. Chem. SOC., 1961,83,2085 ; (c) M. Good, A. Major, J. Nag-Chaudhuri and S. P. McGlynn, J. Amer. Chem. SOC., 1961, 83, 4329. J. Van der Veen and W. Stevens, Rec. Trav. chim. Pays-Bas, 1963,82,287. G. G . Aloisi, G. Cauzzo and U. Mazzucato, Trans. Furaduy SOC., 1967,63,1858 ; G .G. Aloisi, G. Beggiato and U. Mazzucato, Trans. Faraday SOC., 1970, 66, 3075. A. I. Vogel, A Textbook of Practical Organic Chemistry (Longmans Green, London, 3rd edn, 1957), p. 173. see ref. (9, p. 186. smith and W. Dzcubas, Rec. Trau. chim. Pup-Bas, 1952,71, 1104. P. A. D. de Maine, J. Chem. Phys., 1957,26, 1192. H. Yada, J. Tanaka and S . Nagakura, Bull. Chem. SOC. Jupan, 1960, 33, 1660. N. W. Tideswell and J. D. McCullough, J. Amer. Chem. SOC., 1957, 79, 1031. J. Jander and G. Turk, Chem. Ber., 1965, 98, 894. ' C. Van de Stolpe, Thesis (Amsterdam, 1954) ; J. A. A. Ketelaar, C. Van de Stolpe, A. Goud- l 1 J. D. McCullough and I. C . Zimmerman, J. Phys. Chern., 1961, 65, 888. l3 A. K. Chandra and D. C. Mukherjee, Trans. Furaduy SOC., 1964, 60, 62,150 SULPHUR CHARGE TRANSFER COMPLEXES l 4 D.C. Frost, F. G. Herring, A. Katrib, C. A. McDowell and R. A. N. McLean, J . P h j : ~ . CI~crn., 1972, 76, 1030. I A. D. Baker, D. P. May and 1). W. Turner, J . Chern. Soc. B, 1968, 22. * G. Scatchard, A m . N. Y. Acad. Sci., 1949, 51, 660. l 7 D. A. Deranleau, J . Ainer. Cliem. SOC., 1969, 91, 4044 and 4050. 0. Hassel, Proc. Cheni. Soc., 1957, 250. l 9 D. Dodson, R. Foster, A. A. S. Bright, M. I. Foreman and J . Gorton, J. Chem. SUC. B, 1971, 1283. 2o R. M. Keefer and L. J. Andrews, J. Amer. Cliern. SOC., 1955, 77, 2164. 21 J. D. McCullough, G. J. Chao and D. E. Zuccaro, Acta Cryst., 1959, 12, 815. 22 M. Yamada, M. Saruyama and K. Aida, Spectrochim. Acta, 1972, 28A, 439. 23 see for instance D. B.Wayland and R. S. Drago, J. Amer. Chem. Soc., 1964,86, 5240. 24 F. G. Bordwell and H. M. Andersen, J. Amer. C'hem. SOC., 1953, 75, 6019.Charge Transfer Complexes of Iodine with Mercaptans andRelated Sulphur CompoundsBY G. REICHENBACH, S. SANTINI AND U. MAZZUCATOlstituto di Chiinica Fisica, Universith di Perugia, 06100 Perugia, ItalyReceived 26th June, 1972Charge transfer complexes of iodine (u acceptor) with ethyl and phenyl mercaptan, and diphenylsulphide and disulphide (n-donors) have been studied in cyclohexane at 25°C. Spectral data,stability constants, enthalpies and entropies of complexation are reported. As comparison withliterature data for related compounds, the stability of the complexes is correlated with the donorstructure and discussed in terms of electronic and steric factors.The effect of substituents on Kct ofthiophenols has been investigated leading to an Hammett reaction constant of - 1.1 1.In the course of a study of the iodine-catalysed isotopic exchange between diphenyldisulphide (Ph,S,) and thiophenol (PhSH),' it seemed interesting to investigate thepossible role played by the charge transfer (c.t.) complexes formed between thecatalyst and the sulphur derivatives.The c.t. complexes between iodine (a acceptor) and the n-donors dialkyl sulphides(R2S) and disulphides (R2S2) are well known,2 as are those of iodine+alkyl arylsulphides (PhSR) (thioani~oles).~ However, data on aryl mercaptans (thiophenols)are absent and the study of the equilibrium iodine+ethyl mercaptan (RSH) hasnot been completed because of its low stability.2cThe study of the complexes of iodine + thiophenols and other sulphur-derivativeswas thus undertaken along the same lines as a series of iodine complexes of similarw+o* nature which used more basic pyridine donor^.^ Stability constants and thermo-dynamic data are reported for PhSH, EtSH and Ph,S and the results discussed bycomparison with literature data for related complexes.EXPERIMENTALTwice sublimed iodine (Carlo Erba RP) was resublimed with potasium ioaide and kept ina dessicator. Cyclohexane (Carlo Erba RP) was purified following Vogel,s dried and storedunder nitrogen.All the donors were commercial products crystallized before use fromabsolute ethanol or distilled under nitrogen.Nitrogen (99.9 %) was purified by bubblingthrough a Fieser solution,6 dried and then saturated with cyclohexane.The stability constants were calculated from spectrophotometric measurements on theassumption of 1 : 1 complexation (see Discussion).A single beam Unicam SP500/2 spectrophotometer (with 1 cm path quartz cells and athermostatted jacket) was used. The same donor-solvent system was used as a blank whennecessary, to compensate for the free donor absorption ([D]>[I,], weak complex). Prelim-inary spectra were run with a double beam Optica CF4-DR spectrophotometer.The Van de Stolpe equation was applied in the form :1 1 1where E~ is the apparent molar extinction coefficient (observed absorbance/total iodineconcentration), Ef and cc are the molar extinction coefficients for the free iodine and the14144 SULPHUK CHARGE TRANSFER COMPLEXEScomplex, and [D] is the donor molar concentration.Five different donor concentrationswere generally used in each experiment. The initial concentrations were 5x M foriodine, 10-1-10-2 M for thiophenols, 5 x 10-'-1.7 x For EtSH, concen-trations of the donor were 2~ 10-l-4.3~ lop3 M in the ultraviolet and 0.4-0.1 M in thevisible experiments. Equilibrium constants and molar extinction coefficients for thiophenolswere calculated at different wavelengths in the range 320-360nm, at the tail of the chargetransfer band. In the case of EtSH and Ph2S, Kct was calculated also at the c.t. peak, wherethe donor does not overlap. A good linearity of the experimental plots was found in everycase (correlation coefficients were in the range 0.9969 to 0.9999).The reported Kct valuesare averages of 3-5 values calculated at different A (mean deviations are of the order of 1 %).When the perturbed iodine band was sufficiently blue-shifted with respect to the free iodine,measurements were also performed in the visible range, producing good agreement with theultraviolet data.Under these conditionsall the systems were stable over a period of several hours, as shown by the constancy of thespectrum at the experimental temperatures.M for Ph2S.All the measurenients were done under a nitrogen atmosphere.RESULTS AND DISCUSSIONThe n-donor S-derivatives display a basic character intermediate between thecorresponding 0- and N-derivatives.In order to investigate the donor properties ofthe sulphur atom towards iodine and to compare them with those of other hetero-atoms, ethyl and phenyl mercaptan and diphenyl sulphide were first investigated.Table 1 shows the equilibrium constants at different temperatures and the thermo-dynamic parameters for these compounds. Table 2 collects similar data reported inthe literature for related donors together with the donor ionization potentials, forcomparison purposes. For all the three donors of table 1 the perturbed iodine bandshifts towards the blue by complexation and a new c.t. absorption appears in the U.V.region, as shown in fig. 1 for EtSH.TABLE 1 .-TEMPERATURE DEPENDENCE AND KELATED THERMODYNAMIC PARAMETERS OF THE C.T.EQUILIBRIA OF IODINE WITH ETHYL AND PHENYL MERCAPTANS AND DIPHENYL SULPHIDE INCY CLOHEX ANEdonor tl"C KctlM-1 -AH/kcal mol-1 -AS/cal mol-1 K-1EtSH 17 28.7 4.625 23.240 15.7PhSH 16 14.725 11.634.6 9.340 7.9PhzS 17.5 4.925 3.935 2.84.35.79.09.616.6ETHYL A N D PHENYL MERCAPTANIntroduction of an ethyl group in H,S (Kct = 1.3 M-I at 25°C) l2 considerablyincreases the base strength of sulphur. In the aliphatic series EtNH,, EtSH andEtOH (no.4, 1, 3 in table 2), the S-derivative gives weaker c.t. complexes than theamine but stronger than the alcohol, as foreseen from the blue-shift of the iodinepeak 2c and expected from the relative basicity of the donor atoms and from theirionization potentials. The combination of the inductive and mesomeric interactionG.REICHENBACH, S . SANTINI AND U. MAZZUCATO 145between the lone pair and the n-electrons of the aromatic ring, which have a strongelectron-releasing influence on the base strength of the nitrogen atom (Kct decreasesabout fifty times on going from ethyl- to phenyl-amine 13), have, however, a smalloverall effect in this case, the S-phenyl-derivative displaying almost the same stabilityas EtSH. The smaller effect in PhSH compared to PhNHz could be due both to thegreater dimensions of the sulphur centre (the orbital overlap with iodine is lesshindered by the phenyl group) and to a return of charge from the ring to the 3d-orbitals of the sulphur, which enhances its donating ability.TABLE 2.-cHARGE TRANSFER EQUILIBRIA OF IODINE WITH ETHYL AND PHENYL MERCAPTANS ANDOTHER ALIPHATIC AND AROMATIC RELATED COMPOUNDS IN INERT SOLVENT AT 25°Cdonor(1) EtSH(2) MeOH(3) EtOH(5) EtOEt8(6) EtSEt 2b(7) EtSSEt 26(8) PhSH(4) EtNHz’(9) PhSCH3(10) PhzS(12) S(C2H4)2S(1 1) MeSeMeKctIM-123.20.440.390.84560 - 170- 511.612.13.947 177-AH/ -AS/kcal moi-1 cal mol-1 K-14.6 9.01.9 3.32.1 4.17.4 12.34.3 9.97.8 15.94.6 12.34.3 9.65.5 135.7 16.68.6 16.46.2 12.1W n m289242243246249302310(310/sh)3333 50433N 300solventcyclohexaneCC14CC14n-heptanen-hept anen-heptanecyclohexanecyclohexanecyclohexaneCC14CC14cc14I .P.* lev9.310.810.58.99.58.48.38.38.98.28.78.5*Ionization potentials are taken from K.Watanabe, T. Nakayama and J. Mottl, J. Quant.Specfr. Radiative Transfer, 1962, 62, 369, except for PhSCH3 (B. G. Gowenlock, J. Kay and J. R.Mayer, Trans. Faraday SOC., 1963,59,2463), for Ph2S (G. Innorta, personal communication) and forS(C2H4)2S (R. W. Kiser and E. J. Gallegos, J. Phys. Chem., 1962,66,947).0.40.33.23.1FIG. 1.-Absorption spectra of the system ethyl mercaptan+iodine in cyclohexane at 25°C: A,0.1 M ethyl mercaptan ; B, c.t. band (0.025 M EtSH+ 5 x M 12) ; C, 4.5 x M iodine;D, perturbed iodine (4.5 x M I, + 0.3 M EtSH). Cell path 1 cm146 SULPHUR CHARGE TRANSFER COMPLEXESRecent data from photoelectron spectroscopy l4 seem to agree with this behaviour,as the first I.P.of PhSH (assigned to a lone pair electron with considerable amount ofbenzene ring character) is only at 8.3 eV. The situation is different for other aromaticcompounds such as PhOH and PhNH2, where the lowest I.P. values are assigned tothe n-system while n-I.P. values are at rather higher levels (1 1.3 and 10.6 eV for PhOHand PhNHz, respectively).15 These values could explain why phenol acts as a n-donorand aniline acts as both an 12- and n-donor, depending on the type of acceptor.2a Thedecrease of n-I.P. from PhOH to PhSH (already observed in table 2 from EtOH toEtSH) and the consequent interchange of the n and n orbital energy levels seems thusto be responsible for the change of the c.t. interaction from 71 to n.SULPHIDESReplacement of the hydrogen atom of mercaptans by alkyl- or aryl-groups hasdifferent effects.The complex Et,S (no. 6) displays the highest K,, observed for theseS-derivatives, due to the electron-donating effect of the alkyl groups. HoweverPhSCH3 (no. 9) has a stability similar to PhSH. This may indicate that the increaseof charge density on the sulphur atom due to the CH,-group is matched by a partialcharge redistribution on the ring and by a small steric effect.Steric hindrance becomes even more important in the diphenyl-derivative whereK,, is reduced to less than 4 M-l. The close approach to the reaction center of anacceptor molecule as large as iodine is, in fact, prevented by two rings and the freedomof phenyl rotation is decreased in the complex. This should reflect a greater negativecomplexation entropy, as experimentally found (- 16.6 cal mol-l K-l).NATURE OF THE COMPLEXESAll complexes were assumed to be of 1 : 1 stoichiometry. Arguments in favourof this assumption are : (1) the good linearity of the experimental plots (correlationcoefficients in the range 0.9969 to 0.9999) in a rather wide concentration interval ; (2)relatively constant K,, values with analytical wavelength (see experimental) ; (3)linearity of the Van de Stolpe plots was further checked by Scatchard plots l6 forEtSH and Ph2S, and satisfactory correlation coefficients (R = 0.9961 and 0.9982respectively) for 10 measurements in a wide range of saturation fractions (s from0.09 to 0.82 for EtSH and from 0.065 to 0.67 for PhzS) were obtained.For PhSH itwas difficult to extend the donor concentration range because of its strong absorptionin the analytical spectral region.Other minor points in favour of 1 : 1 complexation come from the analogy withthe PhSCH, + Iz complex (for which analysis indicates 1 : 1 a~sociation),~ and withthe (PhCH,),S + I, complex (for which structural data indicate bimolecular associa-tion in the solid state)." The presence of termolecular species cannot be completelyruled out l9 but, in the light of the above arguments, we think that the observed K,,values closely represent the correct values for 1 : 1 association.A linear dependence of the energies of the intermolecular c.t. bands on the I.P.values of the donors, frequently observed for weak interactions,2a is only approxi-mately followed by the compounds investigated.This is not unexpected however,since aliphatic and aromatic donors are involved. Moreover, complex formationmay be controlled not only by electronic factors but also by different steric require-ments of the S-derivatives which may influence the position of the c.t. absorption. Inthis respect, the isoequilibrium relationship of fig. 2 between AH and AS, usuallyshown by series of related equilibria involving moderate changes in structure,2u isalso indicative. The plot, which refers to various compounds of table 2, shows som8 -I L 6 - 2 - (d Y% 4 -1 .\42 -L------ 10 (5 2 0-AS/cal mol-' K-IFIG. 2.-Relationship between enthalpies and entropies of formation for c.t.complexes of a series ofn-donors with iodine as the common acceptor. Numbers refer to the donors in table 2.1The irregularities observed in the AH against A S plot of fig. 2, as well as in otherrelationships predicted by Mulliken's theory, could hence be due both to a primarysteric effect and to a contribution of the mystem in the c.t. interaction of the sulphuratom of the aromatic derivatives.Excluding complexes having aromatic donors in the plot of fig. 2, an isoequilibriumtemperature of 499 K can be evaluated for these iodine complexes, similar to thatreported for the iodine +aliphatic amines complexes (545 K)9 but rather smaller thanthat estimated by the data for iodine+alkylbenzenes (680 In the latter casethe n+tP nature of the complexes leads to a stronger dependence of the complexstability on the energetic, rather than on the entropic, parameter.As to the nature of the donor orbital involved in the S-complexes, we believe thatthe relatively high stability (benzenes have, comparatively, rather lower K,, values)points to donation from the sulphur centre in both aliphatic and aromatic derivatives.The influence of temperature on the extent of complex formation is also consider-able, leading to enthalpy values characteristic of n-donor complexes.It is knownfrom X-ray studies on complexes of iodine with dibenzyl sulphide l8 and 1,4-dithiane 21that the S-1-1 atoms all lie in a straight line, which is probably true also for themercaptan complexes. For the phenyl-derivatives, the spatial distribution may b148 SULPHUR CHARtiE TRANSFER COMPLEXESdifferent from the alkyl-ones, at least as far as the angle between the I-1-S axis andthe plane containing the C-S-C atoms is concerned. That S-derivatives act as n-donors is also suggested by the primary steric effect of substituents adjacent to thesulphur, as well as by the position of the halogen band in the far-infrared.22 Thepossibility of two simultaneous complexes involving both n- and n-electron bindinghas been proposed in other cases,23 but seems improbable here, as a-acceptors givehigh intermolecular overlap and so high degree of association with n-donors.THIOPHENOLSTable 3 shows the stability constants at 25°C for somep- and m-substituted thio-phenols. The substituent effect, which follows the Hammett equation (fig.3), istransmitted rather efficiently to the reaction centre with an increase of the donorproperties of the sulphur atom for thiophenols with elect ron-donating substituents( p = - 1.1 1, linear correlation coefficient R = 0.997). The fact that the two m-derivatives investigated fit the Hammett plot supports a n-a* nature also for thearomatic S-complexes. The reaction constant is very similar to that found in thesame solvent for thioanisoles (- 1.3) but rather smaller than that for pyridines( -2.34),4 as can be expected from the reduced charge localization. A comparisonwith the acid-base equilibrium constants (K, is the dissociation constant of the donor)shows a linear relationship between pK,, and pK,, as expected from the fact that bothprotonation and complexation processes obey the Hammett equation and from theassumption that the reaction centre is the same in the two cases.TABLE 3 .-SUBSTITUENT EFFECT ON CHARGE TRANSFER EQUILIBRIA OF SOME SUBSTITUTEDTHIOPHENOLS WITH IODINE IN CYCLOHEXANE AT 25°C COMPARED WITH THEIR DISSOCIATIONCONSTANTSsubst ituent KctlM-' PKi 'none 11.6 7.76P-CH3 18.2 8.03p-OCHj 24.4 7.99p-c1 6.4 6.96P-F 9.1m-CH, 14.1 7.96m-C1 4.7 6.74t,- 0 .4 - 0 . 2 0 . 2 0.40FIG. 3.-Hammett plot for the stability constants of the charge transfer complexes between thio-phenols and iodine in cyclohexane at 25°CG. REICHENBACH, S. SANTINI AND U. MAZZUCATO 149DIPHENYL DISULPHIDEDiphenyl disulphide, which is the other partner of the iodine-catalysed isotopicexchange reaction,' was also investigated but complex formation was not observed(maximum concentrations : [I2] = 5 x M, [Ph,S2] = 0.1 M).A similar nega-tive result was also obtained with the 4,4'-dichloro-derivative, while for the 4,4'-dimethoxy-derivative complexation was detected by slight spectral modifications.Their extent, however, did not allow reliable equilibrium measurements to be carriedout. A strong decrease of K,, has been reported also for Et2S2 (no. 7) and explainedas being due to a smaller donor-acceptor overlap integral for the disulphide thanfor the monosulphide, owing to the different geometry (1-1 molecule pointing per-pendicularly to the middle of the S-S group in the first case).2c While the reasonfor the practically undetectable complex formation in the case of Ph2S2 is not com-pletely understood, explanations similar to those advanced for Et,S2 are probablyoperative.2 The possibility of two stability constants for this complex, correspondingto the complexation with each S-atom, can be discounted for steric reasons.Also inother cases, only 1 : 1 complexes are formed when iodine concentration is not toohigh, as found e.g. by McCullough et al. with 1,4-dithiane and diselenane."CONCLUSIONSThe donor properties of the S-derivatives investigated depend to a small extenton their structure, the stability constants being in the range 4-23 M-l at room temper-ature for the donors of tables 1 and 3. The possibility of direct n-donation foraromatic derivatives seems to be ruled out by the present results for c.t.equilibria andby recent photoelectron spectroscopy data. It should, however, be recalled that thecharacterization of the complex is only approximate. Like anilines and pyridine~,~inercaptans and sulphides cannot be classified rigorously into n- or n-donors becausethe unshared electrons of the nitrogen or sulphur atoms are more or less extensivelyconjugated with the n-orbitals of the ring. One can thus infer only that the sulphuratom is the donor site in the c.t. interaction with iodine, with a conjugative contribu-tion from the n-system in the aromatic derivatives.We thank Dr. G. Distefano for helpful discussions.work in progress ; see also A. Fava, G. Reichenbach and U. Peron, J. Amer. Chem. SOC., 1967,89, 6696.(a) see for instance R. Foster, Organic Charge Transfer Complexes (Academic Press, London,1969) ; (b) H. Tsubomura and R. P. Lang, J . Amer. Chem. SOC., 1961,83,2085 ; (c) M. Good,A. Major, J. Nag-Chaudhuri and S. P. McGlynn, J. Amer. Chem. SOC., 1961, 83, 4329.J. Van der Veen and W. Stevens, Rec. Trav. chim. Pays-Bas, 1963,82,287.G. G . Aloisi, G. Cauzzo and U. Mazzucato, Trans. Furaduy SOC., 1967,63,1858 ; G . G. Aloisi,G. Beggiato and U. Mazzucato, Trans. Faraday SOC., 1970, 66, 3075.A. I. Vogel, A Textbook of Practical Organic Chemistry (Longmans Green, London, 3rd edn,1957), p. 173.see ref. (9, p. 186.smith and W. Dzcubas, Rec. Trau. chim. Pup-Bas, 1952,71, 1104.P. A. D. de Maine, J. Chem. Phys., 1957,26, 1192.H. Yada, J. Tanaka and S . Nagakura, Bull. Chem. SOC. Jupan, 1960, 33, 1660.N. W. Tideswell and J. D. McCullough, J. Amer. Chem. SOC., 1957, 79, 1031.J. Jander and G. Turk, Chem. Ber., 1965, 98, 894.' C. Van de Stolpe, Thesis (Amsterdam, 1954) ; J. A. A. Ketelaar, C. Van de Stolpe, A. Goud-l 1 J. D. McCullough and I. C . Zimmerman, J. Phys. Chern., 1961, 65, 888.l3 A. K. Chandra and D. C. Mukherjee, Trans. Furaduy SOC., 1964, 60, 62150 SULPHUR CHARGE TRANSFER COMPLEXESl 4 D. C. Frost, F. G. Herring, A. Katrib, C. A. McDowell and R. A. N. McLean, J . P h j : ~ . CI~crn.,1972, 76, 1030.I A. D. Baker, D. P. May and 1). W. Turner, J . Chern. Soc. B, 1968, 22.* G. Scatchard, A m . N. Y. Acad. Sci., 1949, 51, 660.l 7 D. A. Deranleau, J . Ainer. Cliem. SOC., 1969, 91, 4044 and 4050.0. Hassel, Proc. Cheni. Soc., 1957, 250.l 9 D. Dodson, R. Foster, A. A. S. Bright, M. I. Foreman and J . Gorton, J. Chem. SUC. B, 1971,1283.2o R. M. Keefer and L. J. Andrews, J. Amer. Cliern. SOC., 1955, 77, 2164.21 J. D. McCullough, G. J. Chao and D. E. Zuccaro, Acta Cryst., 1959, 12, 815.22 M. Yamada, M. Saruyama and K. Aida, Spectrochim. Acta, 1972, 28A, 439.23 see for instance D. B. Wayland and R. S. Drago, J. Amer. Chem. Soc., 1964,86, 5240.24 F. G. Bordwell and H. M. Andersen, J. Amer. C'hem. SOC., 1953, 75, 6019

