Axial-beam on-column absorption detection for open tubular capillary

Axial-beam on-column absorption detection for open tubular capillary liquid ... Axial-beam laser-excited fluorescence detection in capillary electroph...
2 downloads 0 Views 2MB Size
1580

Anal. Chem. 1990. 62. 1580-1585

divided by the number of moles of histidine found in that run, yielding the number of residues for the other amino acids. On the basis of the number of residues found (106), a molecular weight of 11660 was estimated for the protein. This was calculated by using the rough assumption that each residue contributes, on average, 110 to the molecular weight (22). The identity of the protein was then disclosed to be lysozyme, with an actual molecular weight of 13 930. The weight calculated for the unknown protein is low due to the amino acids that cannot be measured with this method, i.e., tryptophan, cysteine, lysine, and proline. The amino acid composition of lysozyme is given in Table I11 and compared to what was experimentally found. The error for the average of five runs is 8.5%.

CONCLUSIONS The method presented in this report has been shown to be useful for the hydrolysis and analysis of proteins a t the low femtomole level. The key to this endeavor was the development of methods that can easily handle and analyze a few nanoliters of sample. There are a few problems with the technique, however. The most disturbing is the loss of lysine during the derivatization process (23). Another drawback is that the derivatizing reagent (NDA) must be purified prior to use. The purification procedure is necessary simply because NDA is not commercially available in a pure enough form for work on the nanoliter scale. In spite of these limitations, the method was able to determine the amino acid composition of 0.1 ng (4 fmol) of protein with an error of only 5.0%. Further improvements in sample handling methods may decrease the amount of protein required for amino acid analysis to still lower levels. The least amount of protein that can be analyzed with this method may ultimately depend on the detection limit of the electrochemical detector used, which is approximately 1 amol, or M, at the present time. ACKNOWLEDGMENT We thank Waters Associates (Milford, MA) for the gift of the 600E Multisolvent Delivery system.

Registry No. Asp, 56-84-8; Glu, 56-86-0; Ser, 56-45-1; His, 71-00-1;Thr, 72-19-5;Gly, 56-40-6;Ala, 56-41-7;Tyr, 60-18-4;Arg, 74-79-3;Val, 72-18-4; Met, 63-68-3;Ile, 73-32-5;Phe, 63-91-2;Leu, 61-90-5; NDA, 7149-49-7; cyanide, 57-12-5.

LITERATURE CITED Cohen, S. A.; Bidlingmeyer, B. A.; Tarvin, T. L. Nature 1986, 320, 769-770. Hill, D. W.; Waiters, F. H.; Wilson, T. D.; Stuart, J. D. Anal. Chem. 1979, 57, 1338-1341. Lindroth, P.; Mopper, K. Anal. Chem. 1979, 57, 1667-1674. bjong, c.; Hughes, G. J.; van Weringen, E.; Wilson, K. J. J. ChromatOgr. 1982, 247,345-359. Rubenstein, M.; Chen-Kiang, S.;Stein, S.; Undenfriend, S. Anal. Blochem. 1979, 95,117-121. Koop, P. R.; Morgan, E. T.; Tarr, G. E.; Coon, M. J. J. Biol. Chem. 1982, 257, 8472-8480. Roth, M. Anal. Chem. 1971, 43,880-882. Joseph, M. H.; Davies, P. J. J. Chromafogr. 1983, 277, 125-136. Allison, L. A.; Mayer, G. S.;Shoup, R. E. Anal. Chem. 1984, 56, 1089-1096. de Montigny, F.; Stobaugh. J. F.; Givens, R. S.;Carlson, R. G.; Srlnivasachar, K.; Sternson, L. A.; Higuchi, T. Anal. Chem. 1987, 59, 1096-1101. Oates, M. D.; Jorgenson, J. W. Anal. Chem. 1989, 67, 432-435. Kennedy, R. T.; Jorgenson, J. W. Anal. Chem. 1988, 60, 1521-1524. Kennedy, R. T.; St. Claire, R. L.; White, J. G.; Jorgenson, J. W. Mlkrochim. Acta 1986, 7987(II),37-45. Flurer, C.; Borra, C.; Beale, S.;Novotny, M. Anal. Chem. 1988, 60, 1826- 1829. Hskh, Y.yBeale, S.C.; Wiesler, D.; Novotny, M. J. Microcolumn S e p . 1989. 7 . 96-100. Knecht, L. A.; Guthrie, E. J.; Jorgenson, J. W.Anal. Chem. 1984, 56, 479-482. St. Claire. R. L. Ph.D. Thesis, University of North Carolina at Chapel Hill, 1986. White, J. G.; St. Claire, R. L.: Jorgenson, J. W. Anal. Chem. 1988, 56, 293-298. Dupont, D.; Keim, P.; Chui, A,; Bozzini, M.; Wilson, K. J. Appl. Bkasysf. User Bull. 1988, 2. Stobaugh, J. F. Personal communication, Universlty of Kansas, 1988. Pelletier. S.W.; Chokshi, H. P.; Desai, H. K. J. Nat. prod. 1986. 49, 892-900. Lehninger, A. Principles of Biochemistry: Worth: New York, 1982; p 127. Oates, M. D.; Jorgenson, J. W. Unpublished results.

