- Email: [email protected]
Determination of reducing and non-reducing carbohydrates in food products by liquid chromatography with post-column catalytic hydrolysis and derivatization comparison with refractive index detection
Journal of Chromatography, 402 (1987) 127-134 Elsevier Science Publishers B.V., Amsterdam - Printed in The Netherlands CHROM. 19 620
DETERMINATION OF REDUCING AND NON-REDUCING CARBOHYDRATES IN FOOD PRODUCTS BY LIQUID CHROMATOGRAPHY WITH POST-COLUMN CATALYTIC HYDROLYSIS AND DERIVATIZATION COMPARISON
WITH REFRACTIVE
INDEX DETECTION
ROBERT A. FEMIA* and ROBERT WEINBERGER ABI Analytical, fiatos Division, 170 Williams Drive, Ramsey, NJ 07446 (U.S.A.) (First received October 2Oth, 1986; revised manuscript received April 2nd, 1987)
SUMMARY
Post-column catalytic hydrolysis combined with 4-aminobenzoic acid hydrazide derivatization is employed for the determination of both reducing and nonreducing carbohydrates in a variety of complex sample matrices such as dairy products, processed foods and tobaccos. Comparison with refractive index detection shows the post-column method to be superior from, the standpoint of selectivity, sensitivity and simplicity of sample preparation. Limits. of, detection care in the lownanogram range. Quantitative results are presented for the determination of sugars in potato extracts. The method is also applicable .for the separation and detection of carbohydrate oligomers.
INTRODUCTION
Separations of simple and polymeric sugars by liquid chromatography (LC) on bonded-phase columns such as amine, C 1s, cation-exchange and anion-exchange resins have been well characterized. These chromatographic schemes can include elution with acetonitrile/water mobile phases on an amine modified silica column’, elution with water on a counterion loaded cation-exchange resit?+, or elution with slightly basic borate buffer on an anion-exchange resin’+*. Carbohydrate oligomers are well separated on C 18 columns with a totally aqueous eluentg. On the other hand, carbohydrate detection is nowhere as sophisticated. While refractive index (RI) detection works well for determining sugars at high concentrations, the poor sensitivity and selectivity is often limiting in the sub-pg range and for complex sample matrices. The same holds true for low ultraviolet (UV) detection at 190 nm. These techniques seem best suited for the determination of high levels of carbohydrates in samples such as sodas and some fruit juice products. Various schemes aimed at increasing the sensitivity and selectivity of carbohydrate analysis have been attempted through the use of pre- and post-column de0021-9673/87/$03.50
0
1987 Elsevier Science Publishers B.V.
128
R. A. FEMIA, R. WEINBERGER
rivatization. Pre-column tagging with dansyl hydrazine has been reportedlO but the reaction is quite slow. The same holds true for pm-column silylation carried out prior to *gas chromatography i l. : On the other hand, many postcolumn derivatizations procedures: do not detect non-reducing carbohydrates, employ corrosive reagents, aggressive temperatures, or exhibit non-uniform response towards various carbohydrates7-8s1 2-1 5. Some recent work by Vratny et al2 illustrated the use of a catalytic reactor to invert non-reducing sugars to their reducing components. This feature, coupled with the 4-aminobenzoic acid hydrazide (ABH) reagent forms a facile, sensitive, and selective means of determining sugars in complex matrices. The present work elaborates on this technology with further optimization and additional applications. Comparisons are made with RI detection for the determination of simple sugars in complex samples. Applications for the determination of oligosaccharrides are presented along with quantitative results for the determination of simple sugars in various potato extracts. EXPERIMENTAL
A block diagram of the entire system is illustrated in Fig. 1. I 8 LC system j A Spectroflow 400 solvent delivery system (ABI Analytical, Kratos Division, Ramsey, NJ, U.S.A.) equipped with a high-sensitivity membrane pulse dampener (part No. 90014001) and a Rheodyne (Berkeley, CA, U.S.A.) No. 7126 injection valve,tiasused. Injections of 10 ,ul were used for most samples. For the post-column reaction pumps, a Spectroflow 400 equipped with l/16” checkvalves (part No. 90024001) and a low-pressure pulse dampener (part No. 7200-0282) was used. Chromatograms were recorded on either a DS 610 strip chart recorder (Kratos) or a DS 650 chromatography data system (Kratos). Columns and mobile phases Three different columns were used in these studies, all with aqueous mobile phases. For carbohydrate oligomers, a Phase Separations (Norwalk, CT, U.S.A.) Cis 250 x 4.6 mm I.D., S5 0DS2 column was used. The flow-rate was 1 ml/min and the column was maintained at 25°C. Temperature control of the column and the subsequent post-column reaction was provided by a PCRS 520 (ABI Analytical).