ISSN:0300-9599    

DOI:10.1039/F19736900143

出版商:RSC    

年代:1973

数据来源: RSC

摘要:

Hydrogen Bonding in some Adducts of Oxygen Bases with Acids Part 7.-Thermodynamic Study by Infra-red Spectroscopy of the Association of Chloroacetic Acids with Some Oxygen Bases BY D. HADZI AND J. RAJNVAJN Department of Cheniistry and Institute Boris KidriE, University of Ljubljana, Ljubljana, Yugoslavia Received 3rd July, 1972 The association between the chloroacetic acids (mono-, di-, and tri-) and a series of bases (dimethyl, dibenzyl, diphenyl sulphoxide, triphenylphosphine oxide, diphenyl selenoxide) in carbon tetrachloride was studied using the absorbance of the OH stretching band of the free acid. Kas, AGO, AH" and ASo of association were derived from the spectrophotometric data. A linear relationship between AH" and the pKa of the acids was found. The thermodynamic quantities reflect also the basicity of the proton acceptors but not in a simple relation.The strongest hydrogen bond is formed between trichloroacetic acid and diphenyl selenoxide, -AH" being -67 kJ/mol. Chloroacetic acids form strong hydrogen bonds with bases such as sulphoxides, phosphine oxides and pyridine N-oxide.l* The infra-red spectroscopic features indicate an increasing hydrogen bond strength in the acid series from mono- to tri- chloroacetic (MCA, DCA, TCA) and from sulphoxides to diphenyl selenoxide (DPSeO) and pyridine-N-oxide (PyO). This order of hydrogen bond strength corresponds to the acidity and basicity, re~pectively.~ The strongest bonds appear to be formed between TCA and DPSeO, triphenylarsine oxide or PyO and the spectra suggest that the hydrogen bonds in these adducts are nearly symmetrical.The subsequent X-ray structure determination of the crystalline adduct of TCA with PyO has shown that the hydrogen bond is indeed extremely short.4 However, thermo- dynamic data on such hydrogen bonds are limited to bihalide ions.5 Therefore, we have determined the equilibrium constants at four temperatures for the association of the chloroacetic acids with diphenyl sulphoxide (DPSO), dibenzyl sulphoxide (DBSO), dimethyl sulphoxide (DMSO), triphenylphosphine oxide (TPPO) and diphenyl selenoxide. The results are quantitatively in accord with expectations, i.e., that the enthalpies of association are extremely high for the strongest hydrogen bond forming acid-base couples. EXPERIMENTAL Carbon tetrachloride of AP grade (Kemika) was dried over Pz05 in a recycling still, and great care was taken to avoid contamination with atmospheric moisture at all stages.The solvent contained less than mol/l. HzO as estimated from the infra-red water bands. The acids and the bases were of good commercial grade, but were recrystallised or distilled before use. Particular attention was paid to drying. Liquids were kept for several days over fresh molecular sieves and the solids were stored in a dessicator over Pz05. DMSO was dried additionally with CaH,. Purity was checked by infra-red spectroscopy. Stock solutions of acids and bases were prepared shortly before the measurements by weighing. The sample solutions were prepared from stock solutions of components and addition of solvent in a calibrated cylinder which was directly connected with the drying still.Mixing in the cylinder was provided by a glass clad iron rod moved by a magnet. The solutions 151I52 were prepared at 22°C and corrected for volume changes with temperature. The cells had an optical path of 8.95 cm. They were made of brass, internally gold plated and fitted with quartz windows. Water was circulated through the jackets from an ultrathermostat. The temperature was kept stable within _+O.l"C. The Perkin-Elmer model 521 was used for spectrophotometry in a double beam mode with two identical cells, one being filled with the solvent from the same batch as used in making up the sample. Two photometric redordings were made of each sample. If the two did not agree to less the 0.5 % transmission, a third recording was made and the average used for calculations.HYDROGEN BONDING I N ADDUCTS OF OXYGEN BASES CALCULATIONS AND RESULTS In diluted acid-base systems, the carboxylic acid monomers are in equilibrium with the dimers and with the acid-base adducts. The equilibrium constant for the latter has to be derived from the absorbance of the free OH stretching band only because of the solvent absorption in other regions. This requires a separate determination of the monomer-dirner dissociation constant (Kd) under similar conditions. The details of this work will be published elsewhere; here only the results will be given (table 1). They agree well with those obtained by other authors.6 The acid-base association constants Kas was calculated from the expression : &, = (CA - M - 2M2 /&)/M(C, - CA -!- M -!- 2M2 I&) in which CA is the total concentration of the acid, M of the monomer and CB the total concentration of the base.It is assumed (i) that concentrations may be used throughout instead of activities, (ii) that the acid dimerisation constant is not in- fluenced by the presence of the base in the medium and (iii) that there are no other equilibria. Assumptions (i) and (ii) are justified in view of the high dilution. This applies also to the association between base molecules. At least for DMSO, it is known that the self-association constant is very low and thus practically only the free base should be present at high dilutions.' The possibility of the base-acid association in 1 : 2 ratio will be discussed in the last paragraph. TABLE 1 .-DISSOCIATION CONSTANTS OF CHLOROACETIC ACIDS IN CC14 acid temp./"C Kd/mol monochloroacetic 20 (2.59k0.13)~ 25 (3.14+0.23)x 30 (4.15& 0.34) x 35 (5.36+ 0.53) x dichloroacet ic trichloroacet ic 20 (3.10+ 0.22) x lo-' 25 (3.86k0.29)~ 30 (5.21 20.37) x 35 (6.51k0.29)~ 20 (1.50+ 0.13) x 25 (1.95 f 0.14) x 30 (2.6550.21)~ 35 (3.65f0.44)~ The concentrations of the solutes were in the range of 10-4molar.For each acid-base pair, 5 to 8 solutions were measured in which the acid to base molar ratios varied within the limits 1 : 1 and 1 : 1.3. Within these limits, photometric accuracy was maximised. The absorbances were determined at 4 temperatures (20, 25, 30, 35°C). K,, for 25°C given in table 2 are the least squares values.The deviations in K,,D. HAD% AND J . RAJNVAJN 153 are within the limits of 12 1.2 to 1 6 . 6 %. The enthalpies of association were calculated from the (log K,,, T-l) dependence by least squares. AH" values show standard deviations within the limits + 3 to AGO and AS" were calculated from the usual expression. The largest deviations are shown by the entropy changes, but on the whole the deviations in the thermodynamic quantities are well within the limits usual for spectrophotometric determinations. A linear relationship between AH" and AS" for all the three series of acid-base adducts exists as usual in such cases. The correlations obtained by least squares treatment are : 12 %. MCA: -AS0+2.15 (-AH") +6.65 DCA -AS" + 2.08 (-AH") + 4.98 TCA: -AS0+3.10 (-AH")+ 18.61.DISCUSSION The very reasonable deviations of the resulting quantities in table 2, and the parallelism shown by the three series of acids, do not exclude certain systematic errors. The most serious source of experimental error is probably the water content of the solvent. Completely dry carbon tetrachloride can be obtained only after very lengthy procedures which were not practicable with the large quantities used in our experimental work, and we had to tolerate a minimum concentration that was at least one order of magnitude less than that of the species investigated. Water may act both as a proton acceptor and donor. The association of water with DMSO and with phosphine oxides has been investigated by several authors.*-l O The association constants are of the order of 0.5 mol l.-l and thus much smaller than the K,, for the acid-base equilibria.Not much is known about the influence of small water con- centrations in non-polar solvents upon the dimerisation of carboxylic acids,12 but one expects to account for this by using a solvent with roughly the same water content for determining both the dimerisation and the acid-base association constants. As a final check we have deliberately added small amounts of water to some acid-base sample solutions and determined K,, at four temperatures as usual, but the results were within the limits of error. TABLE 2.-THERMODYNAMIC QUANTITIES OF ASSOCIATION OF CHLOROACETIC ACIDS WITH SOME 0x0 BASES IN CC14 (25°C) -AS01 acid base €&/I. mol-1 -AGo/kJ md-1 -AH"/kJ mol-1 J mol-1 K--l monochloroacetic DPSO DBSO DMSO TPPO DPSeO dichloroacetic DPSO DBSO DMSO TPPO DPSeO trichtoroacetic DPSO DBSO DMSO TPPO DPSeO (1.548k0.034)X lo3 (4.105f0.076)~ lo3 (5.631k0.088)~ lo3 (7.161+0.101)x lo3 (2.569+0.089) x lo4 (2.375k0.108)~ lo3 (1.248f0.016)~ lo4 (1.964f0.120) x lo4 (4.361 f 0.098) x lo4 (1.487 0.067) X 1 O5 (4.772+ 0.232) x lo3 (3.501 k 0.067) x lo4 (4.530f 0.113) x lo4 (7.003 f 0.154) X lo4 (2.41 1 k0.068) x lo5 18.20+0.04 20.63 0.04 21.42+ 0.04 22.01 f 0.04 25.19+ 0.08 19.29f 0.12 23.39k0.04 24.52f 0.16 26.48f 0.04 29.54f 0.12 2 1 .oo+ 0.1 2 25.94f 0.04 26.57f 0.04 27.66f0.04 30.71 & 0.08 27.03+ 3.55 32.26f 3.14 36.15f 1.63 42.59f 2.01 49.20f 3.05 28.87f 0.13 38.03f 1.51 39.46+ 1.46 51.17f 2.51 59.455 2.09 31.13f 1.30 40.632 2.1 3 42.34f 1.26 58.45f 5.10 66.69-1 5.48 28.9f 12.1 39.3f 10.5 49.4f 5.4 69.5+ 6.7 80.8f 10.0 31.8+ 6.7 49.4f 5.0 50.2f 5.0 83.3f 8.4 99.9+ 7.1 33.9k4.2 48.9f 7.1 53.1 k4.2 102.5f 17.2 120.l+ 18.4154 HYDROGEN DONDING I N ADDUCTS OF OXYGEN DASES The other disputable point concerns the equilibria not accounted for iii the cal- culations.They are : K : I K P 2AH+B+(AH), . . .B and AH+B+A-. . . HBF. The first type of equilibrium is known in the association of phenols with carbonyl type acceptors l 3 and in isothiocyanic acid with DMSO and trioctylphosphine oxide (TOPO).14 The K21 are smaller than Ka, by one order of magnitude, particu- larly with DMSO, and thus should not disturb the results obtained for the sulphoxides. This may not be so in the case of association of phosphine oxides with chloroacetic acids.We have also made a series of measurements with TOP0 as proton acceptor and obtained much higher values for the thermodynamic quantities than with TPPO. For instance, -AH" for the association with MCA, DCA and TCA came to 54, 96, and 121 kJ/mol, respectively, which seems excessively large ; therefore, they are not included in table 2. Attempts were made to extend the treatment of data so as to include the K,, since this may be important in this case. However, a more extended range of concentrations would be needed for a proper assessment of KZ1. Absorb- ances under such conditions are of low photometric accuracy, so other methods wil have to be used to clarify this point. The proton transfer equilibrium cannot be assessed from the free OH bands only.However, since in the overall spectra of more concentrated solutions in CC14 no indications of proton transfer were observed, the neglect of this equilibrium is justified. The equilibrium constants for the acid-base association. Ka, (25°C) and the negative enthafpies -AH" (table 2) are larger than the corresponding values for the acid dimerisation, except for the weakest base DPSO. However, if the enthalpy per hydrogen bond is considered, then even in this case the acid-base hydrogen bond is stronger than the bond in the dimer. The sequence of proton donorship is clearly mono- < di- < tri-chloroacetic acid. The relationship between - AH" and the pK, of the acids is linear. The appearance of the A, B, C bands in the infra-red spectra of the acid-base adducts,l their frequency and, particularly, the intensity changes when going from mono- to tri-chloroacetic acid are consistent with the increase of the K,, and -AH", and thus with the strengthening of the hydrogen bond in the acid sequence.The sequence of hydrogen bonding ability of the bases expected from their protonation constants and from the features in the infra-red spectra of acid-base adducts is as follows : diphenyl sulphoxide < dibenzyl sulphoxide < dimethyl sulphoxide < triphenyl- phosphine oxide < diphenyl selenoxide. This sequence is borne out also by the K,, and the -AHo values. These values are extremely high for the strongest hydrogen bond forming acid-base pair, but one might have expected a larger difference between TPPO and DPSeO as proton accept- ors.The infra-red spectrum of the adduct of DPSeO and TCA indicates that here the strongest type of hydrogen bonding is that close to the fully symmetrical 0ne.l The change in the spectral features on going from the adduct of TCA with TPPO (type (i), ref. (1)) to that of TCA to DPSeO (type ii) is certainly great, involving a vOH differ- ence of at least 1000 cm-1 whereas the difference in enthalpies is only -8 kJ mol. that the enthalpy of association does not directly reflect the strength of the hydrogen bond because it includes also energy consuming processes of which the main one is the stretching of the 0-H bond. Sellier and Wojtowiak have proposed a correction for this effect." We have tried to apply it It has been stressed beforeD .HADZI AND J . RAJNVAJN 155 to the present examples, putting AvOH in the range 900 cm-1 to 1200 cm-l. HOW- ever, the enthalpies of bond formation calculated in this way were larger than the - AHo of association determined directly by factors of 3 to 5 and thus quite unreason- able. The relationships between the changes in the vibrational spectra due to hydrogen bonding and the thermodynamic quantities have been largely discussed for the bonds of weak and intermediate strength,16 but the conclusions are not necessarily applicable to bonds approaching the symmetrical. Further work is needed to progress in this direction. Thanks are expressed to the Boris KidriE Fund for financial assistance. J. R. is grateful to the Boris KidriE Institute (Belgrade) for leave of absence.l D. Hadii and N. Kobilarov, J. Chem. SOC. A , 1966,439. D. Had& H. Ratajczak and L. Sobczyk, J. Chem. SOC. A, 1967,48. D. Had& C. Klofutar and S. Oblak, J. Chem. SOC. A, 1968,905. L. GoliiS, D. Ha& and F. Lazarini, Chem. Comm., 1971,860 D. G. Tuck, in Progress in Inorganic Chem., ed. F. A. Cotton (Wiley, New York, 1968), vol. 9, p. 161. R. E. Kagarise, Naval Res. Lab. Rep. 1957,4955. ' R. H. Figueroa, E. Roig and H. H. Szmant, Specfrochim. Acta, 1966, 22,587. * J. T. Conocchioli, M. T. Tocher and R. M. Diamond, J. Phys. Chem., 1965,69,1106. J. R. Holmes, D. Kivelson and W. C. Drinkard, J. Amer. Chem. SOC., 1962,84,4677. lo G. Roland and G. Duyckaerts, Specfrochim. Acta, 1966,22,793. l1 S. F. Ting, S. M.Wang and N. C. Li, Canad. J. Chem., 1967,45,425. l2 J. de Villepin, A. Lautik and M.