RECEIVED for review December 27,1989. Accepted April 30, 1990. Support for this work was provided by the National Institute of Health under Grant No. GM39515.

Axial-Beam On-Column Absorption Detection for Open Tubular Capillary Liquid Chromatography Xiaobing Xi and Edward S. Yeung* Ames Laboratory-USDOE and Department of Chemistry, Iowa State University, Ames, Iowa 50011 A novel approach has been developed for absorptlon detection In open tubular caplllary liquld chromatography. The axial coupling of source Hght wfth the caplllary columns and the use of optical waveguide capillary columns made It posslble to utilize the full length of the sample bands Inside the capillary columns as the path length for absorbance measurement. Compared with the cross-beam arrangement, the axlal-beam on-colwnn absorption detectlon provides an Increase of up to 1000 tlmes In path length and In associated improvements on the concentration limits of detectlon. This allows lnvestlgation of substantially lower sample concentrations and quantlties than those conventionally feasible for absorption detectlon In open tubular capillary liquid chromatography.

INTRODUCTION Recently, the subject of open tubular capillary liquid

chromatography (OTCLC) has received considerable interest (1-3). Using “separation impedance”, a criterion introduced by Bristow and Knox (41,to assess the theoretical performance of conventional and capillary columns in liquid chromatography, Knox (5) predicted that OTCLC could have the smallest theoretical plate height and the best performance. It is generally accepted that OTCLC with inner diameters of less than 10 pm and employing detectors with low nanoliter volumes and high sensitivities will produce high separation efficiencies in a reasonable time with superior limits of detection (6-8). The advantages of OTCLC with respect to separation speed and efficiency can be exploited only when the external contribution to peak broadening is minimized (6,9). Since the inner diameters of OTCLC columns are normally 10 Fm or less, the injection and detection volumes have to be substantially reduced compared with conventional liquid chromatography to maintain the column efficiency. As far as the

0003-2700/90/0362-1580$02.50/00 1990 American Chemical Society

ANALYTICAL CHEMISTRY. VOL. 62, NO. 15. AUGUST 1. 1990 * 1581

injection volume is concerned, this is readily achieved by using techniques such as split injection (IO),microsyringe injection (I I ) , or stopped flow injection (12). However, the reduction of detection cell volume to below 1nL is a more difficult task. Controlling the shape of the detector is also a challenge. The ideal focusing of a Gaussian light beam will give (13)

V

= 16.73

where Vis the minimum volume, Z is the path length, and A is the wavelength of light. To provide a volume of the order of 1 nL, the maximum path length is about 0.2 mm a t 300-nm excitation. A variety of detection methods have been utilized in the development of OTCLC, including W-vis absorption (24-1 7), fluorescence (18-20), electrochemistry (21-23), mass spectrometry (24), flame photometry (25),flame ionization (26). and thermal lens detection (27). Most of the reported research in OTCLC has utilized absorption, fluorescence, or electrochemistry detection. Although it has relatively poor sensitivity compared with fluorescence and electrochemistry detection, absorption detection is more versatile and is widely used in OTCLC. The early studies of OTCLC with absorption detection employed the postcolumn scheme by miniaturizing detectors of conventional liquid chromatography. However, the volumes of flow cells, which were about 100 nL, were still far from the optimum for OTCLC. The problems were approached in a different way by Yang (ZS),who developed an on-column absorption detector for OTCLC that utilized a segment of the fused silica capillary column for its detection cell. This on-column absorption detection scheme was later extended to detection for open tubular capillary zone electrophoresis (29). An on-column fluorescence detector was also developed for OTCLC based on the idea of on-column absorption detection (30,3J). The major advantage of on-column detectors over postcolumn detectors based on miniaturization of conventional flow cells is the elimination of zone-broadening effects caused by extra column volumes inherent to detector cells and associated coupling fittings. However, since on-column absorption detectors so far rely on a cross-beam arrangement (i.e., the light beam crosses the capillary column a t right angles), the absorption path length was severely limited to that of the inner diameter of capillary columns, ranging from 10 to 50 pm. In turn, the absorbance and thus the concentration limits of detection were also limited. For example (32),if one assumes an absorbance detection limit of 5 X lo-' for a high-quality commercial chromatogaphic detector interfaced to a capillary column and an analyte with molar absorptivity of le L mot' cm-', then the concentration limits of detection range from 5X to 1 X M for path lengths of 10-50 pm, hardly state of the art in trace analysis. On the other hand, it is known that the theoretical plate height in OTCLC is roughly equal to 2 times the inner diameter of the capillary column (33).and so the number of available theoretical plates N = L/2d, where L is the column length and d is the column inner diameter. Also, N = 5.544tR/m,,9* 2 5.54(L/b)*,where b is the length of the sample zone inside the capillary column. Consequently,

b2 ii 11.08dL

(1)

