Table IV. Data Corresponding to Figure 2 Volume of Concn tris-(TMS)POd of aqueous injected, p1 Peak area, cma phosphate, pprn 1 .o 0.08 0 5.0 0.80 0 1.0 1.05 0.21 1 .o 1.35 0.21 5.0 8.20 0.21 1 .o 3.60 0.53 1.0 3.36 0.53 1 .o 7.73 1.06 1 .o 7.28 1.06
The FPD responded negatively to large (injected) volumes of sample, thereby obscuring the tris-TMS-phosphate peak. Five-microliter portions can be safely injected under the conditions recommended here. Chromatograms 5 and 6 in Figure 2 correspond to 5-pl injections of a blank and a 0.21 ppm PO.,sample, for example, and the normalized response is virtually identical to that obtained with 1 -pl injections. Representative chromatograms are shown in Figure 2, along with the calibration curve corresponding to these chro-
matograms. Data from these chromatograms are plotted in Figure 2 and given in Table IV. Note that both 1- and 5-pl injections are included in this series. Best performance is obtained if the instrument is standardized and samples are run o n the same day. Random changes in detector response are observed from day-to-day. With daily standardization, this method is capable of determining concentrations of aqueous inorganic phosphate as low as 100 ppb with a precision of 7-10%. We are currently investigating procedures for determining other anions by extraction-derivatization-GC. The determination of sulfate by this technique is especially attractive because of the availability of flame photometric detection for sulfur (IS). Several anions should be determinable simultaneously by this technique by use of flame ionization detection. RECEIVED for review April 12,1971. Accepted June 21,1971. This research was sponsored by the U. S. Atomic Energy Commission under contract with the Union Carbide Corporation. (15) H.W.Grice, M. L. Yates, and D. J. David, J . Chromatogr. Sci., 8, 90 (1970).
Separation of Weak Organic Bases Such as Aniline, N-Heterocyclic, and Sulfonamide Derivatives on Cation Exchange Resin T. Carter Gilmer’j* and Donald J. Pietrzyk Department of Chemistry, University of Iowa, Iowa City, Iowa 52240
Distribution coefficient data for aniline, N-heterocyclic, amide, and sulfonamide derivatives on H-form macroporous and gel type resins are reported for several different water-organic solvent mixtures. The organic solvent influences elution by providing aprotic conditions, or differing degrees of a weakly basic property. The weak organic base-resin interaction, depending on water composition, is explained in terms of an acid-base reaction or solubilization. A variety of separations including typical sulfonamide mixtures are described. It i s possible in some cases to use flow rates as high as 34 ml/min for the separation. Separation efficiency in terms of number of plates and resolution was considered for the porous and gel type resins.
IN RECENT YEARS many metal ion separation studies o n exchange resins have been modified or improved by incorporating organic solvents into the eluting mixtures (1, 2 ) . Similar NDEA Fellow, September 1967 to September 1970; taken from the Ph.D. thesis submitted to the University of Iowa. * Present address, Eastman Kodak Co., Rochester, N. Y .
-_
~
Helffer:ch, “Ion Exchange,” McGraw-Hi11, New York, N.Y , 1962. (2) \V Ritman and H. F. Walton, “Ion Exchange in Analytical Chemistry,” Pergainon Press, New York, N. Y . , 1970. (1) F.
applications to the separation of nonelectrolytes have been sparse (1-3). An advance which should aid ion exchange separation technology is the synthesis of the macroporous ion exchange resins (4). These resins, even though they contract and swell? retain much of their porous properties in nonaqueous solvents. Thus, it is possible to use these resins even in nonpolar solvents such as benzene and heptane. In contrast. the gel variety of ion exchange resin must be preswollen, if it is to be used in a nonpolar type solvent. If the gel type resin is placed in a dry nonpolar solvent, the resin will he in a co!!apsed state and its exchange sites unavailable. Solvents play a very important role in determining the properties and interactions of electrolytes and nonelectrolytes in solution. Thus, it would be anticipated that interactions between ion exchange resins and electrolytes or nonelectrolytes would also be affected by nonaqueous solvents. One such kteraction is that which can occlur between weak organic bases and the stronglv acidic H-form exchange resins. The eFect of organic solvents on this interaction mhy be ccnsidered ~~~
(3) J. Inczedy, “Analyticdl Auplicntions of Icn Exchangers ” Pergamon Press. New York, N Y . . 1966. (I R.) Kunin. E. F. Meitzner, 9. A . Clme, S A Ft-her, and N. Frisch, ind. Erzg. Chem., Pyod Res. fiieo., 1, l*b ( 1 5 ~ 2 ) .
