Jan., 1956
BASESTRENGTHS OF AMINESIN NON-PROTOLYTIC SOLVENTS
63
POTENTIOMETRIC DETERMINATION OF THE BASE STRENGTHS OF AMINES IN NON-PROTOLYTIC SOLVENTS BY H. K. HALL,JR. Contribution from the Textile Fibers Department, Pioneering Research Division, E. I . du Pont de Nemours & Company, Inc., Wilmington, Delaware Received Juns 10. 1866
A reproducible titration method was devised and successful potentiometric titrations were performed on a large number of monoamines and diamines in five solvents of varying nature: ethyl acetate, acetonitrile, nitrobenzene, nitromethane and ethylene dichloride. The titrant acids were perchloric, p-toluenesulfonic and perfluorobutyric, all in dioxane solution. Methanesulfonic acid was used as a solution in the given solvent. The mid-point of the titration curve was taken as a measure of the dissociation constant of the amine. The existence of quantitative linear relationships between the mid oint readand the Hammett u constants in water for m- and p-substituted anilines WM established. For the apkylamines ings the observed order of base strength was: NHo < RNH2, RzNH > RIN, an order similar to that observed in water or alcohol. As the length of the alkyl chain increased to four or five carbon atoms, the base strength was lowered. Diamines, such as ethylenediamine, hexamethylenediamine and the piperazines behaved in acetonitrile as rather strong bases. Polar substituents in the iperidine ring as in morpholine, monoacyl iperazines, etc., decreased the base strength of the parent amine as expected. d e order and even the quantitative strengtt of the bases were independent of solvent or of acid for the five organic solvents. When data for organic solvents were plotted against those for water, acceptable linear plots were obtained (with one exception-see below). Even this extreme change of solvent does not affect relative base strength. The exception noted above referred to amines containing a strongly polar group near the N atom, such as morpholines and monoacylpiperazines. These bases are much stronger in organic solvents than their p K . values in water would predict. A discrepancy between the predicted and observed slopes of the plots of E+ against p K , is explained in terms of variations in the activity coefficients of the ammonium ions relative to the corresponding a m e s .
Non-protolytic solvents in which base strengths have been quantitatively studied include benzene,1--3 chl~robenzene,~J acetonitrile16 nitrobenlo chloro~ene~7.8 tricresyl p h o ~ p h a t enitromethane, ,~ form”-13 and carbon tetrach10ride.I~ The methods employed have been chiefly indicator ones. Conductimetric determinations, although precise, offer a more complex problem in media of low dielectric constant than in high dielectric solvents. Reaction kinetics methods have been employed, but, according to Grunwald and B e r k o ~ i t z , ’often ~ give discordant values of pKa. Association phenomena which complicate the kinetics methods will be more serious in non-polar media. Infrared methods were used in one investigation.l 3 Finally potentiometric methods involving either glass or hydrogen electrodes have been employed on only two occasions1Sl6 in non-protolytic media. These difficulties account for the fact that far fewer base strengths in such solvents are known than in protolytic solvents. Because of the nature of the available experimental methods, investigations in these solvents have been limited to only moderately strong acids. Witschonke and Kraus8 concluded that a.ny salt less dissociated than pyridine picrate cannot be studied successfully by conductimetric means. (1) V. K. Lamer and H. C. Downes, J . Am. Chem. Soc., 65, 1810 (1033), and references contained therein. (2) M. M. Davis and H. B. Hetzer, +bid., 76, 4247 (1954), and references contained therein. (3) A. A. Maryott, J . Research Nett. Bur. Standards, 41,7 (1948). (4) D. C. Griffiths, J . Chem. SOC.,818 (1938). (5) R. P.Bell and J. W. Bayles, ibid., 1518 (1952). (6) M. Kilpatrick and M. L. Kilpatrick, Chem. Reus., 13, 131 (1933). (7) I. AI. Kolthoff, D. Stocesoca and T. S. Lee, J . A m . Chem. Soc., 76, 1834 (1053). (8) C. R . Witschonke and C. A. Kraus, dbid., 69, 2472 (1947). (9) M. A. Elliott and R. M. Fuoss, ibid., 61, 294 (1931)). (10) L. C. Smith and L. P. Hammett, ibid., 67, 23 (1945). (11) J. A. Moede and C. Curran, ibid., 71, 852 (1949). (12) M. M.Davis, ibid., 71, 3544 (1949). (13) G. M.Barrow and E. A. Yerger, +bid., 76, 5211, 5247, 5248 (1954). (14) E. Grunwald and B. J. Berkowits, ibid., 73, 4980 (1951). (15) J. S. Fritz, Anal. Chem., IS, 407 (1053).