For a 100-cm-long and 10-pm-i.d. capillary column, b equals 10.5 mm, which is 1050 times larger than the inner diameter of the column. If one can couple a light source into one end of the capillary column and guide the beam through the whole column, the light intensity detected a t the other end of the column will be a measure of the absorption of the samples. One therefore will be able to take full advantage of the length of sample bands inside the capillary column and increase the

Ha-Nc

+

I

\M

LZ PN

0-1

Flgwe 1. Schematic diagram of ttm axiaCbeam on-columnabsorption detector for open tubular capillary liquid chromatography. He-Ne, helium-neon laser; LS, laser power stabilizer; M. mirror: L1. 25-mm focal length lens: TI, 1i32-in. tee: T2, 'fIgin. tee; CT, fused silica capillary tubing: IN, injector; FL, finer; PU. syringe pump: CC. capillary column; PN. optical fiber positioner: L2, 5X microscope obiective: PMT. photomultiplier tube; PS. power supply: DMM. digital multimeter; CM. microcomputer: CR, chart recorder

path length, and so the absorbance, by a factor of lo3. We refer to this detection scheme for OTCLC as "axial-beam" on-column absorption detection, with the potential of providing an improvement of 1000 times on the concentration limits of detection over the cross-beam on-column absorption detection. In this paper, we report the results of preliminary experiments using axial-beam on-column absorption detection for OTCLC. The interface between the light source and the capillary columns is discussed. The intensity profiles of light beams traveling through the capillary columns were measured to explore the possibilities of guiding light through long capillary columns with inner diameters ranging from 10 to 75 pm. The feasibility of applying this detection method to reversed-phase OTCLC using an optical waveguide capillary column is demonstrated.

EXPERIMENTAL SECTION Apparatus. A schematicdiagram of the axial-beam on-column absorption detector for OTCLC is shown in Figure 1. The detector utilized as a light source a 3-mW He-Ne laser (Unipbase Corp., San Jose, CA, Model 1105P),which operated at 632.8 nm and was linearly polarized. The laser beam was passed through a laser power stabilizer (Cambridge Research and Instrumentation, Cambridge, MA, Model LSlOO), was reflected by two mirrors (Newport,Fountain Valley, CA, Model lODlO.ER3, and was then focused by a 25-mm focal length lens (Newport,Model KPDX076) into the end of the capillary column. The laser beam power after the focusing lens was about 1 mW. The transmitted laser beam at the other end of the capillary column was collected by a 5X microscope objective lens. The image was passed through a 632.8-nm line filter (Corion Corp., Holliston, MA, Model P1633-FR) and then was detected by a photodiode (Hamamatsu Corp., Middlesex, NJ, Model S1790) or a photomultiplier tube (Hamamatsu, Model R928), which was operated with a bias potential of 25Cb550 V. The output of the photomultiplier tube was fed to a digital multimeter (Keithley Instruments, Cleveland, OH, Model 160B), which amplified and converted the signal to a voltage output. This output was then sent either to a chart recorder (Houston Instruments Corp., Austin, TX, Model B5117-51) or to a data acquisition system consisting of an analog-to-digital1/0 interface (Data Translation, Marlborough, NH, Model DT2827) and a microcomputer (IBM, Boca Raton, FL, Model PC/AT). The lens was mounted on a precision x-y positioner (Newport, 462 series) for fine alignment of the laser beam focal point. The entrance end of the capillary column was inserted into one port

1582

ANALYTICAL CHEMISTRY, VOL. 62, NO. 15, AUGUST 1, 1990

was used to image the beams at the output end of the column to a white screen about 2 m away. The 30 X 30 and 50 X 50 pixel images were digitized by using a cooled CCD camera system (Photometrics Ltd., Tucson, AZ, Series 200) equipped with a 90-mm macrolens (Tamron, Tokyo, Japan). The images were then transferred to a microcomputer (Compuadd, Austin, TX, Model Standard 286) for storage and analysis. .__, W

Figure 2. Schematic diagram of axial coupling of the light beam with the capillary column. L, 25-mm focal length lens; W, quartz window; T, 1/32-in. tee; CT, capillary tubing; CC, capillary column: F, polyimide ferrule; A, fused silica adapter; N, nut.

