The salt effect on the yields of trioxane in reaction solution and in distillate

Abstract

Batch reaction experiments were performed to investigate the salt effect on the yield of trioxane in the reaction solution. The salts considered include NaHSO₄, Na₂SO₄, NaH₂PO₄, Na₂HPO₄, KCl, NaCl, LiCl, ZnCl₂, MgCl₂, and FeCl₃. The effects of the anionic structure and the cation charge density on the yield of trioxane in the reaction solution were elucidated and the mechanisms that govern such effects were established. It is shown that the first four salts exerted a negative effect on the yield of trioxane in the reaction solution and such an effect increased progressively from left to right. This trend is due to the formation of NaHSO₄, H₃PO₄, or (H₃PO₄ and NaH₂PO₄), which decreased the concentration of H⁺ in the solution. The latter six salts showed a positive effect on the yield of trioxane in the reaction solution. The salt effect paralleled the ability of the salt to decrease the water activity of the reaction solution and followed the order KCl < NaCl < LiCl < ZnCl₂ < MgCl₂ < FeCl₃. Continuous production experiments were performed to investigate the salt effect on the concentration of trioxane in the distillate. The salts considered were KCl, NaCl, LiCl, ZnCl₂, MgCl₂, and FeCl₃, and the salt effect increased progressively from left to right. Such an effect was shown to be determined by the ability of the salt to increase the yield of trioxane in the reaction solution and to increase the relative volatilities of trioxane and water and of trioxane and oligomers.

---

1 Introduction

In the last few decades, trioxane, the cyclic trimer of formaldehyde, has been studied with interest as a starting material for the preparation of acetal resins, anhydrous formaldehyde, pesticides, moulding materials, bonding materials, disinfectant agents, and antibacterial agents.¹ Due to their superior chemical stability, mechanical strength, and plasticity, acetal resins have found broad applications in areas where traditional metals were applied.^2,3 Accordingly, a rapid expansion of acetal resin production has occurred worldwide.^4,5 Such expansion requires a more practical, more efficient, and more economic route for the production of trioxane.

The usual method for the production of trioxane consists of heating aqueous formaldehyde in the presence of an acid catalyst such as sulfuric acid, acidic ion exchange resin, heteropolyacid, and acidic ionic liquid.^1,4,5 Using either acidic ion exchange resin or heteropolyacids as the catalyst yields relatively good results. However, these methods require a high concentration of formaldehyde, a large amount of catalyst or high reaction conditions (e.g., temperature, pressure and material purity).^6,7 Recently, acidic ionic liquids (ILs) such as imidazole- and pyridine-based ILs have been used as catalysts in the preparation of trioxane.^8,9 In addition, an IL catalyst that was successfully used in a pilot plant trial is less corrosive to instruments. However, use of this IL catalyst instead of sulfuric acid lowers the yield of products. Furthermore, the price of this IL catalyst is as high as US $128 000 per ton.¹⁰ Therefore, to date, sulfuric acid is still the most generally used catalyst, as the corresponding processing route is mature and the price of sulfuric acid is low. However, the disadvantage of using sulfuric acid as a catalyst for the synthesis of trioxane is that by-products such as formic acid, methyl formate, and the like are extensively formed. In particular, when the concentration of sulfuric acid exceeds 8 wt% (weight percent), the formation of large amount of by-products will decrease the yield of trioxane.¹¹

Fortunately, a positive salt effect has been found in the palladium(II) catalyzed isomerization of alkylidene cyclopropyl ketones.¹² The salt effect has also been identified in the reaction of the in-mediated allylation of N-tert-butanesulfinyl imines.¹³ More recently, the remarkable role of halide salt additives in the Negishi reaction involving aryl zinc reagents has been reported.¹⁴ Therefore, the present efforts explored the effect of salts on the formation of trioxane. The structure–activity relationship for the salt effect on the synthesis of trioxane was elucidated for the first time. In addition, the batch reaction and continuous production experiments were performed simultaneously for the first time to uncover new mechanisms that govern the salt effect on the yield of trioxane in the reaction solution and in the distillate. The results thus obtained are important for us to use the (sulfuric acid–salt) system in place of sulfuric acid as a catalyst for trioxane synthesis to decrease the concentration of sulfuric acid and to achieve greater yields of trioxane in the reaction solution and in the distillate.