ANALYTICAL CHEMISTRY, VOL. 43, NO. 12, OCTOBER 1971
0
1585
- +
+
Solvent
- +
Resin-S03H B Resin-S08BH (1) to be analogous to that observed in nonaqueous acid-base measurements. Thus, by appropriate changes of solvent o r solvent mixtures, the apparent acidic and/or basic properties of the resin and organic bases, respectively, may be lessened or enhanced which results in shifts in the direction of the equilibrium in Reaction 1. If this is applied to separation of mixtures of organic bases, it would be necessary to use different solvents or solvent mixtures as eluting agents. Many studies in the literature have been devoted to establishing the properties of the gel and macroreticular type ion exchange resins in nonaqueous solvent (4-9). In these studies, however, some separations have also been included. For example, several amides were separated using acetonitrilemethanol (IO). From the work in this laboratory, several eluting conditions can be predicted for the separation of nitroanilines (11). More recently macroreticular anion and cation resins were used in separating certain alkylphenols, naphthenic acids, nitrogen bases, and pyrrolic compounds from petroleum products (12). Gel type cation resins have also been used for separation of bases (9,13-15). The data in this report are intended to illustrate the potential in using nonaqueous solvents for the separation of closely related weak bases on H-form ion exchange resins. Various parameters involved in the separation are evaluated and discussed. It appears that the general technique should be useful for quantitative as well as purification purposes. EXPERIMENTAL
Reagents. AMINES. The organic bases were obtained from Eastman Organic Chemical Co. and other chemical supply houses. Purification was effected by repeated recrystallizations until sharp, acceptable melting points were obtained. RESINS. Dowex 50 X8 (D-50), 100-200 mesh was purchased from the J. T. Baker Chemical Company. Amberlyst15 (A-15), 16-50 mesh was purchased from Rohm and Haas Chemical Company. A more uniform and smaller particle sized A-15 resin was obtained by crushing the resin in a blender. After drying the resin (A-15c), it was sieved and a 40 to 60 and 100 to 200 mesh U.S.standard was used. RESIN TREATMENT. The A-15 resin was washed in a column alternately with ethanol and benzene and then transferred to a Soxhlet extractor where refluxing ethanol and benzene were used until the solvents remained colorless. The ethanol wetted resin was transferred to a column and washed with water, sodium citrate solution, and finally 1.5M HCl. Excess acid was removed with water and finally air-dried. If the resin was to be used for distribution co( 5 ) D. J. Pietrzyk, Tdanta, 16, 169 (1969). (6) [bid., 13, 209 (1966). (7) A. D.Wilks, M. D. Grieser, and D. J. Pietrzyk, unpublished
results.
(8) J. E.Gordon, J . Chromatogr., 18, 542 (1965). (9) H.F. Walton, ANAL.CHEM.,42,86R (1968). (10)J. F. Cassidy and C. A. Streuli, A d . Chim. Acta, 24, 334 (1961). (11) D. J. Pietrzyk, Tdanta, 13, 225 (1966). (12) P. V. Webster, J. N. Wilson, and M. C. Franks, A n d . Chim. Acta, 38, 193 (1967). (13) L. N. Tokareva, G. D. Galpern, A. V. Kotova, T. E. Koscukhova, and A. Y. Lanchuk, Zh. Prikl. Khim. (Leningrad), 43,1403 (1970). (14) H. Hatano, K. Sumizu, S. Rokushika, and F. Murakami, A n d . Biochem., 35, 377 (1970). (15) S. R. Watkins and H.E. Walton, Anal. Chim. Acta, 24, 334 (1961). 1586
efficient studies, it was washed with alcohol before drying in a n oven. The treatment of the D-50 resin was the same except for elimination of the ethanol-benzene washings and Soxhlet extraction. This part of the procedure was replaced with washing with ethanol. All resins were in the H-form except where noted. For the distribution coefficient studies, the resins were dried in a vacuum oven at 3-7 mm H g and 100 “C for over 24 hours, SOLVENTS. Analytical reagent grade solvents were used in all cases. In the studies involving measurement of the distribution coefficients, the solvents were dried and purified by previously described techniques (5, 6, 16). Where mixed solvents are used, the per cent cited is by volume. Procedures. DISTRIBUTION COEFFICIENTS. Approximately 1 gram of resin was accurately weighed into a 125-ml ground glass stoppered Erlenmeyer flask. Aliquots of the solvents, one of which contains the organic base in the organic solvent, were added such that the total volume would add up to 50 ml and the solution would be approximately 6.9 X 10-*M in base. These solutions were allowed to equilibrate at 24 f 1 “C. Standards were made in the same way except that the resin is omitted. After equilibrating, an aliquot of the solution is removed, diluted to volume, and measured spectrophotometrically. Concentrations were established from previously measured calibration curves. A Beckman DBG or DU spectrophotometer was used for these measurements. The distribution coefficient was calculated by KD =
ml of solution X wt of organic base on the resin wt of dry resin X wt of organic base in solution
