Anal. Chem. 1997, 69, 4286-4290
Hyphenation of Ion Exchange High-Performance Liquid Chromatography with Fourier Transform Infrared Detection for the Determination of Sugars in Nonalcoholic Beverages Richard Vonach, Bernhard Lendl, and Robert Kellner*
Institute for Analytical Chemistry, Vienna University of Technology, A-1060 Vienna, Austria
Fourier transform infrared spectroscopy (FT-IR) is presented here as a molecular-specific detection system for high-performance liquid chromatography (HPLC) in an aqueous phase, focusing on the chromatographic separation of sugars in beverages. The separation was achieved with an isocratic HPLC setup using an ion exchange column (counterion, Ca2+). The FT-IR detection of the C-O bands in the mid-IR between 1000 and 1200 cm-1 was performed in real time with a 25 µm flow cell without elimination of the solvent. Characteristic FT-IR spectra of the common sugars sucrose, glucose, and fructose in concentrations of 1 mg/mL could be recorded during the separation. The calibration of these compounds in the 5-100 mg/mL range resulted in a linear correlation with a standard deviation of the method (sx0) of 0.11, 0.07, and 0.11 mg/mL for sucrose, glucose, and fructose, respectively. The method was, furthermore, applied to the analysis of nine soft drinks and fruit juices containing between 6 and 97 mg/mL of each carbohydrate. The accuracy of the method was confirmed by standard ion exchange HPLC with refractive index detection. The average deviation from the reference method was in the range of 0.5-0.9 mg/mL. Furthermore, the method was found to be suitable to identify and quantify also minor components in beverages, such as taurine (4 mg/mL) and ethanol (0.4 mg/mL). The use of Fourier transform infrared spectroscopy (FT-IR) as detection system in high-performance liquid chromatography (HPLC) is of particular interest in analytical chemistry. This is primarily due to the lack of identification capabilities of conventional HPLC detectors. If the analyte is not UV absorbing and mass spectrometry (MS) cannot be applied as a detection system, vibrational spectroscopy and, in particular, FT-IR spectroscopy is the method of choice for multidimensional and molecular-specific detection. Especially the fingerprint region around 1400-900 cm-1 is of certain interest for the identification of molecules, since most organic (and many inorganic) compounds exhibit strong and narrow absorption bands providing a high degree of qualitative and structural information. Among several concerns, the most important, accounting for the relatively poor performance of HPLC-FT-IR, is the IR opacity of the mobile phase.1,2 To overcome this problem, the hyphenation of HPLC and FT-IR is typically performed by two different (1) Fujimoto, C.; Jinno, K. Anal. Chem. 1992, 64, 476A-481A.
4286 Analytical Chemistry, Vol. 69, No. 20, October 15, 1997
approaches. The first one utilizes a standard flow cell as the interface. The two IR-transparent windows are separated by a Teflon or lead spacer, providing a short optical path length which is generally between 10 and 200 µm, dependent on the transparency of the mobile phase. If short path lengths of 20 µm or less are required, there are, in a few cases, also internal reflectance flow cells in use. Typically, cylindrical ZnSe rods serve as an attenuated total reflectance (ATR) element.3-5 Spectral overlap between solvent and analyte can be minimized by