Fig. 1. Instrument configuration for the determination of carbohydrates by LC with catalytic hydrolysis and ABH postcolumn detection.
LC OF CARBOHYDRATES IN FOODS
129
For simple sugars requiring a sucrose-lactose separation, a Kratos carbohydrate column in the Pb’ + ionic form, maintained at 85°C was used. The flow-rate was 0.4 ml/min. Other sugar separations were carried out on a Kratos carbohydrate column in the Ca2+ form, also maintained at 85°C. The flow-rate was 0.7 ml/min. A guard column (Bio-Rad., Richmond, CA, U.S.A.) in the H+ form was used to protect the ion-exchange columns. Catalytic reactor A carbohydrate CHEMphaseTM reactor (ABI Analytical, part No. 2900-0322) maintained at 85°C was used for sugar inversion when required (sucrose, raflinose, etc.) Post-column reaction conditions The reagent was prepared as follows: Solution A, 2.5 g of ABH, finely pulverized was transferred to a loo-ml volumetric flask. The 2.46 ml concentrated hydrochloric acid (J. T. Baker) was added and diluted to the mark with water and sonicated until dissolved. The reagent is stable for 3 days refrigerated. Solution B, 9.6 g sodium hydroxide (Fisher) and 1.46 g sodium chloride (Baker) were added to a 200-ml volumetric flask and diluted to the mark with water and sonicated until dissolved. Both solutions were combined and filtered through a 0.45-pm filter and sparged with helium to form the post-column reagent. A cooling chamber (Fisher Thermoflask) was used to maintain the temperature of this regent between 0 and 5°C to stabilize the reagent. A l-ml reaction coil (ABI Analytical) maintained at 1WC was used for the post-column reaction. The post-column reagent flow-rate was 0.5 ml/min. ‘ Detection For the post-column reaction system, a Spectroflow 757 (GBI Analytical), equipped with a tungsten halogen lamp and set to 410 nm was used. The rise time was 2 s. An RI detector (Varian RI-3) set at 16 - lo-+ RI units was used for the comparative studies.
Sample preparation Beverages. A ten-fold dilution followed by filtration through a 0.45~pm filter is adequate. 1 Cereah~hbgiains. A l-g pulverized sample was blended with 100 ml water for 10 min and &en filtered. The’saine’procedure can be employed for tobacco though pulverization is generally not required. Fatty samples. Extraction was carried out with hexane-water (50:50). The aqueous layer was filtered. Dairy samples. These were diluted lOO-fold and filtered. RI detection. For RI detection, it is often necessary to work with sample concentrations that are 10-100 times higher than for the post-column system. Since many food samples contain high protein levels, a pre-column deproteinization step is necessary to preserve the analytical column. Five ml of sample was mixed with 5 ml of
R. A. FEMIA, R. WEINBERGER
b
Fig. 2. Separation of five dairy sugars. Column: Pb2+ cation exchange, column temperature: 85X!, mobile phase: water, flow-rate: 0.4 ml/mm. Key: a = sucrose, 1.7 pg, b = lactose, 6.8 pg, c = glucose, 3.6 pg, d = galactose, 3.6 erg and e = fructose, 3.6 pg. Detection: ABH post-column chemistry at 410 mn, range: 0.5 a.u.f.s. Fig. 3. Comparison of ABH post-column and RI detection for peanut butter samples. Separation and post-column conditions as in Fig. 2. RI range: 16 . lO-* RIU. Key: (1) RI detection, sample concentration, 100 mg/ml, (2) same sample as in (1) with post-column detection, sample concentration 10 &ml, (3) another brand of peanut butter with post-column detection. a = Raffinose, b = sucrose, c = lactose, d = glucose, e = galactose, f = fructose. Sample preparation: fatty samples, see text. ‘I t !I, ;: 1 ;,*:,‘I “,!I_,‘:
5% aqueous trichloroacetic acid, shaken for 5 min and filtered. Samples with high levels of calcium or other cations were cleaned up off-line with a solid-phase cation-exchange column. RESULTS AND DISCUSSION