(2)
CALIBRATION CURVES. Calibration curves were made in the usual way. Absorption of the parent compound was used except for the amides and the sulfonamides. Amides were determined by converting the amide to the hydroxamic acid-Fe(II1) complex (1 7). Sulfonamide was determined by conversion to the Schiff base by reaction with N,N-dimethyl-p-aminobenzaldehyde(18). I t was necessary to carefully buffer the solution to pH 1.9 and ensure that the benzaldehyde derivative-to-sulfonamide ratio is at least 400 to 1 on a molar basis. For the samples containing dimethylsulfoxide (DMSO), the DMSO concentration in both the calibration curve and the samples was carefully controlled. SEPARATIONS.Columns of resin (1-cm diameter glass tube with a coarse glass frit as support) were made by the slurry technique, using the initial eluting agent as the solvent for the resin slurry. Flow rates were controlled by either of two pumps; AutoAnalyzer Proportioning Pump o r Biichler Polystaltic Pump M-2-6100. Calibration of the latter pump was done by recording the time needed to colIect a known volume of solvent. Column effluent was followed either spectrophotometrically (Beckman Model DB6 using a flow through cell, Hellma Cells, No-175QS) or by refractive index (Waters Associates Model R-4). Once the column settIed and the base line stabilized, pumping was stopped for addition of the sample (10-p1 syringe o r pipet). Pumping was resumed and according to the recorder trace each component of the mixture was collected and analyzed spectrophotometrically.
RESULTS AND DISCUSSION Perhaps the most fruitful way of establishing the proper eluting conditions for a separation in ion exchange chroma(16) D. D. Perrin, W. L. F. Armarego, and D. R. Perrin, “Purification of Laboratory Chemicals,” Pergamon Press, New York,
N. Y., 1968. (17) S. Siggia, “Quantitative Organic Analysis via Function Groups,” Wiley, New York, N. Y.,1963. (18) C. J. 0. Morris, Biochem. J., 35,952(1941).
ANALYTICAL CHEMISTRY, VOL. 43, NO. 12, OCTOBER 1971
tography is to determine distribution coefficients under a variety of experimental conditions. I n this way a strictly trial and error method is avoided. From the batch distribution coefficients, KO, it is possible to predict elution behavior. To separate two substances, conditions should be selected such that the distribution coefficient of one is low (preferably 1 or less) while the other is very large. As the ratio of the two KD’s approaches one, the column must be increased for separation. However, if both K D ’ s are large, the elution peaks will also be broad with extreme tailing. If the mechanism of base retention by the cation resin is a n acid-base reaction as described in Reaction 1, the type of solvent that is chosen will play a vital role in determining the equilibrium of that reaction. For example, solvents with basic properties will compete with the organic base for the acidic sites on the resin. Thus, Reaction 1 would also describe the solvent resin interaction where the solvent has basic properties. To qualitatively predict the direction of the equilibrium, the basic strengths of the solute base and solvent and the mass action effect of the solvent must be considered. Solvents which participate in H-bonding will interact with the H-form resin somewhere between the two cases (A) and (B). Using an alcohol solvent as a n example (A)
0
0
II
+ H . . .O-R H
R-S-0-
I/
0
I1 I/
R-S--0-H..
.O-R H
0
(A)
(B)
represents complete ionization of the sulfonic acid group and solvation of the proton. Although not shown, it is assumed in (A) that solvation of the resinate anion is also possible. At the other extreme, case (B), the sulfonic acid interacts with the solvent through hydrogen bonding. The net result of the interaction of the acidic sulfonic acid group o n the resin with a basic or hydrogen bonding type solvent is a leveling effect of the acidic sites; leveling by the basic solvent would be the most extensive. Thus, by change of solvent the apparent acidic nature of the site can be altered and consequently the direction in the equilibrium in Reaction 1 is changed. The magnitude of the K D will depend on the extent of leveling. In the case of solvents that d o not possess basic properties or are non- or very weak hydrogen bonders, the leveling property is absent. These solvents would cause the equilibrium in Reaction 1 to shift in the direction of the products o r a larger KDwould be measured. Assorted resins swell differently in the various solvents. Since t h t acidic sites are available only when the resin is swollen, the importance of this property is obvious. Broad generalizations regarding the eluting power of solvents are difficult to make, particularly if the comparisons involve different resins. In the studies reported here, two resins, one porous or macroreticular (A-15), and one gel or microreticular (D-50) were used. The former, although it does undergo swelling (7), retains most of its porous properties even in nonpolar-like solvents. In contrast, the gel variety, if it is freed of water by drying, will collapse and remain collapsed (unswollen) in the nonpolar solvents; therefore, its swelling properties are much more dependent on the experimental environment. In previous reports (5, I I ) , it was suggested that two types of eluting conditions are possible ; pure nonaqueous solvents
