Indirect thermal lens detection for capillary electrophoresis

Talanta 71 (2007) 1788–1794

Indirect thermal lens detection for capillary electrophoresis D.A. Nedosekin a,∗ , S.N. Bendrysheva a , W. Faubel b , M.A. Proskurnin a , U. Pyell c a

Analytical Chemistry Division, Chemistry Department, M.V. Lomonosov Moscow State University, Vorob’evy Hills d. 1 Str. 3, 119992 Moscow, Russia b Research Center Karlsruhe, Institute of Technical Chemistry, Water Technology and Geotechnology Division, Karlsruhe, Germany c Chemistry Department, University of Marburg, Marburg, Germany Received 9 February 2006; received in revised form 10 August 2006; accepted 21 August 2006 Available online 18 September 2006

Abstract Thermal lens detection with a 325.0 nm He–Cd excitation laser is used for thermooptical indirect detection in combination with the capillary electrophoretic separation of organic anions. The optimization of indirect thermooptical detection is discussed. With Mordant Yellow 7 (an azo dye) chosen as a probe ion limits of detection for 1-heptane-, 1-pentane-, 1-butane-, 1-propanesulfonic, and acetic acid at a level of n × 10−7 M were achieved with a separation electrolyte containing 50 ␮M of the probe ion and 5 mM Tris pH 9.90. A further increase in the detection sensitivity (twofold decrease in the limit of detection ) was obtained with a separation electrolyte containing a volume fraction of 20% acetonitrile. © 2006 Published by Elsevier B.V. Keywords: Capillary electrophoresis; Indirect thermooptical detection; Thermal lens detection; Alkanesulfonic acids; Organo-aqueous separation electrolyte

1. Introduction The high sensitivity and spatial resolution of laser based thermooptical methods of analysis predetermined the method expansion into the field of flow analysis [1,2]. The combination of photothermal detection with separation techniques such as capillary zone electrophoresis (CZE) [3–7], separation in microfluidic channels [8,9], and liquid chromatography [10,11] takes advantage of a sensitivity increase compared to conventional detection techniques. Thermal lens detection provides not only a detection sensitivity level comparable to or even better than that of laser-induced fluorescence detection but also has the advantage that it can be applied for a larger number of analytes, while laser-induced fluorescence detection is restricted to those substances having a sufficiently large fluorescence quantum yield [1,3,4,12]. Two different approaches to the combination of thermal lensing and CZE are to be named: the first one is the use of ther-

∗

Corresponding author. Tel.: +7 095 939 3514; fax: +7 095 939 4675. E-mail address: [email protected] (D.A. Nedosekin).

0039-9140/$ – see front matter © 2006 Published by Elsevier B.V. doi:10.1016/j.talanta.2006.08.016

mal lens microscopy [5–7] and the second one is the use of a crossed-beam thermal lens detector [3,4]. These approaches provide close levels of the detection sensitivity [2,5], however, thermal lens microscopy cannot be used in the capillary, requiring a microchip interface to be used for the detection [5,6]. The near-field crossed-beam thermal lens detector, on the other hand, can be directly used for the detection in the capillary and proved to be very efficient for the CZE needs [2]. The difficulty common for all laser based photothermal spectrometry methods, however, is the need to apply lasers emitting at the absorption band of the analyte [1,2]. A high cost associated with the most promising UV lasers and the power instability of the gas lasers still are the problems to be solved. Besides the search for new non-laser based excitation sources [1,9] there are ways to overcome this drawback with the use of the indirect detection technique [12] which is well known in CZE in combination with UV–vis or with laser-induced fluorescence detection [13–19]. The information on the use of the indirect mode in thermal lens detection is rather scare, in spite of the fact that the indirect detection makes it possible to detect substances which do