using deuterated mobile phases. Of course, they are only cost-effectively applied in microbore HPLC.6,7 Reversed-phase (RP) and size-exclusion chromatography (SEC) separations in the low-microgram range have been reported so far using CD3CN/D2O and D2O mobile phases.8 The second approach for coupling LC and FT-IR is based on the elimination of the mobile phase. This involves an interface that evaporates the eluent and deposits the analytes onto a medium compatible with FT-IR detection. With this method, solvent-free IR spectra of separated analytes in the nanogram range can be obtained. In the case of aqueous eluents, typically used in RPLC and in the described application of ion exchange LC, the low volatility of water hampers a rapid solvent elimination. Sophisticated interfaces are necessary to enhance the solvent evaporation power. However, the reduction of the eluent flow is often required, using microbore columns instead of analyticalscale columns. The direct elimination of aqueous solvents has been done by mixing nitrogen into the carrier flow9 or using a concentric flow nebulizer10-12 and by ultrasonic13 or electrospray14 nebulization. Other authors have employed a particle beam (2) Griffiths, P. R.; Pentoney, S. L., Jr.; Pariente, G. L.; Norton, K. L. Mikrochim. Acta 1987, 3, 47-62. (3) Sabo, M.; Gross, J.; Wang, J.; Rosenberg, I. E. Anal. Chem. 1985, 57, 18221826. (4) Rein, A.; Wilks, P. Am. Lab. 1982, 14 (10), 152-154. (5) McKittrick, P. T.; Danielson, N. D.; Katon, J. E. J. Liq. Chromatogr. 1991, 14 (2), 377-393. (6) Fujimoto, C.; Jinno, K. Trends Anal. Chem. 1989, 8, 90-96. (7) Jinno, K.; Fujimoto, C. J. Chromatogr. 1990, 506, 443-460. (8) Fujimoto, C.; Uematsu, G.; Jinno, K. Chromatographia 1985, 20, 112116. (9) Gagel, J. J.; Biemann, K. Anal. Chem. 1987, 59, 1266-1272. (10) Yang, J.; Griffiths, P. R. Proc. SPIE-Int. Soc. Opt. Eng. 1993, 2089, 336337. (11) Lange, A. J.; Griffiths, P. R. Appl. Spectrosc. 1993, 47, 403-410. (12) Lange, A. J.; Griffiths, P. R.; Fraser, D. J. J. Anal. Chem. 1991, 63, 782787. (13) Castles, M. A.; Azarraga, L. V.; Carreira, L. A. Appl. Spectrosc. 1986, 40, 673-680. (14) Raynor, M. W.; Bartle, K. D.; Cook B. W. J. High Resolut. Chromatogr. 1992, 15, 361-366. S0003-2700(97)00307-7 CCC: $14.00
© 1997 American Chemical Society
interface15-18 (originally designed for HPLC-MS) or a thermospray interface.19-21 Another approach for the water elimination is based on postcolumn on-line extraction. After extraction with chlorinated solvents (CH2Cl2, CHCl3, CCl4), various interfaces can be applied, such as regular flow cells,22,23 the deposition of the extracts on KBr powder with subsequent diffuse reflectance infrared detection (DRIFT),24 or, more recently, the deposition on a zinc selenide window by a spray jet assembly, followed by the analysis of the residues with a FT-IR microscope.25 Only very few applications dealing with aqueous phase HPLC and flow cell interfaces were available. The first paper in 1985, described the separation of a mixture of acetophenone, ethyl benzoate, and nitrobenzene (1-2%) with an internal reflectance element (Circle cell).3 In another contribution, structural information of metallocene-amino acid adducts was gained after the separation of 0.7 mg of the analytes.26 The separation of caffeine and theophylline using, again, the Circle cell has also been reported, stating detection limits in the order of 0.2% (0.1 mg).5 However, although