2 Experimental

2.1 Materials and reagents

50 wt% aqueous formaldehyde solution was prepared by concentrating ∼37 wt% aqueous formaldehyde solution (Sinopharm Chemical Reagent Co., Ltd) using vacuum distillation at 343.15 K. Analytical grade inorganic salts and sulfuric acid were supplied by Aladdin Industrial Corporation (Shanghai, China) and they were used without further purification.

2.2 Batch reaction

A known amount of solution containing 50 wt% formaldehyde and an inorganic salt was added to a 250 mL spherical reactor equipped with a sampling device. A Pt/100 temperature sensor monitored by a CSOOB digital thermometer was inserted into the reactor to control the temperature with an uncertainty of ±0.1 K. The reactor was pressurized to 1.0 MPa with nitrogen gas via a conduit system to prevent the solution from boiling. Then, the reactor was heated in an electric heating jacket with continuous stirring. When the temperature reached the pre-set value (∼371.15 K), a given amount of sulfuric acid was introduced into the reaction solution to initiate the reaction and the reaction time was recorded. The cyclotrimerization reactions were typically allowed to proceed for 30 min. During the reaction, a known amount of reaction solution was sampled and analyzed at a certain interval (e.g., 3 min.) to monitor the change with reaction time in the composition of the reaction solution. At the same time, another known amount of reaction solution was injected into a conical flask to determine the acid value of the reaction solution by acid–base titration using a potentiometric titrimeter (Leici ZDJ-5).

2.3 Continuous production

For the continuous production of trioxane, a 250 mL glass reactor with a formaldehyde solution feeder and a packed tower was used. The packed tower was filled by stainless steel raschig rings. A reflux condenser, which was added to the top cover of the packed tower, was heated with warm water at approximately 323.15 K so that the vapor phase condensed on the inner face of reflux condenser without the precipitation of paraformaldehyde. A Pt/100 temperature sensor monitored by a CSOOB digital thermometer was inserted into the reactor to control the temperature of reaction with an uncertainty of ±0.1 K. The reaction mixture (100 mL) comprising 50 wt% formaldehyde solution, the acid catalyst, and the inorganic salt was heated in an electric heating jacket. The reaction time was recorded when the temperature reached ∼375.15 K. After 1 h, the 50 wt% formaldehyde solution (100 mL) was continuously fed to the reactor from the feeder, and the feed rate of the formaldehyde solution was adjusted to ensure a constant level of the reaction solution. A reflux ratio of 2 : 1 was maintained. When the formaldehyde solution feed was finished, the distillate was analyzed by gas chromatography with a thermal conductivity detector (TCD).

3 Typical reactions

Aqueous solutions of formaldehyde are inherently reactive complex multicomponent mixtures in which formaldehyde is predominately bound in different oligomers:¹⁵

CH₂O + H₂O ⇌ HOCH₂OH (1)

HO(CH₂O)_n−1H + HOCH₂OH ⇌ HO(CH₂O)_nH + H₂O (2)

The most probable mechanism by which trioxane is formed can be described as:¹⁵

However, information on the true species distribution is not available under different reaction conditions.¹⁵ HO(CH₂O)₃H is usually not individually observable in experiments that involve a high concentration of acid.¹⁵ Accordingly, it must use the calculated concentration of HO(CH₂O)₃H. However, models that can calculate the concentration of HO(CH₂O)₃H in the reaction solution (formaldehyde–H₂SO₄) are not available to our knowledge. Therefore, the formation of trioxane is usually described in the literature as:^15–17

and quantitative results are presented using overall concentrations (neglecting the oligomer formation).^15,16,18

4 Results

Fig. 1 compares the results of the two independent batch experiments performed at 371.15 K.

Fig. 1 The results of the two independent batch reactions in [(50 wt%) formaldehyde–(0.4 mol L⁻¹) H₂SO₄] performed at T = 371.15 K.

It is clear that the reproducibility of the measurements is highly satisfactory. Therefore, a series of batch experiments was carried out to explore the effects of salts on the yield of trioxane and on the formation of by-product(s) in the reaction solution. The results thus obtained by adding various sodium salts are shown in Fig. 2 and 3.