D.A. Nedosekin et al. / Talanta 71 (2007) 1788–1794

not absorb at the wavelength of the excitation laser. Firstly, the application of the thermal lens spectrometry detection in the indirect mode for CZE was described by Ren et al. [20] and Hu et al. [21]. Indirect thermooptical detection was used for the determination of amino acids [20] and several metal cations [21] with methylene blue as a probe ion (excitation laser wavelength 632.8 nm). However, a high level of adsorption of the cationic probe ion on the negatively charged capillary surface and the use of the same excitation and probing laser wavelength reduced the sensitivity of the measurements due to light scattering. In spite of these drawbacks, detection limits at the level of 5 × 10−6 M for lysine and 2 × 10−7 M for Cu(II) were achieved [20,21]. The thermal lens detector designed for capillary electrophoresis with the use of a He–Cd excitation laser (325.0 nm) and a diode probe laser (681.9 nm) described previously [22] increases the sensitivity of direct detection by a factor 10–30 compared to the sensitivity of a conventional UV absorbance detector. However, it is reasonable to expect that the detection in the indirect mode could be less sensitive as the baseline noise in the thermooptical detection is a function of the baseline signal [22]. Thus, an optimization strategy similar to that used for indirect laser-induced fluorescence detection [18] should be applied. In this paper, the description of the optimization of indirect detection for the separation of model compounds employing a mode-mismatched dual-laser crossed-beam thermal lens detector is presented. The most suitable approach for thermooptical detection in combination with CZE [2–4] was selected employing a UV excitation laser (325.0 nm) which was not previously used in an indirect mode. The comparison of conventional UV–vis indirect detection and indirect thermooptical detection is made. 2. Materials and methods 2.1. Reagents Following reagents were used: Alizarin Yellow GG (dye content 50%, Aldrich, Germany); Mordant Yellow 7 (dye content 65%, Aldrich); Naphthol Yellow S (p.a., Fluka, Germany); 9,10anthraquinone-1,5-disulfonic acid (p.a., Fluka); 4-nitrophenol (p.a., Fluka); Tris (p.a., Fluka); histidine (p.a., Fluka); lysine (p.a., Fluka); glycine (p.a., Fluka). Deionised high-purity water (Milli-Q plus 185 (18.0 M), Millipore, Bedford, MA, USA) was used for sample preparation. Sample solutions of alkanesulfonic acids and acetic acid (p.a., Fluka) were prepared in water. Photometric measurements were performed using a Philips PU 8720 UV/VIS (the Netherlands) spectrophotometer. The pH value of the running buffer was measured by an inoLab pH Level 1 pH-meter (Germany). 2.2. CZE unit A SpectraPhORESIS 100 capillary electrophoresis system (ThermoQuest, USA) was used. It was possible to couple it both with the thermal-lens detector and a conventional UV-detector

1789

(TSP SPECTRA UV100, USA). Separations were carried out in fused-silica capillaries (Polymicro technologies, Phoenix, AZ, USA), i.d. 75 ␮m, o.d. 360 ␮m and 71.6 cm total length (41.3 cm to detector). The protection coating at the 5 mm wide zone of the detection window was removed with warm concentrated sulfuric acid (accuracy is required), because this procedure gives a better quality of the capillary rather than the use of flame or a blade, resulting in a better laser beam diffraction picture at the pinhole plane. Sample injection was performed by applying vacuum (200 mbar) for 1 s at the detection side. 2.3. Thermal lens indirect detection The near-field thermal lens detector used in our experiments had a (crossed) dual-beam mode-mismatched optical scheme design [3,4,22]. At this configuration, the excitation beam and the probe beam cross at right angles within the inner volume of the capillary. The pathways of the excitation and probe beams through the capillary and the main elements of the optical scheme are shown in Fig. 1. For excitation a He–Cd laser is used (excitation wavelength, λe , 325 nm, Pe , 50 mW, IK3552RG Kimmon Electronic Co. Ltd. Tokyo, Japan). The spot size at the waist is 7.7 ␮m. The laser beam is modulated by a mechanical chopper (frequency range from 40 to 95 Hz), reflected at a right angle by a mirror and focused into the capillary by the focusing Lens 1. The probe laser beam (Toshiba TOLD 1050, Schaeffer & Kirchhoff, Germany, a laser diode, wavelength, λp 682 nm) is focused by the achromatic Lens 2, passes through the capillary, and changes its divergence dependent on the thermooptical element created in the capillary. The laser power at the sample, Pp , is 30 mW; the beam waist in the capillary is 75 ␮m. The beam center passes through the pinhole and reaches the photo diode. The photo diode signal is amplified by a laboratory made lock-in amplifier and translates to a PC by ADC-DAC. The synchronization of the measurements was implemented by an in-house written software. 2.4. Data treatment The analytical thermal lens signal θ is defined as following: θ = 2.303E0 Pe A,