real-time recording of FT-IR spectra is proposed, no HPLC-FT-IR traces have been presented in the cited paper. The purpose of this paper is to present an improved flow cellbased HPLC-FT-IR system as a convenient and robust tool for gaining qualitative and quantitative information about carbohydrates in aqueous solution. The content of sucrose, glucose, and fructose in soft drinks and fruit juices can be quantified by HPLCFT-IR, and a comparison with HPLC and refractive index (RI) detection as standard reference method for carbohydrate analysis is given. Finally, the capability of HPLC-FT-IR for the identification and quantification of other IR-active minor components (taurine, ethanol) in these beverages is demonstrated. EXPERIMENTAL SECTION Reagents. Sucrose, glucose, and fructose standard solutions were prepared by dissolving an appropriate amount of each compound (concentration, >99%, for biochemical use, Merck) in distilled water. Industrial samples were drawn from various soft drinks and fruit juices. Real samples were drawn from various commercially available soft drinks or fruit juices. Carbonated samples were degassed in an ultrasonic bath, and all were filtered (0.45 µm) prior to injection. Standards and real samples were stabilized with 50 mg/L NaN3 and stored 48 h before use in order to prevent mutarotational effects on the glucose and fructose (15) Robertson, R. M.; de Haseth, J. A.; Kirk, J. D.; Browner, R. F. Appl. Spestrosc. 1988, 42, 1365-1368. (16) Robertson, R. M.; de Haseth, J. A.; Browner, R. F. Appl. Spestrosc. 1990, 44, 8-13. (17) Turula, V. E.; de Haseth, J. A. Anal. Chem. 1996, 68, 629-638. (18) Turula, V. E.; de Haseth, J. A. Appl. Spectrosc. 1994, 48, 1255-1264. (19) Robertson, A. M.; Wylie, L.; Littlejohn D.; Watling, R. J.; Dowle, C. J. Anal. Proc. 1991, 28, 8-9. (20) Robertson, A. M.; Littlejohn D.; Brown, M.; Dowle, C. J. J. Chromatogr. 1991, 588, 15-24. (21) Robertson, A. M.; Farnan, D.; Littlejohn D.; Brown, M.; Dowle, C. J.; Goodwin, E. Anal. Proc. 1993, 30, 268-271. (22) Johnson, C. C.; Hellgeth, J. W.; Taylor L. T. Anal. Chem. 1985, 57, 610615. (23) Hellgeth, J. W.; Taylor L. T. Anal. Chem. 1987, 59, 295-300. (24) Conroy, C. M.; Griffiths, P. R.; Duff, P. J.; Azarraga, L. V. Anal. Chem. 1984, 56, 2636-2642. (25) Somsen, G. W.; Hooijschuur, E. W. J.; Gooijer C.; Brinkman U. A. Th.; Velthorst, N. H.; Visser, T. Anal. Chem. 1996, 68, 746-752. (26) Tartar, A.; Huvenne, J. P.; Gras, H.; Sergheraert, C. J. Chromatogr. 1984, 298, 521-524.
spectra.27 Distilled and ultrapurified water (18 MΩ) was used as mobile phase. Degassing of the water prior to use was found not to be necessary for keeping a constant solvent flow rate. HPLC. The HPLC system of these studies consisted of a Merck/Hitachi L7100 isocratic pump (flow rate, 0.5 mL/min) and a Rheodyne 7725 injection valve (injection volume, 20, 50, and 100 µL). Furthermore, an in-line filter and a 50 mm × 4.6 mm guard column were employed to protect the 300 mm × 7.8 mm analytical column (Sarasep CAR-Ca, purchased from Inovex GmbH, Vienna, Austria). The stationary phase of both analytical and guard columns was a polystyrene-divinylbenzene anionic exchange resin with Ca2+ counterion (particle size, 8 µm; crosslinkage, 8%). Instead of a column oven, a modified glass tube connected with a thermostated water bath was used to keep the column temperature at 85 °C. All connections were made of poly(ether ether