of a l/&. stainless steel tee (Valco Instruments Corp., Houston, TX, Model ZT.5) and was tightened by a one-piece fused silica adaptor (Valco Instruments, Model FS.25, '/32 in.). The output end of the capillary column was held by a fiber optic positioner (Newport, Model FP1) and was adjusted to slightly touch the tip of a drill bit (1/20in.) to direct the liquid flowing out of the capillary column to drain. The capillary column was either sandwiched between two pieces of foam plastic sheets or taped on a solid mount to prevent any vibrations. Chromatography. The chromatographic system employed in this work is shown in Figure 1. A microsyringe pump (ISCO, Lincoln, NE,Model pLC-500) was used in the constant-flow mode to propel the eluent through the column. A 0.5-pm line filter (Alltech,Deerfield, IL, Model SSI 05-1005) was connected before the injector (Rheodyne, Cotati, CA, Model 7520), which had a rotor volume of 0.5 or 1.0 pL. Since the majority of the mobile phase was diverted from the column by a split-flow system, the flow rate in the open tubular capillary column was typically 20 nL/min. A 1/16-in.stainless steel splitting tee (Valco Instruments, Model ZTlC) was positioned after the injector, and a section of fused silica capillary tubing (150-pm 0.d. and 10-25-pm id., Polymicro Technologies Inc., Phoenix, AZ) was inserted into the tee and was used to draw off most of the flow from the open tubular capillary column. Another piece of fused silica capillary tubing was employed to connect the splitting tee and the lightcoupling tee. By the proper choice of the lengths and the inner diameters of both capillary tubings, the desired splitting ratios were obtained. Columns. The open tubular capillary columns used in this work were prepared by using a procedure adapted from Farbrot et al. (34). The reverse-phase stationary phases were coated on the inside of fused silica capillary columns (150-pm 0.d. and 10-21-pm id., Polymicro Technologies) by cross-linking vinylsilicone gums. The details of the column-coating procedure have been described elsewhere (35). Reagents. The mobile phases used in this work were 93% by volume pyridine (Fisher, Fair Lawn, NJ) in water, dimethyl sulfoxide (Fisher), or methanol (Mallinckrodt, Paris, KY). Bromocresol green was supplied by Baker (Phillipsburg,NJ), and azulene was obtained from Aldrich (Milwaukee, WI). The water used was purified by using a Barnstead Nanopure I1 system (Boston, MA). All chemicals were reagent grade and were used as supplied. All solutions were filtered with a disposable Nylon 66 filter (Alltech, 0.2 fim) and were degassed in an ultrasonic bath. Axial Coupling of the Light Source with the Capillary Columns. The interface between the light source and the open tubular capillary column is shown in Figure 2. The column and the fused silica capillary tubing were inserted into a '/32-in. stainless steel tee and were tightened by fused silica adapters as described above. A quartz window that permitted the coupling of He-Ne laser light into the end of the open tubular capillary column was held by a 1/32-in.polyimide ferrule (Valco Instruments, Model ZF.5). The quartz window was homemade by using a high-quality synthetic fused silica rod (Dynasill Corp., Berlin, NJ, Selected Grade 4100) and was 1.6 mm thick and 0.78 mm in diameter. The quartz window, with surfaces of good quality, can transmit about 80% light. The gap between the end of the capillary column and the quartz window was estimated to be 1 mm, which allowed the mobile phase to flow into the column. Beam Profile Measurement. To measure the profiles of laser beams passing through the columns, a 5 X microscopic objective

RESULTS AND DISCUSSION Source Light Coupling w i t h the Capillary Columns. One of the keys for success in developing an axial-beam oncolumn absorption detector for OTCLC was the interface between the light source and the small columns. An interface is required that not only can transmit light with high optical quality, Le., high transmittance and small scattering and wave-front distortion, but also can withstand the relatively high pressure produced in the chromatographic system. The interface used in our experiment was a homemade quartz window. This system readily tolerated 1500 psi. On the other hand, high optical quality in the window was difficult to achieve. We found that the quartz window with low optical qualities provided poor transmittance, large wave-front distortion, and severe scattering when the light source was coupled with columns with inner diameters ranging from 10 to 75 pm. These problems were mainly controlled by the flatness and the scratch and dig numbers of the window surfaces, the refractive index homogeneity, and the numbers of inclusions, bubbles, and seeds in the window material. Compared with the commercial-grade fused silica, the Selected Grade 4100 fused silica provided much better refractive index homogeneity and more than 100 times fewer inclusions, bubbles, and seeds. This in turn efficiently reduced the wave-front distortion and the scattering. To make flat and scratch- and dig-free surfaces for the quartz window, we tried to grind and polish the surfaces using 600 grit paper and then 0.3-pm aluminum oxide abrasive. However, due to the difficulty of handling the small quartz window, grinding and polishing did not produce satisfying surfaces. We then used a normal glass-drawing technique and cut the quartz windows with a sharp knife. With practice, it was possible to make quartz windows with relatively flat and scratch- and dig-free surfaces. Even though it was tedious to prepare, a quartz window with high optical quality could be used for a long period of time. Profiles of Transmitted Laser Beams. Lei et al. (36) developed a long capillary cell with 1-mm i.d. to enhance the detection power of ordinary colorimetry for the determination of phosphorus in natural water. They found that changing the capillary cell shape, i.e., curving or looping the cell, resulted in serious attenuation of the incident light. A 70-cm looped capillary cell could transmit only 2.8 X lo* of the light energy compared to a linear capillary cell of the same length. Similarly, we found it absolutely necessary for capillary columns with 10-75-pm i.d. to be at a linear arrangement to efficiently transmit light. No light energy could be detected for a 85 cm x 75 pm looped capillary column (one turn loop, 12-cm diameter). Figure 3 shows the intensity profile of a laser beam transmitted through a linear 85 cm X 75 pm capillary column. It can be seen that most of the light energy was distributed inside the liquid core of the column; that is, transmission through the fused silica wall played a minor role. To quantitatively determine the percentage of light energy distributed inside the core, the output end of the column was immersed in a cell containing pyridine (refractive index n = 1.51). This caused the light energy distributed inside the wall of the column to refract out, so that only light traveling through the core of the column was detected. We found this percentage to be about 78%. This naturally depends on the alignment of the capillary column and focusing of the light source.