(or mixtures of nonaqueous solvents) and water-organic solvent mixtures. Since water competes favorably with the weak base for the acidic sites of the resin, a decrease in KD is expected upon adding water to the organic solvent. This decrease should be very sharp because of the leveling action of water toward the acidic resin. Indeed, if distribution coefficient data for p-nitroaniline (PNA) are examined, a sharp decrease is observed. [Data for PNA are listed in Table 1 in reference (11) as a function of water-organic solvent ratios in which 14 different organic solvents were used.] A similar trend was observed for o-nitroaniline (ONA), m-nitroaniline, and caffeine (11). From these data, eluting conditions for the separations of PNA and ONA are readily predicted. For example, the eluting mixture, 8 0 x alcohol-20% water, where the alcohol is methanol, ethanol, n-propanol, or isopropyl alcohol can be used for the separation of PNA (KO’sof 38.9, 17.9, 19.5, and 22.8, respectively) from ONA (KO virtually zero in all four mixtures). The eluting mixture that would be used would be determined by the properties of the solvent mixture such as viscosity, rate at which equilibrium is reached (6), solvent availability, and solvent purity. Several other mixtures appear to be useful. For example, in acetonitrile-water mixtures a sharp drop in KD takes place as the per cent water increases, and this mixture can be used for the separation of PNA and ONA. A third type of eluting mixture is one containing solvents with basic properties. The more basic the solvent is for a given water-solvent composition, the lower the KB will be. F o r example, the KD for PNA is 0 from 100% DMSO to 7 0 x DMSO a t which point KD increases as the DMSO concentration decreases. Other solvents with basic pioperties will provide the same effect and the extent of KO decreasing will be greater as the basicity of the solvent increases. Adding acid such as HCl, or CH3S03H to the eluting mixture will cause a decrease in KO. Although not explored in detail, mixtures of this type (acid-water-organic solvent) could also be potential eluting agents. Obviously, it is not possible to measure distribution coefficients under all conditions and for all weak organic bases. Therefore, from a practical point of view what is needed is to be able to predict the optimum elution conditions for the separation of a mixture of weak bases from a minimum amount of KD data. From the previous PNA data in a wide variety of organic solvent-water mixtures, it is possible to qualitatively predict the relative effect of the solvents on other bases using PNA as a reference point. Also, from the previous data ( I ] ) , a qualitative conclusion would be that the retention of weak organic bases will follow their respective order of basicity. In order to establish a quantitative or qualitative relationship between compound structure and eluting conditions, KD’s were determined for a series of aniline derivatives. These data for A-15, H-form resin in ethanol-water mixtures are listed in Table I. It should also be possible to qualitatively predict KD behavior for similar structures in other solvent mixtures from these data. For this reason some data for acetonitrile-water mixtures are also included. Several possible relationships were explored. The first two were attempts at correlating KD with K,(H20) and with Hammett substituent constants, u. A linear relationship was found between KD and CT for several p-substituted acetanilides (8).
If the data are plotted, a qualitative relationship is obtained that suggests that the stronger the base, the greater the reten-
ANALYTICAL CHEMISTRY, VOL. 43, NO. 12, OCTOBER 1971
1587
Table I. Distribution Coemcients for Several Aniline Derivatives in Ethanol-Water Mixtures on A-15,
Compound PKO 100 m-Nitroaniline 2.47 9600 1900 p-Nitroaniline 1.00 398 89.9 59.6 o-Nitroaniline -0.26 18.0 1.9 N-Meth yl-p-nitroaniline 59.3 9.2 N,N-Dimethyl-pnitroaniline 0,61 109 12 4-Methyl-2-nitroaniline 0.43 31.5 4.1 2-Methyl-4-nitroaniline 1.04 249 63.1 2-Methyl-5-nitroaniline 2.32 7560 1040 2-Aminobenzoic acid 2.11 1990 412 278 4-Aminobenzoic acid 2.41 4450 1230 754 3-Aminobenzoic acid 3.12 1080 1110 1080 4-Aminoacetophenone 2.29 c 2060 1340 3-Aminoacetophenone 3.59 c 4410 4220 1-Nitro-2-naphthylamine - 0.85 lo6 >los Purine 2.30 m >lo' >lo4 9500 6800 m 3900 1900 790 533 Pyridine N-oxide 0.79 Pyrimidine 1.23 m 2220 985 515 384 Pyrazine 0.65 2160 481 280 164 138 Quinoxaline 0.72 659 93.8 49.3 29.9 28.7 Benzotriazole 1.6 189 35.2 22.2 12.5 13.5 Phenazine 1.19 1220 123 60.2 39.2 43.0