ketone) (PEEK) tubings with 0.25 and 0.17 mm i.d. FT-IR. Infrared measurements were performed on a Bruker IFS 88 FT-IR spectrometer equipped with a liquid nitrogen-cooled narrow-band MCT detector (D* ) 2 × 1010 cm Hz1/2 W-1). A standard Perkin Elmer transmission flow cell with an optical path length of 25 µm and CaF2 windows (of 2 mm thickness each) was applied, yielding an approximately 1/e attenuation (absorbance, ∼0.4) of the water background absorption at 1100 cm-1. A reduction of the cross-sectional area of the flow cell (385 mm2, leading to a cell volume of less than 10 µL) was not necessary, taking the carrier flow of 500 µL/min into account. Special care was taken to reduce the void volume between the flow cell and the PEEK tubing to prevent peak broadening. As we already showed in previous work by our group,28-30 a significant increase of the S/N ratio of FT-IR measurements in the investigated spectral region (1300-900 cm-1) is achievable by the introduction of an InSb low-wave-pass filter (5% cutoff, 1370 cm-1) into the optical setup. By means of this filter, the radiation intensity in the region of interest can easily be increased by opening the aperture without causing oversaturation of the highly sensitive narrow-band MCT detector. Hence, an improved limit of detection is achieved which is accompanied by a reduction of qualitative information, because the spectral range above 1400 cm-1 is no more accessible when using the optical filter. Nevertheless, for the analysis of carbohydrate absorption patterns, as well as for the general investigation of monovalent C-O bondings, the fingerprint range from 900 to 1400 cm-1 is the region of interest. Data Acquisition and Recording. For each spectrum recorded during an HPLC-FT-IR run, 50 scans were coadded. The resolution of all spectra was 4 cm-1, the scanner velocity was set to 80 kHz (HeNe frequency), and Blackman-Harris threeterm apodization was applied. The total acquisition time, including Fourier transformation and further data processing, was 6 s/spectrum, allowing a sufficient time resolution of 10 spectra/min. The standard Bruker GC/LC software was applied for a convenient 3D data treatment and postrun trace calculation. HPLC traces were evaluated according to the absorption of the analytes, choosing the appropriate peak and baseline wavenum(27) Black, D. M.; Michalska, D. F.; Polavaparu, P. L. Appl. Spectrosc. 1984, 38, 173-180. (28) Krieg, P.; Lendl, B.; Vonach R.; Kellner, R. Fresenius’ J. Anal. Chem. 1996, 356, 504-507. (29) Vonach, R.; Kellner, R.; Lippitsch, M. Proc. Euro. Food Chem. 1995, 8, 573-577. (30) Vonach, R.; Lendl, B.; Kellner, R. Analyst 1997, 122, 525-530.
Analytical Chemistry, Vol. 69, No. 20, October 15, 1997
4287
Figure 1. Three-dimensional plot of a standard solution containing 10 mg/mL each of sucrose, glucose, and fructose, determined by HPLCFT-IR (injection volume, 50 µL). Table 1. Influence of the Injection Volume on the Most Important Chromatographic Parameters: Dispersion D, Peak Width (Full Width at Half-Height) W1/2 (s), and Number of Theoretical Plates per Meter N (m-1)
Dmax (C0/Cmax) W1/2 (s) N (m-1) a
sucrose,a injection volume (µL)
glucose,a injection volume (µL)
fructose,a injection volume (µL)
20
50
100
20
50
100
20
50
100
8.9 16.8 26 400
3.4 18.7 21 300
1.8 20.2 18 700
13.5 28.4 13 800
4.7 29.5 12 800
2.6 30.7 11 800
19.5 37.8 12 100
6.6 38.7 11 600
3.5 39.8 11 100
Concentrations, 10 mg/mL.