Fig. 2 The effect of the addition of a sodium salt on the yield of trioxane in the reaction solution [(50 wt%) formaldehyde–(0.4 mol L⁻¹) H₂SO₄–(1 mol L⁻¹) salt]. (): Na₂HPO₄; (): NaH₂PO₄; (): Na₂SO₄; (): NaHSO₄; (): NaCl. For the sake of comparison, the results obtained for the reaction solution [(50 wt%) formaldehyde–(0.4 mol L⁻¹) H₂SO₄] are also shown and are symbolized by ().

Fig. 3 The effect of the addition of a sodium salt on the acid strength of the reaction solution [(50 wt%) formaldehyde–(0.4 mol L⁻¹) H₂SO₄–(1 mol L⁻¹) salt]. (): Na₂HPO₄; (): NaH₂PO₄; (): Na₂SO₄; (): NaHSO₄; (): NaCl. For the sake of comparison, the results obtained for the reaction solution [(50 wt%) formaldehyde–(0.4 mol L⁻¹) H₂SO₄] are also shown and are symbolized by ().

Fig. 2 shows that, in comparison with the results obtained for (formaldehyde–H₂SO₄), the addition of Na₂SO₄, NaH₂PO₄, or Na₂HPO₄ considerably decreased the concentration of trioxane in the reaction solution (formaldehyde–H₂SO₄–salt) over the entire experimental reaction time. The effect of NaHSO₄ on the yield of trioxane in the reaction solution was also negative. However, this effect was considerably smaller than that exerted by the above-mentioned salts. On the contrary, the presence of NaCl increased the concentration of trioxane in the reaction solution over the entire experimental reaction time. The order observed in Fig. 3 for the increase in acid strength of the reaction solution with the reaction time is Na₂SO₄ < NaH₂PO₄ < Na₂HPO₄.

The experiments were made to uncover the mechanisms that govern the effect of the addition of Cl⁻-based salts on the formation of trioxane in the reaction solution. The results are shown in Fig. 4 and show several features of interest. The concentration of trioxane in the reaction solution increased in the order KCl < NaCl < LiCl < ZnCl₂ < MgCl₂ < FeCl₃, which is exactly the order for the decreasing radius of the constituent cations of the salts considered. However, the mechanism that fundamentally controls the effect of the Cl⁻-based salts considered on the yield of trioxane in the reaction solution is still poorly understood to date. Note that the yield of trioxane in [(50 wt%) formaldehyde–(0.4 mol L⁻¹) H₂SO₄–(1 mol L⁻¹) LiCl/ZnCl₂/MgCl₂/FeCl₃] is greater than the yield in [(50 wt%) formaldehyde–(1.4 mol L⁻¹) H₂SO₄]. The effect of these salts on the change in the acid strength of the reaction solution with reaction time was small {see Fig. 3 for the rapidity of the change with the reaction time in the acid strength of the reaction solution [(50 wt%) formaldehyde–(0.4 mol L⁻¹) H₂SO₄]}. Therefore, the corresponding results are not shown.

Fig. 4 The effect of the addition of a Cl⁻-based salt on the yield of trioxane in the reaction solution [(50 wt%) formaldehyde–(0.4 mol L⁻¹) H₂SO₄–(1 mol L⁻¹) salt]. (): KCl; (): NaCl; (): LiCl; (): ZnCl₂; (): MgCl₂; (): FeCl₃. For the sake of comparison, the results obtained for the reaction solution [(50 wt%) formaldehyde–(1.4 mol L⁻¹) H₂SO₄] are also shown and are symbolized by ().

The values of the rate constants (k) were determined by fitting the experimental data for the concentration of trioxane in the reaction solution (c_p) as a function of reaction time to eqn (5):¹⁹

dc_p/dt = k₁c_A³ − k₂c_p (5)

where c_A is the concentration of formaldehyde. k₁ and k₂ are the rate constants for the forward reaction and the reverse reaction, respectively. Therefore, the influence of the Cl⁻-based salts considered on the formation of trioxane is quantitatively represented by the ratio k_salt/k, where k_salt and k are the rate constants for the salt-containing solution and for the salt free solution, respectively. The results are shown in Table 1.