(1)

where Pe is the excitation power (W), A the absorbance of the sample, and E0 is the enhancement factor of thermal lensing for unit excitation power. E0 = −

dn/dT . λp k

(2)

Here, λp is the probe laser wavelength, dn/dT is the temperature gradient of the medium refractive index, and k is the thermal conductivity of the medium [1,23]. The instrumental and the analytical signal are related to each other by: ϑ = KBθ,

(3)

1790

D.A. Nedosekin et al. / Talanta 71 (2007) 1788–1794

Fig. 1. Optical scheme of the near-field thermal lens detector [22].

where K is a constant which depends on the lock-in amplifier and B is the spectrometer configuration parameter [1]. 3. Results and discussion The limit of detection (LOD) for indirect photometric detection of ionic analytes is given by [12,16,19]: cmin =

cp NBL = , RDr Rεp l

(4)

where cmin is the LOD (molar concentration), cp is the probe ion molar concentration, R is the transfer ratio (the number of probe ions displaced by one analyte ion), Dr is the dynamic reserve (background signal-to-noise ratio), NBL is the background noise, εp is the molar absorbance coefficient of the probe ion, and l is the detection cell pathlength. In general, in the case of indirect thermooptical detection the baseline noise is a function of the probe ion concentration and excitation laser power, i.e. it is a function of signal value, NBL = f(cp , Pe ). Thus, expressing the analytical thermal lens signal θ by replacing A in Eq. (1) with cp εp l, and Dr in Eq. (3) with ϑNBL −1 and taking in consideration that NBL = const cp Pe , the LOD for an analyte ion can be given as: cmin =

cp NBL constcp = R ϑ 2.303KBRE0 εp l

(5)

From this equation it is obvious that the use of a highly absorbing dye as a probe ion decreases the LOD [16,19]. And increasing the probe ion background signal increases the baseline noise so that cmin is inversely proportional to the probe ion concentration [12]. On the other hand, this might increase the electrophoretic

dispersion during separation, i.e. reduce the separation efficiency correspondingly decreasing the peak height [12,16]. Thus, with the use of a highly sensitive detector such as the thermooptical detector, new optimum conditions for the composition of the separation electrolyte are to be found. Direct application of the techniques developed for indirect photometric detection is inefficient [18]. Also it should be noted that thermooptical detection provides a further increase in the sensitivity of analysis in case of improvements in the spectrometer configuration [1,3,22] by reducing the fraction of scattered light and by employing organo-aqueous separation buffers increasing the enhancement factor E0 described in Eq. (2) by improving the thermooptical properties of the medium [1]. 3.1. Detection mode optimization The optimization of the optical scheme of the thermal lens detector and consideration of the detector sensitivity for direct detection mode is described elsewhere [3,22]. However, indirect detection demands additional considerations concerning the dependence of the detector noise on the separation conditions and on the parameters of the signal acquisition. The analysis of the TL detector ‘baseline signal’-to-‘baseline noise’ ratio dependent on the chopper frequency and on the electrolyte velocity (separation voltage) (Fig. 2) shows that the optimum detection conditions are reached with higher chopper frequencies and are relatively independent from the separation voltage and hence the electrolyte flow velocity. These findings strongly deviate from results obtained for the direct TL detection where better signal-to-noise ratios were obtained with lower chopper frequencies compared to results

D.A. Nedosekin et al. / Talanta 71 (2007) 1788–1794

1791

Fig. 3. Mordant Yellow 7 dye.

Fig. 2. Dynamic reserve dependent on the chopper frequency for different separation voltages (electroosmotic flow velocities). The data were calculated from a horizontal part of the electropherogram taking 500 data points. Probe ion: Mordant Yellow 7, 50 ␮M; buffer: Tris 5 mM or lysine 10 mM.

obtained with maximum chopper frequency of 95 Hz [22]. This fact corroborates the assumption that parameters providing the improvement of TL direct detection are not accessible in indirect detection because of the proportionality of the noise of the signal to the signal height.