bers. Gram-Schmidt traces could not be successfully drawn, which can be explained as due to the strong water background absorption. It is assumed that slight drifts in the matrix background absorption cause a greater change in the interferogram domain than the small changes due to the analyte absorption. The quantitative evaluation of the chromatographic peaks was done “manually”, i.e., all data points of the peak in the corresponding trace were summed. Reference Measurements. The reference measurements were made with a standard HPLC employing RI detection. This was done by connecting the same pump and column unit to a differential refractometer (Knauer GmbH, Berlin, Germany). The samples were diluted to 1/4 prior to injection (injection volume, 20 µL) to prevent exceeding of the dynamic range of the RI detector. RESULTS AND DISCUSSION HPLC-FT-IR Chromatograms. Figure 1 shows the threedimensional HPLC-FT-IR separation of a standard solution containing 10 mg/mL each of sucrose, glucose, and fructose (injection volume: 50 µL, without precolumn). It is not necessary to extract spectra from the three-dimensional plot to see the differences of the analyte-specific absorption patterns. From that three-dimensional chromatogram, traces with maximum sensitivity for sucrose (A), glucose (B), and fructose (C) were extracted (Figure 2). The wavenumbers for the trace calculation were chosen corresponding to the absorption maxima of the analytes. By selecting the appropriate baseline points (in brackets of Figure 2), drift effects were reduced. As was already mentioned, the application of interferogram domain evaluation methods, involving 4288
Analytical Chemistry, Vol. 69, No. 20, October 15, 1997
Figure 2. HPLC-FT-IR traces derived from Figure 1, attaining maximum S/N for sucrose (trace A, offset 0.04), glucose (trace B, offset 0.02), and fructose (trace C). Negative peaks in trace B were caused by the baseline points (bpts in parentheses) chosen for the calculation of the traces.
the extraction of Gram-Schmidt traces, did not lead to satisfying results. Separation Quality. For the optimization toward a lower concentration limit, an enlargement of the injection volume seems appropriate, as far as the separation is not affected by excessive peak broadening. Injection volumes of 20, 50, and 100 µL were applied, and their influence on the chromatographic separation was investigated. The concentration used was 10 mg/mL for all components. The dispersion D in Table 1 was determined as the ratio between the analyte concentration in the sample, C0, and the concentration maximum, Cmax, of the chromatographic peak.
Table 2. Detection Limits for Carbohydrates Determined by HPLC-FT-IR (50 µL Injected)
concn limit (µg/mL) mass limit (µg)
sucrose
glucose
fructose
87 4.4
110 5.5
130 6.5
By injecting 100 µL, the dispersion was already reduced to 1.8, which means the sucrose concentration was reduced to approximately 56% during the separation process. Hence, a further enlargement of the injection volume could result in a peak broadening rather than in signal enhancement. The peak widths, W1/2, are not adversely affected by the injection volume. Neither was the number of theoretical plates, N, estimated by N ) 5.54(tR/W1/2)2 and standardized to 1 m column length, affected (Table 1). Only for the first-eluting peak was a decrease of N found, which is in accordance to the low dispersion of sucrose. A comparison of N with the data provided by the column supplier (14 044-35 539 m-1 for carbohydrates) confirms that the chromatographic system including the FT-IR flow cell does not induce significant peak broadening. For the application to real samples, it has to be considered that an increased sample volume affects the lifetime of the column by increasing also the matrix adsorbed on it. Limit of Detection and Identification. The limits of detection (LODs) were estimated, by a rule of thumb, as peak height exceeding 3 times the baseline noise from the traces shown in Figure 2. The 100 µg/mL range of carbohydrates is feasible with the proposed setup (Table 2). The mass detection limit was found to be in the range of 5 µg, but it has be taken into consideration that an analytical-scale (7.8 mm i.d.) column was applied. A further reduction of the mass detection limit is expected if the whole system is down-scaled to microbore HPLC, since the miniaturization of the FT-IR flow cell has already been shown.31 However, miniaturized polystyrene-divinylbenzene cation exchange columns are not yet commercially available. Nevertheless, this detection limit is more than 1 order of magnitude lower than the results reported in the literature3,5,26 dealing with flow cell aqueous phase HPLC-FT-IR (see introduction). The identification limit is not clearly defined in literature, but it can be assumed to be the lowest concentration or amount of analyte from which identifiable spectra can be recorded. The spectra in Figure 3 were extracted from a HPLC-FT-IR run injecting 1.0 mg/mL of each analyte (injection volume, 50 µL). The spectra can be easily distinguished and related to the corresponding carbohydrate, which confirms an identification limit of less than 50 µg. Real Sample Analysis. The applicability of the proposed HPLC-FT-IR system to real-world samples is shown by the following set of measurements. Nine samples containing sucrose, glucose, and fructose in the concentration range between 5 and 100 mg/mL were drawn from various soft drinks and fruit juices. Carbonated samples were degassed in an ultrasonic bath, and all samples were filtered prior to injection. In order to treat the column carefully and to extend its lifetime, the injection volume was reduced to 20 µL. (31) Lendl, B. Development of the Principles of Miniaturised Flow Injection Analysis Systems with FT-IR-Spectroscopic Detection, Ph.D. Thesis, Technical University of Vienna, 1996.
