摘要:

Demonstration of isoelectric focusing on an etched quartz chip with UV absorption imaging detection† Qinglu Mao and Janusz Pawliszyn* Department of Chemistry, University of Waterloo, Waterloo, Ontario, Canada N2L 3G1. E-mail: janusz@sciborg.uwaterloo.ca Received 15th December 1998, Accepted 16th March 1999 The feasibility of isoelectric focusing (IEF) performed on-chip was demonstrated for the first time via absorption imaging detection. Microchannels on a quartz chip were fabricated using photolithography and a chemical etching process. The separation channels were 40 mm long, 100 mm wide and 10 mm deep, and were coated with linear polyacrylamide to reduce electroosmotic flow.A quartz chip cartridge for IEF was assembled in which two pieces of hollow fiber were glued to the two ends of the separation channel to isolate the electrolytes from the samples. Low molecular mass pI markers and myoglobin were selected as model samples which were mixed with 4% carrier ampholyte solution.Samples were injected into the channel via the connection capillary by pressure. A voltage of 3 kV was applied to perform IEF. The IEF current decreased from about 13.4 to 1.3 mA. The focused zones were monitored in real time by absorption imaging detection at 280 nm. The detection limit was about 0.3 mg ml21 or 24 pg for pI marker 6.6, and 30 mg ml21 or 2.4 ng for myoglobin with an optical pathlength of 10 mm. Good reproducibility and resolution were obtained for linear polyacrylamide coated channels.The total analysis time was less than 10 min. This imaged chip IEF provides a fast separation technique with quantitative ability and the potential for increasing throughput. Introduction The microfabrication of analytical instrumentation has attracted great interest, offering compact, reliable, and inexpensive methods for chemical and biological analysis. Chemical separation devices appear to be particularly amenable to microfabrication. The feasibility of the integration of miniaturized separation techniques into compact devices has been demonstrated for gas chromatography (GC),1 capillary electrophoresis (CE),2–8 micellar electrokinetic capillary chromatography (MEKC)9,10 and open channel electrochromatography (CEC).11 Their separation performance has been extensively evaluated.Although much work remains to be done to prove their general suitability for practical analytical applications, they have already been shown to offer some unique features with respect to separation speed, sample injection, consumption of sample and buffer solution and the use of low voltages for highly efficient separations.However, no experimental demonstration has been presented for isoelectric focusing (IEF) performed in free solutions using microfabricated devices. Although on-chip IEF is expected to have some great advantages over the conventional capillary IEF (CIEF) technique with respect to compact size, integration and throughput, the major limitations of developing this technique, as addressed later, are associated with its unique separation mode and the detection methods available.Isoelectric focusing is a high-resolution technique for the separation of complex protein mixtures.12,13 It is routinely used for the characterization of biological extracts, monitoring protein purification, evaluating the stability or microheterogeneity of protein therapeutics and the determination of protein isoelectric points (pI).As in gel IEF, proteins are separated according to their pIs in a pH gradient formed by carrier ampholytes when an electric potential is applied. CIEF14–17 combines the high resolving power of conventional gel IEF with the automation and quantification advantages of CE instrumentation. The excellent heat dissipation of capillaries allows separations to be performed in free solution, without the requirement for an anticonvective gel.Typical CIEF employs 12–60 cm long capillaries with an on-column absorbance detector. Focused sample zones are moved to the detection point by electrophoretic (salt), hydrodynamic or electroosmotic mobilization. Both one-step and two-step approaches have been developed.14,15 However, problems associated with the mobilization process may be encountered, including long analysis time, high risk of protein precipitation and distortion of pH gradient.To overcome the disadvantages caused by mobilization, imaged-CIEF has been developed by ourselves in the past few years.18–21 A specially designed cartridge holding a short capillary was constructed. Charge-coupled devices (CCDs) were used as the imaging sensor. With such a whole column imaging technique, the focused sample zones can be monitored in their stationary state without mobilization. With the development of this imaged-CIEF technique, IEF performed in microchip channels becomes possible. In most CE microchip systems, single-point detection including laserinduced fluorescence and UV absorbance detection has been widely adopted.3,22 However, in the case of IEF, if performed on a microchip, it is preferable to implement whole channel imaging detection so that the focusing process takes place, as in slab gel IEF, free from any disturbance, e.g., the influence of electroosmostic flow (EOF).The mobilization step to facilitate single-point detection for a short channel is expected to be less controllable than with a long capillary. Imaging detection, in which no mobilization is required, is therefore ideal for on-chip IEF.Moreover, imaged CIEF reveals the dynamic focusing process easily and accurately because of its real-time mode, and provides valuable extra information to IEF analysis, especially in fundamental investigations. In this paper, IEF performed on a microchip is reported and the quartz chip cartridge for IEF is described in detail.To † Presented at the 9th Annual Frederick Conference (Maryland), October 19–21, 1998. Analyst, 1999, 124, 637–641 637demonstrate the chip IEF, low molecular mass pI markers and myoglobin were selected as model samples. The chip channel was coated with linear ployacrylamide and the performance of both uncoated and coated channels for IEF of the model samples was explored. Experimental Microchip The microchip separation devices were fabricated by Alberta Microelectronic Corporation (AMC, Edmonton, Alberta, Canada) using photolithography, chemical etching and bonding processes. The channel structures were fabricated on a piece of quartz plate.Another quartz cover plate was then thermally bonded to seal the channels.23 The two quartz plates were 40 3 50 3 0.5 mm each. Access to the channel terminals was provided by ultrasonically drilled 0.5 mm holes. In our experiments, only one channel was selected to perform IEF each time.To minimize the EOF, the channel walls were coated with linear polyacrylamide following Hjerten’s procedure.24 Microchip IEF cartridge A microchip IEF cartridge was assembled to match the imaging detection as shown in Fig. 1(A). In order not to disturb the pH gradient in the separation channel, two pieces of hollow fiber (od 200 mm, id 170 mm, length 5 mm; Spectrum Medical Industries, Los Angeles, CA, USA), one at each end of the channel, were used to isolate the electrolytes from the sample solution. This special design eases the sample injection for the short separation channel. Two connection capillaries (od 160 mm, id 100 mm; Polymicro Technologies, Tucson, AZ, USA) were inserted into and glued to the two pieces of hollow fiber dialysis tubing.The sections of hollow fiber were glued to the two holes in the ends of chip channel. A piece of chemically etched metal with a 50 mm wide slit was glued to the under surface of the substrate plate.The slit ensures that UV light only passes through the channel for absorption imaging detection. Two plastic tubes, each with a volume of about 0.15 ml, were finally glued to the end of the channels as electrolyte reservoirs. Imaged chip IEF procedures The instrumental set-up for imaged chip IEF and sample injection is shown in Fig. 1(B). The UV source was an 80 W Xe lamp with a 280 nm bandpass filter. The UV radiation was projected on to the chip channel by an optical fiber bundle. A linear CCD array was used as detection sensor.The microchip IEF cartridge was fixed in the optical path. Samples were injected by pressure: one end of the capillary connected with the channel was inserted into a 2 ml sample vial with a rubber septum cap. A 0.5 ml volume of air was then injected into the vial using a 3 ml syringe so that sample was forced into the separation channel. After 30–60 s, the capillary was pulled out and inserted into the two balancing vials.The two vials were filled with water to the same level in order to reduce hydrodynamic flow while performing IEF. After 1 min relaxing and stabilizing, IEF was started by applying a high voltage of 3 kV. Images before and during focusing were taken by the absorption imaging detection system. The exposure time was 20 ms for each scan, and 64 scans were averaged for each electropherogram to increase the signal-to-noise ratio. Chemicals All chemicals were of analytical-reagent grade and solutions were prepared using de-ionized, distilled water.Solutions of 10 mm H3PO4 and 20 mm NaOH were used as the anolyte and catholyte, respectively. Low molecular mass pI markers 8.6, 7.4, 6.6, and 5.3 were purchased from Bio-Rad (Missisauga, ON, Canada). Myoglobin was purchased from Sigma (St. Louis, MO, USA). The pI markers or myoglobin were/was mixed with the carrier ampholyte solution (Pharmalyte pH 3–10, Sigma) to a final concentration of 2–40 mg ml21 for low molecular mass pI markers, 400 mg ml21 for myoglobin and 4% for ampholytes.Results and discussion IEF performed in uncoated channel In CIEF, coating of the capillary inner wall is employed to reduce substantially the adsorption of proteins on the capillary wall to avoid poor reproducibility and detection sensitivity, and to minimize the electroosmotic flow to produce a high resolution. In the case of IEF performed on-chip, both uncoated Fig. 1 (A) Schematic diagram of microchip IEF cartridge; (B) schematic diagram of the instrument set-up for chip IEF using absorption imaging detection. 