Determination of Sorbitol as its Hexacetate by Gas Liquid Chromatography Using an Ionization Detector JAMES A. HAUSE, JOSEPH A. HUBICKI, and GEORGE G. HAZEN Process Developmenf Laboratories, Merck Chemical Division, Merck & Co., lnc., Danville, Pa. Until now, a simple reliable method for the quantitative determination of sorbitol in mixtures has not been available. Gas chromatography of sorbitol hexacetate using an ionization detector provides such an assay. Detector response i s variable and not linear, but use of mannitol as an internal standard minimizes this deficiency. Using this technique the error can be reduced to less than 1%. The assay could be easily adapted to the simultaneous determination of mannitol and sorbitol by use of an alternate internal standard. The assay has been applied to natural products and materials to which sorbitol has been added, Procedures for acetylating and preparing column packing are given in detail. The latter i s quite important. Minor deviations often cause tailing and poor resolution.
S
has come into wide use as a vehicle, humectant, and chemical intermediate. -1s its area of application increases so does the need for a n accurate and specific assay. At least two techniques are currently available to provide separation and quantification of sorbitol in the presence of other polyols. One (8) involves chromatographic separation of the borate complexes over a strong anion exchange resin. The eluates are analyzed b y oxidation to formaldehyde and measurement of the colored chromotropic acid complex. Another (4, 6) entails partition over a Florex XXX column, sectioning the column into zones by streaking 1%-ithalkaline permanganate. elution, and titration of the eluates indometrically after treatment with periodate. These techniques are time consuming, involve comples manipulation, and are prone to error. Recently Horning ( 7 ) rrported the separation of the heyitol polyacetates by gas chromatography using the polar fluoralkylsilicone polymer QF-1 (FS1265Dow Corning Corp.) as the liquid phase. This work has been extended b y Eiposito and Swann ( 1 ) to other polyhydric alcohols using Carbowax colurnns. The extraordinary sensitivity of the ionization detector coupled rvith thc superb selectility of the new polar phases provides the basis of the assay procedure d i i c h we have devised. ORBITOL
EXPERIMENTAL
Reagents and Apparatus. All chemicals were N e r c k Reagent Grade unless otherwise specified. The analyses !\-ere performed on a BarberColnian Model 15 high temperature gas chromatograph. Argon was the carrier gas. Flow rates were measured b y a flowmeter a t t h e effluent of t h e column after conditioning and before making t h e connection t o t h e detector. T h e pressure was regulated t o approximately 35 p.s.i.g. t o maintain a flow rate of about 100 ml. per minute. A Lovelock ionization detector containing a 56-pc. radium-226 foil source was used. The columns were glass U-tubes, 6 feet long with 6mm. i.d. Column Packing. T h e support was G a s Chrom P, 100 to 140 mesh, a n d a solution technique was used t o apply a 1% coating of the fluoralkyl silicone polymer QF-I-0065-10,000 cs. supplied b y t h e Dow Corning Corp. This material is now called FS 1265. T h e support, liquid phase and, if desired, t h e coated support can be purchased from the Applied Science Laboratories, Inc., 140 North Barnard St., State College, Pa., or can be made by techniques described in the Procedures section. Chromatography. T h e column was conditioned at 260" C. for 16 hours with a n argon flow rate of 50 ml. per minute. T h e column was not connected to the detector during this period. When running analyses the column was maintained a t 217' C., detector bath 270' C., and the flash heater a t 325' C. Cell voltage was maintained a t 1000 volts. A relative gain of 10 (1 x 10-7 amp.) was employed. Acetone solutions of the samples were injected with a lO-pI. Hamilton syringe (Hamilton Co., Khittier, Calif.). Sample size ranged from 0.5 to 2.0 pl, and was adjusted to give from one- to tn-o-thirds full scale deflection. Procedures. SAMPLE ACETYLATIOX. A . A n h y d r o u s Polyols. .An accurately weighed (2 to 3.5 grams) sample of solid polyol was refluved for 1 hour with a mixture of 30 ml. of pyridine and 30 nil. of acetic anhydride. The mixture was cooled t o 2 5 O C., quantitativeIj- transferred t o a volumetric flask, and diluted to 1000 nil. with acetone. B. Aqueous Solutions of Polyols. An amount of aqueous polyol solution equivalent to 2 to 3.5 grams of sorbitol was accurately weighed into a flask. Sufficient pyridine to remove the