1583

ANALYTICAL CHEMISTRY, VOL. 62, NO. 15, AUGUST 1, 1990

Fwre'.

Intens@ pofile Of the transmittedlaser beam' Cdwnn: 85 cm X 75 Fm, 150-pm 0.d. Mobile phase: methanol. Flow rate: 200 nL/min.

Figure 5. Intensity profile of the transmitted laser beam. Column: 85 cm X 25 pm, 150-pm 0.d.. Mobile phase: 93% pyridine in water. Flow rate: 7o nL/min.

i ' TIME (min)

O k - F - r - 6 '

PIXEL

Figure 4. Intensity profile of the transmitted laser beam. Column: 85 cm X 25 pm, 150-pm 0.d.. Mobile phase: methanol. Flow rate: 70 nL/min.

For applications in OTCLC, columns with inner diameters smaller than 75 pm have to be utilized. Figure 4 shows the intensity profile of a laser beam transmitted through a 85 cm X 25 pm capillary column. In this case, the vast majority of light energy was distributed along the wall of the column. Obviously, this arrangement is useless for absorption detection. One of the reasons for light to preferentially travel through the wall instead of the core was the difficulty in focusing light into the exact center of the column with 25-pm i.d. and 150-pm 0.d. Wave-front distortion and scattering caused by the imperfect window made focusing more difficult. Another reason was that the cross-sectional area of the wall was 35 times larger than that of the core. An optical waveguide using hollow glass fiber as the cladding and liquid (with a higher refractive index) as the core was developed for long-distance optical communication (37,38). Later, this technique was employed for signal enhancement in Raman spectroscopy (39),colorimetry (do), and fluorimetry (41). If one uses mobile phases in liquid chromatography with higher refractive indices than that of the column material, light will travel exclusively through the core of the column by total internal reflection. Figure 5 shows the intensity profile of laser light traveling through a 85 cm x 25 pm column filled with pyridine by total internal reflection. All the light energy was distributed inside the core, and the Gaussian profile was nicely retained. The column can be looped since the transmittance was independent of the column shape. Optical Waveguide Capillary Columns. According to Snyder and Kirkland (42),about 30% of the most often used solvents for liquid chromatography have refractive indices larger than that of the column material (fused silica, refractive index n = 1.458). These all could be used in OTCLC with total internal reflection. In our work, we chose pyridine and dimethyl sulfoxide (refractive index n = 1.477) as the mobile phases based on their relatively high polarity, low viscosity,

l

b

'

l

h

'

l

b

Figure 6. Axil-beam on-column absorption detection of azulene (AZ) and bromocresol green (BG). I,: base-line intensity. I intensity with BO. I,: intensity wffh A2 and BG. AZ: 4 X lo4 M. BG: 5.5 X M. Eluent: 93% pyridine in water. Flow rate: 20 nL/min. Injection volume: 0.93 nL.

,:

!'Lei -40

I*

.50 -60

,

0

,

2

,

,

4

,

,

6

,

,

8

,

,

1

,

0

,

1

,

2

I

1

4

TIME (min) Figure 7. Chromatogram of azulene (B) and bromocresol green (C). A is the system peak. The plot is a 21-point quadratic-smoothed first derivative of Figure 6.

large water solubility, and transparency at the operating wavelength. Figure 6 shows a typical chromatogram of OTCLC using an axial-beam on-column absorption detector with total internal reflection. Io was the light intensity detected when only the mobile phase ran through the column. I , was due to the absorption of the component with the larger retention time. Zzresulted from the absorption of both components. In Figure 6, AZ was azulene and BG was bromocresol green, which had a larger retention time. For quantitative determination, it can be seen that A A z = log (Zl/Zz), ABG = log (ZO/ZJ, and A = AAZ+ ABG = log (Zo/Zz). The retention times of azulene and bromocresol green were measured as the time between the first and second inflection points and the time between the first and third inflection points, respectively. The rise times at the slopes of the two plateaus provide information about the peak widths. Although Figure 6 shows a chromatogram appearing different from the "normal" chromatogram, it is nothing more