638 Analyst, 1999, 124, 637–641and coated channels were tested. First, IEF performed on uncoated channels was investigated. Fig. 2 shows the electropherograms of pI markers 8.6 and 5.3 focused in the chip channel. At 2 min after focusing took place, two peaks are observed. The dynamic focusing process shows that the two peaks merged into one peak after 6 min.This suggests that the two pI markers were not separated. According to our previous experiments,20 the dynamic focusing process reveals a ‘double peak’ focusing pattern for each component in the early focusing stage when the sample is uniformly distributed inside the separation column. This means that the component is focused towards its pI point from both sides. For a two-component system, four peaks are observed in the early stage of a normal focusing process. Hence the experimental results show that a large EOF drove the components out of the short channel before the focusing was complete.In this case, the focusing pattern is not reproducible, as shown by our unpresented experimental results. IEF performed in coated channel IEF performance was assessed in the linear polyacrylamide coated channel. Fig. 3 shows one IEF dynamic process of pI marker 6.6 with a concentration of 8 mg ml21. At 2 min, two focused zones of the pI marker appeared and focused towards the correspondent pI point from both sides.They joined together to form one peak after the focusing came to end. During the focusing process, the current decreased about 10-fold from 13.4 to 1.3 mA. The focusing process reached a steady state after 4 min for one pI marker component. The electropheorgrams show a dynamic IEF process where EOF is not obvious. It is observed that the total focusing time is < 5 min. Hence the channel coating was successful, resulting in a good focusing process.It should be noted that in our experiments the pH gradient formed inside the channel does not correspond to the exact pH range 3–10 for the chosen ampholytes. The real pH range in the channel is narrower because in the two ends of the channel, the holes to which hollow fibers were glued stored a larger amount of ampholytes than that of the channel itself. After the voltage is applied, the pH gradient is partly influenced by this volume.To minimize such an effect, this volume, which can be called the dead volume in connection, must be reduced by improving the method of gluing the hollow fiber to the ends of channel. The reproducibility and the detection limit of chip IEF were checked. Fig. 4 shows two electropherograms of pI marker 6.6 with two different concentrations at 6 min. The peak position shows good reproducibility for the two runs, and a smooth baseline is observed. The noise level of absorption imaging detection in our experiments is 1 31023 arbitrary units when 64 scans of CCD images are averaged and the detection limit can therefore be assumed to be 3 3 1023 arbitrary units.When 4 mg ml21 pI marker 6.6 was injected, a peak with a height of 0.042 arbitrary units was formed after focusing. The detection limit, therefore, corresponds to a concentration of 0.3 mg ml21 of the injected pI marker 6.6. The channel volume is only 0.08 ml, so the amount which can be detected is only 24 pg.The quantitative performance of the IEF chip was studied on the basis of pI marker 6.6. In the concentration range 1–30 mg ml21, the absorption is linearly proportional to the concentration of injected pI marker with a correlation coefficient of 0.999. Compared with the use of a capillary cartridge,25 less spike noise appeared with this chip cartridge, as shown in Fig. 4. With a capillary cartridge, after applying a voltage, the capillary is subject to slight bending owing to an electrostatic force which may cause some unpredictable noise.The chip cartridge shows no such a bending owing to its solid channel structure. Moreover, the chip cartridge, in which each channel was covered by a slit, eases the requirements for optical alignment for the light passing through, and gains a high light throughput that also improves the detection limit, since the noise level in imaged chip IEF is mainly limited by the shot noise of the CCD sensor.26 A better choice for absorption imaging detection will be a photodiode array detector (PDA), the high well capacity of which makes the shot noise insignificant in the detection signal.26, 27 Fig. 2 Electropherograms of pI markers 8.6 and 5.3 at 2, 6 and 15 min. Concentration: 40 mg ml21 of each pI marker. Applied voltage: 3 kV. Fig. 3 The dynamic process of pI marker 6.6 with a concentration of 8 mg ml21 recorded at 2, 3 and 7 min. Left, cathode; right, anode. Applied voltage: 3 kV.Fig. 4 Reproducibility of peak position for chip IEF of pI marker 6.6 at 6 min with concentrations of 8 and 16 mg ml21. Left, cathode; right, anode. Applied voltage: 3 kV. Analyst, 1999, 124, 637–641 639Separation of model samples The electropherograms of pI marker mixture 7.4 and 6.6 separated on a coated channel are shown in Fig. 5. At 7 min, four peaks are observed that finally focus into two peaks correspondent to pI marker 7.4 and 6.6 (as shown at 11 min). The molar absorptivity of pI marker 6.6 is higher than that of pI marker 7.4, and therefore the peak height is correspondingly higher.This demonstrates the successful separation of pI markers by chip IEF. As another example, myoglobin was separated in coated channels as shown in Fig. 6. Myoglobin has two variants: one is pI 7.2 and the other is pI 6.8. The pI 7.2 variant is present at higher levels than the pI 6.8 variant. Correspondingly, the pI 7.2 variant has a higher peak. For myoglobin, the detection limit is 30 mg ml21 or 2.4 ng based on the pI 7.2 variant.The total separation time for the two examples is < 10 min. Resolving power for chip IEF It is feared that the short channel may cause a decrease in resolution for IEF. For this reason, theoretical considerations of the feasibility of a short channel are discussed here. For a sample zone focused in a capillary by the IEF process, concentration has a Gaussian distribution with a variance s:12,13 C = C0 exp(2pEx2)/2D (1) where C = C0 is the maximum concentration, p the mobility slope (2du/dx), E the field strength, D the diffusion coefficient and x the position along the capillary.s = ± - D E x u d d ( ) (2) Using the criterion of three times the variance s for resolved adjacent proteins, the resolving power, D pI, of IEF in terms of s can be expressed as12 Dp d(pH) d d(pH) d I D E u x = - 3 (3) Equation (3) shows that good resolution is favored by high field strength, low diffusion coefficient, high mobility slope du/ d(pH) and a narrow pH gradient. Of the variables, the diffusion coefficient and the mobility slope are intrinsic properties of the analytes, so only pH gradient and the field strength can be varied experimentally.The resolving power in imaged-CIEF employing a short capillary format can be adopted to the chip channel format. In conventional CIEF, where long capillaries (12–60 cm) are used, pH gradients are more shallow than those seen in imaged CIEF with short capillaries.Satisfactory resolution, however, still can be obtained by adopting a higher field strength. Narrow fusedsilica capillaries have excellent heat dissipation and so allow high field strengths (500–800 V cm21) to be applied. When a narrow pH gradient of 6–8 and a high voltage of 3 kV are used, a resolution of about 0.03 pH unit can be achieved with imaged- CIEF performed in a 5 cm long capillary.28 This resolution is slightly lower than that in optimized conventional single-point detection CIEF with a resolution of 0.01–0.02 pH unit,29 but, it is good enough for clinical analysis.Owing to the short capillary, imaged-CIEF attains equilibrium faster, e.g., within a few minutes (2–3 min), resulting in a faster analysis speed. For similar reasons, imaged chip IEF provides a fast separation technique with a good resolution. Conclusions Isoelectric focusing was successfully performed on a quartz microchip for the first time, where low molecular mass pI markers and myoglobin were used as model samples.Absorption imaging detection proved to be an ideal way to record the chip IEF process in real time. Coating of the inner wall of channels played an important role in the chip IEF performance. Good resolution, sensitivity and reproducibility are obtained for IEF performed on coated channels. Imaged chip IEF allows fast separations with a potential for increased throughput. Acknowledgements The authors thank Dr.Jiaqi Wu and Dr. Arthor Watson of Convergent Bioscience for their invaluable help and discussions. This work was supported by the National Science and Engineering Research Council of Canada. References 1 S. C. Terry, J. H. Jerman and J. B. Angell, IEEE Trans. Electron Devices, 1979, 26, 1880. 