tion of the base to the H-form resin. However, it also became apparent that other factors are influencing the retention. Thus, this generality holds only for compounds of similar structures (such as the aniline family). A similar conclusion is arrived at by comparing retention volume with Ku(H20). In this experiment a column of resin is used, a single amine added, and the volume of eluting agent needed to reach the elution peak maxima is recorded. Since several amines, using 80% acetonitrile as eluting agent, were used, the retention volumes could be correlated to K,(H20). The importance of structure in influencing the retention becomes obvious by comparing the KD's of the anilines to representative compounds of other classes of weak organic bases. These data for a variety of N-heterocylic and amide compounds are shown in Table I1 in ethanol-water mixtures. It is obvious that no general relationship between KU(H20)and KD is possible when considering all the different structures. As a guideline, it should be anticipated that KO for derivatives of the parent compounds in Table 11 will change according to how the substituents affect the basicity of the
60 1180 32.5
0
0
2.3
H-Form Resin
5.6 4.9
14.5 5.8
0
PPt 24.9
10.3
ppt
~~
Organic Bases on A-15, H-Form, Resin 50
40
8.6 0.5
>105 > 10'
364 339 140 45.1 21.2 92.0
30 10.3 2.90
20
65.6 27.5
PPt
0
PPt 43.8
17.2
> 106 >lo'
>10'
470 204 89 7 86.7 513
10 12.3 13.1
> 10' > 10' > 10'
575 326 464 264 PPt
compound. Thus, if the substituent increases the basicity, KD will be larger while if a decrease takes place KO will be a smaller value. If the actual Ku's are known, it should be possible to even qualitatively predict a KOrange for the derivative. If all of the data in Table I and I1 are plotted, a set of curves are obtained which are very similar. The KD value is very large in organic solvent, drops sharply as small amounts of water are added, passes through a minimum at about 80% ethanol, and then rises again as the concentration of water increases. This shaped curve was also observed for other solvents (11). A graph showing some of these curves is shown in Figure 1. The reason for the behavior in above 80 organic solvent was discussed earlier. In less than 80% organic solvent, it has been suggested that solubilization is the primary reason for weak base retention ( 2 , I I , 19). Several experiments were carried out to illustrate this and will only be briefly cited here. (19) W. Rieman, J . Chem. Educ., 38, 338 (1961).
ANALYTICAL CHEMISTRY, VOL. 43, NO. 12,OCTOBER 1971
Table 111. Solubility of Several Weak Organic Bases in Ethanol-Water Solvent, % Compound ethanol KD S, m g / d Acetamide 20 11.4 740 50 8.6 760 250 1400 Pyrazine 20 1390 140 50 190 720 Quinoxaline 20 41 870 50 8 5.46 Acetanilide 20 0.5 84 50 Phenazine 20 2000 0.27 50 90 5.5 p-Nitroanilineatb 20 523 1 .OO 50 55 10.0 a Solubility in 0, 10, 30, 70, 80, 90% ethanol is 0.525, 0.660, 1.92, 26.4, 35.7, and 41.5 mg/ml, respectively. * Solubility of m-nitroaniline in 100% water and ethanol is 1.14 and 50 mg/ml, respectively, as compared to 0.8 and 40 mg/ml for p-nitroaniline. The importance of the resin being in the H-form is readily established by using instead Na-form resin. When this was done, retention of the weak base (PNA or MNA) was greatly reduced in all proportions of ethanol-water. These data are listed in Table I. This experiment also suggests that the acid-base interaction is important in solutions containing even large water concentrations. If a weak carboxylic acid resin in the H-form (Amberlite IRC-50) is used, only slight retention of P N A is found ( K D < 1) in 100% ethanol. As the water concentration in the mixture increases, the amine will tend to become less soluble. However, the resin in the H-form, provides a n acidic site which coupled with the reduced solubility causes the amine to seek out the most desirable interaction or a n interaction with the resin. This is established by measurement of the solubility of several of the compounds in ethanol-water mixtures. These data are shown in Table 111. As the solubility increases, the K O decreases. In order for the slopes to be comparable, the solubility data are plotted in Figure 1 as 1jS. If the KO curve is compared to the 1jS curve, it is apparent that a change in KD parallels, qualitatively, the change in solubility. For those compounds whose KO changes rapidly the 1jS follows the same pattern (compounds 2, 8, and 9 in Figure 1). In contrast where there is a gradual change in K D , there is also similar behavior for 1jS (compounds 3,6, and 7 in Figure 1). The influence of an acid-base interaction in the solubilization portion is demonstrated by comparing the solubility of PNA and M N A and the KD for these compounds o n A-15 (H+) and A-15 (Na+). MNA, the stronger base, has a higher K D than PNA for A-15 (H+). However, for the N a form, PNA has the larger K D . Since no acid-base interaction is possible for the Na-form resin, solubility becomes the dominate factor. Upon comparing the solubilities, PNA is less soluble (therefore a larger KDon N a form) than MNA.