Table 1 The influence of the salts considered on the sulfuric acid catalyzed formation of trioxane at ∼371.15 K

Salt (1 mol L^−1)	Cationic radius (pm)	k1,salt/k1	k2,salt/k2
KCl	138	1.08	1.04
NaCl	102	1.22	1.16
LiCl	76	1.50	1.33
ZnCl2	74	1.59	1.37
MgCl2	72	1.70	1.43
FeCl3	65	2.24	1.83

---

The results of the continuous production experiments are summarized in Table 2. It is clear that the addition of the salts considered simultaneously increased the reaction conversion and time/space yield. In addition, such positive effects increase in the order KCl < NaCl < LiCl < ZnCl₂ < MgCl₂ < FeCl₃. Notably, it can be deduced from Table 2 that the reaction conversion and time/space yields in [(50 wt%) formaldehyde–(0.4 mol L⁻¹) H₂SO₄–(1 mol L⁻¹) LiCl/ZnCl₂/MgCl₂/FeCl₃] were greater than in [(50 wt%) formaldehyde–(1.4 mol L⁻¹) H₂SO₄]. This suggests that the salt effects on the reaction conversion and time/space yield cannot be understood solely by the Lewis acid strength of the salt added.

Table 2 Effect of the added salt on the conversion and time/space yield for trioxane production

Acid and salt	Conversion^a (%)	Time/space yield^b (g h^−1 L^−1)
a The percent of trioxane converted.b The grams of trioxane formed within one hour in one liter of solution.
0.4 mol L^−1 H2SO4	18.07	47.51
0.4 mol L^−1 H2SO4 + 1 mol L^−1 KCl	19.08	49.42
0.4 mol L^−1 H2SO4 + 1 mol L^−1 NaCl	22.07	55.17
0.4 mol L^−1 H2SO4 + 1 mol L^−1 LiCl	27.20	63.28
0.4 mol L^−1 H2SO4 + 1 mol L^−1 ZnCl2	28.79	64.56
0.4 mol L^−1 H2SO4 + 1 mol L^−1 MgCl2	30.63	67.89
0.4 mol L^−1 H2SO4 + 1 mol L^−1 FeCl3	38.47	86.89
1 mol L^−1 H2SO4	23.06	58.20

---

5 Discussion

The mechanisms that govern the effect of the sodium salts considered on the formation of trioxane in the reaction solutions are very complex. The value of the acid dissociation constant K^θ_a decreases in the order H₂SO₄ (∼10²) > HSO₄⁻ (1.0 × 10⁻²) > H₃PO₄ (6.7 × 10⁻³) > H₂PO₄⁻ (6.2 × 10⁻⁸).²⁰ Accordingly, the addition of Na₂SO₄, NaH₂PO₄, or Na₂HPO₄ to the reaction solution (formaldehyde–H₂SO₄) must have induced the formation of NaHSO₄, H₃PO₄, or (H₃PO₄ and NaH₂PO₄), respectively. Such formation inevitably decreases the concentration of H⁺ that serves as the catalyst for trioxane synthesis. On the other hand, NaH₂PO₄ and Na₂HPO₄ are kosmotropic salts (water-structuring salts).²¹ The addition of these salts to the reaction solution may induce the formation of an oligomer-rich phase and a salt-rich phase (both of which are aqueous).²¹ Such formation will increase the true concentrations of the oligomers including the reactant. Therefore, their effect on the yield of trioxane in the reaction solution will be a result of the competition between these two factors. Fig. 2 shows that the addition of Na₂SO₄, NaH₂PO₄, or Na₂HPO₄ significantly decreases the yield of trioxane in the reaction solution and that the negative effect increases with decreasing K^θ_a value of the corresponding acid in the order Na₂SO₄ < NaH₂PO₄ < Na₂HPO₄. These results indicate that the effect of decreased concentration of H⁺ in the reaction solution overrides the water-structuring effect.