2 A. Manz, J. C. Fettinger, E. Verpoorte, H. Ludi, H. M. Widmer and D. J. Harrison, Trends Anal. Chem., 1991, 10, 144. 3 D. J. Harrison, A. Manz, Z. Fan H.Ludi and H. M. Widmer, Anal. Chem., 1992, 64, 1926. 4 S. C. Jacobson, R. Hergenroder, A. W. Moore and J. M. Ramsey, Anal. Chem., 1994, 66, 4127. 5 A. T. Woolley, G. F. Sensabaugh and R. A. Mathies, Anal. Chem., 1997, 69, 2181. 6 A. T. Woolley, K. Lao, A. N. Glazer and R. A. Mathies, Anal. Chem., 1998, 70, 684. Fig. 5 Separation of low molecular mass pI markers 7.4 and 6.6 on a coated channel. Concentration of each component: 4 mg ml21. Fig. 6 Electropherogram of myoglobin with a concentration of 400 mg ml21 separated by chip IEF at 11 min.Two variants are pI 7.2 and 6.8. 640 Analyst, 1999, 124, 637–6417 A. W. Moore, Jr., S. C. Jacobson and J. M. Ramsey, Anal. Chem., 1995, 67, 4184. 8 F. von Heeren, E. Verpoorte, A. Manz and W. Thormann, Anal. Chem., 1996, 68, 2044. 9 C. S. Effenhauser, A. Paulus, A. Manz and H. M. Widmer, Anal. Chem., 1994, 66,2949. 10 D. E. Raymond, A. Manz and H. M. Widmer, Anal Chem., 1994, 66, 2858. 11 S. C. Jacobson, R. Hergenroder, L. B. Koutny and J. M. Ramsey, Anal. Chem., 1994, 66, 2369. 12 H. Rilbe, Ann. N. Y. Acad. Sci., 1973, 209, 11. 13 P. G. Righetti, Isoelectric Focusing: Theory, Methodology and Applications, Elsevier, Amsterdam, 1983. 14 S. Hjerten and M. Zhu, J. Chromatogr., 1985, 346, 265. 15 J. R. Mazzeo and I. S. Krull, Anal. Chem., 1991, 63, 2852. 16 P. G. Righetti, C. Gelfi and M. Conti, J. Chromatogr., 1997, 699, 91. 17 R. Rodriguez-Diaz, T. Wehr and M. Zhu, Electrophoresis, 1997, 18, 2134. 18 J. Wu and J. Pawliszyn, Anal. Chem., 1992, 64, 2934. 19 J. Wu and J. Pawliszyn, Anal. Chem., 1994, 66, 867. 20 J. Wu and J. Pawliszyn, Analyst, 1995, 120, 1567. 21 J. Wu and J. Pawliszyn, Anal. Chem., 1995, 67, 2010. 22 Z. Liang, N. Chiem, G. Ocvirk, T. Tang, K. Fluri and D. J.Harrison, Anal. Chem., 1996, 68, 1040. 23 Z. Fan and D. J. Harrison, Anal. Chem., 1994, 66, 177. 24 S. Hjerten, J. Chromatogr. A, 1985, 347,191. 25 J. Wu, S. Li and A. Watson, J. Chromatogr. A, 1998, 817, 163. 26 J. M. Harnly and R. E. Fields, Appl. Spectrosc., 1997, 51, 334A. 27 C. T. Culbertson and J. W. Jorgenson,, Anal. Chem., 1998, 70, 2629. 28 J. Wu, C. Tragas, A. Watson and J. Pawliszyn, Anal. Chim. Acta, 1999, 383, 67. 29 T. Wehr, M. Zhu, R. Rodriguez, D. Burke and K. Duncan, Am. Biotechnol. Lab., 1990, 8, 22. Paper 8/09756I Analyst, 1999, 124, 637–641 641

ISSN:0003-2654    

DOI:10.1039/a809756i

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

Theory of analyte extraction by selected porous polymer SPME fibres† Tadeusz Górecki, Xiaomei Yu and Janusz Pawliszyn* Department of Chemistry and Waterloo Centre for Groundwater Research, University of Waterloo, Waterloo, Ontario, Canada, N2L 3G1 Received 2nd November 1998, Accepted 12th February 1999 Extraction of analytes by the new porous polymer solid phase microextraction (SPME) fibres is based on adsorption rather than absorption. The equilibrium theory developed for the liquid poly(dimethylsiloxane) (PDMS) coating does not apply to these coatings.The paper presents theoretical description of the extraction process for adsorption-type fibres, including PDMS–DVB (divinyl benzene), Carbowax–DVB and Carbowax–TR (template resin). The model is based on Langmuir adsorption isotherm. Expressions describing the amount of analyte extracted by the fibre in two- and three-phase systems are presented and discussed. The effect of selected experimental variables is discussed.In general, there is a non-linear dependence between the amount of an analyte extracted by the fibre and its concentration in a sample. The dependence can be approximated by a straight line for low concentrations only. Matrix composition can significantly affect the amount extracted. Interferences co-extracted with the analyte of interest may reduce the amount extracted and the quasi-linear range of the response. Great care should be exercised therefore when performing quantitative analysis with porous polymer SPME fibres.The phenomena discussed are illustrated on an example of benzene and 4-methyl-2-pentanone extraction from water by PDMS–DVB and Carbowax–DVB fibres. Introduction Solid phase microextraction (SPME) has gained widespread acceptance in many areas in recent years. It has been applied for the determination of a wide spectrum of analytes in a variety of matrices. The most widespread seems to be analysis of volatile and semi-volatile compounds in water.Examples in this area include determination of substituted benzene compounds,1,2 volatile organic compounds,3–5 polycyclic aromatic hydrocarbons and polychlorinated biphenyls,6 pesticides,7–12 phenols, 13,14 fatty acids,15 as well as lead and tetraethyllead.16 There are two distinct types of SPME coatings available commercially. The most widely used is poly(dimethylsiloxane) (PDMS), which is a liquid coating. Even though it looks like a solid, it is in fact a high viscosity rubbery liquid.Poly(acrylate) (PA) is a solid crystalline coating that turns into liquid at desorption temperatures. Both PDMS and PA extract analytes via absorption. The remaining coatings, including PDMS–DVB (divinylbenzene), Carbowax–DVB, Carbowax–TR (template resin—DVB with uniform pore sizes) and Carboxen, are mixed coatings, in which the primary extracting phase is a porous solid, extracting analytes via adsorption. Similarity of the names can be very deceptive, since the fundamentals of absorption and adsorption are different.Fig. 1 illustrates the initial and equilibrium/steady-state stages of the extraction process for absorption- and adsorption-type SPME coatings. Independently of the nature of a coating, analyte molecules initially get attached to its surface. Whether they migrate to the bulk of the coating or remain at its surface depends on the magnitude of the diffusion coefficient of an analyte in the coating.Diffusion coefficients of organic molecules in PDMS are close to those in organic solvents, therefore diffusion into PDMS is relatively fast and this coating extracts analytes via absorption. Diffusion coefficients in poly(acrylate) are lower by about an order of magnitude, but still large enough for absorption to be the primary extraction mechanism. On the other hand, diffusion coefficients of organic molecules in the bulk of divinylbenzene and Carboxen are so small that within the time frame of SPME † Presented in part at Pittcon ’99, March 7–12, 1999, Orlando, Florida, USA.Fig. 1 Comparison of absorption and adsorption extraction mechanisms (cross-sections of the coated fibres). Diagrams on the left illustrate the initial stages of the processes. Diagrams on the right illustrate the steadystate condition. Analyst, 1999, 124, 643–649 643analysis, essentially all the molecules remain on the surface of a coating. Should the organic molecules remain there for a very long time (measured in days or weeks rather than hours), they still might diffuse into the bulk of the coating (over very short distances).This would manifest itself during analysis as persistent carryover, difficult to eliminate even after repeated desorptions. For all practical purposes, however, adsorption is the only extraction mechanism for those coatings. Louch et al.17 developed equilibrium theory for coatings extracting analytes via absorption in two-phase systems (sample and fibre coating), while Zhang and Pawliszyn extended it to three-phase systems (sample-headspace-coating).18 A complete theory for absorption-type coatings was presented in a book by Pawliszyn.19 Recently, Ai developed theoretical descriptions of the dynamics of non-equilibrium direct extraction,20 as well as equilibrium21 and non-equilibrium22 headspace extraction. This paper presents the steady state theory of analyte extraction via adsorption by selected porous polymer fibres (PDMS/DVB, Carbowax/DVB, Carbowax/TR). The theory does not apply to Carboxen fibres.Theory Weak intermolecular interactions (and hydrophobic interactions when sampling from water)23 play the most important role in analyte extraction by the porous polymer SPME coatings available commercially. The number of surface sites where adsorption can take place is limited. When all such sites are occupied, no more analyte can be trapped (unless it can condense into pores by the capillary condensation mechanism). 