The background electrolyte should provide a sufficient electric conductivity of the solution, some buffering capacity at the selected pH and a low level of competition with the probe ion for analyte displacement [12,24]. Lysine, glycine, histidine, Na2 HPO4 , and Tris-based buffer electrolytes were tested, and 5 mM Tris (not titrated with HCl) was chosen for further experiments. With low Tris concentrations there is a low electric current minimizing Joule heating which is reported to result in peak broadening and baseline distributions [12,24]. Tris-based buffers (5 mM, pH 9.90) provided the most stable baseline with a minimal noise and insignificant probe ion adsorption on the walls. These BGE conditions allowed the separation and detection of the anions selected within 10 min using a counter-EOF separation under positive voltage. 3.3. Optimization of the probe ion concentration

3.2. Choice of the background electrolyte The detection sensitivity is expected to be (at a constant transfer rate) directly proportional to the molar absorptivity of the probe ion at the wavelength of the excitation laser (λe = 325.0 nm) (see Eq. (5) [1,16,19,24]. The following acidic dyes and acids were tested: Alizarin Yellow GG, Mordant Yellow 7, Naphthol Yellow S, 9,10-anthraquinone-1,5-disulfonic acid, and 4-nitrophenol. The analytes to be separated and detected are alkanesulfonic acids and acetic acid. The following parameters were taken into consideration for the selection of the probe ion [12]: (a) molar absorptivity at the pH of the separation electrolyte; (b) effective electrophoretic mobility (which should match the effective electrophoretic mobility of the analytes at the pH of the separation electrolyte); (c) photostability; (d) disposition towards adsorption onto the capillary walls. Adsorption onto the walls should be minimized in order to avoid baseline disturbances. The stability of these dyes in the excitation laser beam was tested experimentally in a standard quartz cell (1 cm length) exposed in the dark to the unfocused laser irradiation for 5 h. The change of the absorption spectra was monitored by a spectrophotometer. For all dyes investigated no significant changes were observed. From the listed probe ions, Mordant Yellow 7 (MY7, see Fig. 3) was selected as the most suitable one. This dye possesses a high molar absorptivity (ε325 ∼ 1.2×104 L mol−1 cm−1 ). At the pH of the optimized separation electrolyte (pH 9.9) the sulfonic acid group, the carboxylic acid group, and the phenolic group can be expected to be fully dissociated. The high negative charge of this dye in alkaline solution is assumed to prevent adsorption of the dye molecule onto the negatively charged fused-silica capillary walls.

As it was mentioned above, the use of indirect thermooptical detection in combination with CZE demands the optimization of the probe ion concentration [19–21]. With a given buffer and an optimized detector scheme, a decrease in the baseline noise can be achieved at low probe ion concentrations, however, causing excessive electrophoretic dispersion in the case of a analyteprobe ion-mobility-mismatch. The empirical optimization of the probe ion concentration (and thus the minimization of the LOD) is very time consuming. Therefore, we employed a semiempirical optimization scheme. Eq. (5) can be presented in the following way: cmin(cprobe ) = NBL(cprobe ) H(cprobe ) const,

(6)

where H(cprobe ) is the parameter reflecting the ratio of electrophoretic dispersion for the analytes in the capillary with a decrease of the probe ion concentration, const is the constant overall parameter for the selected optical scheme geometry and buffer composition. The parameter H was measured experimentally as follows: H(cprobe ) = hmax / h(cprobe ) , here h(cprobe ) is the peak height for the analyte under investigation obtained for the selected concentration of the probe ion and hmax is the peak height in the case of minimal electrophoretic dispersion observed experimentally. For the selected range of the probe ion concentrations, the criterion of the minimal electrophoretic dispersion is the achievement of the separation conditions where the analyte peak reaches the maximum height and is independent from the probe ion concentration. Thus, the H parameter equals to 1 in the case of minimal peak broadening, and H > 1 in the case of significant electrophoretic dispersion, increasing the detection limits. Fig. 4 shows the relative noise of the baseline signal NBL value (the maximum noise value corresponds to the highest probe ion