Picric and dichloroacetic acids were the strongest which were studied in acetonitrile.6 No study has been reported on the interaction of amines with very strong acids except that of Fritz. l6 He performed potentiometric titrations of amines in acetonitrile with perchloric acid in dioxane using glass and calomel electrodes. He made the significant observation that the millivoltage reading a t half-neutralization, hereafter called Ell2, could be correlated with the base strength of the amine. It appeared, therefore, that base strengths in nonprotolytic solvents could be measured by this method. We took this as a starting point for the present study. Our first experiments using the procedure of Fritz did not give reproducible values of E I ~the ~, millivoltage reading at half-neutralization. It was necessary, as described in the Experimental Section, to standardize the electrodes against an organic buffer solution. When four different sets of electrodes were standardized in this way, their readings upon immersion in various amine solutions agreed to il mv. Solvents.-The original paper of Fritz, which served as the starting point of the present study, utilized acetonitrile as the solvent. However, there seemed to be no reason why the study should be confined to this solvent. A recent booklet by Fritz16summarizes work on acid-base titrations in a variety of non-aqueous solvents. It appeared that any solvent of even moderate polarity can be used as the solvent. I n the present study, five solvents were found to be perfectly satisfactory. Nitromethane, nitrobenzene and acetonitrile are all highly polar. Ethyl acetate is moderately polar and the fact that it gave sharp titration curves occasioned no surprise. Ethylene chloride, however, is quite non-polar, and it was somewhat unexpected to discover that it behaved as well as the other solvents. Other solvents tried, (16) J. S. Fritz, “Acid-Base Titrations in Non-aquoous Solvents,” The G. Frederick Smith Co., Columbus, Ohio, 1952.
64
H.I(.HALL,JR.
VOl. 60
which gave drifting millivoltage readings, were benzene, anisole, dioxane, chloroform and methylene chloride. The question arose as to reaction of the amines with several of the solvents employed. Reaction of amines with nitrobenzene is unlikely, but they might remove a proton from nitromethane or acetonitrile to form the corresponding anions. Amidine formation from the latter is possible. Ethyl acetate might undergo aminolysis, and several modes of decomposition can be visualized for ethylene chloride. There are several reasons for believing that none of these reactions occur: (1) The amine and solvent are in contact for less than five minutes. (2) Precipitation from ethylene chloride solutions should be observable. (3) If reaction were very rapid with ethyl acetate, the ethanol and amide formed would be non-basic, and any decrease in amine concentration would show up as a steady drift in the electrode reading. (4) Establishment of a mobile equilibrium with acetonitrile or nitromethane would not affect the results, since water participates in just such an equilibrium. Amines.-There was little limitation on the type of amine which could be titrated. Of the aromatic amines, only p- and especially o-nitroanilines were too weak bases to give sharp titration curves. m-Nitroaniline was titrated successfully in several solvents, however, and stronger bases were very easily titrated. At the other extreme, pyrrolidine seemed to be about the strongest base encountered, but there was no reason t o believe stronger bases could not be studied.
sulfonic acid formed precipitates with a variety of amines in each solvent, and this restricted its use-
chloric acid; A, N-allylpiperidine, ethyl acetate, perchloric acid.