24 This means that the dependence between the concentration of the analyte in a sample and the amount of the analyte extracted from this sample by a solid SPME coating cannot be linear over broad concentration ranges.In addition, while absorption is a non-competitive process, adsorption is by definition competitive, and a molecule with higher affinity for the surface can replace a molecule with lower affinity.Thus, the amount of the analyte extracted by the fibre from a sample can be significantly affected by sample matrix composition. The following considerations pertain to PDMS–DVB, Carbowax –DVB and Carbowax/TR coatings. No theory has been developed yet for the Carboxen coating, for which capillary condensation plays an important role. As a result, this coating behaves differently than the other coatings. For example, equilibration times for the DVB-based coatings are usually quite short.In headspace extraction of tetraethyllead (TEL) from water (25 mL sample in a 40 mL vial) these coatings reach equilibrium in less than 30 min. On the other hand, the amount of TEL extracted by the Carboxen coating increases with increased extraction time under similar conditions even after 48 h. Such a long time cannot be explained by poor mass transfer conditions—it can only result from slow filling of the pores with the condensing analyte. Consequently, the assumption that the analyte molecules can only attach themselves to active sites on the coating surface is not valid for the Carboxen coating.The dependence between the equilibrium concentration of a compound associated with the sorbent and its concentration in the solution is commonly referred to as adsorption isotherm. It will be illustrated later in this paper that Langmuir adsorption isotherm well describes equilibrium analyte extraction by PDMS/DVB and Carbowax/DVB coatings, therefore it has been used to develop the theoretical description of the process.In the Langmuir model, the surface has a limited number of adsorption sites that can be occupied by the sorbate. The following assumptions apply: (1) the adsorbing molecule adsorbs into an immobile state; (2) all sites are equivalent; (3) each site can hold at the most one molecule of the adsorbate, and (4) there are no interactions between adsorbate molecules on adjacent sites so that the equilibrium constant is independent of the coverage of the adsorbed species.24 Assumption (3) means that a monolayer of the adsorbate can be formed at the surface at the most.Adsorption is treated as a reaction where a molecule A reacts with an empty site, S, to yield an adsorbed complex Aad: A + S ) Aad (1) At equilibrium, surface concentration of A in mol cm22, [Aad], is described by the following equation: [ ] [ ] [ ] [ ] A S A A ad A A = + 0 1 K K (2) where [S0] is the total concentration of active sites on the surface (maximum surface concentration of the analyte) in mol cm22, KA is the adsorption equilibrium constant, and [A] is the concentration of A in the matrix.It would be cumbersome to use surface concentration expressed in number of moles per cm2 for the description of the SPME process. However, if we assume that the sorbent has a uniform pore size distribution and surface area throughout its bulk, surface concentrations can be replaced by bulk concentrations by multiplying both sides of eqn.(2) by the term F/Vf, where F is the surface area (in cm2). Such an assumption seems reasonable in view of the fact that the fibre-to-fibre reproducibility is usually very good for fibres originating from the same batch. We can now define the concentration of the analyte on the fibre CfA and the maximum concentration of active sites on the coating Cf max in the following way: C V fA ad f A = [ ] F (3) C V f max 0 f S = [ ] F (4) We will also use the symbol C° sA instead of [A] to denote analyte concentration in the sample at equilibrium. From these, we can define the equilibrium concentration of the analyte on the fibre, C° fA: C C K C K C fA f max A sA A sA ¥ ¥ ¥ = + 1 (5) It is evident that C° fA is not a linear function of equilibrium analyte concentration in the sample, except when the product KAC° sA is much smaller than one.This may happen when either the affinity of the analyte towards the coating is low, or its concentration in the sample is very low.The reciprocal of this equation yields: 1 1 1 C C C K C fA f f A sA ¥ ¥ = + max max (6) Therefore the plot of 1/C° fA vs. 1/C° sA should be a straight line with a slope of 1/Cf maxKA and an intercept of 1/Cf max. Eqn. (5) is difficult to use in practice, since it requires knowledge of the analyte concentration in the sample at equilibrium. It is more practical to determine the dependence between the initial concentration of the analyte in the sample (C0A) and the amount extracted.Mass balance can be used for this purpose: C0AVs = C° sAVs + C° fAVf (7) From eqn. (5), equilibrium concentration of the analyte is: C C K C C f sA fA A fA ¥ ¥ ¥ = - ( ) max (8) By combining eqns. (7) and (8), after a few rearrangements one gets: n C V K C V V C C V K V C C f f = = - + - ¥ ¥ ¥ fA f A A s f fA s A f fA 0 ( ) ( ) max max (9) 644 Analyst, 1999, 124, 643–649where n is the amount of the analyte extracted by the fibre at equilibrium.Eqn. (9) is an iterative dependence, since equilibrium analyte concentration on the fibre (C° fA) appears on both its sides. Nevertheless, it gives an insight into the nature of analyte extraction with porous polymer coatings, as will be illustrated later. It is in fact a quadratic equation, which can be solved analytically. Of the two roots obtained, only the following has a physical meaning: n C K V C V K V K C V C V K V C V C V V K f = + + - - + + + f max A 0A s A s A f max f 0A s A s f max f 0A s s 2 2 2 2 2 ( ) ( ) (10) A discussion of this dependence is presented in the Results and discussion section of this paper.In real life situations, one can hardly assume that only one compound will be extracted by the coating. Since adsorption is a competitive process, the presence of other compounds must affect the amount of analyte A extracted by the fibre (nA). In the following derivation, only one competing compound is taken into account.The same reasoning can be applied, however, to any number of compounds present in the sample. The concentration of analyte A on the fibre in the presence of a competing compound B is given by the following equation: C C K C K C K C fA f max A sA A sA B sB 1 + ¥ ¥ ¥ ¥ = + (11) where KB is the adsorption equilibrium constant for compound B, and C° sB is the equilibrium concentration of B in a sample. If more than two compounds were present in the sample, the denominator would contain additional KiC° si terms.Mass balance for A is again described by eqn. (7). A derivation similar to that described above yields the following relationship: n C V K C V V C C K C V K V C C A fA f A 0A s f f max fA B sB s A f f max fA = = - + + - ¥ ¥ ¥ ¥ ( ) ( ) ( ) 1 (12) Eqn. (12) can be solved in the same way as eqn. (9). The only root with a physical meaning has the form: n C V C K V C V K V K C K K C V C V K V K C C V C V V K C K s A fA f f max A f 0A s A s B sB A A f max f 0A s A s B sB f max f 0A s B sB A = = + + + + - - + + + + + ¥ ¥ ¥ ¥ ( ) ( ) ( ) ( ) ( ) 1 2 2 1 1 2 2 2 2 2 (13) Even though this dependence seems very complex, it can give insight into the extraction process, as will be illustrated in the Results and discussion section.SPME extraction can be carried out by immersing the fibre in the sample (direct extraction), or by exposing it to the sample headspace. In fact, when volatile compounds are analysed, headspace extraction is the preferred mode of operation.19 Mathematical description of the headspace extraction process is more complex than that of direct extraction, as in headspace extraction one has to deal with equilibria involving three phases: sample, its headspace, and the fibre coating.When the fibre is exposed to the headspace, partitioning of analytes occurs between the gas phase and the coating, as well as between the sample and the gas phase. Mass balance for such a system (containing one analyte only) can be written in the following way: C V C V C V C V 0A s sA s hA h fA f = + + ¥ ¥ ¥ (14) where C° hA is the equilibrium concentration of the analyte in the sample headspace, and Vh is the headspace volume.Based on the Langmuir model, equilibrium concentration of the analyte on the fibre coating can be defined as: C C K C K C fA f A hA A hA h h ¥ ¥ ¥ = + max 1 (15) where the subscript h in KAh denotes that this is the equilibrium constant for adsorption of the analyte from the gas phase (sample headspace), as opposed to the liquid phase.Equilibrium concentration of the analyte in the sample headspace is determined by dimensionless Henry’s law constant KHA: C K C hA H sA A ¥ ¥ = (16) Let us denote the product of KHA and KAh as KAA. Substituting eqns. (15) and (16) into eqn. (14), after a few rearrangements, yields: n C V K C V V C C V V K K V C C = = ¢ - + + ¢ - ¥ ¥ ¥ fA f A 0A s f f max fA s h H A f f max fA A ( ) ( ) (17) Analytical solution of eqn.(17) yields: n C K V C V K V V K K K C V C V K V C V C V V K V V C V K C V V K K = ¢ + ¢ + + ¢ + - ¢ - + ¢ + + + æ è ç ö ø ÷ + + ¢ f max A f 0A s A s h H A A f max f 0A s A s f max f 0A s hA H f s f max h H 0A s h H A A A A a 2 2 2 2 2 2 ( ) ( ) (18) Taking into account that usually the total volume of the system (e.g., sample vial) is fixed, we can define a = Vh/Vs, and eliminate Vh by substituting it with aVs: n C K V C V K V aK K K C V C V K V C V C V aK V C aV K C V aK K s = ¢ + ¢ + + ¢ + - ¢ - + ¢ + + + + + ¢ f max A f 0A s A s H A A f max f 0A s A s f max f 0A s H f f max s H 0A H A A A A a ( ) ( ) ( ) ( ) 1 2 2 1 2 2 2 2 2 (19) When more than one compound is extracted from the sample headspace, derivation similar to those above yields the following equation for analyte A extracted by the fibre: n C V K C V V C C V K C aV K K C K V C C A fA f A 0A s f f max fA s B sB s H B sB A f f max fA A = = ¢ - + ¢ + + ¢ + ¢ - ¥ ¥ ¥ ¥ ¥ ( ) ( ) ( ) ( ) 1 1 (20) where KAB = KBhKHB.The analytical solution of eqn. (20) is as follows: n C K V C V K V K C aK K K C V C V K V K C C V C V aK V C aV K C V K s A f max A f 0A s A s B sB H A A f max f 0A s A s B sB f max f 0A s H f f max H 0A s A A A = - ¢ + ¢ + + ¢ + + ¢ ¢ - + ¢ + ¢ + + + + + ¢ ¥ ¥ ( )( ) ( ) ( ) ( ) ( 1 1 2 2 1 1 2 2 2 B sB H A a C aK K ¥ + ¢ ) ( ) 2 2 1 2 (21) This dependence is very complicated, and it is not immediately clear when looking at eqn.