1792

D.A. Nedosekin et al. / Talanta 71 (2007) 1788–1794

Fig. 5. Electropherogram of an equimolar (10 ␮M) anion mixture: (1) 1heptanesulfonic acid; (2) 1-pentanesulfonic acid; (3) 1-butanesulfonic acid; (4) 1-propanesulfonic acid; (5) acetic acid. Separation electrolyte: Mordant Yellow 7 50 ␮M, Tris 5 mM, pH 9.9. Fig. 4. Dependence of detection limits for acetate on the probe ion concentration, indirect TLS detection, and aqueous separation electrolyte. Separation electrolyte: Mordant Yellow 7, varied concentrations; Tris 5 mM; pH 9.9. Separation voltage: 30 kV. All parameters are given as a relative value (normalized on the maximum value) in order to improve the readability of the graph.

concentration and was taken as 100%) and the H(cprobe ) parameter for acetate (10 ␮M) as a function of the probe ion (Mordant Yellow 7) concentration. According to Eq. (6) the LOD can be obtained by multiplying these values. The dependence of LODs presented in Fig. 4 by the dotted line was obtained semiempirically from the trends of the noise and dependence of the peak height on the probe ion concentration. This approach is effective, because it is possible to obtain the detailed function of the noise for the detection system on the basis of only a few experiments. Based on this approach, the concentration of Mordant Yellow 7 of 50 ␮M was selected for further experiments making necessary only four test runs with varied probe ion concentrations. The dynamic range calculated from the background thermal lens signal divided by the corresponding noise of the background signal (signals were measured with a separation electrolyte containing 50 ␮M Mordant Yellow 7 and 5 mM Tris, separation voltage 30 kV, and calculation of baseline noise taking 200 data points) equals to 80. In photometric indirect detection the dynamic range reaches 500 [25].

Table 1 Limits of detection (S/N = 3) for thermal lens detection of model mixture components (for experimental conditions see Fig. 5) (n = 8, P = 0.95) Mixture component

LOD (␮mol L−1 )

CH3 COOH C3 H7 SO3 H C4 H9 SO3 H C5 H11 SO3 H C7 H15 SO3 H

0.5 ± 0.1 0.7 ± 0.1 1.1 ± 0.2 0.8 ± 0.1 0.8 ± 0.1

is a less pronounced peak fronting indicating the similarity of the acetate and the probe ion effective electrophoretic mobilities at the conditions given. The ratio of the analyte/probe ion electrophoretic mobilities mismatch causes the increase in the LOD calculated for the analytes selected (see Table 1) except the acetic acid. Due to electrophoretic dispersion there is no linear dependence of the peak height on the analyte concentration (see Fig. 6). The calibration graphs for C3 –C7 alkanesulfonic acids and acetic acid were calculated from the respective peak areas. The linearity range for the calibration graphs at the conditions selected

3.4. Electromigration dispersion and calibration graph linearity In the case of low electrolyte concentration, which has to be employed in indirect thermooptical detection, the local electric field strength in the sample zone differs from that in the buffer zone [12,24]. This discontinuity results in peak fronting or tailing according to the respective analyte/probe ion electrophoretic mobility ratio. In Fig. 5, an electropherogram is presented demonstrating the separation and indirect TL detection of different alkanesulfonic acids (10 ␮M). Peak fronting caused by electrophoretic dispersion can be observed for each component of the mixture. However, in the case of acetate there

Fig. 6. Dependences of the peak height on the concentration of the analyte in the injected sample. For further experimental parameters refer to Fig. 5.

D.A. Nedosekin et al. / Talanta 71 (2007) 1788–1794

Fig. 7. Calibration graph for anionic analytes. For further experimental parameters refer to Fig. 5.