(18) (a) R. 0. Pearson and F. V. Williams, J . Am. Chem. Soc., 76, 258 (1954); @) A. F. Trotman-Diokenson, J . Chem. Soo., 1293 (1949). (19) (a) M. Misutsni, 2. physik. Cham., 116, 360 (1925); (b) 118, 327 (1925). (20) H. F. Herbrindson, R. T. Dickerson, Jr., and J. Weinstein, J . Am. Cham. Soc., 76, 4046 (1954). (21) A. A. Maryott. J . Research Nall. Bur. Standards, 41, I (1948).
fulness. Acids.-Dilute solutions of 70% perchloric acid in pure dioxane were found to be perfectly stable for many months. p-Toluenesulfonic acid, which is readily available in crystalline form, was also chosen as a titrant. Perfluorobutyric acid was strong enough to titrate triethylamine, but not 2,6dimethylpyridine. It appears, therefore, to be relatively weak. Conformity to Titration Equation.-It was felt to be necessary to verify that the titration data followed the usual equation. Figure 1 shows plots of log X/(1 - X ) , where X is the fraction neutralized, against the observed millivoltage value. The plote of all solvents except ethylene dichloride are linear. The simple interpretation of Et/* as a measure of dissociation constant cannot therefore be valid for this very non-polar solvent. The slopes of the lines are not the expected ones of 59 mv. This may be due to lack of constant ionic strength. The titrations are performed a t 0.01 M concentration. Results.-The results obtained are given in Tables I-VII. The relative base strengths were independent of the titrating acid and, to a first approximation, of the solvent. The observed values can, therefore, be discussed in terms of the structures of the amines. From the data in Tables I-VII, it can be seen that the results are in accord with expectations considering their structures. For example, electron-attracting substituents such as allyl or benzyl cause a marked reduction in base strength." Numerous other examples could be cited. Only the case of the alkylamines will be discussed, since this problem has attracted considerable attention. For the alkylamines, the complete sequence of base strengths in each solvent (and in water) is NH3 < RNH2, RzNH > RaN (Table I). The exact relative positions of RNHz and RzNH cannot be established because of experimental inaccuracies, but it can be said that they are always very similar. In the past, claims have been made18that the true order of base strengths should be NHa < RNHz < %NH < RaN, because of increasing electron supply to the nitrogen atom by the alkyl groups. Deviations from this order in water, methan01,'~ and ethanol19 are explained in terms of preferential solvation, through H-bonding, of the several ammonium ions. It does not appear to be reasonable to expect that in inert solvents the ideal order should be observed, however, since hydrogen bond0 4 ing of the ammonium ion will still occur to either -1.0 -0.5 0 +0.5 +1.0 another amine molecule,' another acid molePK~. or other molecules of salt.2.8.21 Therec~le,7.'3~~ Fig. 1.-Conformity of data to titration equation: @, diethylamine, nitrobenzene, perchloric acid; +, n-butyl- fore, it does not appear probable that this order of amine, acetonitrile, perchloric acid; 0,piperidine, ethylene base strengths toward strong acids can be realized. dichloride, p-toluenesulfonic acid (70 mv. added to ob(17) The higher the El/* value, the weaker is the base. served reading); a,N-ethylmorpholine, acetonitrile, per-
A serious limitation lay, however, in the insolubility of certain amine salts. With perchloric acid precipitation occurred only in ethylene chloride, only a few other cases being noted. p-Toluene-
BASESTRENGTHSOF A.MINEB IN NON-PROTOLYTIC SOLVENTS
Jan., 1956
65
T-LE I M I L L I V O ~ AHALF-NEUTEUIZATION GE VALUESFOB ALIPHATICAMINES AND DIAMINES, USINGPEBCHLOBICACID Amine
5%
Nitromethane
Ethylene ckchlonde
9.26
acetate
Ethyl
Nitrobemene
1~Ammonia 2 Methylamine
193 73
Ppt.
Ppt.
10.64
Ppt.
98
94
3 Dimethylamine
10.61
101
..
102
4 Trimethylamine
10.72
150
..
Ppt.
5 Ethylamine
10.75
..
Ppt.
115
6 Diethylamine
11.00
..
175
124
7 Triethylamine
10.74
60
172 165
197 198 129 158 228 127
68
10.59
..
9 Di-n-propylamine 10 Tri-n-prop ylamine 11 Isopropylamine
10.91 10.70 10.63
.. .. ..
..
12 Diisopropylamine 13 n-Butylamine 14 Di-n-butylamine
11.05 10.61 11.31
.. ..
..
..