-L. Josien, Ann. Chim., 1966, 1, 365. l4 S. Detoni, D. Hadii, R. Smerkolj, J. Hawranek and L. Sobczyk, J. Chem. SOC. A, 1970,2851. l 5 G. Sellier and B. Wojtowiak, J. Chim. phys., 1968, 65, 936. l6 E. M. Atnett, L. Joris, E. Mitchell, T. S. S. R. Murty, T. M. Gorrie and P. v. R. Schleyer, G. Nagarajan, Z. Nafurforsch., 27a, 1972,221. J. Amer. Chem. SOC., 1970,92,2365.Hydrogen Bonding in some Adducts of Oxygen Bases with AcidsPart 7.-Thermodynamic Study by Infra-red Spectroscopy of the Association ofChloroacetic Acids with Some Oxygen BasesBY D. HADZI AND J. RAJNVAJNDepartment of Cheniistry and Institute Boris KidriE, University of Ljubljana,Ljubljana, YugoslaviaReceived 3rd July, 1972The association between the chloroacetic acids (mono-, di-, and tri-) and a series of bases (dimethyl,dibenzyl, diphenyl sulphoxide, triphenylphosphine oxide, diphenyl selenoxide) in carbon tetrachloridewas studied using the absorbance of the OH stretching band of the free acid.Kas, AGO, AH" and ASoof association were derived from the spectrophotometric data. A linear relationship between AH"and the pKa of the acids was found. The thermodynamic quantities reflect also the basicity of theproton acceptors but not in a simple relation. The strongest hydrogen bond is formed betweentrichloroacetic acid and diphenyl selenoxide, -AH" being -67 kJ/mol.Chloroacetic acids form strong hydrogen bonds with bases such as sulphoxides,phosphine oxides and pyridine N-oxide.l* The infra-red spectroscopic featuresindicate an increasing hydrogen bond strength in the acid series from mono- to tri-chloroacetic (MCA, DCA, TCA) and from sulphoxides to diphenyl selenoxide(DPSeO) and pyridine-N-oxide (PyO).This order of hydrogen bond strengthcorresponds to the acidity and basicity, re~pectively.~ The strongest bonds appear tobe formed between TCA and DPSeO, triphenylarsine oxide or PyO and the spectrasuggest that the hydrogen bonds in these adducts are nearly symmetrical. Thesubsequent X-ray structure determination of the crystalline adduct of TCA with PyOhas shown that the hydrogen bond is indeed extremely short.4 However, thermo-dynamic data on such hydrogen bonds are limited to bihalide ions.5 Therefore, wehave determined the equilibrium constants at four temperatures for the associationof the chloroacetic acids with diphenyl sulphoxide (DPSO), dibenzyl sulphoxide(DBSO), dimethyl sulphoxide (DMSO), triphenylphosphine oxide (TPPO) anddiphenyl selenoxide.The results are quantitatively in accord with expectations, i.e.,that the enthalpies of association are extremely high for the strongest hydrogen bondforming acid-base couples.EXPERIMENTALCarbon tetrachloride of AP grade (Kemika) was dried over Pz05 in a recycling still, andgreat care was taken to avoid contamination with atmospheric moisture at all stages. Thesolvent contained less than mol/l. HzO as estimated from the infra-red water bands.The acids and the bases were of good commercial grade, but were recrystallised or distilledbefore use. Particular attention was paid to drying.Liquids were kept for several daysover fresh molecular sieves and the solids were stored in a dessicator over Pz05. DMSO wasdried additionally with CaH,. Purity was checked by infra-red spectroscopy. Stocksolutions of acids and bases were prepared shortly before the measurements by weighing.The sample solutions were prepared from stock solutions of components and addition ofsolvent in a calibrated cylinder which was directly connected with the drying still. Mixingin the cylinder was provided by a glass clad iron rod moved by a magnet. The solutions15I52were prepared at 22°C and corrected for volume changes with temperature. The cells hadan optical path of 8.95 cm.They were made of brass, internally gold plated and fitted withquartz windows. Water was circulated through the jackets from an ultrathermostat. Thetemperature was kept stable within _+O.l"C. The Perkin-Elmer model 521 was used forspectrophotometry in a double beam mode with two identical cells, one being filled with thesolvent from the same batch as used in making up the sample. Two photometric redordingswere made of each sample. If the two did not agree to less the 0.5 % transmission, a thirdrecording was made and the average used for calculations.HYDROGEN BONDING I N ADDUCTS OF OXYGEN BASESCALCULATIONS AND RESULTSIn diluted acid-base systems, the carboxylic acid monomers are in equilibrium withthe dimers and with the acid-base adducts.The equilibrium constant for the latterhas to be derived from the absorbance of the free OH stretching band only becauseof the solvent absorption in other regions. This requires a separate determinationof the monomer-dirner dissociation constant (Kd) under similar conditions. Thedetails of this work will be published elsewhere; here only the results will be given(table 1). They agree well with those obtained by other authors.6 The acid-baseassociation constants Kas was calculated from the expression :&, = (CA - M - 2M2 /&)/M(C, - CA -!- M -!- 2M2 I&)in which CA is the total concentration of the acid, M of the monomer and CB thetotal concentration of the base. It is assumed (i) that concentrations may be usedthroughout instead of activities, (ii) that the acid dimerisation constant is not in-fluenced by the presence of the base in the medium and (iii) that there are no otherequilibria.Assumptions (i) and (ii) are justified in view of the high dilution. Thisapplies also to the association between base molecules. At least for DMSO, it isknown that the self-association constant is very low and thus practically only the freebase should be present at high dilutions.' The possibility of the base-acid associationin 1 : 2 ratio will be discussed in the last paragraph.TABLE 1 .-DISSOCIATION CONSTANTS OF CHLOROACETIC ACIDS IN CC14acid temp./"C Kd/molmonochloroacetic 20 (2.59k0.13)~25 (3.14+0.23)x30 (4.15& 0.34) x35 (5.36+ 0.53) xdichloroacet ictrichloroacet ic20 (3.10+ 0.22) x lo-'25 (3.86k0.29)~30 (5.21 20.37) x35 (6.51k0.29)~20 (1.50+ 0.13) x25 (1.95 f 0.14) x30 (2.6550.21)~35 (3.65f0.44)~The concentrations of the solutes were in the range of 10-4molar.For eachacid-base pair, 5 to 8 solutions were measured in which the acid to base molar ratiosvaried within the limits 1 : 1 and 1 : 1.3. Within these limits, photometric accuracywas maximised. The absorbances were determined at 4 temperatures (20, 25, 30,35°C). K,, for 25°C given in table 2 are the least squares values. The deviations in K,D. HAD% AND J . RAJNVAJN 153are within the limits of 12 1.2 to 1 6 . 6 %. The enthalpies of association werecalculated from the (log K,,, T-l) dependence by least squares. AH" values showstandard deviations within the limits + 3 to AGO and AS" were calculatedfrom the usual expression. The largest deviations are shown by the entropy changes,but on the whole the deviations in the thermodynamic quantities are well within thelimits usual for spectrophotometric determinations.A linear relationship betweenAH" and AS" for all the three series of acid-base adducts exists as usual in such cases.The correlations obtained by least squares treatment are :12 %.MCA: -AS0+2.15 (-AH") +6.65DCA -AS" + 2.08 (-AH") + 4.98TCA: -AS0+3.10 (-AH")+ 18.61.DISCUSSIONThe very reasonable deviations of the resulting quantities in table 2, and theparallelism shown by the three series of acids, do not exclude certain systematic errors.The most serious source of experimental error is probably the water content of thesolvent.Completely dry carbon tetrachloride can be obtained only after verylengthy procedures which were not practicable with the large quantities used in ourexperimental work, and we had to tolerate a minimum concentration that was at leastone order of magnitude less than that of the species investigated. Water may actboth as a proton acceptor and donor. The association of water with DMSO andwith phosphine oxides has been investigated by several authors.*-l O The associationconstants are of the order of 0.5 mol l.-l and thus much smaller than the K,, for theacid-base equilibria. Not much is known about the influence of small water con-centrations in non-polar solvents upon the dimerisation of carboxylic acids,12 but oneexpects to account for this by using a solvent with roughly the same water content fordetermining both the dimerisation and the acid-base association constants.As afinal check we have deliberately added small amounts of water to some acid-basesample solutions and determined K,, at four temperatures as usual, but the resultswere within the limits of error.TABLE 2.-THERMODYNAMIC QUANTITIES OF ASSOCIATION OF CHLOROACETIC ACIDS WITH SOME0x0 BASES IN CC14 (25°C)-AS01acid base €&/I. mol-1 -AGo/kJ md-1 -AH"/kJ mol-1 J mol-1 K--lmonochloroacetic DPSODBSODMSOTPPODPSeOdichloroacetic DPSODBSODMSOTPPODPSeOtrichtoroacetic DPSODBSODMSOTPPODPSeO(1.548k0.034)X lo3(4.105f0.076)~ lo3(5.631k0.088)~ lo3(7.161+0.101)x lo3(2.569+0.089) x lo4(2.375k0.108)~ lo3(1.248f0.016)~ lo4(1.964f0.120) x lo4(4.361 f 0.098) x lo4(1.487 0.067) X 1 O5(4.772+ 0.232) x lo3(3.501 k 0.067) x lo4(4.530f 0.113) x lo4(7.003 f 0.154) X lo4(2.41 1 k0.068) x lo518.20+0.0420.63 0.0421.42+ 0.0422.01 f 0.0425.19+ 0.0819.29f 0.1223.39k0.0424.52f 0.1626.48f 0.0429.54f 0.122 1 .oo+ 0.1 225.94f 0.0426.57f 0.0427.66f0.0430.71 & 0.0827.03+ 3.5532.26f 3.1436.15f 1.6342.59f 2.0149.20f 3.0528.87f 0.1338.03f 1.5139.46+ 1.4651.17f 2.5159.455 2.0931.13f 1.3040.632 2.1 342.34f 1.2658.45f 5.1066.69-1 5.4828.9f 12.139.3f 10.549.4f 5.469.5+ 6.780.8f 10.031.8+ 6.749.4f 5.050.2f 5.083.3f 8.499.9+ 7.133.9k4.248.9f 7.153.1 k4.2102.5f 17.2120.l+ 18.154 HYDROGEN DONDING I N ADDUCTS OF OXYGEN DASESThe other disputable point concerns the equilibria not accounted for iii the cal-culations.They are :K : I K P 2AH+B+(AH), . . .B and AH+B+A-. . . HBF.The first type of equilibrium is known in the association of phenols with carbonyltype acceptors l 3 and in isothiocyanic acid with DMSO and trioctylphosphineoxide (TOPO).14 The K21 are smaller than Ka, by one order of magnitude, particu-larly with DMSO, and thus should not disturb the results obtained for the sulphoxides.This may not be so in the case of association of phosphine oxides with chloroaceticacids. We have also made a series of measurements with TOP0 as proton acceptorand obtained much higher values for the thermodynamic quantities than with TPPO.For instance, -AH" for the association with MCA, DCA and TCA came to 54, 96,and 121 kJ/mol, respectively, which seems excessively large ; therefore, they are notincluded in table 2.Attempts were made to extend the treatment of data so as toinclude the K,, since this may be important in this case. However, a more extendedrange of concentrations would be needed for a proper assessment of KZ1. Absorb-ances under such conditions are of low photometric accuracy, so other methods wilhave to be used to clarify this point.The proton transfer equilibrium cannot be assessed from the free OH bands only.However, since in the overall spectra of more concentrated solutions in CC14 noindications of proton transfer were observed, the neglect of this equilibrium is justified.The equilibrium constants for the acid-base association.Ka, (25°C) and thenegative enthafpies -AH" (table 2) are larger than the corresponding values for theacid dimerisation, except for the weakest base DPSO. However, if the enthalpy perhydrogen bond is considered, then even in this case the acid-base hydrogen bond isstronger than the bond in the dimer. The sequence of proton donorship is clearlymono- < di- < tri-chloroacetic acid. The relationship between - AH" and the pK,of the acids is linear.The appearance of the A, B, C bands in the infra-red spectra of the acid-baseadducts,l their frequency and, particularly, the intensity changes when going frommono- to tri-chloroacetic acid are consistent with the increase of the K,, and -AH",and thus with the strengthening of the hydrogen bond in the acid sequence.Thesequence of hydrogen bonding ability of the bases expected from their protonationconstants and from the features in the infra-red spectra of acid-base adducts is asfollows :diphenyl sulphoxide < dibenzyl sulphoxide < dimethyl sulphoxide < triphenyl-phosphine oxide < diphenyl selenoxide.This sequence is borne out also by the K,, and the -AHo values. These values areextremely high for the strongest hydrogen bond forming acid-base pair, but onemight have expected a larger difference between TPPO and DPSeO as proton accept-ors. The infra-red spectrum of the adduct of DPSeO and TCA indicates that herethe strongest type of hydrogen bonding is that close to the fully symmetrical 0ne.lThe change in the spectral features on going from the adduct of TCA with TPPO (type(i), ref.(1)) to that of TCA to DPSeO (type ii) is certainly great, involving a vOH differ-ence of at least 1000 cm-1 whereas the difference in enthalpies is only -8 kJ mol.that the enthalpy of association does not directlyreflect the strength of the hydrogen bond because it includes also energy consumingprocesses of which the main one is the stretching of the 0-H bond. Sellier andWojtowiak have proposed a correction for this effect." We have tried to apply itIt has been stressed beforD . HADZI AND J . RAJNVAJN 155to the present examples, putting AvOH in the range 900 cm-1 to 1200 cm-l. HOW-ever, the enthalpies of bond formation calculated in this way were larger than the- AHo of association determined directly by factors of 3 to 5 and thus quite unreason-able.The relationships between the changes in the vibrational spectra due to hydrogenbonding and the thermodynamic quantities have been largely discussed for the bondsof weak and intermediate strength,16 but the conclusions are not necessarily applicableto bonds approaching the symmetrical. Further work is needed to progress in thisdirection.Thanks are expressed to the Boris KidriE Fund for financial assistance. J. R. isgrateful to the Boris KidriE Institute (Belgrade) for leave of absence.l D. Hadii and N. Kobilarov, J. Chem. SOC. A , 1966,439.D. Had& H. Ratajczak and L. Sobczyk, J. Chem. SOC. A, 1967,48.D. Had& C. Klofutar and S. Oblak, J. Chem. SOC. A, 1968,905.L. GoliiS, D. Ha& and F. Lazarini, Chem. Comm., 1971,860D. G. Tuck, in Progress in Inorganic Chem., ed. F. A. Cotton (Wiley, New York, 1968), vol. 9,p. 161.R. E. Kagarise, Naval Res. Lab. Rep. 1957,4955. ' R. H. Figueroa, E. Roig and H. H. Szmant, Specfrochim. Acta, 1966, 22,587. * J. T. Conocchioli, M. T. Tocher and R. M. Diamond, J. Phys. Chem., 1965,69,1106.J. R. Holmes, D. Kivelson and W. C. Drinkard, J. Amer. Chem. SOC., 1962,84,4677.lo G. Roland and G. Duyckaerts, Specfrochim. Acta, 1966,22,793.l1 S. F. Ting, S. M. Wang and N. C. Li, Canad. J. Chem., 1967,45,425.l2 J. de Villepin, A. Lautik and M.-L. Josien, Ann. Chim., 1966, 1, 365.l4 S. Detoni, D. Hadii, R. Smerkolj, J. Hawranek and L. Sobczyk, J. Chem. SOC. A, 1970,2851.l 5 G. Sellier and B. Wojtowiak, J. Chim. phys., 1968, 65, 936.l6 E. M. Atnett, L. Joris, E. Mitchell, T. S. S. R. Murty, T. M. Gorrie and P. v. R. Schleyer,G. Nagarajan, Z. Nafurforsch., 27a, 1972,221.J. Amer. Chem. SOC., 1970,92,2365