1584

ANALYTICAL CHEMISTRY, VOL. 62, NO. 15, AUGUST 1, 1990

than an integrated plot derived from a normal chromatogram. T o demonstrate this, we took a 21-point quadratic first derivative on Figure 6, using a least-squares differentiation method introduced by Savitzky and Golay (43). Figure 7 shows the result, which is in the form of a normal chromatogram. Peak A in Figure 7 was a system peak due to injecting the sample into the column. It is not surprising that one obtains an “integrated” chromatogram by axial-beam oncolumn detection. Only when absorbing components enter or leave the column will there be a change in transmission. At any given time, the cumulative absorbance within the column is recorded. Synovec and Yeung ( 4 4 ) developed an integration method to improve the limits of detection (LOD) in chromatography. The relation between Figure 6 and 7 here is identical with the transformation discussed in ref 44, except that integration is inherent to the optical scheme here and not performed mathematically. In the earlier work, a LOD enhancement ratio of 5 to 20 could be obtained for a typical chromatographic system by using an integration procedure. We have confirmed that conclusion in this work. Figure 7 shows noticeably larger noise levels than does Figure 6 even though a 21-point smooth has been applied to the former. The superior signal-to-noise ratio ( S I N ) in Figure 6 shows an inherent advantage of using the raw data rather than converting back to a normal chromatogram. As pointed out earlier ( 4 4 ) ,chromatographic resolution and quantitative accuracy are retained in the integrated chromatogram. Finally, it should be noted that for simplicity the intensities were plotted in Figures 6 and 7. For quantitation, these chromatograms should first be converted to absorbance units. One would then be able to avoid nonlinearity when a trace analyte is present with other strongly absorbing components. Naturally, the dynamic range will eventually be affected if the base-line stability begins to degrade when the transmitted intensity becomes too low. Limit of Detection. The LOD of the axial-beam on-column absorption detector is dependent on the base-line stability of the transmitted intensity. Several factors affecting the base-line stability were identified. First, flow turbulence in the mobile phase immediately before entering the column will change the focal point of the incident light and cause fluctuations in light intensity. However, we found that the base-line stability was independent of flow rate ranging from 0 to 35 nL/min for a 10-12-pm-i.d. column in the given split-flow arrangement. Second, any vibrations in the column will change the alignment and introduce noise. Taping the column on a solid mount overcomes this problem. Finally, the formation, the growth, and the release of the liquid drop a t the exit end of the column can produce a periodic optical distortion. The noise level became larger with increasing mobile-phase viscosity. This problem was solved by touching the column end to the tip of a drill bit to direct the liquid flow. The base-line stability in our work was most likely limited by the residual fluctuations at this interface. A laser power stabilizer was used to minimize the intensity fluctuations. In the end, we were only able to obtain a base-line stability of 1 part in 500 instead of the specified 1 part in 2000 after the stabilizer. This is partly due to the low powers used. This equals a base-line stability of 8.7 x IO-* absorbance unit (AU) or a LOD of 2.6 x AU at S I N = 3. To determine the concentration LOD of the detector for a given absorber (given molar absorptivity), we use the ratio of absorbance to path length, A / b , as the criterion. For a 10-pm capillary column, a LOD of 5 X lo-* AU at S I N = 3 is estimated for a cross-beam on-column absorption detector, so A / b is 0.5 cm-’ for that system. For the axial-beam oncolumn absorption detector with a 10-mm sample zone length, this ratio is 2.6 x cm-I. Therefore, even though the

base-line stability in our system is worse, the concentration LOD is better by a factor of 192, due to the increased path length. It should be noted that although a He-Ne laser was used in our experiment, such a source is not essential and can be replaced by a noncoherent source such as a xenon lamp, provided one can couple sufficient light energy to the entrance of the capillary. In using UV lasers or the UV output of arc lamps, one will have to be concerned about photobleaching because the analytes will be exposed to radiation for much longer periods than they would be in normal detectors. Other Considerations. From Figure 7, the number of plates was calculated as 3500, which was about 7% of the theoretical one. The band broadening was caused partly by the column-to-injector connection and the dead volume inside the light-coupling tee. No attempt was made in this work to further minimize the band broadening and to optimize the system. Further effort on developing axial-beam on-column absorption detection has to address this problem. The requirement that mobile phases with large refractive indices must be used limits the applicability of axial-beam on-column absorption detection. To overcome this limitation, a couple of approaches can be pursued. Tsunoda et al. (45) have recently employed a poly(tetrafluoroethy1ene-co-hexafluoropropylene) (FEP) tubing as a waveguide cell for liquid absorption spectrometry. The refractive index of FEP is only 1.338, close to that of water (1.333). If FEP capillary columns with inner diameters of about 10 pm are available, most solvents, including aqueous solutions, can be used in the optical waveguide mode. On the other hand, the possibility of successive total reflection at the outer wall of a capillary cell has been demonstrated in an application to colorimetry (46). It was shown that a considerable fraction of the light energy could be distributed in the core of the column if all the light was introduced into the center of the column. As previously indicated, this is confirmed here even for capillary columns with inner diameters smaller than 25 pm. We expect that the percentage of light energy distributed in the core will be enhanced if the ratio of the cross-sectional area of the core to the wall of the column is larger. With capillary columns with smaller o.d., say 50 pm, the total reflection at the outer column surface may be employed for axial-beam on-column absorption detection using mobile phases of any refractive index. Although the axial-beam arrangement was only applied to OTCLC in this work, it might be applicable to other chromatographic systems using open tubular capillar columns, such as supercritical fluid chromatography and capillary electrophoresis. Finally, we note that transmission here is by total internal reflection. There is no fundamental limit to the length of the columns used. When gradient elution is used, provided that at no time the solvent refractive index becomes smaller than that of the capillary wall, the chromatogram will maintain its stable base line. Contributions due to absorption by the changing solvents, however, will be magnified in this scheme.