(21) how the particular terms affect Analyst, 1999, 124, 643–649 645the amount extracted. Nevertheless, it can be used relatively easily to model equilibrium extraction conditions for various sets of input variables, as will be illustrated in the Results and discussion section. Experimental All the reagents were of analytical reagent grade. Benzene was purchased from Caledon Laboratories (Ontario, Canada), while i-propanol and 4-methyl-2-pentanone from Aldrich Chemical Co.(Milwaukee, WI). SPME holder and fibres (PDMS–DVB and Carbowax–DVB) were purchased from Supelco (Bellefonte, PA). Aqueous standard solutions of benzene and 4-methyl- 2-pentanone were prepared from primary dilution standard solutions of the analytes in methanol. Concentrations of the primary dilution standard solutions were such that 25 mL of a given solution added to 25 mL of water produced the desired concentration of the aqueous standard.In this way, the amount of methanol in each aqueous standard was constant. Aqueous standard solutions were prepared in 40 mL amber vials (Supelco). They were stirred during extraction with a digital magnetic stirrer (VWR model HPS 400, VWR Scientific of Canada, Ltd., Mississauga, ON) at 1200 rpm. Extraction was carried out from sample headspace. Extraction times were set in such a way that they were slightly longer than equilibration times of the analytes.For benzene, extraction time was 2 min, and for 4-methyl-2-pentanone it was 12 min. All analyses were performed using a Varian Star 3500 GC (Varian Associates, Sunnyvale, CA) equipped with a 30 m 3 0.25 mm 3 0.25 mm SPB-5 column (Supelco). The column was equipped with a 1 m fused silica precolumn. Hydrogen at 20 psi was used as the carrier gas. Injector temperature was held at 210 °C for SPME injections, and was temperature programmed from 55–250 °C at 250 °C min21 for syringe injections.Flame ionization detector (FID) was held at 250 °C. Oven temperature program for SPME injections was the following: 35 °C for 1 min, ramped to 120 °C at 15 °C min21, held for 1 min. For syringe injections, the initial oven temperature was 55 °C. Detector response factors were determined by injecting 0.5 mL of standard methanolic solutions of benzene (0.995 mg mL21) and 4-methyl-2-pentanone (0.887 mg mL21). The same benzene solution was used for quality control injections performed at least daily.Results and discussion Fig. 2 presents calibration curves obtained for i-propanol in the presence of 4-methyl-2-pentanone (methyl-isobutyl ketone; MIBK).25 Sampling was carried out from sample headspace with a PDMS/DVB fibre [for details, see ref. (25)]. The affinity of MIBK for the fibre coating was much higher than the affinity of i-propanol. As long as MIBK concentration remained low (10 times lower than the concentration of i-propanol; points represented by squares), the calibration curve remained linear up to ~ 75 mg L21, and the deviation from linearity at higher concentrations was not very significant. However, when MIBK concentration at each point was equal to that of i-propanol (circles), the dependence could be approximated by a straight line only up to ~ 25 mg L21.Moreover, at higher MIBK concentrations, displacement of i-propanol was evident. The amount of i-propanol extracted from the sample at 150 mg L21 was lower by almost 50% when MIBK concentration was also 150 mg L21, compared to the case when it was 15 mg L21.Fig. 2 illustrates therefore that the presence of interfering compounds can affect both the amount extracted and the linear range of the method for porous polymer fibres. Table 1 illustrates the effect of the interfering compound on the amount of the analyte extracted from the sample for the PDMS–DVB and Carbowax–DVB fibres. In this experiment, the concentration of the interfering compound was kept constant for all the analyte concentration levels.It is evident from this table that in all cases the presence of the interfering compound caused a reduction in the amount of the analyte extracted by both fibres. In general, the effect of MIBK on the extraction of Table 1 The effect of an interfering compound on the amount of analyte extracted by the fibre. Extraction of benzene in the presence of MIBK, and extraction of MIBK in the presence of benzene Mass of analyte extracted by the fibre/ng PDMS/DVB fibre Carbowax/DVB fibre Benzene MIBK Benzene MIBK Analyte concentration/ mg L21 No MIBK 3.5 mg L21 MIBK Difference (%) No benzene 0.85 mg L21 benzene Difference (%) No MIBK 5 mg L21 MIBK Difference (%) No benzene 0.85 mg L21 benzene Difference (%) 85 104 80 23 39 22 43 31 23 27 6 3 50 350 296 243 18 122 76 38 87 67 23 20 12 40 850 481 435 10 228 153 33 209 166 20 43 28 35 1700 792 704 11 343 249 27 292 218 25 69 51 26 3450 1097 1005 8 502 397 21 471 431 8 114 86 25 Fig. 2 Calibration curves for i-propanol in the presence of methyl-isobutyl ketone (MIBK) (PDMS/DVB fibre, headspace sampling). Squares—MIBK concentration 10 3 lower than i-propanol concentration; circles—MIBK concentration equal to i-propanol concentration; error bars represent ± one standard deviation of the measurement. 646 Analyst, 1999, 124, 643–649benzene was less pronounced than the effect of benzene on the extraction of MIBK (note that in benzene extraction MIBK concentration was higher than benzene concentration in all but the most concentrated samples).This can be easily explained taking into account that MIBK revealed lower affinity to the fibres examined than benzene did. As a result, benzene effect on MIBK extraction was more significant even at lower concentrations (0.85 mg L21). Eqn. (6) predicts that the plot of 1/CfA ° vs. 1/CsA ° should be a straight line. This was verified for headspace extraction of methyl-isobutyl ketone (MIBK) and benzene.For the Carbowax/ DVB fibre, the dependences were linear, with R2 values of 0.9992 and 0.9933 for MIBK and benzene, respectively. For the PDMS/DVB fibre, the R2 values were 1.000 and 0.9954, respectively. The high R2 values indicate that within the concentration ranges examined, the Langmuir isotherm model is suitable for the description of analyte adsorption on the fibres examined. The slopes of the lines were proportional to 1/Cf maxKAA (where KAA = KAhKHA), while the intercepts to 1/Cf max.The latter allowed estimation of Cf max. For the PDMS–DVB fibre, the estimated values were 33.3 and 14.3 mmol mL21 for benzene and MIBK, respectively. For the Carbowax/DVB fibre, those values were 16.8 and 5.0 mmol mL21. The differences between the estimates obtained for the same fibres with two different compounds are not surprising in the light of the fact that the fibres also extracted water.Even though it had much lower affinity to those fibres than the organic molecules, it was present in the headspace in concentrations higher by orders of magnitude than the analytes of interest. Thus, it was able to compete for the active sites on the surface, and effectively reduced their number [see the discussion of eqn. (12)]. Additionally, standard aqueous solutions contained also methanol in concentrations much higher than those of the analytes. In spite of its very good aqueous solubility, the amount of methanol extracted by the fibres was significant.Of the two analytes of interest, benzene had higher affinity to the coatings examined, therefore the reduction in the effective number of active sites was less significant for this compound. Determination of the ‘true’ Cf max value would require conditions bearing little resemblance to those usually encountered in SPME. To make sure that only one compound is sorbed, it would be necessary to expose the fibre to vacuum containing pure vapors of an organic compound, since it cannot be excluded that even permanent gases can cause effective reduction of the number of active sites. For these reasons, estimation of KAA was not carried out, since no meaningful results were expected.The form of eqn. (9) is very similar to that of the equation for n when coatings extracting analytes by absorption rather than adsorption are used:19 n KC V V V KV = + 0A s f s f (22) where K is the partition coefficient of the analyte between the sample and the coating.The main difference between eqn. (9) and (22) is the presence of the fibre concentration term (Cf max 2 C° fA in the numerator and denominator of eqn. (9) (also, note that the meaning of KA is entirely different than that of K:KA is adsorption equilibrium constant, while K is the partition coefficient). For very low analyte concentrations on the fibre, it can be assumed that Cf max > > C° fA. For this condition to be fulfilled, analyte concentration in the sample and/or its affinity for the fibre must be very low.When these requirement(s) are met, a linear dependence should be observed. If, however, the amount of the analyte on the fibre is not negligible compared to the total number of active sites, the dependence cannot be linear any more. Eqn. (12) indicates that the amount of analyte A extracted from the sample containing more than one compound (nA) must be lower than n from eqn.