Per- c e n t
ethanol
Figure 1. Distribution coefficients us. ethanol on A-15 Hform resin and comparison to the weak base solubility 1. o-Nitroaniline 2.
3. 4. 5.
6. 7.
Acetanilide Acetamide 4-Methyl-2-nitroaniline N-Methyl-4-nitroaniline Pyrazine Quinoxaline
p-Nitroaniline Phenazine 2-Aminobenzoicacid 4-Aminobenzoic acid 2-Methyl-4-nitroaniline m-Nitroaniline 14. Purine
8. 9. 10. 11. 12. 13.
As the KO increases, it becomes more difficult to elute the base from the column in sharp well-defined bands in a reasonable volume of eluting agent. Thus, even the weak bases such as m-nitroaniline or purine, pK, of 2.47 and 2.30, respectively, would be difficult to elute. An extension of the general’separation method to stronger bases is accomplished by using a solvent with basic properties in the eluting mixture. If two organic solvents are used, for example DMF-ethanol mixtures, the KO’s should be between the values found for the pure solvents. To illustrate the K D change in DMF-ethanol, data for several amines are listed in Table IV. To illustrate the use of a more basic eluting mixture K D data in DMSO-water mixtures were collected for several sulfonamides which are stronger bases than many of the aniline derivatives. These data are plotted in Figure 2. Data for two anilines are included. Thus, the two solvent mixtures, DMSO-H20 and ethanol-H20 can be compared by considering the aniline data in Figure 2 and Table I, respectively. Retention of the sulfonamides by the H-form resin can not be represented by Reaction 1. On the basis of the pKb values for the aromatic amine group which are all about 12, it would be predicted that all the sulfonamides would give
Table IV. Distribution Coefficients for Several Organic Bases in Dimethylformamide-Ethanol Mixtures on A-15 Resin DMF-WOH, Compound 90-10 80-20 60-40 40-60 20-30 10-90 0-100 N,N-Dimethyl-4nitroaniline 0 0 0 0 0 1.8 112 p-Nitroaniline 0 0 0 1.9 8.97 25.9 354 rn-Nitroaniline 12.5 19.1 36.4 96.0 348 882.0 8340
ANALYTICAL CHEMISTRY, VOL. 43, NO. 12, OCTOBER 1971
*
1589
n
I
C
C .c
/
E
C
C
c
a
Chart distance Figure 2. Distribution, coefficients for sulfonamide derivatives us. % dimethyl sulfoxide on A-15 H-form resin 1. 2-Methyl-5-nitroaniline
Figure 3. Actual tracings for elution of (a) 2,ddichloro4-nitroaniline, (6) N,N-dimethyl-p-nitroaniline,and (c) pnitroaniline on 1 X 32 cm column of 100-200 mesh, A-15, H-form resin using 70 CH,CN
2. 5-Nitro-1-naphthylamine
3. Sulfabenzamide (4.57p 4. Sulfacetamide (5.38) 5. Sulfadiazine(6.48) 6. Sulfathiazole (7.12) 7. Sulfamerazine (7.06) 8. Sulfamethazine (7.37) 9. Sulfanilamide (2.02) 10. Sulfapyridine (8.43) 11. Sulfaguanidine (11.2) “PK, (HzO)
the same K D k Experimentally, this is not observed. However, the sulfonamides also have an acidic group and its pK, will depend on the nature of the R group. As the molecules become more acidic, it should exist to a larger extent in a zwitterion state, or
Y This equilibrium will effectively reduce the availability of the basic site for retention. Thus, KD can be qualitatively correlated to pKa and in Figure 2 it is observed that the stronger the acid, the lower the K O . Separations. From the KD data, it is possible to predict the eluting conditions for a particular separation. In many cases, several different eluting mixtures can be used with equal success One experimental parameter which proved to be of interest was the flow rate. In order to establish an optimum flow rate, a mixture of two amines was separated as a function of eluting agent flow rate. It was observed that rapid flow, up to 34 ml/min, could be used. Faster flow rates were not tried because of an unfavorable pressure build-up across the column. To illustrate the elution behavior, the elution curve for the separation of three anilines as a function of flow rate is shown in Figure 3. 1590
0
ml‘minI
ml/min 0.1 in./min chart speed 15.9 ml/min 0.5 in./min chart speed E. 33.5 ml/min B.. A C.