Reagent stabilization The ruggedness of the method was tested with regard to the stability of the ABH reagent. A mixture of sucrose and glucose (1 nmole/pl) was injected 20 times over a 10-h period with the reagent maintained at room temperature. Linear regression of the measured signals showed a response decay of 0.925%/h for sucrose and 0.562%/h for glucose. This was attributed to reagent decomposition. By cooling the ABH reagent to 0-5°C the response decay rates were lowered to 0.018%/h and 0.002%/h, respectively. Cooling the reagent eliminates the need for frequent restandardization. Comparison with RI detection The inherently poor sensitivity of RI detection requires a high sample concentration or a large injection volume if sub-pg detectability is required. The separation illustrated in Fig. 2 for five standard dairy sugars is from a lo-p1 injection containing
LCOF CARBOHYDRATES IN FOODS
131
1.7 pg sucrose, 6.8 pg lactose, and 3.6 pg each of glucose, galactose and fructose. A limit of detection of less than 20 ng is found for all of these sugars under the given chromatographic conditions and ABH post-column derivatization with W detection. Such high sensitivity is often not the issue for food samples. However, for the majority of these samples, the high protein content requires a deproteinization step to be performed prior to injection. This is usually accomplished with either 5% trichloroacetic acid (TCA) or acetonitrile followed by centrifugation and filtration. With the post-column scheme described here, the sensitivity of the method allows for a dilution factor of at least ten-fold compared to RI detection, without compromising performance. This obviates the need for removing the proteins since a cation-exchange guard column has ample capacity for on-line clean-up of the diluted sample. In this regard, sample preparation is simplified and column lifetime is probably prolonged. The issue of selectivity provides an even greater rationale for applying postcolumn technology to sugar analysis. Specifically, interferences from sugar alcohols, starches, amino acids and small peptides may be possible depending on the precise conditions of analysis. As illustrated in Figs. 3 and 4, for peanut butter and tobacco samples, respectively, substantial interferences are found using RI and yet with the post-column system, the sugars are clearly resolved. Similar results were found for fatty samples such as bacon and sausage. For all of the samples used in this study, none were found to require deproteinization when using post-column ABH detection. For highly proteinaceous food products, the sample can either be diluted or treated with 5% TCA. Shown in Fig.. 5 are several chromatograms of a low lactose infant feeding formula. Comparisons of Fig. 5 (TCA pre-treatment) and (simple dilution) shows no discernable differences between the two methods of sample pre-treatment. A blank TCA clean-up produced
d d
I
s
min
I WI
Fig. 4. Comparison of ABH post-column and RI detection for tobacco samples. Separation, concentration and post-column conditions as in Fig. 2. Key: (1) RI detection of pipe tobacco, (2) post-column detection of pipe tobacco, (3) post-column detection of cigarette tobacco. a = Sucrose, b = lactose, c = glucose, d = fructose. Sample preparation: cereals and grains, see text.
R. A. FEMIA, R. WEINBERCER
132
1
I
8)
min
Fig. 5. Comparison of ABH post-column and RI detection for low lactose infant feeding formula. Column: Bio-Rad HPX-87P, column temperature, 85”C, flow-rate: 0.4 ml/mm, injection size: 20 fl. Key: (1) RI detection, sample concentration, 100 mg/ml, TCA protein precipitation, (2) post-column detection, sample concentration, 10 mg/ml, TCA protein precipitation, (3) post-column detection, sample concentration, 10 mg/ml, no protein precipitation. a = Sucrose, b = lactose, c = glucose, d = galactose, e = fructose. ..
nq peaks by the post-Column method. Thus the method is compatible with existing sample pre-treatment methods. Similar results were found for acetonitrile deproteinization. ..I Carbohydrate oligomers
Oligomers can be separated on either a 4% cross-linked Na+ cation-exchange Me_%=
44.2dJ I
99_
:
8e_ 7e_ 6e_ 5e_ 4e 38 28 18 !