ISSN:0300-9599    

DOI:10.1039/F19736900151

出版商:RSC    

年代:1973

数据来源: RSC

摘要:

Interactions in Lanthaiiide Systems Part 2.-Raman Study of Aqueous Cerium(II1) Nitrate j- Chemistry Dept., University of Waterloo, Waterloo, Ontario, Canada Receiiied 7th July, 1972 Raman spectra of aqueous cerium(II1) nitrate solutions, and compIementary infrared spectra, have been obtained for a wide range of compositions. The intensities of the 718 cm-' line of solvated nitrate ion have been measured and used to evaluate the average ligand number %. The data yield stability constants PI = 1.68 and B2 = 1.17 for Ce(NO,)'+ and Ce(N0,);. Spectral features suggest that the most prevalent structure of the dinitratocerium(II1) species involves one nitrate ion bound in monodentate fashion and one nitrate ion bound in bidentate fashion. Band assignments are suggested for the nitrate ligands assuming C,, symmetry.Objectives for investigation of aqueous lanthanide nitrate solutions, methods for obtaining and analysing vibrational spectral data and detailed results for gadolinium nitrate were presented in the first paper of this series.' The results of a similar study of Ce(NO,), are now presented. EXPERIMENTAL Solutions were prepared from 99.99 % Ce(N0&xH20 supplied by the American Potash Corporation, or from Ce203 and perchloric acid. Raman spectra were recorded for stoichiometric solutions ranging in concentration from 0.50 to 3.00 M and for solutions prepared for Job and molar ratio analyses. Experimental procedures were identical to those previously described.l RESULTS AND DISCUSSTON SPECTRAL FEATURES The spectra (fig. 1) are similar to those of Gd(N03)3 illustrated in ref.(1). In addition to the Rainan lines generated by the aquated nitrate ion which occur at 697, 718, 1049, 1348, and 1410 cm-' new hies are apparent at 742, 820, 1044, 1299, 1325,1456, and 1487 cm-l. The 1044 cm-' line is a shoulder of the 1049 cm-1 line and increases in intensity as the concentration of bound nitrate increases. The six lines in the region 1299-1487 cm-I have been inferred from a computer analysis and their positions and half-widths are essentially constant over a wide concentration and com- position range (cf. ref. (I)). The 1325 cm-' and 1487 cm-l bands are polarized. The 742 cm-l line is intense in the infrared spectrum but the 718 cm-l line is too weak to be seen above the water background. A prominent 820cm-l infrared line is asymmetric on the high frequency side at 830 cm-l (c = 3.51 M).The spectral -f- presented in part at the XXI Mid American Symposium on Spectroscopy, Chicago, Illinois, $ present address : Department of Chemistry, Rensselaer Polytechnic Institute, Troy, New York. June 2-5, 1970. 156D . L . NELSON A N D D. E. IRISH 157 features and their concentration dependence are consistent with the occurrence of equilibria involving solvated nitrate ion and nitrate ion bound to the Ce3+ cation. The intensity analysis which follows confirms this interpretation. . . , , , , , , , I 1400 lz00 1000 800 rn t/cm-' FIG. 1.-(a) Raman and (b) infrared spectrum of a 3.51 M stoichiometric solution of cerium(lI1) nitrate. INTENSITY ANALYSIS The band at 718 cm-1 is a feature of the " free " solvated nitrate ion.Its inteiisity has been used to obtain species concentrations. The molar intensity, JT18, was obtained from the linearity between intensity and concentration observed for NaN03. * in this study an internal intensity standard equivalent to 1 mol 1.-l ClOi was used. The concentration of the " free " nitrate ion, [N0;IF, was then obtained from the integrated intensity 1, divided by the molar intensity. The concentration of associated nitrate ion is given by [NO& = [NO,],-[NO,],. The concentrations so obtained are presented in table 1. The stability constants p1 and /.I2 were computed from the values of the average coordination numbers, ii and [NO;]F. A computer programme which generates a set of p values for which the difference between the computed ii and the observed ii is minimal was used with 21 " best " sets of data, as inferred from inspection of the magnitude of the differences obtained when all the data sets were employed.As suggested by Bond the differences between 6 calculated from the stability constants and the observed fi, denoted Aii, are shown in table 1. The values obtained for p1 and /I2 are 1.68 and 1.17 where concentrations are in mol l.-I. The step stability constants are : KI = 1.68 ; K2 = 0.70. The values of the stability constants obtained by the computer refinement are somewhat larger than those initially obtained * In ref. (1) JVl8 is given as 1.63 M-I, relative to the intensity of the 935 cm-I line of 1 M ClO,. This value is compatible with the data in the present paper as well but it should be noted that an instrument sensitivity correction is necessary to obtain a value which is universal.Application of this Factor gives JVl8 = 0.0532relative to the 935 cm-I line of 1 M ClO, used as an internal standard, when excitation is with unpolarized 435.8 nm radiation from a mercury Toronto arc.158 INTERACTIONS I N LANTHANIDE SYSTEMS from the coininon graphical slope-intercept method and are considered to be in closest harmony with the observation^.^ The consistency is apparent in the plot of the formation curve, 5, against the concentration of free nitrate, cF (fig. 2). All data scatter smoothly about the curve computed from p1 and b2. The constancy of the concentration quotients over a wide range of concentration and ionic strength is noteworthy and similar to that observed in other Rainan studies.6 The degree of formation of the species, cc, is plotted against cF in fig.3 to provide the distribution curve. TABLE MEASURED CONCENTRATIONS OF FREE AND BOUND NITRATE FROM SPECTRAL DATA (a) molar ratio analysis with [Ce3+]~ = 1.00 M R 0.50 0.75 1 .oo 1.25 1.50 1.75 2.00 2.50 3 .OO 0.20 0.30 0.40 0.50 0.60 0.64 0.70 0.75 0.80 1 .oo “O~IT 0.50 0.75 1 .oo 1.25 1 S O 1.75 2.00 2.50 3.00 [CeJ+lT 2.20 1.93 1.65 1.38 1.10 1 .00 0.83 0.70 0.55 0 [WNOs)J 0.50 1 .oo 1.50 2.00 2.50 [NO& 0.26 0.33 0.37 0.61 0.67 0.89 1.04 1.33 1.71 * Afi = jicdculated- [NOJIB 0.24 0.42 0.63 0.64 0.83 0.86 0.96 1.17 1.29 ’ nobserved (6) Job analysis 0.55 0.83 1.10 1.38 1.65 1.75 1.93 2.10 2.20 2.75 0.06 0.20 0.28 0.53 0.67 0.89 0.99 1.25 1.44 2.75 0.49 0.63 0.82 0.85 0.98 0.86 0.94 0.85 0.76 0 (c) stoichiometric solutions mO;lF m0ilB n 0.92 0.58 1.16 1.71 1.29 1.29 2.33 2.17 1.45 2.91 3.09 1.55 3.03 4.47 1.79 3obs 0.24 0.42 0.63 0.64 0.83 0.86 0.96 1.17 1.29 ii0btl 0.23 0.33 0.50 0.62 0.89 0.86 1.13 1.21 1.38 0 Aii 3.0.06 + 0.03 - 0.05 + 0.07 0.0 + 0.07 + 0.07 + 0.02 + 0.03 Aii - 0.04 - 0.01 - 0.03 + 0.03 - 0.04 + 0.07 - 0.06 - 0.03 -0.11 - Aii -0.13 + 0.03 + 0.02 + 0.01 - 0.17 The value of 1.68 which we obtain for is in good agreement with the value 1.63f0.10 (25°C) reported by Choppin and Strazik and 1.3k0.3 (22°C) reported by Peppard et aZ.,8 obtained by solvent-extraction methods.Choppin and Strazik also recognized the existence of higher complexes when nitrate concentration was increased but were not confident in the accuracy of their data in that composition region so did not estimate p2.They suggested the species were outer-sphere ionD . L. NELSON AND D . E. IRISH 159 pairs although Ce(II1) did not conform to their proposed test in the way that other lanthanide cations did. The Raman spectra clearly suggest contact ion pairs.g Cation exchange studies yielded a value of about 1.6 prior to inclusion of an activity coefficient correction; a value of about 2.5 was then obtained and preferred.1° 0 I .o 2 .o 3.0 CP Fic;. 2.-Formation curve for the cerium(lI1) nitrate system : 0, Job data ; 0, molar ratio data ; 0, stoichionietric solutions ; -, calculated from and p2. 1 I I I I I C F FIG.3.-Distribution curve showing the degree of formation of thc species against the " free " nitrate concentration. A, E M L ~ ; B, KML ; C, a ~ . For the systems discussed above there is no suggestion of polynuclear formation, unlike Gd(N03)3.1 However, when systems composed of Ce(C10,), +HN03 + H,O with [NO,] : [Ce3+] > 4.0 were measured the solutions appeared to be unstable and the apparent ii values decreased as if polynuclear species might be forming. The lack of stability of such systems was also noted by Quill and Robey.ll These data were not incorporated into the above analysis, but could warrant further investigation. The intensities of the 718 and 740 cm-' lines of 2.0 and 3.0 M solutions were also measured over the temperature range 15 to 85°C. The ratio 1,,,/1,1, increased to a small extent suggesting that the degree of association is enhanced by increasing the temperature, which is comparable to the effect observed for ZII(NO,)~ in methanol.12 BAND ASSIGNMENTS AND SPECIES GEOMETRY No metal-oxygen vibration has been detected in this study and it therefore appears that the interactions are primarily ionic.l 3 The measured low degrees of depolariza- tion of the Raman lines at 1325 and 1487 cm-l provide a clue to the species geometries. It is probable that nitrate acts as a monodentate ligand in the 1 : 1 species, the other coordination sites being occupied by water. This ligand generates the polarized 1325 CII'L Stalistically, a number of orientations iiiay be adopted by two nitrate ions in contact with Ce3 1..The 1456 cm-l intensity :ind depolarized 1456 cm-1 lines.160 INTERACTIONS I N LANTHANIDE SYSTEMS is greater when substantial amounts of both Ce(N03)2+ and Ce(NO3); are present than when Ce(N03)2+ predominates. A monodentate interaction in Ce(NO,); is thus indicated. But the presence of a polarized 1487 em-1 line and a depolarized, less intense 1299 cm-1 line suggests that, on the average, a bidentate orientation is preferred by the second nitrate ion. Presumably two water molecules are thereby displaced from the primary solvation sphere. Thus for both classes of bound nitrate a Czv model is consistent with the data. Assignments are therefore analogous to those tabulated for Gd(N03),. l This work was supported by the National Research Council of Canada arid by the Province of Ontario in the form of Ontario Graduate Fellowships to D. L.N. The authors are grateful to Dr. R. S . Tobias for helpful discussion and for providing the computer programme; they are also grateful to Dr. T. G. Chang for modifying and executing the programme. Discussions with D. R. Williams are also gratefully acknowledged. D. L. Nelson and D. E. Irish, J. Chem. Phys., 1971,54,4479. A. R. Davis, D. E. Irish, R. B. Roden and A. J. Weerheim, Appl. Spectr., 1972,26, 384. T. G. Chang, Ph.D. Thesis (University of Waterloo, 1972). M. T. Beck, Chemistry of Complex Equilibria (Van Nostrand, Reinhold Co., London, 1970), p. 58. J. D. Riddell, D. J. Lockwood and D. E. Irish, Canad. J. Chem., 1972,50,2951. D. F. Peppard, G. W. Mason and I. Hucher, J.Inorg. NucZeur Chem., 1962, 24, 881. D. E. Irish, G. Chang and D. L. Nelson, Inurg. Chem., 1970, 9, 425. lo R. E. Connick and S. W. Mayer, J. Amer. Chem. SOC., 1951, 73, 1176. l 1 L. L. Quill and R. F. Robey, J. Amer. Chern. Soc., 1937,59, 2591. l2 S, A. AI-Baldawi, M. H. Brooker, T. E. Gough and D. E. Irish, Canad. J. Chem., 1970,48,1202. l 3 J. H. B. George, J. A, Rolfe and L. A. Woodward, Tram. Farday SOC., 1953, 49, 375. 4A. M. Bond, Coord. Chem. Rev., 1971,6, 377. ' G. R. Choppin and W. F. Strazik, Znorg. Chem., 1965, 4, 1250.Interactions in Lanthaiiide SystemsPart 2.-Raman Study of Aqueous Cerium(II1) Nitrate j-Chemistry Dept., University of Waterloo, Waterloo, Ontario, CanadaReceiiied 7th July, 1972Raman spectra of aqueous cerium(II1) nitrate solutions, and compIementary infrared spectra,have been obtained for a wide range of compositions. The intensities of the 718 cm-' line of solvatednitrate ion have been measured and used to evaluate the average ligand number %.The data yieldstability constants PI = 1.68 and B2 = 1.17 for Ce(NO,)'+ and Ce(N0,);. Spectral featuressuggest that the most prevalent structure of the dinitratocerium(II1) species involves one nitrate ionbound in monodentate fashion and one nitrate ion bound in bidentate fashion. Band assignmentsare suggested for the nitrate ligands assuming C,, symmetry.Objectives for investigation of aqueous lanthanide nitrate solutions, methods forobtaining and analysing vibrational spectral data and detailed results for gadoliniumnitrate were presented in the first paper of this series.' The results of a similar studyof Ce(NO,), are now presented.EXPERIMENTALSolutions were prepared from 99.99 % Ce(N0&xH20 supplied by the AmericanPotash Corporation, or from Ce203 and perchloric acid.Raman spectra were recordedfor stoichiometric solutions ranging in concentration from 0.50 to 3.00 M and for solutionsprepared for Job and molar ratio analyses. Experimental procedures were identical tothose previously described.lRESULTS AND DISCUSSTONSPECTRAL FEATURESThe spectra (fig. 1) are similar to those of Gd(N03)3 illustrated in ref. (1). Inaddition to the Rainan lines generated by the aquated nitrate ion which occur at697, 718, 1049, 1348, and 1410 cm-' new hies are apparent at 742, 820, 1044, 1299,1325,1456, and 1487 cm-l.The 1044 cm-' line is a shoulder of the 1049 cm-1 line andincreases in intensity as the concentration of bound nitrate increases. The six linesin the region 1299-1487 cm-I have been inferred from a computer analysis and theirpositions and half-widths are essentially constant over a wide concentration and com-position range (cf. ref. (I)). The 1325 cm-' and 1487 cm-l bands are polarized.The 742 cm-l line is intense in the infrared spectrum but the 718 cm-l line is too weakto be seen above the water background. A prominent 820cm-l infrared line isasymmetric on the high frequency side at 830 cm-l (c = 3.51 M). The spectral-f- presented in part at the XXI Mid American Symposium on Spectroscopy, Chicago, Illinois,$ present address : Department of Chemistry, Rensselaer Polytechnic Institute, Troy, New York.June 2-5, 1970.15D .L . NELSON A N D D. E. IRISH 157features and their concentration dependence are consistent with the occurrence ofequilibria involving solvated nitrate ion and nitrate ion bound to the Ce3+ cation.The intensity analysis which follows confirms this interpretation.. . , , , , , , , I1400 lz00 1000 800 rnt/cm-'FIG. 1.-(a) Raman and (b) infrared spectrum of a 3.51 M stoichiometric solution of cerium(lI1)nitrate.INTENSITY ANALYSISThe band at 718 cm-1 is a feature of the " free " solvated nitrate ion. Its inteiisityhas been used to obtain species concentrations. The molar intensity, JT18, wasobtained from the linearity between intensity and concentration observed for NaN03.*in this study an internal intensity standard equivalent to 1 mol 1.-l ClOi was used.The concentration of the " free " nitrate ion, [N0;IF, was then obtained from theintegrated intensity 1, divided by the molar intensity. The concentration ofassociated nitrate ion is given by [NO& = [NO,],-[NO,],. The concentrationsso obtained are presented in table 1.The stability constants p1 and /.I2 were computed from the values of the averagecoordination numbers, ii and [NO;]F. A computer programme which generates aset of p values for which the difference between the computed ii and the observed iiis minimal was used with 21 " best " sets of data, as inferred from inspection of themagnitude of the differences obtained when all the data sets were employed.Assuggested by Bond the differences between 6 calculated from the stability constantsand the observed fi, denoted Aii, are shown in table 1. The values obtained for p1and /I2 are 1.68 and 1.17 where concentrations are in mol l.-I. The step stabilityconstants are : KI = 1.68 ; K2 = 0.70. The values of the stability constantsobtained by the computer refinement are somewhat larger than those initially obtained* In ref. (1) JVl8 is given as 1.63 M-I, relative to the intensity of the 935 cm-I line of 1 M ClO,.This value is compatible with the data in the present paper as well but it should be noted that aninstrument sensitivity correction is necessary to obtain a value which is universal.Application ofthis Factor gives JVl8 = 0.0532relative to the 935 cm-I line of 1 M ClO, used as an internal standard,when excitation is with unpolarized 435.8 nm radiation from a mercury Toronto arc158 INTERACTIONS I N LANTHANIDE SYSTEMSfrom the coininon graphical slope-intercept method and are considered to be inclosest harmony with the observation^.^ The consistency is apparent in the plotof the formation curve, 5, against the concentration of free nitrate, cF (fig. 2). Alldata scatter smoothly about the curve computed from p1 and b2. The constancy ofthe concentration quotients over a wide range of concentration and ionic strength isnoteworthy and similar to that observed in other Rainan studies.6 The degree offormation of the species, cc, is plotted against cF in fig.3 to provide the distributioncurve.TABLE MEASURED CONCENTRATIONS OF FREE AND BOUND NITRATE FROM SPECTRAL DATA(a) molar ratio analysis with [Ce3+]~ = 1.00 MR0.500.751 .oo1.251.501.752.002.503 .OO0.200.300.400.500.600.640.700.750.801 .oo“O~IT0.500.751 .oo1.251 S O1.752.002.503.00[CeJ+lT2.201.931.651.381.101 .000.830.700.550[WNOs)J0.501 .oo1.502.002.50[NO&0.260.330.370.610.670.891.041.331.71* Afi = jicdculated-[NOJIB0.240.420.630.640.830.860.961.171.29’ nobserved(6) Job analysis0.550.831.101.381.651.751.932.102.202.750.060.200.280.530.670.890.991.251.442.750.490.630.820.850.980.860.940.850.760(c) stoichiometric solutionsmO;lF m0ilB n0.92 0.58 1.161.71 1.29 1.292.33 2.17 1.452.91 3.09 1.553.03 4.47 1.793obs0.240.420.630.640.830.860.961.171.29ii0btl0.230.330.500.620.890.861.131.211.380Aii3.0.06 + 0.03- 0.05 + 0.070.0 + 0.07 + 0.07 + 0.02 + 0.03Aii- 0.04- 0.01- 0.03 + 0.03- 0.04 + 0.07- 0.06- 0.03-0.11-Aii-0.13 + 0.03 + 0.02 + 0.01- 0.17The value of 1.68 which we obtain for is in good agreement with the value1.63f0.10 (25°C) reported by Choppin and Strazik and 1.3k0.3 (22°C) reportedby Peppard et aZ.,8 obtained by solvent-extraction methods.Choppin and Strazikalso recognized the existence of higher complexes when nitrate concentration wasincreased but were not confident in the accuracy of their data in that compositionregion so did not estimate p2.They suggested the species were outer-sphere ioD . L. NELSON AND D . E. IRISH 159pairs although Ce(II1) did not conform to their proposed test in the way that otherlanthanide cations did. The Raman spectra clearly suggest contact ion pairs.gCation exchange studies yielded a value of about 1.6 prior to inclusion of an activitycoefficient correction; a value of about 2.5 was then obtained and preferred.1°0 I .o 2 .o 3.0CPFic;. 2.-Formation curve for the cerium(lI1) nitrate system : 0, Job data ; 0, molar ratio data ;0, stoichionietric solutions ; -, calculated from and p2.1 I I I I IC FFIG. 3.-Distribution curve showing the degree of formation of thc species against the " free "nitrate concentration.A, E M L ~ ; B, KML ; C, a ~ .For the systems discussed above there is no suggestion of polynuclear formation,unlike Gd(N03)3.1 However, when systems composed of Ce(C10,), +HN03 + H,Owith [NO,] : [Ce3+] > 4.0 were measured the solutions appeared to be unstable andthe apparent ii values decreased as if polynuclear species might be forming. Thelack of stability of such systems was also noted by Quill and Robey.ll These datawere not incorporated into the above analysis, but could warrant further investigation.The intensities of the 718 and 740 cm-' lines of 2.0 and 3.0 M solutions were alsomeasured over the temperature range 15 to 85°C.The ratio 1,,,/1,1, increased to asmall extent suggesting that the degree of association is enhanced by increasing thetemperature, which is comparable to the effect observed for ZII(NO,)~ in methanol.12BAND ASSIGNMENTS AND SPECIES GEOMETRYNo metal-oxygen vibration has been detected in this study and it therefore appearsthat the interactions are primarily ionic. l 3 The measured low degrees of depolariza-tion of the Raman lines at 1325 and 1487 cm-l provide a clue to the species geometries.It is probable that nitrate acts as a monodentate ligand in the 1 : 1 species, the othercoordination sites being occupied by water. This ligand generates the polarized1325 CII'L Stalistically, a number of orientationsiiiay be adopted by two nitrate ions in contact with Ce3 1..The 1456 cm-l intensity:ind depolarized 1456 cm-1 lines160 INTERACTIONS I N LANTHANIDE SYSTEMSis greater when substantial amounts of both Ce(N03)2+ and Ce(NO3); are presentthan when Ce(N03)2+ predominates. A monodentate interaction in Ce(NO,); isthus indicated. But the presence of a polarized 1487 em-1 line and a depolarized,less intense 1299 cm-1 line suggests that, on the average, a bidentate orientation ispreferred by the second nitrate ion. Presumably two water molecules are therebydisplaced from the primary solvation sphere. Thus for both classes of bound nitratea Czv model is consistent with the data. Assignments are therefore analogous tothose tabulated for Gd(N03),. lThis work was supported by the National Research Council of Canada arid bythe Province of Ontario in the form of Ontario Graduate Fellowships to D. L. N.The authors are grateful to Dr. R. S . Tobias for helpful discussion and for providingthe computer programme; they are also grateful to Dr. T. G. Chang for modifyingand executing the programme. Discussions with D. R. Williams are also gratefullyacknowledged.D. L. Nelson and D. E. Irish, J. Chem. Phys., 1971,54,4479.A. R. Davis, D. E. Irish, R. B. Roden and A. J. Weerheim, Appl. Spectr., 1972,26, 384.T. G. Chang, Ph.D. Thesis (University of Waterloo, 1972).M. T. Beck, Chemistry of Complex Equilibria (Van Nostrand, Reinhold Co., London, 1970),p. 58.J. D. Riddell, D. J. Lockwood and D. E. Irish, Canad. J. Chem., 1972,50,2951.D. F. Peppard, G. W. Mason and I. Hucher, J. Inorg. NucZeur Chem., 1962, 24, 881.D. E. Irish, G. Chang and D. L. Nelson, Inurg. Chem., 1970, 9, 425.lo R. E. Connick and S. W. Mayer, J. Amer. Chem. SOC., 1951, 73, 1176.l 1 L. L. Quill and R. F. Robey, J. Amer. Chern. Soc., 1937,59, 2591.l2 S, A. AI-Baldawi, M. H. Brooker, T. E. Gough and D. E. Irish, Canad. J. Chem., 1970,48,1202.l 3 J. H. B. George, J. A, Rolfe and L. A. Woodward, Tram. Farday SOC., 1953, 49, 375.4A. M. Bond, Coord. Chem. Rev., 1971,6, 377.' G. R. Choppin and W. F. Strazik, Znorg. Chem., 1965, 4, 1250

ISSN:0300-9599    

DOI:10.1039/F19736900156

出版商:RSC    

年代:1973

数据来源: RSC

摘要:

Similar Excited State pK Behaviour of Xanthone and theBenzophenonesBY J. F. IRELAND AND P. A. H. WYATTDepartment of Chemistry, The University, St. AndrewsReceived 10th July, 1972Benzophenone, like xanthone, shows the excited state pK order pK(Tl) > pK(S1) > pK(So), whichis readily understood in terms of the S1-Tl splitting to be expected from the n-n* and n - d states ofthe unprotonated and protonated forms respectively. Both fluorescence and phosphorescence occurfrom the protonated form at 77 K, but in fluid solutions the pK* values were not accessible to directdetermination but were estimated by the Forster cycle. Transients were observed from both B andBH+ forms by flash spectroscopy. Several substituted benzophenones behaved similarly, but thep-phenyl compound has a different pK order (due to the larger T-T* Sl-Tl splitting in the B form)and with the N-protonation of the p-amino compound the order is completely reversed.The pK*values for the phenolic ionization of the p-hydroxy compound are also reassessed.We recently reported that the excited state pK values of xanthone (I) are in theunusual order pK(Tl) > pK(S1) > pK(So) but an explanation was withheld until thepossibility could be discounted of complications due to protonation at the etheroxygen. A similar study was therefore attempted on benzophenone (11, R = R‘ = H)in which this possibility is absent. Although direct determination from [BH+]/[B]ratios proved impracticable in this case, pK values for benzophenone and substitutedbenzophenones have been obtained from the Forster Cycle.2R RII0The results show that the pK order for benzophenone is the same as that for xanthone,but the excited singlet and triplet states both show a smaller increase in basic strength.This behaviour, which i s probably quite common, can now be interpreted in terms ofthe characteristics of the types of transition involved in the neutral and protonatedmolecules.In the unprotonated form (B) benzophenone has a weak first absorption band ofn-rc* type (near 320nm) and a second, much more intense, n-n* band at shorterwavelengths.Since the S,-T, splitting is small for n-rc* ~ t a t e s , ~ Tl for the unproton-ated base does not lie far below S1. The protonated molecule, however, has a firstabsorption band at a longer wavelength than that of B and of the intense n-n* type,with the important consequence that the Sl-Tl splitting is greater than that for then-rc* transition of the base.The lower energy of the SltSo transition for BH+ thanfor B means that B becomes a stronger base upon excitation to S1 and the greater1-6 16162 BENZOPHENONE pK* VALUESSl-Tl splitting means that B becomes an even stronger base in the TI state; hencethe observed order, pK(Tl) > pK(S1) > pK(S,) (see fig. 1).It is in accord with the above interpretation that ring substitution in benzophenonecan alter the pK order, since substituents affect the n-n* and n-n* energies and hencecan alter the position and type of the excited states involved and the relative S,-T,splittings in the B and BHf forms.Substitution of a phenyl group in the paraposition of the benzophenone molecule has little effect on the n-n* transition but, aswould be expected from the extended n system, has a marked effect on the n-n*tran~ition.~ Thus, although phenyl substitution results in insufficient shift to invertthe singlet n-n* and n-n* levels, the greater Sl-Tl splitting of the n-n* system resultsin the lowest triplet of the base form being of n-n* character. The S,-T, splittingof the n-n* level of the B form of p-phenylbenzophenone is in fact found to be largerthan the Sl-Tl splitting of the BH+ form and therefore the Forster Cycle pK valueshave the order pK(S,)>pK(T,)>pK(S,). (The S,-Tl splitting of the BH+ forms ofthe molecules studied generally lies in the range 4000-5000 cm-l which, although it issmall for a n-n* transition, is always larger than the splitting found for n-n* states.)BH+So\P K ( S J \ 5 oB+H+ 2 BH+ T--protonated form unprotonated formFIG.1.-Schematic representation of the relative energies OF B and BH I- For benzophenoiic in the So,S1, and TI states.Xanthone, like phenylbenzophenone, also has the n-n* and the n-n* first excitedsinglet states very close together and at first sight might therefore be expected to showthe pK order ofp-phenylbenzophenone rather than that of benzophenone itself. But,although the first excited triplet state of xanthone is of n-7~'~ characterY6 in this case theobserved S,-T, splitting of the n-n* state is smaller than that for the B form ofp-phenylbenzophenone, possibly because the first n-n* transition has some charge-transfer character.Thus, while the lowest triplet state is of n-z* character in xan-thone, the S,-T, splitting is still less than that for the n-n* state of the BH-k form andhence we obtain from the Forster Cycle the pK order pK(T,)> pK(S,)> pK(So).EXPERIMENTALThe apparatus and methods used were described previously.1 The benzophenones wereobtained commercially and benzophenone itself was recrystallized several times from ethanol.For the Forster Cycle calculations the absorption and phosphorescence spectra of the B formswere obtained in ethanol and those of the BH+ forms in 98 % H2S04, except for the proton-ated amino form of p-aminobenzophenone which was recorded in 20 % H2S04/ethanol, iJ .F . IRELAND AND P . A. H . WYATT 163which only the NH2 group is protonated. For p-hydroxybenzophenone, 20 % 1 M NaOH/ethanol was used for the phenolate ion form (-0-) and ethanol for the -OH form. Noneof the compounds listed in table 1 was very unstable in the strong acid although the absorptionspectra showed slight changes on standing as would be consistent with slow sulphonation.The phosphorescence spectra also changed slightly on standing but the values in table 1 wereobtained using freshly prepared solutions which were immediately cooled to liquid nitrogentemperature.RESULTSBENZOPHENONEABSORPTION AND EMlSSION SPECTRAThe absence of fluorescence from the unprotonated benzophenone (B) moleculeprecludes the location of the 0-0 transition in the n-dk band by averaging the absorp-tion and fluorescence maxima,’ but fortunately the phosphorescence excitation spec-trum shows considerable structure and a direct determination of the (1-0 frequency istherefore available for Forster Cycle calculations.At 77 K in solid solution in ice or ethanol benzophenone phosphoresces but doesnot fluoresce, the intersystem crossing having a high triplet quantum yield.* Thefrequency of the 0-0 band of the triplet n-n* transition is easily obtained from the well-defined phosphorescence spectrum.Though quenched by oxygen, this emission isdetectable in undegassed aqueous solutions with its intensity reduced by an amountconsistent with Ledger and Porter’s quenching constant for dissolved oxygen.(See also Parker and Joyce lo for other cases of emissions from fluid solution.) Theemission in aqueous solution at 20°C is similar to the phosphorescence at 77 K but hasa less well defined structure.It is markedly acid-quenched and at pH 5 has only abouthalf its full intensity, as also observed by Porter and Ledger.gwavelength InmFK,. 2.--Benzophenone absorption spectra: (a) B form (1.1 x mol dm-j) in water; (b) BH+form (8 x mol dm-9 in 98 % H,SO,.Fig. 2 shows the absorption spectrum of protonated benzophenone (BH+), withthat of the unprotonated form for comparison. The breadth of the first band (n-n*)and the absence of structure both in the absorption and excitation spectrum madelocation of the 0-0 transition difficult.In this case, however, fluorescence as well asphosphorescence occurred at 77 K (see fig. 3) and the average of absorption and fluo-rescence maxima could therefore be resorted to.4 The BHf phosphorescence is muc164 BENZOPHENONE pK* VALUbSlonger lived than that of B, as would be expected from a n-z* triplet.'' At roomtemperature BH+ gave a broad emission band (maximum 470nm) which was notresolvable into fluorescence and phosphorescence on our apparatus but which, bycomparison with the low temperature emission, appeared to be mainly phosphor-escence. The excitation spectrum was identical with that for fluorescence and phos-phorescence at 77K, whence (unless some form of energy transfer from BH+ toanother species is occurring) all these emissions are attributable to BH+.(Otherpossible interpretations of this emission, e.g. excimer formation, are discounted by thefact that the emission spectrum is unchanged by dilution to very low concentrations.)Fig. 4 shows the effect of H2S04 concentration upon the relative intensity of BH+emission at room temperature. The quantum yield of the emission is so low thatrelatively concentrated solutions had to be used and correction for the inner filtereffect l2 was necessary. Corrections were also made for medium effects upon opticaldensity and for the different absorption coefficients of B and BH+. According to theForster Cycle calculations (see table 1) the BHf singlet and triplet might be expected**100-80 -1420 460 5 0 0 540 580 6120wavelength/nmFIG.3.-Benzophenone emission in !X3 % H2S04 at 77 K : (a) total emission ; (h) phosphorescence.FIG. 4.-Acid dependence of relative emission intensity of protonated benzophenone (BH)'J . F . IRELAND AND P . A . H . WYATT 165to occur at lower acidities than the pK(S,) (i.e. lower than 74 % H2S04),13 but inpractice the emission is, to a first approximation, only observed when BH+ is presentin sufficient concentrations for excitation from the ground state. Presumably acidquenching of excited B is so efficient in this composition range that excited BH+ is notformed in the encounter : this could amount to a case of a non-adiabatic reaction,14in which electronic energy is lost during the transfer of a proton to the excited base.h$ 0 .5 -Wc .-CI 3 .I Y 3 0.30.7TRANSIENTS OBSERVED IN FLASH PHOTOLYSISComparison of the absorption spectra of the transients observed in benzophenonesolutions in benzene and water confirms that the spectrum in water is attributable tothe B triplet, as explained by Ledger and P ~ r t e r . ~ As expected from phosphorescencequenching effects, this transient disappears in dilute acid solutions. At higher acidconcentrations, beyond pK(S,), a new transient absorption spectrum appears,attributable to the BH+ triplet. At a fixed delay setting after the photoflash, theoptical density of this transient is greater when the solution is degassed (fig. 5), justas expected from an oxygen quenching effect; but surprisingly the intensity of theemission (apparently phosphorescence) from BHf in 98 % H2S04, unlike that from Bin water, seems insensitive to the presence of oxygen.While the apparent discrepancycould possibly be attributable to a fortuitous compensation between the effect ofoxygen upon the triplet quantum yield and the triplet lifetime, further work is neededfor clarification.-----4 5 0 5 0 0 550 600wavelengthlnmFIG. 5.-Effects of degassing upon triplet-triplet absorption spectrum of protonated benzophenone(BH+) : (a) before and (6) after degassing.These results show that there is a gap in the acidity scale (presumably due to Hfquenching) between the regions where triplet B and BH+ are detectable and directdetermination of excited state pK values from measured [BH+]/[B] ratios is thereforenot possible.FORSTER CYCLE CALCULATIONSBENZOPHENONE AND DERIVATIVESTable 1 shows the spectral data used in the Farster Cycle calculations and thederived Aij values for the singlet and triplet transitions on protonation: for thosTABLE 1 .-SPECTRAL DATA FOR SUBSTITUTED BENZOPHENONES AND pK* VALUES FROMphosphorescencelmax/nm for BH+ maxima4,4'-substituents in (11)Amax/nm fluores- AV/cm-1 lmax/nm lmaxlnrn AV/cm-1R R' for B absorption cence mean for SpSo for B for BH+ for TI-&Hc1CH300-OHC6HSNH2NH,+Hc1CH30HHHHH372373364388*366374387*369346378404372386334-42641 7451430480427--386 lo00398 1700428 4100366T -1500401 2400433 3600369t -1300380 800415 462419 506412 515469 422422 485471 535481 415415 470250041004700-240031002500 - 33002700* Average of absorption and fluorescence maxima.Other values in this column are for t Values taken from phosphorescence excitation spectra.a protonation of 0- ; b protonation of NH2J . F. IRELAND AND P . A . H . WYATT 167compounds for which pK(S,) is known pK(S1) and pK(T,) are reported. Thoughthe ApK values are subject to the usual uncertainties in Forster Cycle applica-t i o n ~ , ~ . 1 5 9 l6 the calculated excited state pK values should certainly be reliable as tothe order for So, S1, and T,, just as was found to be the case for xanthone in which adirect comparison between the Forster Cycle and experimental values could be made.The pK values indicate that an electron-withdrawing group, e.g.C1, increases thebasicity of the triplet state more than that of the singlet, whereas electron-donatinggroups cause a decrease in the pK(S,)-pK(T,) separation. Within this series ofcompounds, the protonation of the amino group of p-aminobenzophenone provides acontrasting case to that of carbonyl protonation, with the pK(T,) still outside thepK(S,)-pK(Sl) range but with a reversal of the order. The lowest singlet and tripletstates here (of charge transfer type 5, are replaced on protonation by a lowest state ofn-n* character and the same argument as for the protonation of the carbonyl groupin benzophenone readily explains why pK( Tl) again lies outside the pK(So)-pK(Sl)range.Unlike the cases of carbonyl protonation, the excited states of the aminocompound are weaker bases and hence the pK order is pK(S,)>pK(S,)>pK(T,).Further protonation of the benzophenoneammonium cation, but now at the carbonylgroup, shows the pK order expected from carbonyl protonation with an increase inpK(S,)-pK(T,) separation due to the electron-withdrawing effect of the NH'; group.The difference in direction of the excited state bascicity change of these two groupsmeans that the species involved would not be experimentally obtainable. p-Hydroxy-benzophenone presents a similar case to the amino compound: the pK valuesreported by Godfrey, Porter and Suppan l7 for the deprotonation of the hydroxylgroup have been revised in table 1 using the 0-0 position from our phosphorescenceexcitation spectrum. It is possible that the low quantum yield of photoreductionfound by Porter and Suppan for 4hydroxybenzophenone, even in solutions of pH = 1,could be due to the increased basicity of the carbonyl group in the triplet state, whichthen becomes protonated and prevents hydrogen abstraction.All the above cases of carbonyl protonation, except for p-phenylbenzophenone,show the pK order first found for xanthone,l but the precise order is clearly fortuitous.For example, while o-hydroxybenzophenones l9 give the same general picture, in onecase, 2-hydroxy-5-nitrobenzophenone, we find (cf.Kysel 20) that there is an increasein basic strength in the first singlet state but that the first triplet state neverthelessbecomes a weaker base, giving the order pK(Sl) > pK(So)> pK(T,).We thank the Carnegie Trust for a postgraduate Scholarship to J.F. T. and theS.R.C. for an apparatus grant.J. F. Ireland and P. A. H. Wyatt, J.C.S. Faraday I, 1972, 68, 1053.A. Weller, Prog. Reaction Kinetics, 1961, 1, 189.J. N. Murrell, The Theory of the Electronic Spectra of Organic Molecules (Methuen, London,1963).E. Vander Donckt, Prog. Reaction Kinetics, 1970,5,273.G. Poiter and P. Suppan, Trans. Fwaday Soc., 1965, 61,1664.R. N. Nurmukhametov, L. A. Mileshina and D. M. Shegorin, Opt. Spektr., 1967,22,404.E. L. Wehry and L. B. Rogers, Spectrochim. Acta., 1965, 21A, 1967.M. B. Ledger and G. Porter, J.C.S. Faraday I, 1972, 68, 539.R. S. Becker, Theory and Interpretation of Fluorescence and Phosphorescence (Wiley-Interscience,New York, 1969).* C. A. Parker and C. G. Hatchard, Analyst, 1962, 87, 664.lo C. A. Parker and T. A. Joyce, Trans. Faraday SOC., 1969, 65,2823.l2 C. A. Parker, Photoluminescence of Solutions (Elsevier, Amsterdam, 1968), p. 220.l 3 T. G. Bonner and J. Phillips, J. Chem. SOC. B, 1966, 650168 BENZOPHENONE pK* VALUES14Th. Forster, Pure Appl. Chem., 1970, 24,443.l5 S. G. Schulman, P. T. Tidwell, J. J. Cetorelli and J. D. Winefordner, J. Amer. Chem. Soc., 1971,l6 G. J. Yakatan and S. G. Schulman, J. Phys. Chem., 1972,76, 508.l7 T. S. Godfrey, G. Porter and P. Suppan, Disc. Faraduy SOC., 1965, 39, 194.l 8 G. Porter and P. Suppan, Pure Appl. Chem., 1964,9,499.l9 J. F. Ireland and P. A. H. Wyatt, unpublished work.*O 0. Kysel, Int. Symp. Macromol. Chem., 1969,5, 263.93, 3179

ISSN:0300-9599    

DOI:10.1039/F19736900161

出版商:RSC    

年代:1973

数据来源: RSC

摘要:

Studies in the System Calcium Sulphate/Water Part 4.-Rehydration of Hexagonal CaSO, BY M. C. BALL* AND L. S. NORWOOD Department of Chemistry, University of Technology, Loughborough, Leics. LEll 3TU The rehydration of hexagonal CaS04 has been studied in the temperature range 298-393 K and at water vapour pressures of 3 x 103-8.2 x lo3 N m-'. The reaction is controlled by diffusion of water, following the parabolic law. The rate of rehydration is strongly affected by the external water vapour pressure, probably through an adsorbed multilayer of water. This effect has been corrected for in terms of the B.E.T. relationship and an activation energy of 46 kJ mol-1 has been calculated for the rehydration. Relatively few rehydration reactions have been studied compared with dehydra- tions.This is probably due to the non-reversible nature of many dehydrations, but may also be due to the requirement of relatively low temperatures and hence slow reaction rates. Two worthy of mention are those on the hydration of magnesium oxide and the rehydration of manganese(I1) formate.2 Both of these rehydrations show large effects by water vapour on the reaction rate, together with a threshold effect where a minimum water vapour pressure is required for rehydration to start. The dehydration reactions of the various hydrates of calcium sulphate have been extensively studied. Hexagonal CaSO, is the primary anhydrous product produced from both the dihydrate (gypsum) and the hemihydrate, as illustrated in the reaction scheme. CaS04, 2H20 4 / hexagonal CaSO,-, orthorhombic CaSO, 11 11 CaSO,, 3H20 The crystal structures of the hemihydrate and hexagonal CaSO, are very similar,, consisting of diamond shaped channels some 4A across, formed by a Ca-OSO, network.These channels have been likened to those in zeolite^,^ but it seems likely that because the channels are so small, either all sites for water molecules are filled, or none. Hexagonal CaSO, is extremely reactive towards water vapour, and the re- hydration has been little studied. Holridge and Walker studied the effects of humidity and sample ageing on the rehydration, using differential thermal analysis as the diagnostic technique. They found evidence of an extra endothermic peak in the differential thermal analysis (d.t.a.) traces which they suggested was caused by re- moval of a sorbed monolayer of water.EXPERIMENTAL Two starting materials were used as sources of hexagonal &SO4. Most runs were carried out on material derived from calcium sulphate dihydrate and these are the results which will be quoted throughout. A few runs were carried out using so-called P-CaS04,3H20. This was produced by dehydrating the dihydrate at 390 K and allowing this anhydrous material to rehydrate in the air. The reason far considering this somewhat cyclic treatment is that the 169170 STUDIES I N CALCIUM SULPHATE/WATER hemihydrate is known to " age ", which is a blanket term used to cover various crystallisa- tion phenoiiiena which usually result i n a reduction of surface area, etc. No significant differences were noted between the two sources. HexagonaI CaS04 rehydrates rapidly in normal atmospheres, so an apparatus was constructed (fig.1) which allowed the starting materials to be hydrated under controlled A 0 FIG. 1 .-Diagram of apparatus : A, B, heated flasks containing solvents ; C, C1, condensers ; T, thermometers ; D, nitrogen inlet ; E, sample suspension from balance arm ; F, siphon with two-way tap ; X, Y, constrictions sealed by movable inserts. conditions and also allowed the hexagonal CaS04 to be rehydrated in situ, under varying conditions. Samples were dehydrated in flowing nitrogen by allowing solvent A to reflux up to the double-walled reaction zone (condenser C, replaced by a stopper, X open, Y closed). Rehydrations were carried out by first switching solvents (X closed, Y open and condenser CB replaced by a stopper).The sample thermometer (Ti) usually indicated a constant temperature within 2-3 min. After about 5 min, to allow the sample to reach the new temperature, the nitrogen flow was switched through a thermostatted water bubbler so that the rehydration could start. Sample weights of 100 mg+ 10 % of the dihydrate were used throughout and the dehydra- tion conditions were standardised at 365 K and 1.3 x N m-2 partial pressure of water vapour. The heating time was also standardised at 5 h. This time was very close to 30 min longer than the time for dehydration. Other times, temperatures and partial vapour pressures were used for the dehydration but made no observable differences to the behaviour of the anhydrous material.Samples of the hemihydrate were heated for only 2.5 h, since these lost the smaller quantity of water more rapidly than the dihydrate. The heating times for the anhydrous compound were still 30 min in this case.M . C . BALL AND L . S. NORWOOD 171 The temperature range used for the rehydration experiments was limited by the hexagonal- The rehydration products were identified by X-ray diffraction and d.t.a.' orthorhombic CaSO, phase change, which is known to occur above 415 K. RESULTS The rehydration product in all cases was P-CaSO4,+H20, with the d.t.a.* exotherm occurring at 573 K. No orthorhombic CaSO, could be detected by X-rays. The rehydration curves were all deceleratory in shape and reduced time plots, based on the time for 50 % reactioi~,~ showed that the best fit for the experimental data was in terms of a one dimensional diffusion process, following the relation a2 = kt.Rate constants were derived from graphs of a2 against time (a = proportion decomposed). Table 1 gives the temperature and partial pressure of runs carried out as well as the derived rate constants. TABLE 1 .-EXPERIMENTAL CONDITIONS AND DERIVED RATE CONSTANTS (min-') FOR THE REHYDRATION OF HEXAGONAL as04 TIK 298 304 332 341 349 355 360 365 375 384 394 pressure of water vapour, P/(lO3 N m-2) -_ 3.0 4.4 5.9 0.027 0.057 0.067 0.042 0.073 0.091 0.093 0.062 0.083 0.108 0.050 0.060 0.082 0.03 1 0.055 0.087 0.01 6 0.049 0.069 0.014 0.040 __ 8.2 0.128 0.135 0.169 0.130 0.118 0.101 0.076 0.007 DISCUSSION The one dimensional diffusion process suggested by the reduced time plots is in line with the known crystal structures of both the hemihydrate and hexagonal CaSO,.The dehydration reaction of the hemihydrate has been studied and shown to be diffusion controlled also. EFFECT OF WATER VAPOUR PRESSURE ON REACTION RATE Fig. 2 shows a plot of reaction rate constant against relative pressure (PIP,) of water vapour for the middle range of temperatures. To a first approximation the plots are linear and when extrapolated intersect the PIP, axis in the region of P/P, = 0.01-0.02. This extrapolation suggests that the rate of rehydration would be zero (or very close to zero) at a value of PIPo !z 0.01 5. This value corresponds very closely with the values recorded by Kelley lo as the equilibrium vapour pressures for P-CaSO,, 3H,O.It can therefore be said that the rehydration will only proceed when PIP, for the water vapour exceeds the equilibrium pressure of the product, and suggests that adsorption of water plays an important part in the rehydration, presumably via an adsorbed film. EFFECT OF TEMPERATURE ON REACTION RATE By inspection of table 1 it can be seen that the rate constants reach a maximum at 355 K. An Arrhenius plot derived from these data is shown in fig. 3 and it is obvious172 STUDIES IN CALCIUM SULPHATEIWATER that the Arrhenius equation is not obeyed by the reaction as at present described. It is possible to explain deviations from the Arrhenius equation in terms of parallel or consecutive reactions,' ' but these explanations usually refer to plots which show two or more straight line portions.The log k against 1 /T plots shown in fig. 3 can be seen to be more complex than these relatively simple suggestions allow. 0 0.05 0.10 0.15 0.20 0 . 2 5 0.30 PIP0 FIG. 2.-Rate constant as a function of PIPo for various temperatures : ., 384 K ; T7 , 375 K ; U, 365K; x , 3 5 5 K ; 0 , 3 6 0 K ; 0 , 3 4 9 K . 4 1.0- 1.6 ' 1.9 * 6 2 . 7 2.9 3. I 3,3 2.2 2.5 1 0 3 ~ 1 ~ FIG. 3.-LOg rate constant against lfTas a function of water vapour pressure : A, 8.20 x lo3 N m-2 ; 8, 5.92 x lo3 N m-* ; x ,4.44 x lo3 N m-2 ; 0,2.95 x lo3 N m-'. It seemed reasonable that the pressure dependence of the rate constant was in some way complicating the temperature dependence. Inspection of the results in table 1 or fig. 3 show that the high temperature behaviour is unexpected, where the rates fall off rapidly as the temperature is increasing.This is contrary to normal kinetic behaviour but can be explained if it i s assumed that adsorption of water isM. C . BALL AND L. S . NORWOOD 173 critical during the rehydration. At constant partial pressure of water vapour (P = constant), as the temperature is raised, the equilibrium pressure (Po) of water increases, therefore PIPo decreases. Since the volume of water adsorbed on any surface is proportional to P/Po, then fig. 3 can be explained if it is assumed that rehydration only occurs on surfaces covered by at least a monolayer of adsorbed water. At low temperatures a complete monolayer or multilayer is present and the rates increase as the temperature increases.At about 355 K, however, the complete monolayer breaks down and only part of the surface is covered. This coverage decreases rapidly as the temperature rises, giving the maximum observed in fig. 3. The precise form of the dependence of rate on partial pressure is difficult to decide a priori. If no more than a monolayer was necessary for the rehydration, then presumably a Langmuir-type dependence would be observed. This type of behaviour has been noted by Messier l2 for the high temperature hydrolysis of magnesium fluoride. Plots of rate constant against pressure for the present data were approxi- mately linear suggesting non-langmuir behaviour at least within the pressure range studied. If more than a monolayer is necessary for the rehydration then a more complex relationship should be observable, but should include the monolayer possibility.The B.E.T. relationship l 3 is the most widely used multilayer adsorption equation, and can be applied to the present results if it is assumed (a) that the rate of rehydration is proportional to the volume of water adsorbed (V,) and that the monolayer volume (V,) is constant, (b) that CBSEaT. = 1 ; water vapour adsorption commonly gives low values of CBeEeT.14. These assumptions lead to the relationship kexp = k' x (PlPo)/(1 -P/Po) where k,,, is the measured rate of rehydration and k' is the theoretical rate corrected for the effects of water vapour. The experimental rate is also the difference between rehydration and possible dehydration at a given temperature. Extrapolation of published data on the dehydra- tion l 5 allowed these corrections for possible dehydration to be made.These were very small and did not alter the turn-down in rate. Plots of k,,, against (P/Po)/(l -P/P,) are shown in fig. 4, which shows that for the lower temperatures, reasonable straight lines passing near the origin are obtained. The lines for the two highest temperatures do not pass through the origin, suggesting that the B.E.T. derivation does not hold at these temperatures. It seems reasonable therefore to suggest that the monolayer/multilayer breaks down at some temperature between 365 and 375 K. The corrected rates were used to derive the second Arrhenius plot shown in fig. 5. Most of the points lie on a curve which is virtually independent of vapour pressure. As expected from fig.4, those points corresponding to 375 and 383 K lie off this curve. For both fig. 4 and 5 it is useful to note that the B.E.T. relationship is usually taken to hold between the limits of PIPo = 0.05-0.30.13 The PIPo limits for the present data lie between 0.027 and 0.286, with the lower values at the higher temperatures. The breakdown occurs in fig. 5 at values of P/Po <0.06. The best straight line through the low temperature points in fig. 5 gives an activa- tion energy of close to 46 kJ mol-1 ; this value is virtually independent of water vapour pressure. This activation energy is close to that previously determined for the de- hydrationY3 and is close to that for the isosteric heat of sorption of water determined by Flanagan as 50 kJ mol-l.The frequency factor A calculated on the basis of this activation energy is 3.5 x lo6 s-l. This is a low value corresponding to reduction174 STUDIES IN CALCIUM SULPHATE/WATER in the degrees of freedom possessed by the activated complex relative to a gaseous molecule (A 21 1013 s-l). It corresponds to an entropy loss in the formation of the transition state of - 117 J inol-1 K-l and agrees very closely with the loss of entropy associated with the condensation of water (- 110 J mol-l K-’ for liquefaction, - 132 J iii01-~ K-l for solidification). (P/Po)/(l - W o ) FIG, 4.--Ylot of rate constant against (P/Po)/(l -P/Po) for various temperatures : M, etc. as fig. 2. 1 0 3 ~ 1 ~ FIG. 5.-Plot of corrected rate constant against 1 /T: 0, 8.20 x lo3 N m-’ ; A, 5.92 x lo3 N m-’ ; x ,4.44 x lo3 N ni-’ ; 0,2.95 x lo3 N m-’.When the number of simplifications used in the derivation are considered, the improvement in the experimental data to give the results outlined suggests that the approach used is basically sound, although not yet complete.M . C. BALL A N D L. S . NORWOOD 175 G. K. Layden and G. W. Brindley, J. Amer. Ceram. SOC., 1963,46,518. R. C. Eckhardt, P. M. Fichte and T. B. Flanagan, Trans. Faraday SOC., 1971,67, 1143. M. C. Ball and L. S . Norwood, J. Chem. SOC. A , 1970,1476. D. W. Florke, Neues Jahrb. Miner., 1952,84, 189. W. A. Caspari, Proc. Roy. SOC. A, 1936,155,41. D. A. Holdridge and E. G. Walker, Proceedings 10th Int. Ceramic Congress, Stockholm, 1966, p. 89. J. H. Sharp, G. W. Brindley and B.N. N. Achar, J. Amer. Ceram. SOC., 1966,49, 379. 1941. ' A.S.T.M. Index, Card number 2.0675. * W. E. P. Fleck, M. H. Jones, R. A. Kuntze and H. G. McAdie, Cunad. J. Chem., 1960,38,936. lo K. K. Kelley, J. C. Southard and C. T. Anderson, U.S. Bureau ufMines Tech. Paper No. 625, l 1 B. G. Gowenlock, Quart. Rev., 1960, 14, 133. l2 D. R. Messier, J. Amer. Ceram. SOC., 1965, 48, 452. l3 P. W. M. Jacobs and F. C. Tompkins in Chemistry ofthe Solid State, ed. W. E. Garner (Butter- l4 J. Hagymassy, S. Brunauer and R.. Sh. Mikhail, J. Colloid Interface Sci., 1969, 29, 485. worths, London, 1955), p. 94. M. C. Ball and R. G. Urie, J. Chem. SOC. A, 1970,528.Studies in the System Calcium Sulphate/WaterPart 4.-Rehydration of Hexagonal CaSO,BY M. C. BALL* AND L.S. NORWOODDepartment of Chemistry, University of Technology, Loughborough,Leics. LEll 3TUThe rehydration of hexagonal CaS04 has been studied in the temperature range 298-393 K and atwater vapour pressures of 3 x 103-8.2 x lo3 N m-'. The reaction is controlled by diffusion of water,following the parabolic law. The rate of rehydration is strongly affected by the external watervapour pressure, probably through an adsorbed multilayer of water. This effect has been correctedfor in terms of the B.E.T. relationship and an activation energy of 46 kJ mol-1 has been calculatedfor the rehydration.Relatively few rehydration reactions have been studied compared with dehydra-tions. This is probably due to the non-reversible nature of many dehydrations, butmay also be due to the requirement of relatively low temperatures and hence slowreaction rates.Two worthy of mention are those on the hydration of magnesiumoxide and the rehydration of manganese(I1) formate.2 Both of these rehydrationsshow large effects by water vapour on the reaction rate, together with a threshold effectwhere a minimum water vapour pressure is required for rehydration to start.The dehydration reactions of the various hydrates of calcium sulphate have beenextensively studied. Hexagonal CaSO, is the primary anhydrous product producedfrom both the dihydrate (gypsum) and the hemihydrate, as illustrated in the reactionscheme.CaS04, 2H204/hexagonal CaSO,-, orthorhombic CaSO, 11 11CaSO,, 3H20The crystal structures of the hemihydrate and hexagonal CaSO, are very similar,,consisting of diamond shaped channels some 4A across, formed by a Ca-OSO,network. These channels have been likened to those in zeolite^,^ but it seems likelythat because the channels are so small, either all sites for water molecules are filled, ornone.Hexagonal CaSO, is extremely reactive towards water vapour, and the re-hydration has been little studied. Holridge and Walker studied the effects ofhumidity and sample ageing on the rehydration, using differential thermal analysis asthe diagnostic technique. They found evidence of an extra endothermic peak in thedifferential thermal analysis (d.t.a.) traces which they suggested was caused by re-moval of a sorbed monolayer of water.EXPERIMENTALTwo starting materials were used as sources of hexagonal &SO4. Most runs were carriedout on material derived from calcium sulphate dihydrate and these are the results which willbe quoted throughout.A few runs were carried out using so-called P-CaS04,3H20. Thiswas produced by dehydrating the dihydrate at 390 K and allowing this anhydrous material torehydrate in the air. The reason far considering this somewhat cyclic treatment is that the16170 STUDIES I N CALCIUM SULPHATE/WATERhemihydrate is known to " age ", which is a blanket term used to cover various crystallisa-tion phenoiiiena which usually result i n a reduction of surface area, etc. No significantdifferences were noted between the two sources.HexagonaI CaS04 rehydrates rapidly in normal atmospheres, so an apparatus wasconstructed (fig.1) which allowed the starting materials to be hydrated under controlledA 0FIG. 1 .-Diagram of apparatus : A, B, heated flasks containing solvents ; C, C1, condensers ; T,thermometers ; D, nitrogen inlet ; E, sample suspension from balance arm ; F, siphon with two-waytap ; X, Y, constrictions sealed by movable inserts.conditions and also allowed the hexagonal CaS04 to be rehydrated in situ, under varyingconditions. Samples were dehydrated in flowing nitrogen by allowing solvent A toreflux up to the double-walled reaction zone (condenser C, replaced by a stopper, X open,Y closed). Rehydrations were carried out by first switching solvents (X closed, Y open andcondenser CB replaced by a stopper). The sample thermometer (Ti) usually indicated aconstant temperature within 2-3 min.After about 5 min, to allow the sample to reach thenew temperature, the nitrogen flow was switched through a thermostatted water bubbler sothat the rehydration could start.Sample weights of 100 mg+ 10 % of the dihydrate were used throughout and the dehydra-tion conditions were standardised at 365 K and 1.3 x N m-2 partial pressure of watervapour. The heating time was also standardised at 5 h. This time was very close to 30 minlonger than the time for dehydration. Other times, temperatures and partial vapour pressureswere used for the dehydration but made no observable differences to the behaviour of theanhydrous material. Samples of the hemihydrate were heated for only 2.5 h, since these lostthe smaller quantity of water more rapidly than the dihydrate.The heating times for theanhydrous compound were still 30 min in this caseM . C . BALL AND L . S. NORWOOD 171The temperature range used for the rehydration experiments was limited by the hexagonal-The rehydration products were identified by X-ray diffraction and d.t.a.'orthorhombic CaSO, phase change, which is known to occur above 415 K.RESULTSThe rehydration product in all cases was P-CaSO4,+H20, with the d.t.a.* exothermoccurring at 573 K. No orthorhombic CaSO, could be detected by X-rays.The rehydration curves were all deceleratory in shape and reduced time plots,based on the time for 50 % reactioi~,~ showed that the best fit for the experimentaldata was in terms of a one dimensional diffusion process, following the relationa2 = kt.Rate constants were derived from graphs of a2 against time (a = proportiondecomposed). Table 1 gives the temperature and partial pressure of runs carried outas well as the derived rate constants.TABLE 1 .-EXPERIMENTAL CONDITIONS AND DERIVED RATE CONSTANTS (min-') FOR THEREHYDRATION OF HEXAGONAL as04TIK298304332341349355360365375384394pressure of water vapour, P/(lO3 N m-2) -_3.0 4.4 5.90.0270.0570.0670.042 0.073 0.0910.0930.062 0.083 0.1080.050 0.060 0.0820.03 1 0.055 0.0870.01 6 0.049 0.0690.014 0.040__8.20.1280.1350.1690.1300.1180.1010.0760.007DISCUSSIONThe one dimensional diffusion process suggested by the reduced time plots is inline with the known crystal structures of both the hemihydrate and hexagonal CaSO,.The dehydration reaction of the hemihydrate has been studied and shown to bediffusion controlled also.EFFECT OF WATER VAPOUR PRESSURE ON REACTION RATEFig.2 shows a plot of reaction rate constant against relative pressure (PIP,) ofwater vapour for the middle range of temperatures. To a first approximation theplots are linear and when extrapolated intersect the PIP, axis in the region of P/P, =0.01-0.02. This extrapolation suggests that the rate of rehydration would be zero(or very close to zero) at a value of PIPo !z 0.01 5. This value corresponds very closelywith the values recorded by Kelley lo as the equilibrium vapour pressures for P-CaSO,,3H,O. It can therefore be said that the rehydration will only proceed when PIP, forthe water vapour exceeds the equilibrium pressure of the product, and suggests thatadsorption of water plays an important part in the rehydration, presumably via anadsorbed film.EFFECT OF TEMPERATURE ON REACTION RATEBy inspection of table 1 it can be seen that the rate constants reach a maximum at355 K.An Arrhenius plot derived from these data is shown in fig. 3 and it is obviou172 STUDIES IN CALCIUM SULPHATEIWATERthat the Arrhenius equation is not obeyed by the reaction as at present described.It is possible to explain deviations from the Arrhenius equation in terms of parallelor consecutive reactions,' ' but these explanations usually refer to plots which showtwo or more straight line portions.The log k against 1 /T plots shown in fig. 3 canbe seen to be more complex than these relatively simple suggestions allow.0 0.05 0.10 0.15 0.20 0 . 2 5 0.30PIP0FIG. 2.-Rate constant as a function of PIPo for various temperatures : ., 384 K ; T7 , 375 K ; U,365K; x , 3 5 5 K ; 0 , 3 6 0 K ; 0 , 3 4 9 K .41.0-1.6 '1.9 *62 . 7 2.9 3. I 3,3 2.22.51 0 3 ~ 1 ~FIG. 3.-LOg rate constant against lfTas a function of water vapour pressure : A, 8.20 x lo3 N m-2 ;8, 5.92 x lo3 N m-* ; x ,4.44 x lo3 N m-2 ; 0,2.95 x lo3 N m-'.It seemed reasonable that the pressure dependence of the rate constant was insome way complicating the temperature dependence. Inspection of the results intable 1 or fig.3 show that the high temperature behaviour is unexpected, where therates fall off rapidly as the temperature is increasing. This is contrary to normalkinetic behaviour but can be explained if it i s assumed that adsorption of water iM. C . BALL AND L. S . NORWOOD 173critical during the rehydration. At constant partial pressure of water vapour (P =constant), as the temperature is raised, the equilibrium pressure (Po) of waterincreases, therefore PIPo decreases. Since the volume of water adsorbed on anysurface is proportional to P/Po, then fig. 3 can be explained if it is assumed thatrehydration only occurs on surfaces covered by at least a monolayer of adsorbedwater. At low temperatures a complete monolayer or multilayer is present and therates increase as the temperature increases.At about 355 K, however, the completemonolayer breaks down and only part of the surface is covered. This coveragedecreases rapidly as the temperature rises, giving the maximum observed in fig. 3.The precise form of the dependence of rate on partial pressure is difficult to decidea priori. If no more than a monolayer was necessary for the rehydration, thenpresumably a Langmuir-type dependence would be observed. This type of behaviourhas been noted by Messier l2 for the high temperature hydrolysis of magnesiumfluoride. Plots of rate constant against pressure for the present data were approxi-mately linear suggesting non-langmuir behaviour at least within the pressure rangestudied.If more than a monolayer is necessary for the rehydration then a more complexrelationship should be observable, but should include the monolayer possibility.TheB.E.T. relationship l 3 is the most widely used multilayer adsorption equation, andcan be applied to the present results if it is assumed (a) that the rate of rehydration isproportional to the volume of water adsorbed (V,) and that the monolayer volume(V,) is constant, (b) that CBSEaT. = 1 ; water vapour adsorption commonly gives lowvalues of CBeEeT.14.These assumptions lead to the relationshipkexp = k' x (PlPo)/(1 -P/Po)where k,,, is the measured rate of rehydration and k' is the theoretical rate correctedfor the effects of water vapour.The experimental rate is also the difference between rehydration and possibledehydration at a given temperature. Extrapolation of published data on the dehydra-tion l 5 allowed these corrections for possible dehydration to be made.These werevery small and did not alter the turn-down in rate.Plots of k,,, against (P/Po)/(l -P/P,) are shown in fig. 4, which shows that for thelower temperatures, reasonable straight lines passing near the origin are obtained.The lines for the two highest temperatures do not pass through the origin, suggestingthat the B.E.T. derivation does not hold at these temperatures. It seems reasonabletherefore to suggest that the monolayer/multilayer breaks down at some temperaturebetween 365 and 375 K.The corrected rates were used to derive the second Arrhenius plot shown in fig.5.Most of the points lie on a curve which is virtually independent of vapour pressure.As expected from fig. 4, those points corresponding to 375 and 383 K lie off this curve.For both fig. 4 and 5 it is useful to note that the B.E.T. relationship is usually takento hold between the limits of PIPo = 0.05-0.30.13 The PIPo limits for the presentdata lie between 0.027 and 0.286, with the lower values at the higher temperatures.The breakdown occurs in fig. 5 at values of P/Po <0.06.The best straight line through the low temperature points in fig. 5 gives an activa-tion energy of close to 46 kJ mol-1 ; this value is virtually independent of water vapourpressure. This activation energy is close to that previously determined for the de-hydrationY3 and is close to that for the isosteric heat of sorption of water determinedby Flanagan as 50 kJ mol-l.The frequency factor A calculated on the basis ofthis activation energy is 3.5 x lo6 s-l. This is a low value corresponding to reductio174 STUDIES IN CALCIUM SULPHATE/WATERin the degrees of freedom possessed by the activated complex relative to a gaseousmolecule (A 21 1013 s-l). It corresponds to an entropy loss in the formation of thetransition state of - 117 J inol-1 K-l and agrees very closely with the loss of entropyassociated with the condensation of water (- 110 J mol-l K-’ for liquefaction, - 132 J iii01-~ K-l for solidification).(P/Po)/(l - W o )FIG, 4.--Ylot of rate constant against (P/Po)/(l -P/Po) for various temperatures : M, etc. as fig. 2.1 0 3 ~ 1 ~FIG. 5.-Plot of corrected rate constant against 1 /T: 0, 8.20 x lo3 N m-’ ; A, 5.92 x lo3 N m-’ ;x ,4.44 x lo3 N ni-’ ; 0,2.95 x lo3 N m-’.When the number of simplifications used in the derivation are considered, theimprovement in the experimental data to give the results outlined suggests that theapproach used is basically sound, although not yet completeM . C. BALL A N D L. S . NORWOOD 175G. K. Layden and G. W. Brindley, J. Amer. Ceram. SOC., 1963,46,518.R. C. Eckhardt, P. M. Fichte and T. B. Flanagan, Trans. Faraday SOC., 1971,67, 1143.M. C. Ball and L. S . Norwood, J. Chem. SOC. A , 1970,1476.D. W. Florke, Neues Jahrb. Miner., 1952,84, 189.W. A. Caspari, Proc. Roy. SOC. A, 1936,155,41.D. A. Holdridge and E. G. Walker, Proceedings 10th Int. Ceramic Congress, Stockholm, 1966,p. 89.J. H. Sharp, G. W. Brindley and B. N. N. Achar, J. Amer. Ceram. SOC., 1966,49, 379.1941.' A.S.T.M. Index, Card number 2.0675.* W. E. P. Fleck, M. H. Jones, R. A. Kuntze and H. G. McAdie, Cunad. J. Chem., 1960,38,936.lo K. K. Kelley, J. C. Southard and C. T. Anderson, U.S. Bureau ufMines Tech. Paper No. 625,l 1 B. G. Gowenlock, Quart. Rev., 1960, 14, 133.l2 D. R. Messier, J. Amer. Ceram. SOC., 1965, 48, 452.l3 P. W. M. Jacobs and F. C. Tompkins in Chemistry ofthe Solid State, ed. W. E. Garner (Butter-l4 J. Hagymassy, S. Brunauer and R.. Sh. Mikhail, J. Colloid Interface Sci., 1969, 29, 485.worths, London, 1955), p. 94.M. C. Ball and R. G. Urie, J. Chem. SOC. A, 1970,528