Up to the fastest flow rate used, separation is still complete. Obviously, the time needed for the entire separation is greatly reduced (35 min for A and 8 min for E in Figure 3). Equilibrium times for the base-resin interaction are moderately fast (6). Apparently, this permits the application of such fast flow rates. When HCI-water-acetone mixtures were used for the separation of Fe(II1) and Cu(II), it was not possible to resolve the mixture at flow rates above 5 ml/min (20). Not so obvious in Figure 3, but expected, is an increase in peak broadening with increased flow rate. Even though broadening occurs, the elution peaks retain their well-defined appearance and no change in retention volume was observed. Whether these fast flow rates can be used o r not for a separation will depend on the closeness of the elution peaks. If they are close together, broadening results in peak overlap. The number of theoretical plates for a given column was calculated by the equation
N
=
16 ( R , / w ) ~
(4)
where R, is the retention volume of the peak and W is the peak width at its base. If N i s plotted us. flow rate, Figure 4, the extent of peak broadening is illustrated as a function of flow rate. The greatest change in the number of plates occurs as the flow rate approaches 10 ml/min. Thereafter, the change is relatively small. In general, acetonitrile-water provides more plates than ethanol-water mixtures. Also, there are (20) J. S. Fritz and T. A.
ANALYTICAL CHEMISTRY, VOL. 43, NO. 12, OCTOBER 1971
Rettig, ANAL.CHEM., 34, 1562 (1962).
Table V.
Sample 1. 2.
3.
4. 5. 6. 7.
Data for the Separation of Several Weak Organic Base Mixtures of A-15 Resin
Mixture" o-Nitroanilinec p-Nitroaniline o-Nitroaniline p-Nitroaniline o-Nitroaniline p-Nitroaniline 2,4-Dinitroaniline o-Nitroaniline 2,4-Dinitroaniline o-Nitroaniline 2,4-Dinitroaniline o-Nitroaniline 2,6-Dichloro-4-nitroaniline
N,N-Dimethyl-4-nitroaniline 8.
N-Methyl-4-nitroanilinec
p-Nitroaniline 9.
2,6-Dichloro-4-nitroanilinec
o-Nitroaniline 4-Methyl-2-nitroaniline N,N-Dimethyl-4-nitroaniline
R,,ml ...
Column height, cma 10
Flow rate, ml/min 1 to 20
13.2
3.4
80% Ethanol or 80% Acetonitrile 80% Ethanol
13.2
3.4
80% Ethanol
13
3.0
100% Ethanol
53
3.9
100% Ethanol
10
1 to 20
100% Ethanol
13
7.4
95 % Acetonitrile
10
0.57
80% Ethanol
43
0.73
80% Acetonitrile
.,.
19 107 15 111 11 38 18 59 13 55 9 19
...
Eluting agent
...
27 32 44 66
Added, mg 0.054 0.078 24.4 0.13 0.22 25.0 0.14 27.3 0.189 0.186 0.0268 0.0596 0.295 0.141 1.08 1.14 0.769 0.897 0.775 0.858 0.846 0.224 0,0775 0.463 0.243 0.240 3.65 2.07 2.13 2.37 0.476 0.483 0.472 0.480 0.471 0.480 0.436 0.413 0.323 0.341 0.412
Found, mg 0,051 0.078 24.6d 0.15 0.29 26.3 0.15 27.4 0.189 0.187 0.0345 0.0545 0.280 0.144 1.19 1.21 0.779" 0.905e 0.778 0.890 0.865 0.216 0,766 0.453 0.232 0.230 3.93 2.33 2.08 2.39 0.466 0.476 0,491 0.464 0,475 0.492 0.462 0.420 0.340 0.324 0.405
p-Nitroaniline 188 10. 2,4-Dinitroaniline 14 6.5 80% Ethanol 18 p-Nitroaniline 150 80% Ethanol m-Nitroaniline 90% DMSO 318 11. N,N-Dimethyl-4-nitroaniline 15 4.6 14 80:20 Ethano1:DMF 65 p-Nitroaniline 11 1.5 12. Acetamidec 10 80% Ethanol 47 Acetanilide 17 13. 15 1 .o 50% Ethanol Acetamide Benzotriazole 130 13 14. Sulfadiazine 13 8.2 90% DMSO 225 Sulfaguanidine 13 15. Sulfanilamide 21 90% DMSO 8.3 223 Sulfaguanidine 13 16. 65% DMSO 3.0 16 Sulfabenzarnide 65 Sulfamerazine 17. 13 Sulfathiazole 2.9 23 80% DMSO 73 Sulfapyridine 18. Sulfadiazine 1.7 50 40 70% DMSO Sulfamerizine 107 304 Sulfamethazine a Order listed is the order of elution. * Column diameters are 1.1 cm except sample 5 which is 0.6 cm. Separation was also done on D-50 resin. Separation was also done at 50: 1 ratios. e Beoause of peak overlap, the analysis of the two fractions was done by simultaneous equations using absorbances at two wavelengths.