:
e&_,
1~““~~“,‘~“~(~~‘1’~“““‘1~“““~~1”~
3.75
7.51
11.25
is.18
18.75
Fig. 6. Separation of carbohydrate ohgomers in high-fructose corn syrup with post-column detection. Total carbohydrate content: 750 mg/g, concentration: 10 mg/g, concentration: 10 mg/ml, injection size: 20 4. Column: Cls, flow-rate: 1 ml/mitt. Key: 1 = fructose, 2 = degree of polymerization (DP) 2,3 = DP 3 anomers, 4 = DP 4 anomers, 5 = DP 5 anomers, 6 = DP 6 anomers, 7 = DP 7 anomers, 8 = DP 8 anomers.
LC. OF CARBOHYDRATES
IN FOODS
133
TABLE I SUGAR CONTENT OF SELECTED POTATO TYPES Sugar content (mgjg) I&h0
sllerose Glucose Fructose l
1.07 0.94 0.58
Red
1.89 6.59 5.22
Maine
0.47 1.45 1.28
Sample* PCRS
GC
0.31 0.09 0.02
0.30 0.07 0.03
Proprietary potato sample.
column or by reversed-phase chromatography with a C1s column. Both separations employ an aqueous mobile phase which is compatible with the ABH chemistry. The order of elution of the oligomers switches between these two chromatographic modes; the high-molecular-weight compounds eluting last by reversed-phase chromatography. Since these oligomers have reducing functionality, the catalytic CHEMphase reactor is not required. This separation is illustrated in Fig. 6 for an injection of 200 pg of corn syrup. It is possible to measure both anomers of up to DP 10 in less then 30 minQ. Anomers are not resolved with the cation-exchange column. Potatoes The sugar content of several types of store bought potatoes was determined by the post-column method and is summarized in Table I. These results are influenced by the age of the potato after harvest. Also reported in Table I are the analytical results for sugar determinations of a proprietary sample by both the post-column and a gas chromatographic method ll. The results were comparable and were within the experimental errors of both procedures. The relative standard deviation of the post-column method is approximately 2% at the I-pg level (n = 5). REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14
N. Watanabe, J. Chromatogr., 330 (1985) 333. P. Vartny, U. A. Th. Brinkman and R. W. Frei, Anaf. Chem., 57 (1984) 224. 0. D. Bonn, J. Chromatogr., 322 (1985) 411. L. A. Th. Verhaar and B. F. M. Kuster, J. Chromutogr., 210 (1981) 279. J. Schmidt, M. John and C. Wandrey, J. Chromatogr., 213 (1981) 151. M. Sinner and J. Puls, J. Chromatogr., 156 (1978) 197. K. Mopper, R. Dawson, G. Liebezelt and H. P. Hansen, Anal. Chem., 52 (1980) 2018. S. Honda, Y. Matsuda, M. Takahashi, K. Kakehi and S. Ganno, Anal. Chem., 52 (1980) 1079. E. Rajakylii, J. Chromtugr., 353 (1986) 1. K. Blau and G. King, Handbook of Derivativesfor Chromatograpky, Heyden, London, 1978, pp. 6 and 350. G. E. Bacon and S. Kokenakes, J. Lob. C/in. Med., 73 (1969) 1030. R. B. Kessler, Anal. Chem.. 39 (1967) 1416. M. Sinner and J. Puls, Ger.‘O&n., 2,657,516 (Cl. GOlN31/08) June (1978), Appl., Dec. (1976), 26 pp. M. H. Simatupang and H. H. Dietrichs, Chromatographia. 11 (1978) 89.
134
K. A. CCMIA,
K. ,WWE~BEKtiEK
15 H. Kikami and Y. Ishida, 10th International Symposium on Column Liquid Chromatography, San Francisco, CA, May 18-23.1986, Paper No. 3506. 16 P. G. Koski, L. Elrod and R. E. Shoup, 10th International Symposium on Column Liquid Chiomatography, San Francisco, CA, May 18-23.1986, Paper No. 3502. 17 J. R. Benson and D. J. Wo, J. Chromutogr. Sci., 22 (1984) 386.