Figure 3. FT-IR spectra extracted from the peak maxima of HPLC peaks of sucrose (A), glucose (B), and fructose (C). Here, 50 µL of a standard solution containing 1 mg/mL of each component was injected. Offsets of 0.004 were chosen for (A) and 0.002 for (B) for better visualization.
Beginning with the calibration, five standard solutions containing 5, 25, 50, 75, and 100 mg/mL of each carbohydrate were prepared and injected into the system. Reasonable results concerning the standard deviation of the method (sx0 ) 70-110 µg/mL) were obtained. The determination of a test set of standard solutions gave an average deviation of 0.3 mg/mL (Table 3). Using these calibration data, the soft drink and fruit juice samples were analyzed. The results of the HPLC-FT-IR measurements, as well as those of the reference measurements using the same HPLC but a RI detector, are listed in Table 4. The deviation of the results of the HPLC-FT-IR technique from those of the reference method is in the range of 0.5-0.9 mg/mL (1-2%). The authors are aware of the fact that the HPLC-RI technique is not a completely independent reference method since both methods use the same separation system. The accuracy, which is defined as the deviation from the “true value”, might be less than that specified above. However, HPLC-RI is a well-established and accurate method for carbohydrate analysis and, therefore, is supposed to be suitable to demonstrate the ability of the proposed method for quantitative real sample analysis. Determination of Components Other Than Sugars. By means of HPLC-FT-IR, the determination and identification of minor components is also feasible. The advantage of HPLC-FTIR is that the combination of retention time and spectrum allows an easy and unequivocal identification. However, both absorption in the fingerprint region and an analyte concentration in the milligrams-per-milliliter range are required. This is illustrated by the HPLC-FT-IR determination of a taurine (2-aminoethanesulfonic acid)-containing soft drink (trade name, Red Bull). Two selected traces of this beverage in Figure 4 show the presence of other components in addition to the three major components, sucrose, glucose, and fructose. Peak 4 can easily be identified as taurine (Figure 5). From the less intense peak 5, a spectrum exhibiting the two characteristic ethanol peaks at 1046 and 1090 cm-1 was extracted. The analysis of standard solutions of both components allowed us to determine a concentration of 4 mg/ mL for taurine and 0.4 mg/mL for ethanol. CONCLUSION From the results of this study, it seems appropriate to state that the potential of advanced flow-cell-based HPLC-FT-IR is, by far, not yet exploited. Despite the high background absorption Analytical Chemistry, Vol. 69, No. 20, October 15, 1997
4289
Table 3. Calibration Data and Test Set Results
calibration concentrations (mg/mL) slope (mL/mg) intercept regression coefficient, r2 residual standard deviation standard deviation of the method, sx0 (mg/mL) standard test set (mg/mL) standard A (30, 10, 100 mg/mL S, G, F) standard B (100, 30, 10 mg/mL S, G, F) standard C (10, 100, 30 mg/mL S, G, F) a
sucrose
glucose
fructose
a 4.27 -0.08 0.999 991 0.46 0.108
a 2.13 0.34 0.999 996 0.156 0.073
a 0.93 -0.14 0.999 991 0.104 0.112
29.9 98.9 10.12
10.03 29.9 100.6
99.5 9.97 30.1
Each calibrated at 5, 25, 50, 75, and 100 mg/mL.