aiiiount of TI ater prewnt n a- adtled and the mixture IT as distilled atniospherically to a volume of about 30 ml. The vapor temperature reached 115' C. indicating that all of the water was remo\ ed. *icetic anhydride (30 ml.) was added and the system v a s heated a t reflux for 1 hour. The mixture was cooled to 25' C.,quantitatilcly transferred to a volumetric flask and diluted to 1000 ml. with acetone. Removal of the bulk of the awtic anhydride and pyridine by vacuum distillation prior to dilution eliminates solvent tailing which may interfere n i t h peaks appearing early in the chromatogram. C. Reference Standards. Pure hexitol hexacetates were made from Fisher Reagent sorbitol and mannitol using Procedure -4, After treatment with acetic anhydride the solvents were removed by evaporation in vacuo. The crystalline mass was dissolved in chloroform and the resultant solution was washed well n i t h nater. The chloroform solution was dried over sodium sulfate and concentrated to a thick slurry in vacuo. The hevacetates were recrystallized to a constant melting point (sorbitol 99' to 100' C., mannitol 124" to 125' C.) from ethyl acetate. COLUMNPACKIKG. The procedures used in the preparation of the paching are essentially those of Horiiing ( 3 developed a t the Sational Institutes of Health. The techniques I\ ere modified to some extent. TTe include complete details because the techniques are highly critical and minor deviations result in unsatisfactory paching. A. Preparation of S u p p o r t . .Ilthough Gas Chrom P is acid and alkali mashed in its manufacture. n e prefer to repeat this treatment. Some color 1' rtmovecl especially by the acid extraction. Sizing. The support was resized t o 100 t,o 110 mesh using the appropriate
screens. Sizing after coating might he an improvement; howwer, the possibility of producing bare spots which might he deleterious must be considered. Acid Kashing. About 55 grams of sized support was added to :I beaker and overla)-ed t o a depth of 2 or 3 inches with concentrated h>-drochloricacid. The mixture was swirled occasionally during a &minute period. T h e liquid was removed using a sintered glass filter stick. The washing operation was repeated until the supernatant acid ivas free of color ( 5 to 6 times). The support was then n-ashed free of acid by decantation using distilled water. The final xash was removed by filtration through R sintered glass funnel. The filtrr cake x i s washed VOL. 34, NO. 12, N O V E M B E R 1962
1567
MANNITOL HEXACETATE
ENTAERYTHRITOL TETRAPROPIONATE
,MANNITOL HEXACETATE
Fiaure 1. SeDaration of sorbitol and mannitol hexacetaies on a FS 1 265 column
with 3 X 200-ml. portions of acetone, spread out on a filter paper, and air-dried a t 60" C. Base Washing. The dried acid washed support was gently stirred for 15 minutes with 400 ml. of Reagent or Spectrograde methanol made 0.5N with potassium hydroxide pellets. The support was collected by filtration, washed free of base with methanol, and air dried on filter paper at 60" C. The dried support was reclassified by flotation using methanol and decanting the fines. The unfragmented portion was collected by filtration and redried as above. Coating. QF-1 (0.52 gram) was dissolved in 100 ml. of redistilled methylethylketone at 25' C. Sixteen grams of the washed and dried support was added slowly t o the QF-1 solution with swirling. After 2 minutes' slvirling, the slurry was poured rapidly into a sintered glass filter, applying just enough vacuum t o keep the funnel from overflowing but not so much that the cake became dry before the transfer was complete. ilir was drawn
Table 1.
Mannitol Sorbitol hexhexacetate acetate concn. concn. (g./l.) (g./l.)
Separation of polyol esterson a
through the cake until no more foam appeared on the lower side of the sintered filter. The cake, a damp powder, was dried in air at 60" C. to constant weight. The amount of liquid phase deposited on the support can be approximated by evaporating the filtrate in a tared flask. The flask is then heated in vacuo a t 100' C. to constant weight. Various lots of support exhibit varying degrees of adsorbtivity toward the liquid phase. The concentration of the QF-1 solution can be varied to obtain the 1% coating. Some success has been achieved in applying the phase by an evaporative technique. Low percentages of liquid phase (0.5 to 1.5%) can be successfully applied in this manner if the new siliconized supports ( b ) are used. To apply mixed phases (2)' this procedure is quite necessary. RESULTS A N D DISCUSSION
The development of gas liquid chromatographic separation of high molecu-
Determination of Proportionality Constant
Mannitol peak height imm.)
Run I Sorbitol peak height (mm.1
MH/SX k,:t
34 0 57 0 79 5 101 0 113 5
28.0 47.0 66.0 82.5 93.0
1.21 1.21 1.20 1.22 1.22
1 2
3
Figure 2.