... .
‘
DETERMINATION OF REDUCING AND NON-REDUCING CARBOHYDRATES IN FOOD PRODUCTS BY LIQUID CHROMATOGRAPHY WITH POST-COLUMN CATALYTIC HYDROLYSIS AND DERIVATIZATION COMPARISON
WITH REFRACTIVE
INDEX DETECTION
ROBERT A. FEMIA* and ROBERT WEINBERGER ABI Analytical, fiatos Division, 170 Williams Drive, Ramsey, NJ 07446 (U.S.A.) (First received October 2Oth, 1986; revised manuscript received April 2nd, 1987)
SUMMARY
Post-column catalytic hydrolysis combined with 4-aminobenzoic acid hydrazide derivatization is employed for the determination of both reducing and nonreducing carbohydrates in a variety of complex sample matrices such as dairy products, processed foods and tobaccos. Comparison with refractive index detection shows the post-column method to be superior from, the standpoint of selectivity, sensitivity and simplicity of sample preparation. Limits. of, detection care in the lownanogram range. Quantitative results are presented for the determination of sugars in potato extracts. The method is also applicable .for the separation and detection of carbohydrate oligomers.
INTRODUCTION
Separations of simple and polymeric sugars by liquid chromatography (LC) on bonded-phase columns such as amine, C 1s, cation-exchange and anion-exchange resins have been well characterized. These chromatographic schemes can include elution with acetonitrile/water mobile phases on an amine modified silica column’, elution with water on a counterion loaded cation-exchange resit?+, or elution with slightly basic borate buffer on an anion-exchange resin’+*. Carbohydrate oligomers are well separated on C 18 columns with a totally aqueous eluentg. On the other hand, carbohydrate detection is nowhere as sophisticated. While refractive index (RI) detection works well for determining sugars at high concentrations, the poor sensitivity and selectivity is often limiting in the sub-pg range and for complex sample matrices. The same holds true for low ultraviolet (UV) detection at 190 nm. These techniques seem best suited for the determination of high levels of carbohydrates in samples such as sodas and some fruit juice products. Various schemes aimed at increasing the sensitivity and selectivity of carbohydrate analysis have been attempted through the use of pre- and post-column de0021-9673/87/$03.50
0
1987 Elsevier Science Publishers B.V.
128
R. A. FEMIA, R. WEINBERGER
rivatization. Pre-column tagging with dansyl hydrazine has been reportedlO but the reaction is quite slow. The same holds true for pm-column silylation carried out prior to *gas chromatography i l. : On the other hand, many postcolumn derivatizations procedures: do not detect non-reducing carbohydrates, employ corrosive reagents, aggressive temperatures, or exhibit non-uniform response towards various carbohydrates7-8s1 2-1 5. Some recent work by Vratny et al2 illustrated the use of a catalytic reactor to invert non-reducing sugars to their reducing components. This feature, coupled with the 4-aminobenzoic acid hydrazide (ABH) reagent forms a facile, sensitive, and selective means of determining sugars in complex matrices. The present work elaborates on this technology with further optimization and additional applications. Comparisons are made with RI detection for the determination of simple sugars in complex samples. Applications for the determination of oligosaccharrides are presented along with quantitative results for the determination of simple sugars in various potato extracts. EXPERIMENTAL
A block diagram of the entire system is illustrated in Fig. 1. I 8 LC system j A Spectroflow 400 solvent delivery system (ABI Analytical, Kratos Division, Ramsey, NJ, U.S.A.) equipped with a high-sensitivity membrane pulse dampener (part No. 90014001) and a Rheodyne (Berkeley, CA, U.S.A.) No. 7126 injection valve,tiasused. Injections of 10 ,ul were used for most samples. For the post-column reaction pumps, a Spectroflow 400 equipped with l/16” checkvalves (part No. 90024001) and a low-pressure pulse dampener (part No. 7200-0282) was used. Chromatograms were recorded on either a DS 610 strip chart recorder (Kratos) or a DS 650 chromatography data system (Kratos). Columns and mobile phases Three different columns were used in these studies, all with aqueous mobile phases. For carbohydrate oligomers, a Phase Separations (Norwalk, CT, U.S.A.) Cis 250 x 4.6 mm I.D., S5 0DS2 column was used. The flow-rate was 1 ml/min and the column was maintained at 25°C. Temperature control of the column and the subsequent post-column reaction was provided by a PCRS 520 (ABI Analytical).