is in the range 1–100 ␮M for acetic acid, and 1–50 ␮M for the alkanesulfonic acids (see Fig. 7). 3.5. Indirect detection in organo-aqueous separation electrolytes The use of organo-aqueous solutions is well known in conventional thermooptical spectroscopy and allows increasing of the detection sensitivity changing the thermooptical properties of the medium, Eq. (2), [1]. In CZE with conventional photometric UV detection organo-aqueous media have been used to improve the separation selectivity [12,26]. It should be also mentioned that the analyte and/or probe ion absorbance coefficients can be also affected by the composition of the separation electrolyte [26]. Thus, a positive impact on the detection sensitivity via a change in the thermooptical properties of the medium can be counteracted by a negative impact on separation parameters. For further comparison, acetonitrile–water mixtures were selected as solvent mixtures because of their advantageous thermooptical properties and the low viscosity of acetonitrile. At the pH chosen (9.90) the molar absorptivity of Mordant Yellow 7 at 325 nm is the same for solutions of the dye in acetonitrile–water mixtures (10–40%) and in water. Electropherograms for a sample containing several alkanesulfonic acids employing an aqueous separation electrolyte and a separation electrolyte containing water and acetonitrile are given in Fig. 8. Under the conditions optimized for the aqueous separation electrolyte and thermooptical detection, the use of the separation electrolyte containing 20% acetonitrile (v/v) decreased the electroosmotic flow velocity by 25% and increased the baseline signal for a factor of 2.6 with an increase in the baseline noise for a factor of 1.3. The choice of the probe ion concentration was done according to the procedure described in Section 3.1 (see Fig. 9). The optimum concentration of Mordant Yellow 7 calculated from this data is in the range 35–55 ␮M; thus, the MY7 concentration of 50 ␮M was kept for further experiments. The Dr factor calculated for these conditions was about 160 (a solution of MY7, 50 ␮M; Tris,

1793

Fig. 8. Electropherograms for a sample containing 10 ␮M of (1) 1heptanesulfonic acid; (2) 1–pentanesulfonic acid; (3) 1–butanesulfonic acid; (4) 1–propanesulfonic acid. Separation electrolyte: Mordant Yellow 7 50 ␮M, Tris 5 mM, pH 9.9, in (A) water and (B) 20% (v/v) acetonitrile; separation voltage: 30 kV.

5 mM; Acetonitrile, 20%; separation voltage, 30 kV; 200 data points). In order to estimate the influence of acetonitrile added to the separation medium on the separation efficiency, UV detection under the same conditions was implemented. The comparison of LODs for 1-propanesulfonic acid obtained with TLS and UV detection in water and in water–acetonitrile is presented in Table 2. The LODs in the case of UV–vis detection both in water and in the water–ACN mixture are very close to each other. The small difference observed can be attributed to a decrease in the electroosmotic flow velocity and to an increase in the vacuum injected sample volume [12,26]. However, in the case of the thermal lens detection the LOD for the water–acetonitrile mix-

Fig. 9. Detection limit for acetate dependent on the probe ion concentration for indirect TLS detection employing a water–acetonitrile 20% mixture as separation medium. Separation electrolyte: Mordant Yellow 7, varied molar concentration; Tris 5 mM; pH 9.9; ACN 20% (v/v); separation voltage: 30 kV. All parameters are given as a relative value (normalized on the maximum value) in order to improve the readability of the graph. For further experimental parameters refer to Fig. 5.

1794

D.A. Nedosekin et al. / Talanta 71 (2007) 1788–1794

Table 2 Limits of detection (S/N = 3) for 1-propanesulfonic acid with TLS and UV indirect detection (for experimental conditions see Fig. 8) (n = 8 for H2 O and n = 6 for ACN/H2 O, P = 0.95). Medium

LOD (␮mol L−1 )

TLS

H2 O ACN/H2 O (20:80, v/v)

0.7 ± 0.1 0.4 ± 0.1

UV

H2 O ACN/H2 O (20:80, v/v)

6 ± 0.1 7 ± 0.1

ture is two times lower than that for water. From these data, it can be concluded that the decrease in the LODs for TLS detection was achieved only because of the increase in E0 which is increasing the dynamic range (see Eqs. (4) and (5)). However, a more detailed study of the impact of the thermooptical properties of the separation medium on the LOD in indirect thermal lens detection (e.g. with water–acetonitrile mixtures) is needed and will be studied in future work. 4. Conclusions A procedure for the optimization of method parameters for a method employing indirect thermal lens detection combined with CZE is developed. With the use of the proposed semiempirical approach, the concentration of the probe ion corresponding to a minimum LOD for a mixture of alkanesulfonic acids was optimized. Selecting Mordant Yellow 7 in alkaline solution as a probe ion provided limits of detection for the model mixture components at the level of n × 10−7 M. These values are by an order of magnitude better than those obtained with UV detection at the same detection wavelength and identical composition of the separation electrolyte. It was shown that limits of detection in indirect TLS detection can be further reduced by improving the thermal lens enhancement factor E0 via use of organic–aqueous separation electrolytes. The data achieved show that the thermal lens detector can be used for the determination of analytes non-absorbing at the wavelength of the excitation laser. However, it should be noted that the LOD obtained in this work for the alkanesulfonic acids