(9), as there is an additional term in the denominator of eqn. (12), which can only be greater than one. The difference does not have to be dramatic if the second term in the denominator of eqn. (12) is much larger than the first one, which can occur when the interfering compound is either present at a very low concentration, and/or is characterized by low affinity to the coating. In all other cases, one can expect that nA will be significantly lower than n.What is less obvious when looking at equation eqn. (12) is the fact that adsorption of interfering compounds affects also the linear range of the calibration curve. The term C° sB is the equilibrium concentration of B. Unless the volume of the sample is very large, in which case the equilibrium concentration of B is practically equal to its initial concentration, C° sB depends on the initial concentration of B and A in the same complex way in which C° sA depends on C0A and C° sB (thus C0B).Incorporating this dependence into eqn. (12) would make it very complex. Instead, we can picture this dependence in the following way: when B adsorbs on the surface of the coating, it reduces the number of adsorption sites available for A. This means that, effectively, Cf max is lower for A, hence the nonlinearity becomes significant at lower concentrations of A compared to the case when the sample contains no interfering compounds. Compared to eqn.(9), eqn. (1) contains an additional term in the denominator, VhKHA. Since this term can only be equal to or greater than 0, at equilibrium the amount of the analyte extracted from the sample headspace can only be equal to or lower than the amount extracted directly from the sample. It is obvious therefore that sensitivity in headspace sampling is usually lower than in direct sampling. To minimize the sensitivity loss, headspace volume should be kept small. On the other hand, headspace sampling eliminates many matrix-related problems, and is usually faster.19 Compared to SPME with liquid coatings, eqn.(17) contains the same additional term as eqn. (9), i.e., (Cf max 2 C° fA). The consequences are similar to those described for eqn. (9). Linear response can be expected only for equilibrium concentrations on the fibre that are much lower than Cf max. From eqn. (20), it is clear that the presence of additional compounds which are co-extracted from the headspace reduces the amount of analyte A extracted by the fibre, unless those compounds are present at very low concentrations in the sample at equilibrium, and/or the product of their Henry’s law constant and adsorption equilibrium constant for extraction from the gas phase is very small.Again, it is not immediately obvious when looking at this equation that additional compounds present in the headspace affect also the linear range of the method. For explanation of this phenomenon, see the discussion of eqn.(12) above. Fig. 3 illustrates the predicted (theoretical) dependence of the amount of the analyte extracted by the fibre vs. the initial concentration of the analyte in the sample for direct extraction when a single analyte is present in the sample, for three different equilibrium constant (KA) values. The plots were determined using eqn. (10). At low analyte concentrations, the dependencies can be approximated by straight lines.At higher concentrations they cease to be linear, and finally they level off when all active sites on the fibre surface are occupied by the analyte molecules. The shapes of the isotherms, and particularly their linear ranges, depend strongly on the KA value. When it is large (see the curve for KA = 1 000 000), the response remains practically linear until the fibre becomes saturated with the analyte. After this point, the curve levels off rather abruptly. When KA is low (see the curve for KA = 10 000), the initial quasi-linear range is narrower, but n changes with the initial analyte concentration C0A in a broader concentration range.Fig. 4 presents the theoretical dependence of the amount of analyte n extracted by the fibre on the initial concentration of the analyte in the sample for a single analyte and direct extraction, Analyst, 1999, 124, 643–649 647for three different Cf max values. It is clear that the concentration of active sites on the fibre has a profound effect on linearity of the response.The higher the number of active sites, the broader is the linear range of the isotherm. This is quite obvious when looking at eqn. (9). When Cf max is high, the value of the difference (Cf max 2 C° fA) is very close to Cf max for a broader range of C° fA values than when Cf max is low. For illustration purposes, the Cf max value estimated for the PDMS/DVB fibre with benzene (33.3 mmol mL21) corresponds to ~ 2.6 g L21.Fig. 5 presents the theoretical relationship between the amount of analyte A extracted by the fibre and the initial concentration of the analyte in the sample when two compounds are present in the sample (direct extraction), for three different KB values. It is intuitively obvious that when the interfering compound has high affinity for the fibre coating, the displacement effects are more pronounced. Indeed, Fig. 5 illustrates that for the same CA, nA decreases when KB increases.When the affinity for the fibre coating is similar for both compounds, the displacement effect is not very significant, especially when the concentration of the interfering compound(s) is low (see the curve for KB = 10 000). On the other hand, when KB is high, displacement is significant. It should be emphasized that, as already mentioned in the discussion of eqn. (12), the curves in Fig. 5 illustrate the effect of equilibrium concentration of the interfering compound in the sample on the amount of analyte A extracted by the fibre.When all other parameters are constant, for a given (small) volume Vs, the higher is the KB value, the lower is the equilibrium concentration of B for the same initial concentration C0B (i.e., more compound is extracted when the affinity for the coating is higher). In order for C° sB to be the same for the three curves in Fig. 5 corresponding to the case of two compounds that undergo extraction, the initial concentration of B would have to be higher for higher KB values.The curves do not illustrate therefore directly what is the effect of the interfering compound(s) when KB changes, while C0B remains constant. Fig. 6 illustrates a similar dependence for constant KB ( = 100 000) and three different C° sB values. It should be noted that in this case C° sB is proportional to C0B. It is clear from Fig. 6 that the amount of analyte A extracted by the fibre decreases when the concentration of the interfering compound increases.This is intuitively obvious, since at higher concentrations of interfering compound(s), a larger fraction of adsorption sites is occupied, therefore fewer sites are available for analyte A. Fig. 7 presents relationships between the amount of the analyte extracted by the fibre and the initial concentration of the analyte in the sample for a single analyte extracted from headspace. The total volume of the system (sample plus its headspace) was set to 40 mL, with the sample volume of 15 mL.Henry’s law constant value of KH = 1 was assumed for the calculations. The amount extracted decreases with decreasing KA value, and so does the initial linear range. The shapes of the relationships do not change much when the sample volume is increased or KH is decreased (not illustrated). The amount extracted increases slightly when sample volume increases, with the change being the most pronounced for analytes with the lowest KA.The amount extracted increases also when KH decreases, due to the fact that lower KH value means that the headspace capacity for the analyte is lower (i.e., at equilibrium fewer analyte molecules are present in the headspace of the sample, therefore more molecules are available for the coating Fig. 3 Amount of analyte extracted by the fibre vs. initial concentration of the analyte in the sample for a single analyte and direct extraction, for three different equilibrium constants.Assumptions: Cf max = 1.0 g L21, Vf = 0.5 mL, Vs = 2 mL. Fig. 4 Amount of analyte extracted by the fibre vs. initial concentration of the analyte in the sample for a single analyte and direct extraction, for three different Cf max values. Assumptions: KA = 10 000, Vf = 0.5 mL, Vs = 2 mL. Fig. 5 Amount of analyte A extracted by the fibre vs. initial concentration of the analyte in the sample when two compounds are present in the sample (direct extraction), for three different KB values.Assumptions: Cf max = 1.0 g L21, KA = 10 000, Vf = 0.5 mL, Vs = 2 mL, = 10 mg L21. Fig. 6 Amount of analyte A extracted by the fibre vs. initial concentration of the analyte in the sample when two compounds are present in the sample (direct extraction), for three different C° sB values. Assumptions: Cf max = 1.0 g L21, KA = 10 000, Vf = 0.5 mL, Vs = 2 mL, KB = 100 000. 