ISSN:0300-9599    

DOI:10.1039/F19736900169

出版商:RSC    

年代:1973

数据来源: RSC

摘要:

Deficiency in the Analysis of Physical Adsorption Data BY LINDA A. BRUCE AND MARGARET H. SHERIDAN C.S.I.R.O. Division of Tribophysics, University of Melbourne, Parkville, Victoria 3052, Australia Received 24th July, 1972 The criterion of fit to adsorption data advocated by Baker and Head is shown to be too lax, and likely to lead to spurious analysis. Computer-generated parameters fitting the derived heats and entropies of adsorption yield isotherms in disagreement with the experimental isotherms from which these derived quantities were evaluated. Reasons for this discrepancy are considered, and an improved criterion of fit is suggested. Impediments to establishing a unique description of a surface by this type of analysis are examined. 1. INTRODUCTION Several procedures have been described 1-3 for analysis of adsorption isotherms on the basis that the surface studied displays a continuous range of adsorption energies.Such treatments assume that the entropy of vibration retained by the adsorbed species is constant over the entire surface and hence inherently require that the entropy of vibration be independent of the energy of adsorption. Secondly, there is no possibility of changing the " local isotherm " equation with changing energy of adsorption, so that adsorption over the entire surface must be described either by a localised isotherm or by a non-localised isotherm. Where the computed energy distribution has any finite width, neither of these requirements is necessarily justified. An alternative method 4a for analysis of adsorption isotherms assumes that the surface studied displays only a small finite number of " patches " of different adsorption energies.Adsorption on any patch may be either localised or non-localised, and the entropy of vibration of the adsorbate may also vary from patch to patch. Although the authors feel that the patch model is a useful and realistic one, the purpose of the present paper is to show that the method of analysis, as it was originally applied, 4-6 provided an inadequate description of the experimental surface. 2. BASIC PROCEDURE There is no analytical solution to the competitive filling of the constituent patches of such a model, and a computer programme was devised 4a to evaluate the filling for a given set of patches. Each patch, i, was characterized by four parameters : (1) the relative proportion, ni, of the patch occurring in the surface, (2) the initial heat of adsorption, iqst, (3) an interaction constant, A i , allowing the heat of adsorption to increase linearly with coverage, and (4) an entropy constant, Ci, where for a patch on which adsorption is localised iASdiff.= iS,-S: = -Rln(0~/(1-8J)-Ci while if adsorption is non-localisedL. A. BRUCE AND M. H. SHERIDAN 177 An " educated guess " was used in assigning initial numerical values to these para- meters for the chosen number of patches. At selected values of the total coverage, the computer programme evaluated the filling of the individual patches determined by the condition that the adsorbed phase on all patches must have the same chemical potential; the isosteric heat and differential entropy of adsorption on the model surface werc theq returned as the weighted averages of these qwntitieg on each individual patch at its compvted coverage, and comparison was made with the experi- mental values on the real surface at the same total coverage.After changing the first parameter of one of the proposed patches, the computations were repeated, and automated decision of improvement of fit to experiment resulted in acceptance or rejection of the change. The programme cycled, testing each parameter in turn until no significant improvement could be made using this particular set of patches. The experimental values for the heat and entropy were evaluated by thermodynamic analysis of the set of experimental isotherms using the integrated form of the Clausius- Clapeyron equation, In Po = - (qst/RT) + K, where K is the integration constant.Treating this equation as linear in In Po and ( l / T ) , least-squares analysis gives the isosteric heat, qst, and a standard Ceviation of its estimate; the standard differential entropy is given by &&iff. = s, - S i = R In po - RK, also with a calculated standard deviation (s, is the differential entropy of the adsorbed phase and is the molar entropy of the gas phase at pressure p"). A model was accepted as a successful fit if the sum of fourth powers of the differences between experimental and model heats and entropies came within the sum of fourth powers of the standard deviations on the experimental values derived from such an analysis.Thus, the original concept of Baker and Head 4a involved the fact that the information in a set of isotherms is summarized in the values of isosteric heat and differential entropy obtained by thermo- dynamic analysis. Their criterion of fit seems very plausible, but will now be shown to be inadequate. 3 . THE DISCREPANCY The above procedure was applied to the measurements of adsorption of xenon on films of gold, aluminium and silver deposited in various ways in order that the metal exposed different proportions of major crystallographic planes.' The aim was to identify the parameters relevant to adsorption on individual crystal planes, intuitively identified with the patches of the model. It was disturbing to discover that several quite different but equally acceptable computer-generated " solutions " for any one set of experimental data could not be distiaguished by the criterion of fit, even when the number of patches was restricted to two.Such conflicting " solutions " arose for each of the metals, and fig. l(a) gives an example of the fit to the derived experi- mental data, qsr and ASdiff., for xenon on gold/Pyrex, for one of the '' solutions ". In order to check the accuracy of fit to the primary data, isotherms were calculated from these computed values of heat and entropy, from These computed isotherms, fig. l(b), reveal that the fit to the primary data is very poor. A similar situation arose when isotherms were computed from other solutions " acceptable " on the basis of fitting qst and ASdiff., including those already pub- l i ~ h e d .~ - ~ For example, fig. 2 shows a comparison between the experimental isotherms measured for xenon on nickel/Pyrex and isotherms constructed from the published analysis of the heats and entropies according to a patch model,6 using eqn (3.1). Again, the accepted fit to the deriuedexperimntal data, qst and ASdiff. (fig. 6, ref. (6)), is not at all a good fit to the primary experimental data, the set of isotherms.ANALYSIS OF ADSORPTION DATA 24 22 -60-80 pressure/N nr2 (b) FIG. 1.-(u) Computed “ fit ” (- - -) to derived experimental data ( x ) for xenon on gold/Pyrex.’ (b) Isotherms calculated from the “ fit ” shown in (a) (- - -), compared with three of the original experimental isotherms (-) from which the derived data were evaluated.I I I 4 I I I I l I l l I I - . I o+ lo5 I o4 to - pi*essure/N rn-’ FIG. 2.-Xenon on nickel/Pyrex : experimental isotherms (-) and isotherms computed using “ best-fit ” parameters (- - -). 4. ORIGIN OF THE DISCREPANCY This seeming inconsistency arises from the way in which errors in pressure are From eqn (3. l), we find that the error Ap in p determined by errors in qst and ASdiff.. arising from errors A(qst) and A(Asdiff.) is given by p+Ap = exp([-(%,$-A(%,))/RT1-[(ASdiff. +A(Asdiff.>)/Rl +InPo) (4*1) i.e., This shows that, for example, an increase either of 4 J K-I niol-l in ASdiff. or of 0.4 kJ mol-l in qst at 84 K (mid-range of the experimental temperatures) would give ( p + Ap)/ ~ ~ 0 . 6 , while decreases of the same magnitude would give (p+Ap)/p= 1.7, yet the values used to calculate such relative errors in p are typical of the standard deviations on qst and ASdlff, evaluated from experimental isosteres in which the experimental (P+AP)/P = exp{ -rA(%t)/~Tl- [A(A~dlff.)/Rll.(4.2)L. A. BRUCE AND M . H . SHERIDAN 179 pressures are certainly not more than 20 % from a straight i.e. 0.8 < {(p + Ap)/ p>,,,t. < 1.2. The vital point is that the experimental estimates of qst and ASdiff are not mutually independent. In the above example, such a change in the estimate of qst or of ASdiff, would not arise in isolation, but would be counterbalanced by a change of opposing sign in the other quantity. However, the procedure of fitting the heat and entropy data, discussed above, involves summing the fourth powers of the differences between experimental and model values, and ignores any effect of the sign of these differences.Thus the fitting procedure takes no account of the inverse correlation that must exist between the variations in qst and ASdiff. if isotherms are to be deduced in reasonable agreement with experiment. Close examination of fig. l(a) shows that in this particular " fit ", the experimental values of qst and ASdiff. at most coverages differ in the same sense from the computed values, leading to the discrepancy shown in fig. l(b). The discrepant fit shown in fig. 2 is also accounted for in these terms. The analyses of adsorption results on several surface^,^-^ based on this criterion of fit, must be taken as spurious. 5. MODIFICATIONS TO THE FITTING PROCEDURE Evidently the criterion of fitting the heat and entropy values is much too lax as it stands.While acceptance of non-compensating differences between the computed and experimental values of the heat and entropy of adsorption could simply be penal- ised in the computer programme, we decided that in any future use of the model it would be more logical to fit the primary data, the set of isotherms. The experimental error (say +20 %) in measuring a pressure could then be used directly as a standard of acceptability of fit by summing the fourth powers of the relative errors between the computed and experimental values of the pressure. (In this case the accepted computed value of pressure may of course be either greater or less than the experi- mental one.) We further decided that close attention be given to assessing the numerical values of parameters on grounds other than those of merely fitting the isotherms.For example, the value of Ci, the entropy constant, is easily shown to depend on the type of adsorption, i.e. localised or non-localised, on patch i;' C, is also related to the isosteric heat of adsorption iqst through the entropy of vibration@) of the ad~orbate.~ A * , the interaction constant, also varies with the type of adsorption and with postu- lated arrangements of adsorbate atoms on particular crystal planes (intuitively the patches) of an adsorbent; acceptable values for Al should be justifiable in terms of Lennard-Jones summations of the interactions between adsorbate atoms spaced apart at definite distances.While none of the estimates of these parameters can be absolute, quite narrow limits can be placed on reasonable values for a particular gas/metal system, and any values returned by the computer which are far outside these limits should be regarded with scepticism. However, it is important to appreciate that since a change in Cl of 4 J K-l mol-1 or in iqst of 0.4 kcal mol-1 alters the calculated pressure (eqn (2.1) or (2.2) and (4.2)) at constant coverage on a patch by up to 70 %, even seemingly small variations in the allowed relative values of Ci and iqst can result in substantial changes in the competitive filling of the set of patches under consider- ation. Computation of the heat and entropy of adsorption at one temperature, followed by the use of eqn (3.1) to evaluate isotherms at other temperatures, has also been found likely to lead to errors in fitting.Since isotherms generated by the parameters of each individual patch vary differently with temperature, the patch coverages at which the180 ANALYSIS OF ADSORPTlON DATA equal chemical potential conditions are met are a function of temperature, and so also is the rate of filling of the patches, dQt/dO, which determines the computed isosteric heat and differential entropy for the In the revised fitting procedure, the competitive filling of the set of patches under test is calculated at each temperature for which an experimental isotherm is to be fitted. To ensure that these computations are made as accurately as possible, the initial heat of adsorption, iqst, on each patch is adjusted with temperature (by RAT) and the entropy constant Ci is also adjusted with temperature for the changing entropies of translation and vibration contributing to its value.The overall result is of slightly different computed heats and entropies of adsorption at each temperature, which, using eqn (3. l), give pressures varying gen- erally by some 10-20 % from those calculated assuming temperature invariance; in particular circumstances, this figure could be much greater. The precise effect is unique to a set of patches, and the extent cannot be calculated analytically. Huang lo has recently discussed essentially this effect in a very restricted case. 6 . APPLICATION TO EXPERIMENTAL RESULTS Using the fully modified procedure, attempts have been made to establish a patch model describing the adsorption of xenon on aluminiuni/Pyrex,' by matching com- puted isotherms to a set of three experimental ones.The entropy of adsorption on the (100) and (1 11) planes, believed to predominate in the surface of the film, was described either by eqn (2.2) for non-localised adsorption, or by polynomials established by Baker l1 to describe localised adsorption with blocking of sites when a large atom is adsorbed on these planes. Since higher index planes might be repre- sented in the surface to a small extent, eqn (2.1) for localised adsorption was used to describe adsorption on their more widely separated sites. The only calculations fully relevant to the acceptability of particular entropy functions are those of Bacigalupi and Neustadter.12 Their results indicate that in the case of xenon adsorption on aluminium, the activation barrier to diffusion on a (1 11) plane would be less than RT for adsorption energies below 23 kJ mol-l in the experimental range of temperatures, suggesting that xenon would be mobile under such circumstances. On the other hand, on a (100) plane, even at an adsorption energy as low as 14 kJ mol-l, the activation energy required for diffusion would be prohibitively high, and xenon would be localised.Nevertheless, trials using numerous sets of starting parameters have been run, involving all possible combinations of these entropy functions. The results returned by the computer as the best-fit values (quoted for 84 K) on three of these attempts to fit the data are listed in table 1." solution " I I1 I11 TABLE 1 .-" BEST-FIT " PARAMETERS RETURNED BY COMPUTER fraction of patch total surface 1Ystl At1 ci I i ni (kJ mol-1) (kJ mol-1) (J K - 1 mol-1) cntropy function 1 0.670 15.14 4.26 68.9 (2.2) 2 0.305 16.86 2.91 79.6 (100) polynomial 3 0.025 28.84 0 87.2 (2.1) 1 0.339 15.29 4.04 69.1 (2.2) 2 0.645 16.74 2.18 80.4 (100) polynomial 3 0.026 27.93 0 87.4 (2.1) I 0.481 15.86 4.12 75.4 (2.2) 2 0.467 17.15 2.21 85.4 (100) polynomial 3 0.05 1 27.92 0 92.4 (2.1)L . A. BRUCE AND M. H . SHERIDAN 181 In the set of three isotherms fitted, no computed pressure differed froin the corresponding experimental value by more than 25 % in the case of solution I, and 15 % in the case of solution 11.For these two solutions, the values returned for the thermodynamic parameters identified with (1 1 l), (100) and higher index planes (patches 1, 2 and 3 respectively) are very similar and fulfil the criteria suggested in section 5. However, the values returned as the fraction of the total surface repre- sented by patches 1 and 2 actually inverts between solutions I and 11, in the one case requiring 67 % of the surface to be (1 11) planes and 30.5 % (100) planes, whereas in the other case only 34 % is (111) planes but 64.5 % is (100). (In each instance, only the very minor residue is of high index planes.) Obviously, both solutions cannot be “ real ”. Since no preferred orientation was observed for this film by electron micro- scopy, there is no way of absolutely rejecting one of them.While the second solution is a better fit with experiment, there remains the suspicion that a still better fit could be found in which the thermodynamic parameters are still very similar, but the relative proportions of major planes differ again. In fact, due to a mistake in the control of values for Ci, solution I11 was obtained in which there was only 10 % relative error in fitting these three isotherms. Because of the possible inadequacies of Hill’s approxi- mation to vibrational entropies, this could well be the “ real ” solution. Thus, even using the more exacting criterion of fitting the isotherms within experimental error, we have been unable to obtain a unique analysis in terms of proportions of the surface which are specific crystal planes with particular thermo- dynamic parameters describing the adsorption characteristics of each.7. FUTURE OF THE PATCH MODEL The failure we report above may be due in part to the size of error which is accept- able, and increased experimental accuracy alone might permit such an analysis to be established. We suggest, however, that a further independent source of information as to the number and relative proportions of patches in the surface will be necessary reliably to establish such a unique solution. For example, in the present procedure, the number of patches is increased until the minimum number required to approach a fit is established. If this number is incorrect, then the other parameters can clearly have no significance, and the fitting is only a mathematical exercise.In this respect, it is probable that a patchwise analysis could not be substantiated for a non-crystalline heterogeneous surface, where an indefinite increase in the number of patches would be permissible, but in the case of a surface exposing only a small number of crystal planes, the model is still an attractive one. Recent work by Baker, Johnson and Maire suggests that in favourable circumstances, information of the necessary type can be extracted from initial photoelectric work function measurements on the experi- mental surface. Baker and Head 4a have previously pointed out that the success of their procedure for fitting by a patch model does depend on the starting values for the parameters of the patches. This remains as a fundamental flaw, which originates at least in part in the search procedure.Changes in the parameters are tested for improvement of fit in a rigid order. If the change tried results in an improved fit, this change is accepted. However, a much greater improvement might have resulted from a change in another parameter of the starting set, and in making prior changes this could become ob- scured and the solution might never move in that initially “ obvious ” direction. The ideal procedure would be to calculate the matrix of percentage improvement of fit on changing each parameter by an equivalent amount, and then to accept that change giving maximum improvement. However, the increase in computing time would be182 ANALYSIS OF ADSORPTION DATA unacceptable to most users. As a possible alternative, we suggest random selection of the parameter to be tested next.Such a change in search procedure, together with the incorporation of independent information on the number and proportions of patches present in a particular experimental situation, might allow a unique patch model to be established. If this could be done in one contrived experiment to estab- lish values for iqst, Ai and Ci on particular planes, subsequently the proportion of patches present could be computed for other less artificial experimental results. In the present situation, there is still too little information to fix any of the parameters, and a unique solution cannot be established. The authors thank B. G. Baker, A. K. Head and J. F. Nicholas for helpful discus- sion.S. Ross and J. P. Olivier, On Physical Adsorption (Interscience, New York, 1964), p. 165. A. W. Adamson, The Physical Chemistry of Surfaces (Interscience, London, 1967), p. 625. J. P. Hobson, Canad. J. Phys., 1965, 4, 1934. B. G. Baker, L. A. Bruce and P. G. Fox, Trans. Faraday Soc., 1968,64,477. (a) Appendix by B. G. Baker and A. K. Head, p. 485. B. G. Baker and P. G. Fox, Trans. Faraday SOC., 1965,61,2001. B. G. Baker and L. A. Bruce, Tram. Faraday Soc., 1968,64,2533. ' L. A. Bruce and M. H. Sheridan, J.C.S. Faraday I, 1972,68,997. L. A. Garden and G. L. Kington, Proc. Roy. SOC. A, 1956,234,24. T. L. Hill, J. Chem. Phys., 1948,16,181. Yun-Yang Huang, J. Catalysis, 1972,25, 13 1. l1 B. G. Baker, personal communication. l2 