15 Tri-n-butylamine 16 Isobutylamine
10.89 10.42
8 +Prop ylamine
Diisobutylamine see-Butylamine Di-see-butylamine t-Butylamine Di-n-amylamine Tri-n-amylamine Neopentylamine Hexamethylenediamine Ethylenediamine Ethanolamine Monometylethylenediamine Monobemoylethylenediamine Determined in the present work. 17 18 19 20 21 22 23 24 25 26 27 28
a
..
Solvents
10.59 10.56
...
10.45 11.18
... 10.40 11.11 10.08 9.45 9.28" 9.13"
Since the present study has shown that the same order as in water is maintained in such solvents aa nitromethane and acetonitrile, it appears more attractive to maintain that the order of base strengths toward H+ is determined solely by some property of the molecules themselves. The B-strain hypothesis of H. C. Brown82is in accord with these results. Relative Base Strengths of Amines in Water and in Organic Solvents.-The base strengths in water of a great many amines have been determined in the past, but there are very few values available for non-aqueous solvents. It has, therefore, been difEcult to assess the importance of the solvent in determining amine base strength. With the large quantity of data collected in the present investigation, it becomes possible for the first time to make an adequate comparison. wz- and p-Substituted Anilines.-This class of amine will be dealt with separately since theoretical interpretat.ionof its behavior is particularly simple. (22) H. C. Brown. Rse. Clrsm. Prop.. 14, 83 (1953).
..
..
..
..
..
.. ..
148 149
261 259
..
..
.. ..
..
.. .. ..
..
..
..
.. ..
..
.. .. .. .. .. .. ..
.. ..
..
Acetonitrile
100 82
83 59 63 87 94 100 112 78 75 87 86 129 121 108 131 127 110
73 88
.. .. 57 75
.. .. ..
..
..
150
..
85 55
210 162 178 172 207 148 182 130 164 263 188
..
..
150
105 114
.. ..
.. .. .. ..
.. ..
..
145 130
..
60
Ppt. Ppt. 90
..
..
.. ..
..
..
.. .. .. ..
.. ..
..
fi2 119 118
147
If the E l / , values of the present study should be proportional to @a, it was expected that a linear relationship should hold between the former and the Hammett U-value of the substituent. In Fig. 2 the data obtained using perchloric acid have been plotted. In ethylene chloride precipitation of the aromatic amine perchlorates occurred and no data could be obtained. Similar effects hampered the study of p-toluenesulfonic acid, where only rudimentary plots were obtained. It is clear that the expected linear relationships hold very well. The lines were determined by the method of least squares and the equations of the lines are given in Table IX. The linear Hammett relationships afford confirmatory evidence that the glass electrode is measuring the activity of H+ in the solutions, for linearity could not hold otherwise. Although theory predicts that the slopes of the lines should all be +59 mv., it is apparent that this is not the case. The slope depends on the solvent. This must be related to the activity coefficients of the amine am1 its ion, as will be discussed below.
66
H. K. HALL,JR.
Vol. 60
TAB^ I1 MILLIVOLTAGE HALF-NEUTRALIZATION VALUESFOR NON-AROMATIC CYCLICBASES,USINQPERCHLORIC ACID Amine
Structure
29 < ~ ) - N H ~ --CHZ
31 32 P
a g $(CH3) a
Solvents Ethylene Ethyl NitroNitro- Apetomethane dichloride acetate benzene nitnle
10.79
..
..
132
..
..
...
..
..
..
73
Pyrrolidine
11.32
..
..
75
.. ..
..
Piperidine
11.20
80 77
166 170
130 131 132
78 70
36 47
Hexamethyleneimine
11.10"
..
..
118
*.
..
2,2,6,6-Tetramethylpiperidine
11.10"
..
..
134
..
*.
8.36"
..
..
191
205
164
63 67
187 190
191 192
..
83 94 76
.. ..
298
..
..
290
,,
221
Bis-(p-aminocyclohexy1)-methane
H
33 34
Ha0
Cyclohexylamine
30 ( H q * > 2
C
pKa,
Name
)Z
35 O N -H
Morpholine
36 o c 2 H 6
N-Ethylpiperidine
\-,
'-\
10.40
N-CHa
N-Methylmorpholine
7.41"
..
NCe&
N-Et hylmorpholine
7.70"
*.