648 Analyst, 1999, 124, 643–649phase, since the total number of molecules in the system remains constant).The courses of the relationships between the amount of analyte A extracted by the fibre and the initial concentration of the analyte in the sample when two compounds are present in the sample and extraction is carried out from sample headspace are very similar to those presented in Fig. 5 for direct sampling, therefore will not be discussed herein. Conclusions Adsorption is a competitive process, therefore matrix composition, as well as extraction conditions, all affect the amount of analyte extracted by the fibre.This makes quantitative analysis using solid coatings more difficult compared to liquid coatings. The equilibrium theory developed for selected porous polymer coatings (PDMS/DVB, Carbowax/DVB) presented here sheds light on the effect of a number of experimental variables on the amount of the analyte extracted by the fibre coating. The theory applies also to Carbowax/TR coatings, which chemically are Carbowax/DVB coatings.In general, porous polymer coatings can be expected to perform well for relatively clean matrices or matrices of constant composition, provided that the concentration of the analyte of interest is low (otherwise, the quasi-linear range of the calibration curve can be easily exceeded and nonlinear calibration is required). Special strategies can be applied when interfering compounds with high affinity to the coating are present in the sample (see ref. 25).It should be remembered that in most practical cases, one has to deal with systems where more than one compound undergoes adsorption on the fibre coating (e.g., the analyte of interest plus water, trace organics accompanying the analyte, etc.) Carboxen coating is a special case. It extracts analytes via adsorption, therefore general description of the extraction process is similar to that for porous polymer coatings. The main difference is that the pores in Carboxen are small enough to cause capillary condensation to occur.As a result, one cannot talk about reaching equilibrium when Carboxen fibres are used. Also, quite obviously, one of the basic assumptions of the Langmuir isotherm model stating that the surface of the adsorbent can be covered by a monomolecular layer of analyte molecules at the most is not fulfilled, therefore the model presented herein is not applicable to Carboxen coatings. References 1 C. L. Arthur, L. M. Killam, S. Motlagh, M.Lim, D. W. Potter and J. Pawliszyn, Environ. Sci. Technol., 1992, 26, 979. 2 B. L. Wittkamp and D. C. Tilotta, Anal.Chem. 1995, 67, 600. 3 C. L. Arthur, K. Pratt, S. Motlagh, J. Pawliszyn and R. P. Belardi, J. High Resolut. Chromatogr., 1992, 15, 741. 4 J. J. Langenfeld, S. B. Hawthorne and D. J. Miller, Anal. Chem., 1996, 68, 144. 5 T. Nilsson, F. Ferrari and S. Facchetti, Proc. 18th Int. Symp. on Capillary Chrom., H�uthig Verlag, Germany, Riva del Garda, 1996, p. 618. 6 D. Potter and J. Pawliszyn, Environ. Sci. Technol., 1994, 28, 298. 7 A. A. Boyd-Boland and J. Pawliszyn, J. Chromatogr., 1995, 704, 163. 8 R. Eisert, K. Levsen and G. Wuensch, J. Chromatogr., 1994, 683, 175. 9 P. Popp, K. Kalbitz and G. Oppermann, J. Chromatogr., 1994, 687, 133. 10 X. Lee, T. Kumazawa, T. Taguchi, K. Sato and O. Suzuki, Hochudoku, 1995, 13, 122. 11 K. N. Graham, L. P. Sarna, G. R. B. Webster, J. D. Gaynor and H. Y. F. Ng, J. Chromatogr., 1996, 725, 129. 12 T. Górecki, R. Mindrup and J. Pawliszyn, Analyst, 1996, 121, 1381. 13 K. Buchholz and J. Pawliszyn, Anal. Chem., 1994, 66, 160. 14 B. Schaefer and W. Engewald, Fresenius’ J. Anal. Chem., 1995, 352, 535. 15 L. Pan, M. Adams and J. Pawliszyn, Anal. Chem., 1995, 67, 4396. 16 T. Górecki and J. Pawliszyn, Anal. Chem., 1996, 68, 3008. 17 D. Louch, S. Motlagh and J. Pawliszyn, Anal. Chem., 1992, 64, 1187. 18 Z. Zhang and J. Pawliszyn, Anal. Chem., 1993, 65, 1843. 19 J. Pawliszyn, Solid Phase Microextraction, John Wiley & Sons, Inc., New York/Chichester/Brisbane/Toronto/Singapore, 1997, ch. 3, p. 43. 20 J. Ai, Anal. Chem., 1997, 69, 1230. 21 J. Ai, Anal. Chem., 1997, 69, 3260. 22 J. Ai, Anal. Chem., 1998, 70, 4822. 23 R. P. Schwarzenbach, P. M. Gschwend and D. M. Imboden, Environmental Organic Chemistry, John Wiley, New York/Chichester/ Brisbane/Toronto/Singapore, 1993, ch. 11, p. 255. 24 R. I. Masel, Principles of Adsorption and Reaction on Solid Surfaces, John Wiley, & Sons, Inc., New York/Chichester/Brisbane/Toronto/ Singapore, 1996. 25 T. Górecki, P. Martos and J. Pawliszyn, Anal. Chem., 1998, 70(1), 19. Paper 8/08487D Fig. 7 Amount of analyte extracted by the fibre vs. initial concentration of the analyte in the sample for a single analyte and headspace extraction, for three different KA values. Assumptions: Vf = 0.5 mL, Cf max = 1 g L21, Vs = 15 mL, Vh = 25 mL, KH = 1. Analyst, 1999, 124, 643–649 6

ISSN:0003-2654    

DOI:10.1039/a808487d

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

In-tube solid phase micro-extraction–gas chromatography of volatile compounds in aqueous solution Boon Chong Dennis Tan,a Philip J. Marriott,*a Hian Kee Leeb and Paul D. Morrisona a Royal Melbourne Institute of Technology, Department of Applied Chemistry, GPO Box 2476V, Melbourne, Victoria 3001, Australia. E-mail: Philip.Marriott@rmit.edu.au b National University of Singapore, Department of Chemistry, 10 Kent Ridge Crescent, Singapore 119260, Singapore Received 2nd March 1999, Accepted 30th March 1999 This paper describes the use of conventional coated capillary gas chromatography columns for sorption of organic solutes from aqueous solution, with subsequent gas chromatographic analysis. The essential principles are similar to those of solid phase extraction (SPE) and solid phase micro-extraction (SPME); this approach may be referred to as in-tube solid phase micro-extraction (ITSPME).The technique was evaluated using toluene in water as the initial test solute, and a mixture of BTEX solutes (benzene, toluene, ethylbenzene, xylenes) in Milli-Q water was used to further characterise ITSPME.A 1 m length of capillary GC column was used for sorption of analytes from aqueous solution passed through the capillary by using nitrogen pressure. Collection of small fractions of aqueous solution issuing from the capillary enabled a sorption profile to be generated, with initial fractions depleted in analyte. A Boltzmann curve could be fitted to the sorption profile data, exhibiting good agreement with experimental data.For recovery of sorbed toluene, a single 100 mL aliquot of hexane was passed through the column as a stripping solvent. The back-extraction step was quantitative. Equilibrium extraction of solutes shows that the total amount of recovered solute is proportional to its initial concentration in the extracted aqueous solution and allows distribution constants to be readily estimated. For BTEX solutes, K values were similar to those reported for SPME and literature Kow values.For toluene, log K decreases from 2.47 to 1.48 when the sorption column temperature increases from 20 to 30 °C; adding salt or reducing the pH of the aqueous solution increases the degree of extraction of phenols, agreeing with general considerations on solute partitioning behaviour. Introduction Solvent free sample preparation methods or those employing less organic solvent are becoming increasingly important1 and may induce a major change in analytical methodology.2 Practical alternatives to existing sample preparation methods may therefore need to be formulated.Solid phase extraction (SPE) and solid phase micro-extraction (SPME) have emerged as efficient, popular alternative extraction techniques.3 A broad array of applications have been reported, such as determining octanol–water partition coefficients,4 detecting BTEX in water5 and analysing pesticide solutions.Gas-phase extraction and a variety of headspace applications are available. Whilst these show that SPE and SPME are good alternative methods, basic principles still require evaluation.6 An effective alternative to these techniques is presented in this paper, in-tube solid phase micro-extraction (ITSPME). Recently, Pawliszyn and coworkers have reported the automation of a similar approach involving coupling of the extraction column to high-performance liquid chromatography (HPLC).6 By extension, ITSPME should also have the potential to be coupled to other analytical instruments, such as gas chromatography (GC) or capillary electrophoresis (CE).ITSPME utilizes similar principles to SPME. Exhaustive extraction might not occur, with equilibrium partitioning of the analyte between the sample solution and the extraction medium (the stationary phase coated to the inner walls of the extracting capillary column). The sorbing phase can be selected according to the type of analytes to be determined, e.g., a non-polar phase if the analyte of interest is non-polar.1,7 In ITSPME, analyte solution is passed through the capillary at a reasonably slow flow rate.The amount of analyte sorbed by the stationary phase at equilibrium is directly related to its concentration in the sample solution, which can be described in a similar manner to SPME1,4 by equation (1): M KV C V KV V s sample sample s s sample = + (1) where Ms is the mass of an analyte sorbed by the stationary phase, Vs and Vsample are the volumes of the stationary phase and the sample passing through the capillary column, respectively, K is the partition coefficient of the analyte between the stationary phase and the sample matrix, and Csample is the initial concentration of the analyte in the sample in mass per unit volume.As in SPME, if Vsample is large (Vsample > KVs), the amount of the analyte extracted8 is: Ms = KCsample Vs (2) Following aqueous sample extraction, the sorbed analyte can be stripped from the stationary phase with minimum amount of organic solvent, and the extract analysed by GC-FID.In addition, the aqueous solution can be monitored both prior to extraction and also in the stream which issues from the extraction capillary. The problem of analysing aqueous samples has been addressed.9 Conceptually, ITSPME should preserve the advantages of SPE and SPME, but may offer potential benefits regarding quantitation and automation.ITSPME may use conventional capillary GC columns for sorption, with thermally stable, non-extractable bonded phases making the phase more robust than thick film phases on fibres. Conversely, thick films are not readily prepared for wall-coated capillary columns. Thus, ITSPME possesses complementary advantages to SPME, as outlined in this paper. Analyst, 1999, 124, 651–655 651Experimental Reagents All reagents used were of analytical reagent grade.Benzene, toluene, ethylbenzene, o-xylene, m-xylene, p-xylene, m-nitrophenol, p-cresol, p-tert-butylphenol, 2,4-dichlorophenol, 2,4,6-trichlorophenol, acetone, hexane, methanol, potassium chloride, sodium chloride, concentrated hydrochloric acid and 2-hexanone were purchased from BDH (Sydney, Australia) or Aldrich (Sydney, Australia); Milli-Q water (Millipore, Bedford, MA, USA) was used throughout. The following stock standard solutions were prepared in water and hexane (BTEX analysis) or methanol (phenols analysis): 100 mg L21 toluene, 100 mg L21 ethylbenzene, and mixtures of 100 mg L21 of each component of BTEX and 100 mg L21 of each component of phenols. 