Fig. 1. Instrument configuration for the determination of carbohydrates by LC with catalytic hydrolysis and ABH postcolumn detection.
LC OF CARBOHYDRATES IN FOODS
129
For simple sugars requiring a sucrose-lactose separation, a Kratos carbohydrate column in the Pb’ + ionic form, maintained at 85°C was used. The flow-rate was 0.4 ml/min. Other sugar separations were carried out on a Kratos carbohydrate column in the Ca2+ form, also maintained at 85°C. The flow-rate was 0.7 ml/min. A guard column (Bio-Rad., Richmond, CA, U.S.A.) in the H+ form was used to protect the ion-exchange columns. Catalytic reactor A carbohydrate CHEMphaseTM reactor (ABI Analytical, part No. 2900-0322) maintained at 85°C was used for sugar inversion when required (sucrose, raflinose, etc.) Post-column reaction conditions The reagent was prepared as follows: Solution A, 2.5 g of ABH, finely pulverized was transferred to a loo-ml volumetric flask. The 2.46 ml concentrated hydrochloric acid (J. T. Baker) was added and diluted to the mark with water and sonicated until dissolved. The reagent is stable for 3 days refrigerated. Solution B, 9.6 g sodium hydroxide (Fisher) and 1.46 g sodium chloride (Baker) were added to a 200-ml volumetric flask and diluted to the mark with water and sonicated until dissolved. Both solutions were combined and filtered through a 0.45-pm filter and sparged with helium to form the post-column reagent. A cooling chamber (Fisher Thermoflask) was used to maintain the temperature of this regent between 0 and 5°C to stabilize the reagent. A l-ml reaction coil (ABI Analytical) maintained at 1WC was used for the post-column reaction. The post-column reagent flow-rate was 0.5 ml/min. ‘ Detection For the post-column reaction system, a Spectroflow 757 (GBI Analytical), equipped with a tungsten halogen lamp and set to 410 nm was used. The rise time was 2 s. An RI detector (Varian RI-3) set at 16 - lo-+ RI units was used for the comparative studies.
Sample preparation Beverages. A ten-fold dilution followed by filtration through a 0.45~pm filter is adequate. 1 Cereah~hbgiains. A l-g pulverized sample was blended with 100 ml water for 10 min and &en filtered. The’saine’procedure can be employed for tobacco though pulverization is generally not required. Fatty samples. Extraction was carried out with hexane-water (50:50). The aqueous layer was filtered. Dairy samples. These were diluted lOO-fold and filtered. RI detection. For RI detection, it is often necessary to work with sample concentrations that are 10-100 times higher than for the post-column system. Since many food samples contain high protein levels, a pre-column deproteinization step is necessary to preserve the analytical column. Five ml of sample was mixed with 5 ml of
R. A. FEMIA, R. WEINBERGER
b
Fig. 2. Separation of five dairy sugars. Column: Pb2+ cation exchange, column temperature: 85X!, mobile phase: water, flow-rate: 0.4 ml/mm. Key: a = sucrose, 1.7 pg, b = lactose, 6.8 pg, c = glucose, 3.6 pg, d = galactose, 3.6 erg and e = fructose, 3.6 pg. Detection: ABH post-column chemistry at 410 mn, range: 0.5 a.u.f.s. Fig. 3. Comparison of ABH post-column and RI detection for peanut butter samples. Separation and post-column conditions as in Fig. 2. RI range: 16 . lO-* RIU. Key: (1) RI detection, sample concentration, 100 mg/ml, (2) same sample as in (1) with post-column detection, sample concentration 10 &ml, (3) another brand of peanut butter with post-column detection. a = Raffinose, b = sucrose, c = lactose, d = glucose, e = galactose, f = fructose. Sample preparation: fatty samples, see text. ‘I t !I, ;: 1 ;,*:,‘I “,!I_,‘:
5% aqueous trichloroacetic acid, shaken for 5 min and filtered. Samples with high levels of calcium or other cations were cleaned up off-line with a solid-phase cation-exchange column. RESULTS AND DISCUSSION