ACKNOWLEDGMENT We thank T. Forre for assistance in preparing the quartz windows, L. Koutny for operating the CCD camera system, and J. Taylor for programming the Savitzky-Golay differentiation. LITERATURE CITED

-

(1) Kucera, P., Ed. Microcolumn High Performance Liquid Chromatogra

phy; Elsevier: Amsterdam, 1984. (2) Yang, F. J.. Ed. Microbore Column Chromatography: A Unified Approach to Chromatography; Marcal Dekker: New York, 1989. (3) Novotny, M. Anal. Chem. 1988. 60. 500A-510A. (4) Bristow, P. A.; Knox, J. H. Chromatographia 1977, 10, 279-289. (5) Knox, J. H. J. Chromatogr. Sci. 1980, 18, 453-461. (6) Knox, J. H.; Gilbert, M. T. J. Chromatogr. 1979, 786, 405-418. (7) Yang, F. J. J. Chromatogr. Sci. 1982, 2 0 , 241-251. (8) Jorgenson, J. W.; Guthrie, E. J. J. Chromatogr. 1983, 225, 335-348.

1585

Anal. Chem. 1990, 62, 1585-1587 (9) (10) (11) (12) (13) (14)

(20) (21) (22) (23) (24) (25) (26) (27) (26) (29) (30)

Sternberg, J. C. A&. Chromafogr. (N.Y.) 1966, 2, 205-209. Yang, F. J. J. Chromafogr. 1982, 236, 265-277. Kennedy, R. T.; Jorgenson, J. W. Anal. Chem. 1988, 60,1521-1524. Manz, A.; Simon, W. J. J. Chromafogr. 1987, 387, 167-196. Yeung, E. S. I n Mlcrobore Column Chromatography: A Unified Approach to Chromatography; Yang, F. J., Ed.; Marcel Dekker, New York, 1989; pp 117-127. Hirata. Y.: Novotnv. M.: Tsuda. T.: Ishii. D. Anal. Chem. 1979. 57, 1807-1809. Krejci, M.: Tesarik, K.; Pajurek, J. J. Chromafogr. 1980, 191, 17-24. Ishii, D.; Takeuchi, T. J. Chromafogr. Sci. 1980, 18, 462-472. Tsuda, T.; Nakagawa, G. J. Chromatogr. 1983, 268, 369-374. Hirata, Y.; Novotny. M. J. Chromafogr. 1979, 786,521-528. Tsuda, T.; Katsugoshi, T.; Nakagaw, G. J. Chromafogr. 1981, 274, 283-290. Foiestad, S.; Johnson, R.; Josefsson, B.; Galle, B. Anal. Chem. 1984, 54, 925-929. Hirata. Y.: Lin. P. T.: Novotnv, M.; Wiahtman, R. M. J. Chromafogr. 1980, 187, 287-294. Slais, K.; Kreci, M. J. Chromafogr. 1982, 235, 21-29. Knecht. L. A.; Guthrie, E. T.; Jorgenson, J. W. Anal. Chem. 1984, 56, 479-482 . . - .- - . Niessen, W. M. A.; Poppe, H. J. Chromafogr. 1985, 323, 37-46. McGuffin, V. L.; Novotny, M. Anal. Chem. 1981, 53, 946-951. Krejci, M.; Tesarik, K.; Rusik, M.; Pajurek, J. J. Chromatogr. 1981, 218, 167-178. Sepaniak, M.; Vargo, J. D.; Kettier, C. N.; Maskarinec, M. P. Anal. Chem. 1984, 56, 1252-1257. Yang, F. J. HRC 8 CC,J . High Resoluf. Chromafogr. Chromafogr. Commun. 1981, 4 , 83-85. Waibroechi. Y.; Jorgenson, J. W. J. Chromafogr. 1984, 375, 135-143. .- - . .- . Guthrie, E. J.; Jorgenson, J. W. Anal. Chem. 1984, 56, 483-486.