The presence of Cl⁻-based salts in the reaction solution may influence the yield of trioxane in several ways. First, according to the model proposed by Debye et al.²² and Gross et al.,²³ the molecules of water tend to congregate in the vicinity of the salt ions, in effect forcing the reactant molecules into the portions of solution remote from the ion fields and therefore raising the true concentration of the reactant in the latter regions. Second, the addition of a salt to the reaction solution decreases water activity of the solution and therefore favors the forward reaction of eqn (3). The water activity of the binary solution (salt–H₂O) with the molality of the salt being 1 mol kg⁻¹ at 298.15 K is 0.9682, 0.9669, 0.9640, 0.9572, 0.9419, and 0.9268 for KCl, NaCl, LiCl, ZnCl₂, MgCl₂, and FeCl₃, respectively.²⁴ This indicates that the ability of these salts to decrease the water activity of the reaction solution (formaldehyde–H₂SO₄–salt) increases in the order KCl < NaCl < LiCl < ZnCl₂ < MgCl₂ < FeCl₃,²⁵ which is exactly the order for the value of (k_1,salt/k₁)/(k_2,salt/k₂) (= 1.04, 1.05, 1.13, 1.16, 1.19, 1.22 from left to right, see Table 1) and the order for the effect of these salts on the yield of trioxane in the reaction solution (Fig. 4). In other words, the effect of these salts on the yield of trioxane relies primarily on their ability to decrease the water activity of the reaction solution, as is demonstrated in Fig. 5, which shows that the rapid increase in the value of (k_1,salt/k₁)/(k_2,salt/k₂) corresponds to the rapid decrease in water activity (i.e., the rapid increase in the ability of the salt to decrease the water activity of the reaction solution). The Lewis acid strength of the cations of the salts considered has been characterized by a D value related to the diagonal elements and an O value related to the off diagonal elements of the secular determinant.²⁶ The values of D and O increase in the order K⁺ < Na⁺ < Li⁺ < Mg²⁺ < Zn²⁺.²⁶ This order means that the effect of the present salts on the yield of trioxane in the reaction solution is not dominated by the Lewis acid strength of their cations. In addition, an unexpected decrease in the yield of trioxane has been observed by the addition of NaHSO₄ to the reaction solution (Fig. 2). These comparisons reveal that the salt effect on the yield of trioxane in the reaction solution does not rely solely on the Lewis acid strength of the salt.

Fig. 5 The changes with cation radius of the Cl⁻-based salts in (k_1,salt/k₁)/(k_2,salt/k₂) and water activity.

It is notable that salt effects on the palladium(II)-catalyzed regioselective isomerization of methylenecyclopropyl ketone apparently increase from LiBr to NaI and from Bu₄NBr to Bu₄NI (Table 1 of ref. 12), indicating that salt effects increase with increasing radius of the constituent cation or/and anion of the salts. On the contrary, salt effects on the yield of trioxane in the reaction solution decrease with increasing radius of the cations of the salts. The salt effect on asymmetric synthesis of chiral homoallylic amines by the in-mediated allylation of chiral N-tert-butanesulfinyl imines in aqueous media at room temperature follow the order LiBr = KBr, NaCl < NaI < NaBr (Table 1 of ref. 13). Therefore, no clear connection between the salt effect and the radius of the constituent cation or/and anion of the salt is observed.

During the continuous production of trioxane, it is removed as a distillate from the reaction solution in the distillation tower. Therefore, the effect of salts on the conversion and time/space yield will emerge as manifestations of much more influential factors. In addition to their effect on the yield of trioxane in the reaction solution as discussed above, the salts can also increase the relative volatility of trioxane and water and of trioxane and HO(CH₂O)_nH, as the specific interactions between trioxane and the coexisting ions of the salt are considerably smaller than those between water (or HO(CH₂O)_nH) and these ions due to the lack of the –OH group(s) in the molecular structure of trioxane. The heat of vaporization of the mixture (water–formaldehyde) is considerably greater than that of trioxane, and almost all the trioxane present in the vapor phase from the reaction solution can be concentrated into the distillate in the distillation tower under proper refluxing conditions.⁴ Therefore, increasing the concentration of trioxane in the vapor phase by adding a salt is a very important method for decreasing the energy consumed in trioxane synthesis.

6 Conclusions

Batch reaction experiments were performed to investigate the salt effect on the yield of trioxane in the reaction solution. The addition of NaHSO₄, Na₂SO₄, NaH₂PO₄, or Na₂HPO₄ considerably decreased the concentration of trioxane in the reaction solution and the negative effect increased in the order NaHSO₄ < Na₂SO₄ < NaH₂PO₄ < Na₂HPO₄. The Cl⁻-based salts had a positive effect on the yield of trioxane in the reaction solution and their effect increased in the order KCl < NaCl < LiCl < ZnCl₂ < MgCl₂ < FeCl₃.

Systematic studies have been performed to uncover the mechanisms that govern the salt effects on the yield of trioxane in the reaction solution. The results show that the negative effect exerted by the addition of Na₂SO₄, NaH₂PO₄, or Na₂HPO₄ arises from the formation of NaHSO₄, H₃PO₄, or (H₃PO₄ and NaH₂PO₄), which inevitably decreases the concentration of H⁺ in the solution. The effect of KCl, NaCl, LiCl, MgCl₂, ZnCl₂, and FeCl₃ on the yield of trioxane in the reaction solution relies primarily on their ability to decrease the water activity of the reaction solution.