R. J. Bacigalupi and H. E.Neustadter, Surface Sci., 1970, 19, 396. l 3 B. G. Baker, B. B. Johnson and G. L. C. Maire, Surface Sci., 1971, 24,572.Deficiency in the Analysis of Physical Adsorption DataBY LINDA A. BRUCE AND MARGARET H. SHERIDANC.S.I.R.O. Division of Tribophysics, University of Melbourne,Parkville, Victoria 3052, AustraliaReceived 24th July, 1972The criterion of fit to adsorption data advocated by Baker and Head is shown to be too lax, andlikely to lead to spurious analysis. Computer-generated parameters fitting the derived heats andentropies of adsorption yield isotherms in disagreement with the experimental isotherms from whichthese derived quantities were evaluated. Reasons for this discrepancy are considered, and animproved criterion of fit is suggested.Impediments to establishing a unique description of a surfaceby this type of analysis are examined.1. INTRODUCTIONSeveral procedures have been described 1-3 for analysis of adsorption isotherms onthe basis that the surface studied displays a continuous range of adsorption energies.Such treatments assume that the entropy of vibration retained by the adsorbed speciesis constant over the entire surface and hence inherently require that the entropy ofvibration be independent of the energy of adsorption. Secondly, there is no possibilityof changing the " local isotherm " equation with changing energy of adsorption, sothat adsorption over the entire surface must be described either by a localised isothermor by a non-localised isotherm. Where the computed energy distribution has anyfinite width, neither of these requirements is necessarily justified.An alternativemethod 4a for analysis of adsorption isotherms assumes that the surface studieddisplays only a small finite number of " patches " of different adsorption energies.Adsorption on any patch may be either localised or non-localised, and the entropy ofvibration of the adsorbate may also vary from patch to patch. Although the authorsfeel that the patch model is a useful and realistic one, the purpose of the present paperis to show that the method of analysis, as it was originally applied, 4-6 provided aninadequate description of the experimental surface.2. BASIC PROCEDUREThere is no analytical solution to the competitive filling of the constituent patchesof such a model, and a computer programme was devised 4a to evaluate the fillingfor a given set of patches.Each patch, i, was characterized by four parameters : (1)the relative proportion, ni, of the patch occurring in the surface, (2) the initial heat ofadsorption, iqst, (3) an interaction constant, A i , allowing the heat of adsorption toincrease linearly with coverage, and (4) an entropy constant, Ci, where for a patch onwhich adsorption is localisediASdiff. = iS,-S: = -Rln(0~/(1-8J)-Ciwhile if adsorption is non-localiseL. A. BRUCE AND M. H. SHERIDAN 177An " educated guess " was used in assigning initial numerical values to these para-meters for the chosen number of patches. At selected values of the total coverage,the computer programme evaluated the filling of the individual patches determined bythe condition that the adsorbed phase on all patches must have the same chemicalpotential; the isosteric heat and differential entropy of adsorption on the modelsurface werc theq returned as the weighted averages of these qwntitieg on eachindividual patch at its compvted coverage, and comparison was made with the experi-mental values on the real surface at the same total coverage.After changing thefirst parameter of one of the proposed patches, the computations were repeated, andautomated decision of improvement of fit to experiment resulted in acceptance orrejection of the change. The programme cycled, testing each parameter in turn untilno significant improvement could be made using this particular set of patches. Theexperimental values for the heat and entropy were evaluated by thermodynamicanalysis of the set of experimental isotherms using the integrated form of the Clausius-Clapeyron equation, In Po = - (qst/RT) + K, where K is the integration constant.Treating this equation as linear in In Po and ( l / T ) , least-squares analysis gives theisosteric heat, qst, and a standard Ceviation of its estimate; the standard differentialentropy is given by &&iff. = s, - S i = R In po - RK, also with a calculated standarddeviation (s, is the differential entropy of the adsorbed phase and is the molarentropy of the gas phase at pressure p").A model was accepted as a successful fit ifthe sum of fourth powers of the differences between experimental and model heats andentropies came within the sum of fourth powers of the standard deviations on theexperimental values derived from such an analysis.Thus, the original concept ofBaker and Head 4a involved the fact that the information in a set of isotherms issummarized in the values of isosteric heat and differential entropy obtained by thermo-dynamic analysis. Their criterion of fit seems very plausible, but will now be shownto be inadequate.3 . THE DISCREPANCYThe above procedure was applied to the measurements of adsorption of xenon onfilms of gold, aluminium and silver deposited in various ways in order that the metalexposed different proportions of major crystallographic planes.' The aim was toidentify the parameters relevant to adsorption on individual crystal planes, intuitivelyidentified with the patches of the model.It was disturbing to discover that severalquite different but equally acceptable computer-generated " solutions " for any oneset of experimental data could not be distiaguished by the criterion of fit, even whenthe number of patches was restricted to two. Such conflicting " solutions " arosefor each of the metals, and fig. l(a) gives an example of the fit to the derived experi-mental data, qsr and ASdiff., for xenon on gold/Pyrex, for one of the '' solutions ". Inorder to check the accuracy of fit to the primary data, isotherms were calculated fromthese computed values of heat and entropy, fromThese computed isotherms, fig.l(b), reveal that the fit to the primary data is very poor.A similar situation arose when isotherms were computed from other solutions" acceptable " on the basis of fitting qst and ASdiff., including those already pub-l i ~ h e d . ~ - ~ For example, fig. 2 shows a comparison between the experimental isothermsmeasured for xenon on nickel/Pyrex and isotherms constructed from the publishedanalysis of the heats and entropies according to a patch model,6 using eqn (3.1).Again, the accepted fit to the deriuedexperimntal data, qst and ASdiff. (fig. 6, ref. (6)),is not at all a good fit to the primary experimental data, the set of isothermsANALYSIS OF ADSORPTION DATA24 22 -60-80pressure/N nr2(b)FIG. 1.-(u) Computed “ fit ” (- - -) to derived experimental data ( x ) for xenon on gold/Pyrex.’(b) Isotherms calculated from the “ fit ” shown in (a) (- - -), compared with three of the originalexperimental isotherms (-) from which the derived data were evaluated.II I 4 I I I I l I l l I I - .I o+ lo5 I o4 to -pi*essure/N rn-’FIG. 2.-Xenon on nickel/Pyrex : experimental isotherms (-) and isotherms computed using“ best-fit ” parameters (- - -).4.ORIGIN OF THE DISCREPANCYThis seeming inconsistency arises from the way in which errors in pressure areFrom eqn (3. l), we find that the error Ap in p determined by errors in qst and ASdiff..arising from errors A(qst) and A(Asdiff.) is given byp+Ap = exp([-(%,$-A(%,))/RT1-[(ASdiff. +A(Asdiff.>)/Rl +InPo) (4*1)i.e.,This shows that, for example, an increase either of 4 J K-I niol-l in ASdiff.or of 0.4 kJmol-l in qst at 84 K (mid-range of the experimental temperatures) would give ( p + Ap)/~ ~ 0 . 6 , while decreases of the same magnitude would give (p+Ap)/p= 1.7, yet thevalues used to calculate such relative errors in p are typical of the standard deviationson qst and ASdlff, evaluated from experimental isosteres in which the experimental(P+AP)/P = exp{ -rA(%t)/~Tl- [A(A~dlff.)/Rll. (4.2L. A. BRUCE AND M . H . SHERIDAN 179pressures are certainly not more than 20 % from a straight i.e. 0.8 < {(p + Ap)/p>,,,t. < 1.2. The vital point is that the experimental estimates of qst and ASdiff arenot mutually independent. In the above example, such a change in the estimate ofqst or of ASdiff, would not arise in isolation, but would be counterbalanced by achange of opposing sign in the other quantity.However, the procedure of fitting theheat and entropy data, discussed above, involves summing the fourth powers of thedifferences between experimental and model values, and ignores any effect of the signof these differences. Thus the fitting procedure takes no account of the inversecorrelation that must exist between the variations in qst and ASdiff. if isotherms are tobe deduced in reasonable agreement with experiment. Close examination of fig. l(a)shows that in this particular " fit ", the experimental values of qst and ASdiff. at mostcoverages differ in the same sense from the computed values, leading to the discrepancyshown in fig.l(b). The discrepant fit shown in fig. 2 is also accounted for in theseterms.The analyses of adsorption results on several surface^,^-^ based on this criterion offit, must be taken as spurious.5. MODIFICATIONS TO THE FITTING PROCEDUREEvidently the criterion of fitting the heat and entropy values is much too lax as itstands. While acceptance of non-compensating differences between the computedand experimental values of the heat and entropy of adsorption could simply be penal-ised in the computer programme, we decided that in any future use of the model itwould be more logical to fit the primary data, the set of isotherms. The experimentalerror (say +20 %) in measuring a pressure could then be used directly as a standardof acceptability of fit by summing the fourth powers of the relative errors between thecomputed and experimental values of the pressure.(In this case the acceptedcomputed value of pressure may of course be either greater or less than the experi-mental one.)We further decided that close attention be given to assessing the numerical valuesof parameters on grounds other than those of merely fitting the isotherms. Forexample, the value of Ci, the entropy constant, is easily shown to depend on the type ofadsorption, i.e. localised or non-localised, on patch i;' C, is also related to theisosteric heat of adsorption iqst through the entropy of vibration@) of the ad~orbate.~A * , the interaction constant, also varies with the type of adsorption and with postu-lated arrangements of adsorbate atoms on particular crystal planes (intuitively thepatches) of an adsorbent; acceptable values for Al should be justifiable in terms ofLennard-Jones summations of the interactions between adsorbate atoms spaced apartat definite distances.While none of the estimates of these parameters can be absolute,quite narrow limits can be placed on reasonable values for a particular gas/metalsystem, and any values returned by the computer which are far outside these limitsshould be regarded with scepticism. However, it is important to appreciate thatsince a change in Cl of 4 J K-l mol-1 or in iqst of 0.4 kcal mol-1 alters the calculatedpressure (eqn (2.1) or (2.2) and (4.2)) at constant coverage on a patch by up to 70 %,even seemingly small variations in the allowed relative values of Ci and iqst can resultin substantial changes in the competitive filling of the set of patches under consider-ation.Computation of the heat and entropy of adsorption at one temperature, followedby the use of eqn (3.1) to evaluate isotherms at other temperatures, has also been foundlikely to lead to errors in fitting.Since isotherms generated by the parameters of eachindividual patch vary differently with temperature, the patch coverages at which th180 ANALYSIS OF ADSORPTlON DATAequal chemical potential conditions are met are a function of temperature, and so alsois the rate of filling of the patches, dQt/dO, which determines the computed isostericheat and differential entropy for the In the revised fitting procedure, thecompetitive filling of the set of patches under test is calculated at each temperature forwhich an experimental isotherm is to be fitted.To ensure that these computationsare made as accurately as possible, the initial heat of adsorption, iqst, on each patch isadjusted with temperature (by RAT) and the entropy constant Ci is also adjusted withtemperature for the changing entropies of translation and vibration contributing to itsvalue. The overall result is of slightly different computed heats and entropies ofadsorption at each temperature, which, using eqn (3. l), give pressures varying gen-erally by some 10-20 % from those calculated assuming temperature invariance; inparticular circumstances, this figure could be much greater.The precise effect isunique to a set of patches, and the extent cannot be calculated analytically. Huang lohas recently discussed essentially this effect in a very restricted case.6 . APPLICATION TO EXPERIMENTAL RESULTSUsing the fully modified procedure, attempts have been made to establish a patchmodel describing the adsorption of xenon on aluminiuni/Pyrex,' by matching com-puted isotherms to a set of three experimental ones. The entropy of adsorption onthe (100) and (1 11) planes, believed to predominate in the surface of the film, wasdescribed either by eqn (2.2) for non-localised adsorption, or by polynomialsestablished by Baker l1 to describe localised adsorption with blocking of sites when alarge atom is adsorbed on these planes.Since higher index planes might be repre-sented in the surface to a small extent, eqn (2.1) for localised adsorption was used todescribe adsorption on their more widely separated sites. The only calculations fullyrelevant to the acceptability of particular entropy functions are those of Bacigalupiand Neustadter.12 Their results indicate that in the case of xenon adsorption onaluminium, the activation barrier to diffusion on a (1 11) plane would be less than RTfor adsorption energies below 23 kJ mol-l in the experimental range of temperatures,suggesting that xenon would be mobile under such circumstances. On the otherhand, on a (100) plane, even at an adsorption energy as low as 14 kJ mol-l, theactivation energy required for diffusion would be prohibitively high, and xenon wouldbe localised. Nevertheless, trials using numerous sets of starting parameters havebeen run, involving all possible combinations of these entropy functions.Theresults returned by the computer as the best-fit values (quoted for 84 K) on three ofthese attempts to fit the data are listed in table 1." solution "II1I11TABLE 1 .-" BEST-FIT " PARAMETERS RETURNED BY COMPUTERfraction ofpatch total surface 1Ystl At1 ci Ii ni (kJ mol-1) (kJ mol-1) (J K - 1 mol-1) cntropy function1 0.670 15.14 4.26 68.9 (2.2)2 0.305 16.86 2.91 79.6 (100) polynomial3 0.025 28.84 0 87.2 (2.1)1 0.339 15.29 4.04 69.1 (2.2)2 0.645 16.74 2.18 80.4 (100) polynomial3 0.026 27.93 0 87.4 (2.1)I 0.481 15.86 4.12 75.4 (2.2)2 0.467 17.15 2.21 85.4 (100) polynomial3 0.05 1 27.92 0 92.4 (2.1L .A. BRUCE AND M. H . SHERIDAN 181In the set of three isotherms fitted, no computed pressure differed froin thecorresponding experimental value by more than 25 % in the case of solution I, and15 % in the case of solution 11. For these two solutions, the values returned for thethermodynamic parameters identified with (1 1 l), (100) and higher index planes(patches 1, 2 and 3 respectively) are very similar and fulfil the criteria suggested insection 5. However, the values returned as the fraction of the total surface repre-sented by patches 1 and 2 actually inverts between solutions I and 11, in the one caserequiring 67 % of the surface to be (1 11) planes and 30.5 % (100) planes, whereas inthe other case only 34 % is (111) planes but 64.5 % is (100).(In each instance, onlythe very minor residue is of high index planes.) Obviously, both solutions cannot be“ real ”. Since no preferred orientation was observed for this film by electron micro-scopy, there is no way of absolutely rejecting one of them. While the second solutionis a better fit with experiment, there remains the suspicion that a still better fit couldbe found in which the thermodynamic parameters are still very similar, but the relativeproportions of major planes differ again. In fact, due to a mistake in the control ofvalues for Ci, solution I11 was obtained in which there was only 10 % relative error infitting these three isotherms.Because of the possible inadequacies of Hill’s approxi-mation to vibrational entropies, this could well be the “ real ” solution.Thus, even using the more exacting criterion of fitting the isotherms withinexperimental error, we have been unable to obtain a unique analysis in terms ofproportions of the surface which are specific crystal planes with particular thermo-dynamic parameters describing the adsorption characteristics of each.7. FUTURE OF THE PATCH MODELThe failure we report above may be due in part to the size of error which is accept-able, and increased experimental accuracy alone might permit such an analysis to beestablished. We suggest, however, that a further independent source of informationas to the number and relative proportions of patches in the surface will be necessaryreliably to establish such a unique solution.For example, in the present procedure,the number of patches is increased until the minimum number required to approacha fit is established. If this number is incorrect, then the other parameters can clearlyhave no significance, and the fitting is only a mathematical exercise. In this respect,it is probable that a patchwise analysis could not be substantiated for a non-crystallineheterogeneous surface, where an indefinite increase in the number of patches would bepermissible, but in the case of a surface exposing only a small number of crystalplanes, the model is still an attractive one. Recent work by Baker, Johnson andMaire suggests that in favourable circumstances, information of the necessary typecan be extracted from initial photoelectric work function measurements on the experi-mental surface.Baker and Head 4a have previously pointed out that the success of their procedurefor fitting by a patch model does depend on the starting values for the parameters ofthe patches.This remains as a fundamental flaw, which originates at least in part inthe search procedure. Changes in the parameters are tested for improvement of fitin a rigid order. If the change tried results in an improved fit, this change is accepted.However, a much greater improvement might have resulted from a change in anotherparameter of the starting set, and in making prior changes this could become ob-scured and the solution might never move in that initially “ obvious ” direction. Theideal procedure would be to calculate the matrix of percentage improvement of fit onchanging each parameter by an equivalent amount, and then to accept that changegiving maximum improvement. However, the increase in computing time would b182 ANALYSIS OF ADSORPTION DATAunacceptable to most users. As a possible alternative, we suggest random selectionof the parameter to be tested next. Such a change in search procedure, together withthe incorporation of independent information on the number and proportions ofpatches present in a particular experimental situation, might allow a unique patchmodel to be established. If this could be done in one contrived experiment to estab-lish values for iqst, Ai and Ci on particular planes, subsequently the proportion ofpatches present could be computed for other less artificial experimental results. Inthe present situation, there is still too little information to fix any of the parameters,and a unique solution cannot be established.The authors thank B. G. Baker, A. K. Head and J. F. Nicholas for helpful discus-sion.S. Ross and J. P. Olivier, On Physical Adsorption (Interscience, New York, 1964), p. 165.A. W. Adamson, The Physical Chemistry of Surfaces (Interscience, London, 1967), p. 625.J. P. Hobson, Canad. J. Phys., 1965, 4, 1934.B. G. Baker, L. A. Bruce and P. G. Fox, Trans. Faraday Soc., 1968,64,477. (a) Appendix byB. G. Baker and A. K. Head, p. 485.B. G. Baker and P. G. Fox, Trans. Faraday SOC., 1965,61,2001.B. G. Baker and L. A. Bruce, Tram. Faraday Soc., 1968,64,2533. ' L. A. Bruce and M. H. Sheridan, J.C.S. Faraday I, 1972,68,997.L. A. Garden and G. L. Kington, Proc. Roy. SOC. A, 1956,234,24.T. L. Hill, J. Chem. Phys., 1948,16,181.Yun-Yang Huang, J. Catalysis, 1972,25, 13 1.l1 B. G. Baker, personal communication.l2 R. J. Bacigalupi and H. E. Neustadter, Surface Sci., 1970, 19, 396.l 3 B. G. Baker, B. B. Johnson and G. L. C. Maire, Surface Sci., 1971, 24,572

ISSN:0300-9599    

DOI:10.1039/F19736900176

出版商:RSC    

年代:1973

数据来源: RSC