Reagent stabilization The ruggedness of the method was tested with regard to the stability of the ABH reagent. A mixture of sucrose and glucose (1 nmole/pl) was injected 20 times over a 10-h period with the reagent maintained at room temperature. Linear regression of the measured signals showed a response decay of 0.925%/h for sucrose and 0.562%/h for glucose. This was attributed to reagent decomposition. By cooling the ABH reagent to 0-5°C the response decay rates were lowered to 0.018%/h and 0.002%/h, respectively. Cooling the reagent eliminates the need for frequent restandardization. Comparison with RI detection The inherently poor sensitivity of RI detection requires a high sample concentration or a large injection volume if sub-pg detectability is required. The separation illustrated in Fig. 2 for five standard dairy sugars is from a lo-p1 injection containing
LCOF CARBOHYDRATES IN FOODS
131
1.7 pg sucrose, 6.8 pg lactose, and 3.6 pg each of glucose, galactose and fructose. A limit of detection of less than 20 ng is found for all of these sugars under the given chromatographic conditions and ABH post-column derivatization with W detection. Such high sensitivity is often not the issue for food samples. However, for the majority of these samples, the high protein content requires a deproteinization step to be performed prior to injection. This is usually accomplished with either 5% trichloroacetic acid (TCA) or acetonitrile followed by centrifugation and filtration. With the post-column scheme described here, the sensitivity of the method allows for a dilution factor of at least ten-fold compared to RI detection, without compromising performance. This obviates the need for removing the proteins since a cation-exchange guard column has ample capacity for on-line clean-up of the diluted sample. In this regard, sample preparation is simplified and column lifetime is probably prolonged. The issue of selectivity provides an even greater rationale for applying postcolumn technology to sugar analysis. Specifically, interferences from sugar alcohols, starches, amino acids and small peptides may be possible depending on the precise conditions of analysis. As illustrated in Figs. 3 and 4, for peanut butter and tobacco samples, respectively, substantial interferences are found using RI and yet with the post-column system, the sugars are clearly resolved. Similar results were found for fatty samples such as bacon and sausage. For all of the samples used in this study, none were found to require deproteinization when using post-column ABH detection. For highly proteinaceous food products, the sample can either be diluted or treated with 5% TCA. Shown in Fig.. 5 are several chromatograms of a low lactose infant feeding formula. Comparisons of Fig. 5 (TCA pre-treatment) and (simple dilution) shows no discernable differences between the two methods of sample pre-treatment. A blank TCA clean-up produced
d d
I
s
min
I WI
Fig. 4. Comparison of ABH post-column and RI detection for tobacco samples. Separation, concentration and post-column conditions as in Fig. 2. Key: (1) RI detection of pipe tobacco, (2) post-column detection of pipe tobacco, (3) post-column detection of cigarette tobacco. a = Sucrose, b = lactose, c = glucose, d = fructose. Sample preparation: cereals and grains, see text.
R. A. FEMIA, R. WEINBERCER
132
1
I
8)
min
Fig. 5. Comparison of ABH post-column and RI detection for low lactose infant feeding formula. Column: Bio-Rad HPX-87P, column temperature, 85”C, flow-rate: 0.4 ml/mm, injection size: 20 fl. Key: (1) RI detection, sample concentration, 100 mg/ml, TCA protein precipitation, (2) post-column detection, sample concentration, 10 mg/ml, TCA protein precipitation, (3) post-column detection, sample concentration, 10 mg/ml, no protein precipitation. a = Sucrose, b = lactose, c = glucose, d = galactose, e = fructose. ..
nq peaks by the post-Column method. Thus the method is compatible with existing sample pre-treatment methods. Similar results were found for acetonitrile deproteinization. ..I Carbohydrate oligomers
Oligomers can be separated on either a 4% cross-linked Na+ cation-exchange Me_%=
44.2dJ I
99_
:
8e_ 7e_ 6e_ 5e_ 4e 38 28 18 !