selected are of the same order of magnitude as those published recently for indirect photometric UV–vis detection of alkanesulfonic acids [19]. Thus, advantages of direct TLS detection over direct UV–vis detection in capillaries [3,4] cannot directly be transferred to indirect detection due to differences in the baseline noise characteristics. References [1] S.E. Bialkowski, Photothermal Spectroscopy Methods for Chemical Analysis, Wiley Interscience, New York, 1996. [2] T. de Beer, N.H. Velthorst, U.A.Th. Brinkman, C. Gooijer, J. Chromatogr. A 971 (2002) 1. [3] B.S. Seidel, E. Steinle, W. Faubel, H.J. Ache, SPIE 2836 (1996) 283. [4] B.S. Seidel, W. Faubel, H.J. Ache, J. Biomed. Opt. 2 (1997) 326. [5] K. Uchiyama, A. Hibara, K. Sato, H. Hisamoto, M. Tokeshi, T. Kitamori, Electrophoresis 24 (2003) 179. [6] K. Uchiyama, M. Tokeshi, Y. Kikutani, A. Hattori, T. Kitamori, Anal. Sci. 21 (2005) 49. [7] M. Yamauchi, M. Tokeshi, J. Yamaguchi, T. Fukuzawa, A. Hattori, A. Hibara, T. Kitamori, J. Chromatogr. A 1106 (2006) 89. [8] T. Kitamori, M. Tokeshi, A. Hibara, K. Sato, Anal. Chem. 76 (2004) 52. [9] E. Tamaki, A. Hibara, M. Tokeshi, T. Kitamori, Lab on a Chip 5 (2005) 129. [10] M. Sikovec, M. Novic, M. Franko, J. Cromatogr. A 739 (1996) 111. [11] M. Sikovec, M. Novic, Vida Hudnik, M. Franko, J. Cromatogr. A 706 (1995) 121. [12] H. Engelhardt, W. Beck, T. Schmitt Kapillarelektrophorese, Vieweg. Braunschweig. Wiesbaden. (1994). [13] W.G. Kuhr, E.S. Yeung, Anal. Chem. 60 (1988) 2642. [14] Y. Xue, E.S. Yeung, Anal. Chem. 65 (1993) 2923. [15] M.C. Breadmore, P.R. Haddad, J.S. Fritz, J. Chromatogr. A 920 (2001) 31. [16] J.E. Melanson, C.A. Boulet, C.A. Lucy, Anal. Chem. 73 (2001) 1809. [17] C. Johns, M. Macka, P.R. Haddad, Electrophoresis 23 (2002) 43. [18] N. Ragozina, M. Putz, W. Faubel, U. Pyell, Electrophoresis 24 (2003) 567. [19] C. Johns, M.J. Shaw, M. Macke, P.R. Haddad, Electrophoresis 24 (2003) 557. [20] J. Ren, B. Li, Ya. Deng, J. Cheng, Talanta 42 (1995) 1891. [21] Yo. Hu, J. Cheng, Ya. Deng, Analyst 122 (1997) 1089. [22] M.A. Proskurnin, S.N. Bendrysheva, N. Ragozina, S. Heissler, W. Faubel, U. Pyell, Appl. Spectrosc. 59 (2005) 1470. [23] M. Fischer, J. Georges, Anal. Chim. Acta 322 (1996) 117. [24] P. Doble, M. Macka, P.R. Haddad, J. Chromatogr. A 804 (1998) 327. [25] Y. Ma, R. Zhang, J. Chromatogr. 625 (1992) 341. [26] C. Schwer, E. Kenndler, Anal. Chem. 63 (1991) 1801.