Table 4. Determination of Soft Drinks and Fruit Juices by HPLC-FT-IR and with HPLC-RI Reference Method (mg/mL) sucrose
soft drink 1 (Red Bull) soft drink 2 (Blaue Sau) soft drink 3 (Full Speed) soft drink 4 (Dreh & Drink) soft drink 5 (Almdudler) soft drink 6 (iced tea) grapefruit juice apple juice 1 apple juice 2 average deviationa a
glucose
fructose
FT-IR
RI
FT-IR
RI
FT-IR
RI
72.32 9.85 11.44 96.63 59.46 68.93 49.52 16.17 25.22
73.21 10.13 11.58 98.36 58.51 69.59 49.51 17.36 25.40
25.70 57.81 57.49 10.61 16.43 9.33 31.74 21.08 24.02
25.36 57.90 57.03 10.40 15.75 8.65 31.91 20.23 23.72
5.98 42.94 57.43 10.53 16.29 9.35 32.20 60.20 69.93
6.14 43.76 58.05 11.21 16.72 9.18 32.67 60.66 70.37
0.86
0.49
0.51
Root mean square.
Figure 4. HPLC-FT-IR traces of a taurine-containing soft drink (20 µL). (A) Fructose trace (peak, 1065 cm-1; baseline points, 1145, 1005 cm-1), (B) taurine trace (peak, 1045 cm-1; baseline point, 1065 cm-1). Peak identification: 1, sucrose; 2, glucose; 3, fructose; 4, taurine; 5, ethanol.
of the aqueous eluent, an appropriate tuning of the FT-IR setup in the 1400-900 cm-1 range allows the recording of identifiable spectra of aqueous carbohydrate solutions in the low- or submilligrams-per-milliliter concentration range. Hence, the proposed method has the potential to be a promising technique for the analysis of carbohydrates and alcohols, which are only poorly detectable with UV/visible detection. The common assumption (32) White, R. Chromatography/Fourier Transform Infrared Spectroscopy and Its Applications; Marcel Dekker: New York, 1990; Chapter 3, p 103. (33) Jinno, K. In Detectors for Liquid Chromatography; Yeung, E. S., Ed.; Chemical Analysis 89, Wiley: New York, 1986; Chapter 3, p 78. (34) Kalasinsky, V. F.; Kalasinsky, K. S. In HPLC Detection, Newer Methods; Patonay, F., Ed.; VCH: New York, 1992; p 127 ff.
4290 Analytical Chemistry, Vol. 69, No. 20, October 15, 1997
Figure 5. Taurine (A) and ethanol (B) spectra extracted from HPLC-FT-IR (Figure 4) run of a taurine-containing soft drink. An offset of 0.004 was applied for (A), and a 10-fold enlargement was applied for (B).
about aqueous phase HPLC-FT-IR, that flow cell approaches are incompatible32,33 or too insensitive (LODs of 1-2%)34 for practical applications, will have to be revised. The mass detection limit of solvent elimination methods is about 3 orders of magnitude lower than that of the flow cell approach presented in this work. But, considering micro-HPLC systems, the injection volumes of solvent elimination HPLC techniques are also approximately 2 orders of magnitude lower, resulting in concentration limits which do not differ so much from those of the proposed method. Nevertheless, the ease of handling, the low investment and maintenance costs, and the applicability for routine analysis are the major advantages of this flow-cell-based HPLC-FT-IR approach. ACKNOWLEDGMENT The authors gratefully thank the Fonds zur Fo¨rderung der wissenschaftlichen Forschung, P11338O ¨ CH, for the financial support of this work.
Received for review March 19, 1997. Accepted July 18, 1997.X AC970307P
X
Abstract published in Advance ACS Abstracts, September 1, 1997.