Mannitol peak height (mm.1 24.0 54.0 85.0 121 0
173 5 225 5 298.0 354.0
421.0
(kIz1)for Run I1 Sorbitol peak height (mm.1 20.0 43.5 69.0 97.0 138.0
183.0 239.0 284.0 340.0
ANALYTICAL CHEMISTRY
265
lar weight compounds has been accelerated by the introduction of highly sensitive ionization detectors. These devices allow detection of samples in the range of 1 to 20 pg. Separation of such small samples can be achieved on packed columns containing relatively small liquid phase coatings (0.5 to 1.5%). Thin coatings allow successful operation a t lower temperatures with relatively shorter retention times than when a thicker coating is used. The lower temperatures decrease the degree of thermal degradation of the substance being chromatographed and the liquid phase. K e have had columns in continuous service for 6 months with no sign of deterioration. Mannitol and sorbitol hexacetates separate cleanly on QF-1 using the conditions described
Mannitol and Sorbitol Hexacetates
MH/SH ki:i 1.20 1.24 1.23 1.25 1 26 1 23 1.25 1.25 1.24
From 1 to 5 grams per liter sample size was 2 pl., and peak heights were measured directly. From 6 to 9 grams per liter sample size was 1 pl., and peak height was multiplied by 2.
1568
Fs
Mannitol peak height (mm.) 24.0 58,O
102.0 139,O 190.0 257.0 340.0 360.0 474.0
Run I11 Sorbitol peak height (mm.) 20 0 48 0 82 5 110 0 150 5 206 0 272 0 287 0 380.0
;WHISH ki:i 1.20 1.21 1.24 1.26 1.26 1 25 1 25 1.25 1.25
under Chromatography. The retention times differ by 2 minutes (Figure 1). The ionization characteristics of the hexacetates were investigated b y determining whether the observed component peak heights and areas were proportional to their concentration in synthetic mixtures. Pure mannitol and sorbitol hexacetates were prepared as described above and mixed in equal weights from 1 t o 9 grams per liter. The mixtures were chromatographed. peak height of each component was measured and related to its concentration. h relationship exists but i t is not linear. I n this work, no improvement in linearity or reproducibility mas noted when the areas under the curves were related to concentrations. The daily variation of the slopes of standard curves were not in steady descending order which would h a r e been indicative of liquid phase degradation. I n every instance the slope of the mannitol curve was greater than that of the sorbitol curve indicating slightly greater detector response for mannitol hexacetate. Response increased with concentration becoming most linear in the 5 to 10 grams per liter range. Comparison to knomns furnishes the basis for an approximate assay. Cse of a n internal standard improves the quantization. The ratio of the peak height of mannitol hexacetate and sorbitol hexacetate was quite consistent when 1 to 1 mixtures were injected in concentration ranges of 4 to 10 grams per liter. Because our need was for an assay for sorbitol in the absence of mannitol or with exceedingly small concentrations of mannitol i t appeared that mannitol could be used as a n internal standard for the precise determination of sorbitol. This avoids errors due to size of sample injected and nonlinear detector response. The proportionality constant ( k l mas determined. The data are reported in Table I. Proportionality constant kl ' 1 = Peak ht. mannitol hexacetate - KH Peak ht. sorbitol hexacetate - S H when concentration of mannitol hesacetate = concentration of sorbitol hexacetate. As can be seen from Table I there is fair agreement in the determined values of ICl over the whole range of concentration (1 to 10 grams per liter). Repeated injections of duplicate samples in the range of 4 to 10 grams per liter showed deiintions of 1%. Larger deviations result when several days or weeks separate the analysis, as was true between Runs I, 11, 111. Thus IC1 must be determined each time conditions are changed. At lower concentrations of 1 or 2 grams per liter the deviation increases to about 5%.
10 knowns each. The results are recorded in Table 111. A t 4 to 5 grams per liter the accuracy is best (0.6 to 1.0 yo error). This can be expected because the mannitol and sorbitol concentrations are equal or close to equal. The conversion of sorbitol to its hexacetate appears to range from 92.5 t o 95.3 yo of theory using the procedure reported above. Sorbitol was acetylated at several concentrations and mannitol hexacetate was introduced a t the same level. The ratio of peak heights (M=/SLI)was then determined. These were compared to the ratio of peak heights when equal quantities of the pure hexacetates were mixed at the proper concentrations. The per cent conversion is reported in Table IV. Work is continuing to examine alternate internal standards which can be used for sorbitol and mannitol mixtures. Pentaerythritol tetrapropionate is cleanly separated from sorbitol and mannitol hexacetates (Figure 2), and is being investigated further.