39 Hd-NCOOCzH6
1-Carbethoxypiperazine
8.28"
205 201
Ppt.
260 262
230
207 196 184
40
1-Benzoylpiperazine
7. 78"
..
..
..
..
210
41
1-Tosylpiperazine
7.44"
..
..
285
..
252
42
Anhydrous piperazine
9.81
Ppt.
Ppt.
..
trane-2,5-Dimethylpiperazine
9.69"
122
Ppt.
103 101 158' 148
44 \Nh/ -
trans-2,5-Dimethylpiperazine Mono-ion
5.25"
425
Ppt.
338 347
..
65 64 85 101 85
45
,NCH~CH=CH,
N-Allylpiperidine
9.68"
,
,
..
248
NCHzCH=CHz
N-Allylmorpholine
7.05"
..
..
354
37 0
\-/ '-\
38 0
\-/
\-/
43
p-toluenesulfonic > methanesulfonic > perfluorobutyric. The latter gave reproducible E l / , values toward triethylamine in acetonitrile (24 and 30 mv.) but 2,6-dimethylpyridine gave no detectable break. Procedure.-The titrations were performed using 50-75 ml. of solvent in a 200-ml. tall form beaker. Magnetic stirring was provided. A Gilmont 1-ml.microburet-pipet was used in the titration. The uantity of amine was such tm to require 0.3-1.1 ml. of 0.5 #perchloric acid. The standard buffer solution was prepared by weighing out 1.5000 g. of Sharples diisobutylamine, b.p. 139.7140.0°, which had been fractionated in a spinning band column,m and making it up to 350.0 ml. in volumetric flasks. To this was added by pipet 5.00 ml. of 0.5032 N erchloric acid in dioxane. The solution is stable for at east two days. The electrodes were Beckman #119042 and #1170, respectively.81 They were connected to a Beckman model G pH meter by means of a Beckman #8700 six-point manual switching box. Adjustment of the millivoltage reading was made on the latter, the position of the zero adjuster dial on the pH meter being immaterial. A small dry cell in the switching box required replacement only at long intervals. The electrodes were allowed to stand in water when not in use. To standardize, they were wiped with tissue and placed in the buffer solution. The mv. reading (taking care to reset dials I and especially I1 when changing from pH scale) was then adjusted by means of the switching box to +170 mv. After standardizing, a titration was performed in the usual fashion, using increments of 0.030 ml. until near the end-point (no steady reading may be obtainable until the first increment has been added). A plot was then made of E va. ml. added and E l / , read from the graph in the usual way. After the titration the electrode was returned to water. A second titration was performed using another pair of electrodes. The first pair was then used again. When checkcd in the organic buffer, i t was always found to read +170 mv. It was difficult to measure a suitable small amount of the gaseous amines into the solvent. This was overcome by adding 0.030 ml. of acid to the solution and passing in amine until the reading reached a value previously determined as indicating a suitable quantity. The pKe values in water marked with an "a" were determined in the present work. This was done by potentiometric titration of 100 ml. of ea. 0.01 M aqueous splution of amine at 25.0' and 0.5 N hydrochloric acid, adding the latter from a Gilmont 1-ml. pipet. Other p K . values are from standard literature references, chiefly references 32-35.
P
Acknowledgments.-We are deeply indebted to Mr. Donald G. Preis for performing most of the titrations, to Miss Helen Anderson for performing the remainder and t o Dr. P. W. Morgan for helpful criticism and encouragement. (30) Careful fractionation waa actually unnecessary since Sharples material dried over potassium hydroxide pellets gave identical results to those obtained using purified material. (31) These electrodes have been shown (B. Gutbeeahl end E. Grunwald, J . Am. CAem. SOC..71,559 (1053)) to be the preferred type? for non-aqueous media. (32) N. F. Hall and M. R. Sprinkle, ibid.. 14, 3469 (1932). (33) J. D. Roberts, R. L. Webb and E. A . hlcElilill, ibid., 11, 408 (1 950). (34) R. G . Pearson and F. V. Williams, ibid., IS, 258 (1954). (35) R. J. L. Andon. J. D. Cox and E. F. C . Harrington, Trans. Faraday Soc., SO, 918 (1954).