(31) Guthrie, E. J.; Jorgenson, J. W.; Diuzneski, P. R. J. Chromatogr. Sci. 1984, 22, 171-176. (32) Dovichi, N. J.; Zarrin, F.; Nolan, T. G.; Bornhop, D. J. Specfrochim. Acta 1888, 438, 639-649. (33) Novotny, M. I n Microcolumn High Performance Liquid Chromafogra phy; Kucera, P., Ed.; Elsevier: Amsterdam, 1984; Chapter 7. (34) Farbrot, A.; Foiestad, S.; Larson, M. HRC 8 CC, J . High Resoluf. Chromatogr. Chrornafogr. Commun. 1986, 9 , 117-119. (35) Pfeffer, W. D.; Yeung, E. S. J. Chromafogr. 1990, 506, 401-408. (36) Lei, W.; Fujiwara, K.; Fuwa, K. Anal. Chem. 1983, 55, 951-955. (37) Stone, J. Appl. fhys. Leff. 1972, 20, 239-240. (38) Stone, J. IEEEJ. Quantum Electron. 1972, €0-8, 386-388. (39) Stone, J. J. Chem. fhys. 1978, 69, 4349-4356. (40) Fuwa, K.; Wei, L.; Fujiwara, K. Anal. Chem. 1984, 57, 1640-1644. (41) Fujiwara, K.;Simeonsson, J. B.; Smith, B. W.; Winefordner, J. D. Anal. Chem. 1988, 60, 1065-1068. (42) Snyder, L. R.; Kirkland, J. J. Introduction to Modern Liquid Chromafography; John Wiiey: New York, 1979; pp 249-250. (43) Savitzky, A.; Goiay, M. J. E. Anal. Chem. 1964, 36, 1627-1639. (44) Synovec, R. E.; Yeung, E. S. Anal. Chem. 1985, 57, 2162-2167. (45) Tsuncda, K.; Nomura, A.; Yamada, J.; Nishi, S. Appl. Spechosc. 1990, 44, 163-165. (46) Tsunoda, K.; Nomura, A,; Yamada, J.; Nishi, S. Appl. Spectrosc. 1989, 43,49-55.

RECEIVED for review February 8, 1990. Accepted April 20, 1990. The Ames Laboratory is operated by Iowa State University for the US. Department of Energy under Contract W-7405-Eng-82. This work was supported by the Office of Basic Energy Science, Division of Chemical Sciences.

pH Hysteresis Effect with Silica Capillaries in Capillary Zone Electrophoresis William J. Lambert* and David L. Middleton Drug Delivery Research & Development-Sterile Michigan 49001

Products, The Upjohn Company, 301 Henrietta Street, Kalamazoo,

Electroosmotlc moblllty was determined as a functlon of pH at constant Ionic strength by using a neutral marker. It was observed that the electroosmotlc moblllty was dependent on the precondltlonlng of the fused slllca column. The electroosmotlc mobility values that were obtained from a column previously exposed to acldlc condltlons for varying periods of time were consistently lower than the values from columns previously exposed to alkallne conditlons, partlcularly in the pH 4-6 region. The equlllbratlon of the surface charge on the fused slllca surface appears to be a relatively slow phenomenon. Since the peak area In caplllary zone electrophoresis varies Inversely with the solute mobility, attentlon should be pald to caplllary precondltlonlng, particularly In the pH 4-6 range.

INTRODUCTION Buffer pH is one of the key parameters used to optimize separations in capillary zone electrophoresis (CZE). The charge and, thus, the electrophoretic mobility of ionizable solutes are affected by pH. In addition, the charge of the capillary surface is influenced by pH, giving rise to differences in the electroosmotic flow. This is particularly true with fused

* Corresponding author.

silica capillaries due to the ionization of acidic silanol groups (1). Therefore, the rational development of CZE assays should be based on a fundamental understanding of the solute and capillary surface charge. While the effects of pH on the net mobility of several solutes were studied, dramatic shifts in the migration time (2-3-fold) were observed under constant conditions. These shifts appeared to be dependent on the preconditioning of the column. The objective of the present report is to demonstrate the hysteresis effect that was observed and to discuss the ramifications of this effect on the qualitative and quantitative use of CZE.

EXPERIMENTAL SECTION An Applied Biosystems (Foster City, CA) Model 270A capillmy

electrophoresissystem with a 50-prn4.d. fused silica capillary (AB1 0602-0014) with a Spectra Physics (San Jose, CA) Chromjet Model SP4400 integrator were used in the present study. On-column ultraviolet (UV) detection was at 230 nm, and the temperature was held constant at 30 "C. Flushing of the capillary was performed by the built-in vacuum system at 20 in.Hg for 20 min. The samples were also injected by the vacuum technique; however, a pressure difference of 5 in.Hg for 1 s was utilized (approximately 3.6 nL using Poiseville's equation). Electroosmotic mobility (pW) was calculated as follows:

where L is the total capillary length (72 cm), 1 is the injector to 0003-2700/90/0362-1585$02.50/00 1990 American Chemical Society