Table II. Ratio of Peak Heights at Constant Mannitol Composition and Varying Sorbitol Content Mannitol concn.,
4
no
Ratio peak heights, mannitol/ sorbitol 9.23 4.25 2.38 1 72
5 7 9 10
00 00 00 00
0 813 0 583 0 49s
Sorbitol concn., g.L
1 .oo 2.00
5,oo
5.00
3.00
5 00
5.00 5.00 5 00 5 00 5 00
i .23
T o obtain the greatest precision a preliminary assay is run. i l n amount of mannitol approximately equal to the sorbitol present is added to a separate sample and the mixture is acetylated and diluted to the range of 4 to 10 grams per liter. A known using equal amounts of mannitol and sorbitol is run to determine kl:l (the ratio of peak heights at equal concentrations). The following equation is used to calculate the concentrations of sorbitol: Concentration of sorbitol = k1:1
(observed peak ht. sorbitol) (concn. of mannitol) -. , [SI observed peak ht. mannitol
If many samples of various but similar compositions are t o be analyzed, i t is possible to save time at some sacrifice in accuracy. A given amount of mannitol is added to all samples. Because detector response is nonlinear. IC determined for a l to l mixture is not valid. However, determination of k a t 1 to 2 grams per liter intervals in the region of 1 to 10 grams per liter sorbitol concentration allows sufficiently accurate interpolations to be made to form the basis of assay. This is illustrated below. Knowns were prepared containing 5 to 1, 5 to 2. 5 to 3, 5 to 4,5 to 5, 5 to 7 . 5 to 9 and 5 to 10 grams per liter of mannitol and sorbitol converted to their hexacetates. The ratios of peak heights determined for each concentration are recorded in Table 11. The concentration of sorbitol us. the ratio of peak heights from Table I1 were plotted and the grams per liter of sorbitol calculated from two sets of
Table IV. Sorbit a1 hexacetate theoretically present
k2.A.) L
4 6 8
= k1:1
SH [MI MH
APPLICATION
Since the development of the assay i t has been put t o practical application in the determination of sorbitol in coconut. Five hundred and twenty-five milligrams of mannitol was added t o 50
Table 111. Sorbitol Found vs. Sorbitol Present in Known Solutions. Series I Series I1 Sorbitol Sorbitol Sorbitol present, found, % found, % g./l. g./l. Error g./l. Error 0.97 - 3 . 0 1 . 0 0 0 . 9 8 -2.0 2.00 2 . 0 5 $ 2 . 5 1.96 - 2 . 0 3.00 2.95 -1.7 2.93 - 2 . 3 4.00 4 . 0 3 + 0 . 8 3.97 - 0 . 8 5.00 4.95 - 1 . 0 4.97 - 0 . 6 6.00 6.07 + 1 . 6 6.10 + 1 . 7 7.00 7.10 1 1 . 4 7.10 + i . 4 800 815 f 1 3 810 +13 9 00 8 90 -0 9 -1 1 8 92 10 00 9 75 -2 5 9 75 -2 5
Percentage Conversion of Sorbitol to Its Hexacetate Mannitol hexacetate added as standard W1.) 2
4 6 8
A'fH/SH
1.28
1.34 1.30 1.33
MH/SH from pure hexacetates 1.22 1.24 1.24 1.24
Per cent acet,ylated 95.3 92.5 95.3 93.2
VOL. 34, NO. 12, NOVEMBER 1962
1569
grams of shredded coconut. The mixture was extracted with 3 X 150 ml. of water. The extract was dried by azeotropic distillation with pyridine and acetylated as described above. Gas chromatography followed b y the appropriate calculation revealed that the coconut meat contained 0.5 % sorbitol. By elimination of the internal standard it was shown that mannitol, if present, is in such a lorn concentration that it does not interfere with the accuracy. I n another experiment, additional sorbitol was added and the determination repeated. The results are recorded in Table T’. The Chromatography must be performed the same day as the acetylation because products appear which interfere with the mannitol hexacetate peak introduced as the internal standard.
LITERATURE CITED
Table
V.
Run Controla Controla
Determination of Sorbitol in Shredded Coconut
Sorbitol added Sorbitol Difference (5) found (ci) 0 00 0 00 1 00 1 00
0 50 0 50
1 50 0 00 1 52 -0 02 Control experiments were performed on fresh meat from different coconuts. (5
(1) Esposito, G. G., Smann, 11. H., AYAL.CHEM.33, 1854 (1961). (2) Haahti, E. 0. A4.,\-anden Heuval, 11.J. A , , Homing, E. C., d n a l . Bzochenz.