:
e&_,
1~““~~“,‘~“~(~~‘1’~“““‘1~“““~~1”~
3.75
7.51
11.25
is.18
18.75
Fig. 6. Separation of carbohydrate ohgomers in high-fructose corn syrup with post-column detection. Total carbohydrate content: 750 mg/g, concentration: 10 mg/g, concentration: 10 mg/ml, injection size: 20 4. Column: Cls, flow-rate: 1 ml/mitt. Key: 1 = fructose, 2 = degree of polymerization (DP) 2,3 = DP 3 anomers, 4 = DP 4 anomers, 5 = DP 5 anomers, 6 = DP 6 anomers, 7 = DP 7 anomers, 8 = DP 8 anomers.
LC. OF CARBOHYDRATES
IN FOODS
133
TABLE I SUGAR CONTENT OF SELECTED POTATO TYPES Sugar content (mgjg) I&h0
sllerose Glucose Fructose l
1.07 0.94 0.58
Red
1.89 6.59 5.22
Maine
0.47 1.45 1.28
Sample* PCRS
GC
0.31 0.09 0.02
0.30 0.07 0.03
Proprietary potato sample.
column or by reversed-phase chromatography with a C1s column. Both separations employ an aqueous mobile phase which is compatible with the ABH chemistry. The order of elution of the oligomers switches between these two chromatographic modes; the high-molecular-weight compounds eluting last by reversed-phase chromatography. Since these oligomers have reducing functionality, the catalytic CHEMphase reactor is not required. This separation is illustrated in Fig. 6 for an injection of 200 pg of corn syrup. It is possible to measure both anomers of up to DP 10 in less then 30 minQ. Anomers are not resolved with the cation-exchange column. Potatoes The sugar content of several types of store bought potatoes was determined by the post-column method and is summarized in Table I. These results are influenced by the age of the potato after harvest. Also reported in Table I are the analytical results for sugar determinations of a proprietary sample by both the post-column and a gas chromatographic method ll. The results were comparable and were within the experimental errors of both procedures. The relative standard deviation of the post-column method is approximately 2% at the I-pg level (n = 5). REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 14
N. Watanabe, J. Chromatogr., 330 (1985) 333. P. Vartny, U. A. Th. Brinkman and R. W. Frei, Anaf. Chem., 57 (1984) 224. 0. D. Bonn, J. Chromatogr., 322 (1985) 411. L. A. Th. Verhaar and B. F. M. Kuster, J. Chromutogr., 210 (1981) 279. J. Schmidt, M. John and C. Wandrey, J. Chromatogr., 213 (1981) 151. M. Sinner and J. Puls, J. Chromatogr., 156 (1978) 197. K. Mopper, R. Dawson, G. Liebezelt and H. P. Hansen, Anal. Chem., 52 (1980) 2018. S. Honda, Y. Matsuda, M. Takahashi, K. Kakehi and S. Ganno, Anal. Chem., 52 (1980) 1079. E. Rajakylii, J. Chromtugr., 353 (1986) 1. K. Blau and G. King, Handbook of Derivativesfor Chromatograpky, Heyden, London, 1978, pp. 6 and 350. G. E. Bacon and S. Kokenakes, J. Lob. C/in. Med., 73 (1969) 1030. R. B. Kessler, Anal. Chem.. 39 (1967) 1416. M. Sinner and J. Puls, Ger.‘O&n., 2,657,516 (Cl. GOlN31/08) June (1978), Appl., Dec. (1976), 26 pp. M. H. Simatupang and H. H. Dietrichs, Chromatographia. 11 (1978) 89.
134
K. A. CCMIA,
K. ,WWE~BEKtiEK
15 H. Kikami and Y. Ishida, 10th International Symposium on Column Liquid Chromatography, San Francisco, CA, May 18-23.1986, Paper No. 3506. 16 P. G. Koski, L. Elrod and R. E. Shoup, 10th International Symposium on Column Liquid Chiomatography, San Francisco, CA, May 18-23.1986, Paper No. 3502. 17 J. R. Benson and D. J. Wo, J. Chromutogr. Sci., 22 (1984) 386.
... .
‘