more plates for 100% ethanol than for 80% ethanol and for smaller resin particles. More plates are found for the more porous A-15 resin than for the D-50 resin. This should be expected since the A-15 resin should be more suitable for organic eluting agents. It can also be demonstrated by calculation from elution curves that resolution is better with the A-15 resin. Distribution data were also measured for many of the organic bases on D-50 resin but are not reported here. Frequently, the KO values for a given base are similar for the two resins. Although the A-15 resin is better, the D-50 resin can also be used for many of the separations. Several different synthetic mixtures of weak organic bases
were separated in order to illustrate the scope of the eluting conditions. A summary of these separations and the experimental conditions used are listed in Table V. Typical elution curves are shown in Figures 3 and 5. Separations were shown to be quantitative at several levels of concentration. Ratios of 250 :1 for binary mixtures were successfully separated. In general the flow rates in Table V are optimum flow rates based on column dimensions, pressure build-up, band resolution, and concentration levels. Often different eluting mixtures can be used for the separation of the same mixture. By switching eluting agents, an earlier elution of the next component is possible. For example, the five compound mixture of sulfabenzamide, sulfacet-
ANALYTICAL CHEMISTRY, VOL. 43, NO. 12, OCTOBER 1971
1591
x
8 U
c .Q .> c
188 ml E=
0
e
+
a
19
51 Volume collected, ml
34
67
Figure 5. Separation of a five-component aniline mixture using 80 CHICN(SeeSample 9 in Table V) A . 2,6-DichloroQnitroaniline B. o-Nitroaniline C. 4-Methyl-2-nitroaniline
\ *I-
\
20
1
I I I I 20 IO Flow r a t e , mllmin
I
I
30
Figure 4. Number of theoretical plates as a function of flow rate for the separation of o-nitroaniline and pnitroaniline (1,2,5-8), 2,4dinitroaniline and o-nitroaniline (3), and 2,6dichloro4nitroaniline, N,N-dimethyl-p-nitroaniline,and pnitroaniline (4). (The second peak is used for calculation of N ; resolution as a function of flow rate is given for 3) 1. D-50; 100-200 mesh; 80% EtOH; 11.5 X 1 cm 2. A-15; 100-200 mesh; 80% EtOH; 11.5 X 1 cm 3. A-15; 100-200 mesh; 100% EtOH; 11.5 X 1 cm 4. A-15; 100-200 mesh; 70% CHXN; 32 X I cm 5. A-15; 40-60 mesh; 80% EtOH; 11.5 X 1 cm 6 . A-15; 40-6Omesh; 80%CHaCN; 11.5 X l c m 7. D-50; 100-200 mesh; 80% CHKN; 11.5 X 1 cm 8. A-15; 100-200 mesh; 80% CHBCN; 11.5 X 1 cm
amide, sulfadiazine, sulfamerazine, and sulfapyridine was successfully separated using as eluting agents 40 %, 52 %, 64 %, 77 %, and 90 DMSO, respectively. Since there was interest in recording retention volumes, the second and third component 111 many cases in Table V could be eluted faster by 3 a n g i n g the eluting mixture. AI: the analyses were done by collecting each band, diluting t ; toiurne, and determining the affine content by spectropnotometry. No attempt was made to do gradient elution.
7.592
a
ANALYTICAL CHEMISTRY, VCL. 4s.
No.
D. N,N-Dimethyl-pnitroaniline E. pNitroaniline
The KD values in 80% ethanol for acetamide and benzotriazole are about the same, 14.5 and 12.5, respectively, and separation is not possible with this eluting mixture. Because of their different solubilities, plots of KD us. ethanol are not parallel, see Figure 1. Thus, at 5 0 z ethanol, the KD value for acetamide is much less than for benzotriazole (8.5 and 24, respectively) and the separation (Sample 13) is possible with this mixture. A variety of synthetic sulfonamide mixtures were separated. Sample 18 represents a common pharmaceutical combination. The separation was also tried on two different commercial brands of tri-sulfa tablets. After tablet work-up, the separation characteristics were similar to the synthetic mixture and quantitative results were obtained. The significance of this elution scheme is illustrated by considering the following advantages. Compounds that are separated are often in a solvent mixture that is convenient for recrystallization. No electrolyte is present in the eluting agent. This is important if purification is the primary goal of the separation. The column of H-form resin can be used over and over again since it needs only to be washed with a suitable solvent; resin form will not be changed during the separation. The separation scheme should be useful for all types of very weak and moderately weak bases. It should be a useful purification technique where removal of organic bases from the sample is required. Qualitatively basic strengths may be estimated for a particular compound on the basis of its elution order. The potential of fast flow rates is present.
RECEIVED for review Maich 29, 1971. Accepted June 21, 1971. Presented to the Analytical Division, 160th National iueeting, American Chemical Society, September 1970, Chicago, Ill. The authors gratefully acknowledge the financial support cf NDEA and National Institute of Health (6M-15851).
12, OCTOBER 1971