2,344-52 (1961).
(3) Homing. E. C., Moscatelli, E. -I., Sweeley, C. C., C h e m . I n d . ( L o n d o n ) 1959, 751. ( 4 ) Lew, B. 1Y. Kolfram, 11.L., Goepp, R. M.,J . A n i . ?hem. Soc. 6 8 , 144953 (1946). ( 5 ) Sjovall, J., Meloni, C R , Turner, I). *4.> J . Lzpid Res. 2 , 31’7-20 (1961,. (6) The Pharmacopeia of the United States of bmerica, Sixteenth Revision, pp. 692-3, hlack Printing Co , Easton,
P a , 1960.
ACKNOWLEDGMENT
( 7 ) 1-anden Heuval, JY. J. ;i,Homing, E. C Bzochem. Bzophys Res. Communs. ~
TT’e are indehted to Robert De Valeria
who contributed to various phases of the work and to Franklin Baker Coconut, General Foods Corp., for the generous supply of coconut.
4,399-403 (1961). (8) Zill, L. P., Khrm, \-, S.,Cheneae, G. ll,,J . A m . Chem. Soc. 75, 1339-42 (1953). RECEIYEDfor review May 7, 1962. Arcepted August 15, 1962.
Chromatographic Separation of Peptides on Ion Exchange Resins Separation of Peptides from Enzymatic Hydrolyzates of the a, and 7 Chains of Human Hemoglobins
p,
W. A. SCHROEDER, RICHARD T. JONES, JEAN CORMICK, and KATHLEEN McCALLA California lnsfifufe of Technology, Pasadena, Cc /if.
b Isolation of components in a mixture is most effectively accomplished by great alteration of conditions from step to step in the separation. This principle has been applied to the separation of complex mixtures of peptides in protein hydrolyzates. Excellent results have been obtained by chromatographing first on the cation exchanger Dowex-50 and then rechromatographing on the anion exchanger Dowex-1. Volatile developers were employed throughout. Experimental procedures are presented, and results are given of their application to the separation of peptides from human hemoglobins A and F.
r
r
HE 5UCCLSSFUL DETERMIXATIOK O f
the amino acid sequence in a protrin depends much upon the adequacy with JT hich the peptides, however obtained, may be separated. Unless the amount of protein available is small and p a p u chromatographic methods must 1~ used, the separation of peptides by column chromatography in amounts wfficient for complete characterization is clearly the method of choice. Although peptides may be adequately separated in many inPtances (17) by
1570 *
ANALYTICAL CHEMISTRY
sodium citrate or sodium acetate buffers of the type that are used in the analytical determination of amino acids by ion exchange chromatography ( I I ) , the product from pooled fractions is essentially only buffer salts that are contaminated with the peptide. The separation of the desired peptide from the buffer salts has been achieved in various ways. If the peptide is dinitrophenylated in the presence of the salts, it may usually be extracted from the solution (17). This procedure suffers from the disadvantage that the free peptide is no longer available for the determination of structure by step15 ise degradation. Other procedures have also been used. For example, peptides have been separated (19) by means of ammonium formate or acetate buffers which were then removed by wblimation ( 5 ) . A q an alternatile, the peptides have been separated \{ ith sodium-containing buffers which w r e converted to ammonium salts by passage through a column in the ammonium form. The ammonium salts were then sublimed (6). .Ilthough effective, these procedures are laborious and time consuming. The recent application of volatile orqanic developers has been a most sig-
nificant step in the separation of peptides on an ion exchange chromatogram, Vanecek et. al. (go), Margoliash and Smith ( I O ) , and Kimmel et al. (8) have described the separation of peptides by means of Do~vex-50 v i t h pyridineacetic acid developers, \Thereas Rudloff and Braunitzer (14) have used columns of Domex-1 with pyridine-collidineacetic acid developers. The separation of peptides from hydrolyzates of CY, p , and y chains of human hemoglobins A and F has been achieved by the use of both Dowex-1 and Doirex-50. I n the earliest experiments, the peptides were separated first on Dowex-50 and further purified by paper electrophoresis or paper chromatography. The purification by paper methods, in general, was unsatisfactory and has been superseded by rechromatography on Don-ex-1 ; paper methods always resulted in losses of 50 to 75% of the material and usually gave a product that still contained impurities. The use of Dowe.t-50 and Dowex-1 in succession (the reverse would probably be equally effective) takes advantage of the fact that vastly different conditions prevail in the two types of chromatography because one is a cation and the other an anion ex-