Continuous production experiments were performed to investigate the salt effects on the concentration of trioxane in the distillate. The results showed that the addition of KCl, NaCl, LiCl, MgCl₂, ZnCl₂, and FeCl₃ to the reaction solution can considerably increase the concentration of trioxane in the distillate. The salt effect increased progressively from left to right. Such an effect is a manifestation of the synthetic performance of the salt to increase the yield of trioxane in the reaction solution and to increase the relative volatility of trioxane and water and of trioxane and oligomers.

Acknowledgements

We acknowledge the National Natural Science Foundation of China (21076224 and 21276271) and Science Foundation of China University of Petroleum, Beijing (qzdx-2011-01) for financial support.

References

1. C. G. Xia, Z. H. Tang, J. Chen, X. Z. Zhang, Z. Li and E. X. Guo, US Pat., 7244854 B2, 2007.
2. T. Suzumori, I. Masumoto, M. Nakagawa and Y. Iha, US Pat., 4814424, 1989.
3. H. Benabdallah and D. Olender, Wear, 2006, 261, 1213.
4. J. Masamoto, K. Hamanaka, K. Yoshida, H. Nagahara, K. Kagawa, T. Iwaisako and H. Komaki, Angew. Chem., Int. Ed., 2000, 39, 2102.
5. T. Grützner, H. Hasse, N. Lang, M. Siegert and E. Ströfer, Chem. Eng. Sci., 2007, 62, 5613.
6. O. Kashihare and M. Akiyama, US Pat., 5929257, 1999.
7. H. Hirohisa, US Pat., 5962702, 1999.
8. C. G. Xia, H. Y. Song, J. Chen, X. Z. Zhang, Z. Li and E. X. Guo, CN Pat., 101311154A, 2007.
9. J. Chen, H. Y. Song, C. G. Xia and Z. H. Tang, CN Pat., 102020629A, 2011.
10. J. Chen, H. B. Wang and R. Cheng, Technol. Dev. Chem. Ind., 2013, 42, 60.
11. K. Yoshida, T. Iwaisako, J. Masamoto, K. Hamanaka and H. Komaki, US Pat., 4563536, 1986.
12. S. M. Ma and J. L. Zhang, Angew. Chem., 2003, 115, 193.
13. X. W. Sun, M. Liu, M. H. Xu and G. Q. Lin, Org. Lett., 2008, 10, 1259.
14. L. C. McCann and M. G. Organ, Angew. Chem., Int. Ed., 2014, 53, 4386.
15. M. Maiwald, T. Grützner, E. Ströfer and H. Hasse, Anal. Bioanal. Chem., 2006, 385, 910.
16. I. Hahnenstein, H. Hasse, C. G. Kreiter and G. Maurer, Ind. Eng. Chem. Res., 1994, 33, 1022.
17. J. F. Walker, in Formaldehyde, Reinhold ACS Monograph Series, New York, 3nd edn, 1964, ch. 7, pp. 191−195.
18. F. X. Jin, MEng Thesis, University of Lanzhou, 2014.
19. Y. L. Cui, Petrochem. Technol., 1990, 19, 214.
20. C. G. Meng, in Acid–Base Equilibrium, ed. J. Jian, Dalian, 5nd edn, 2006, ch. 5, vol. 1, p. 108.
21. K. E. Gutowski, G. A. Broker, H. D. Willauer, J. G. Huddleston, R. P. Swatloski, J. D. Holbrey and R. D. Rogers, J. Am. Chem. Soc., 2003, 125, 6632.
22. P. Debye and J. McAulay, Phys. Z., 1925, 26, 22.
23. O. Halpern and P. Gross, J. Phys. Chem., 1934, 2, 184.
24. R. A. Robinson and R. H. Stokes, in Electrolyte Solutions, Butterworths Scientific Publications, London, 2nd edn, 1959, ch. 8, vol. 72, pp. 450–480.
25. Y. F. Hu, S. S. Fan and D. Q. Liang, J. Phys. Chem. A, 2006, 110, 4276.
26. W. B. Jensen, Chem. Rev., 1978, 78, 1.

---