2-Hexanone (100 mg L21) in hexane solution and 100 mg L21 o-xylene in methanol were also prepared. In order to ensure that the chemicals were adequately dissolved in the solvent, 0.5 mL of acetone was used to dissolve the reagent first, followed by dilution as required with solvent Milli-Q water or hexane.5,10 The effect of pH and salt on efficiency of extraction in the ITSPME technique was examined.A pH 2 buffer was prepared with 25 mL of 0.2 m KCl and 6.5 mL of 0.2 m HCl in 100 mL of water, and saturated salt solutions were prepared with NaCl. Instrumentation A Shimadzu GC 17A with an autosampler and FID detector (Shimadzu Scientific Instruments, Rydalmere, NSW, Australia) was used for all gas chromatographic analyses. The fused silica capillary column used for GC was 30 m 3 0.25 mm with 0.25 mm film thickness BPX5 phase (SGE International, Ringwood, Australia). The conditions for the analysis were as follows.BTEX analysis: column flow approximately 1.9 mL min21; linear velocity approximately 35.6 cm s21; column oven at 93 °C (this temperature allowed acceptable analysis time without causing the toluene peak to overlap with trace acetone solvent peak); injection port at 250 °C with split injection (split ratio 1 : 20); FID detector at 320 °C. Phenols analysis: column flow approximately 1.4 mL min21; linear velocity approximately 30 cm s21.The injector, operated in splitless mode, was maintained at 200 °C; the FID detector was at 275 °C. The temperature program was 50 °C for 1 min, ramp to 190 °C at 10 °C min21, final hold time 1 min at 190 °C. In-tube solid phase micro-extraction (ITSPME). To perform extraction of BTEX and phenols using ITSPME, 2 different types of capillary GC columns were used.The first was 1 m long, 0.25 mm internal diameter with 3.5 mm thick BP1 (100% methylsiloxane) stationary phase, while the second was 1 m long, 0.32 mm internal diameter with 1 mm thick BP20 (polyethylene glycol) stationary phase (both columns from SGE International, Ringwood, Australia). Nitrogen gas was used to provide head pressure to the sample vial to force the aqueous solution through the capillary (Fig. 1). Most extractions were carried out at 20 °C and the capillary may be immersed in a water bath for temperature control.Two forms of extraction were performed, forward extraction and back extraction, as described below. Forward extraction. A volume of solution (in water) was prepared at the desired concentration from the stock solution. It was then forced through the capillary by applying nitrogen head pressure. Solution was passed through as a continuous stream, and collected in separate vials in 100 mL volumes or fractions. A suitable internal standard was added in the fractions.The analytical results of the collected aliquots could be used to generate a sorption profile. A separate standard was prepared in water to serve as a reference solution against which the collected aliquots could be compared. The extraction experiment may be conducted with different linear velocities of aqueous solution passing through the capillary column, by means of controlling the head pressure of nitrogen in the vial using control gauge pressure.Back extraction. After a specific volume of solute in water at a fixed concentration was passed through the extraction capillary column, the capillary was dried with a nitrogen flow and then a minimum volume of organic solvent was passed through the extracting capillary to strip the sorbed solute. A suitable internal standard was added, and the solution was analysed by using GC-FID. The result was then compared with the calibration plot from a series of standard solutions in the same organic solvent to estimate the amount of recovered solute.Results and discussion Sorption profiles Using forward extraction, four sorption profiles of aqueous toluene passing through the 1 m capillary GC column at flow rates of 20, 30, 50 and 70 mL min21 were obtained. The 100 mL fractions collected were analysed by GC-FID after addition of the internal standard. Each flow rate was repeated 3 times and Fig. 2 presents representative results from individual studies. Fig. 1 Experimental set up for the ITSPME technique. Fig. 2 Experimental sorption profiles for toluene at different flow rates. Lines of best fit based on a Boltzmann distribution are shown for each set of data. (2) 20 mL min21; (!) 30 mL min21; (/) 50mL min21; (½) 70 mL min21. 652 Analyst, 1999, 124, 651–655Variation in peak areas or area ratios may arise from (i) injection volume uncertainty, or (ii) variation in volume of either the collected toluene fraction or the added ethylbenzene volume.The toluene volume has the greatest uncertainty since the 100 mL volumes collected could not be controlled with precision. An alternative procedure would involve weighing the collection vial. For the 20 mL min21 flow data, the absence of toluene in fractions up to fraction 4 is noted; fraction 13 and later fractions have essentially the same level of toluene as that in the reference solution, shown as a broken line in Fig. 2, indicating 100% ‘breakthrough’.As the flow rate, and hence velocity, of the aqueous toluene through the extracting capillary increases, extraction is less complete for the early collected fractions, and traces of toluene could be detected in fraction 1 of the sorption profiles for faster flow rates. The curves also become less steep than those at slow velocities. Experimental uncertainties meant that data for the extraction profile did not precisely fit a smooth curve; however, a Boltzmann-type curve could be fitted to experimental data, as seen for the sorption profile data in Fig. 2. The steepness of the curve increased for slower flow rates. Integration of the Boltzmann equation can be used to give an estimate of the total sorbed solute amount. Determination of partition coefficient, K, for toluene between water and BP1 phase The 1 m BP1 capillary has a stationary phase volume of 2750 nL. Each 100 mL fraction of aqueous solution passing through the capillary has a maximum amount of 2 mg of toluene that can be sorbed, with less sorbed when breakthrough occurs.Using results for a 20 mL min21 flow of aqueous toluene, the total amount of toluene sorbed by the capillary is 16.2 mg (Table 1). Using K = Cs/Cm (where Cs is the concentration in the stationary phase and Cm the concentration in the mobile phase) the K value of toluene between water and stationary phase can be calculated. At equilibrium (i.e., at 100% breakthrough) Cs = 16.2 mg per 2750 nL = 5801 mg L21 and Cm is the concentration of the original toluene standard, i.e., Cm = 20 mg L21.Thus, log K = 2.47, which is in good agreement with the K value determined by equation (2), log Kow and log K determined from SPME (Table 2)8,11–16 reported for the same temperature. Recovery of the sorbed toluene by back extraction should yield 81 mg L21 of toluene (16.2 mg per 100 mL of hexane, diluted 1 : 1 with IS solution). The 100 mL hexane strip, with an added IS, was found to contain 66 mg L21 toluene, indicating a recovery of about 80% toluene; again, the volume uncertainty can lead to error in the calculated value.Determination of partition coefficient, K, for BTEX compounds K values for BTEX compounds were determined as above, in triplicate. A 2.5 mL mixture of BTEX (each 20 mg L21) in Milli-Q water was passed through the capillary and it was assumed that each solute reached 100% sorption. Results showed that different analytes had different recoveries, and hence a different affinity for the stationary phase of the extraction column.K values are reported and compared to literature log Kow and log K (SPME) values (for a BP1-like coating fibre) (Table 2). Good agreement was found, so apparently the methyl siloxane stationary phase behaves similarly to octanol in the octanol–water partition experiment. Dependence of ITSPME on extraction temperature, solution pH and salt content The back extraction procedure for 20 mg L21 toluene was repeated at least 4 times with the same piece of capillary to ensure that the extraction was consistent. The extraction was then carried out at a temperature of 30 °C, with a decrease in the amount of toluene sorbed expected and confirmed (Fig. 3). Increasing the temperature decreases the analyte Cs; in other words, there is less affinity for the stationary phase. Log K at 30 °C is estimated to be 1.48 and a temperature increase of 10 °C decreases the value of K by a factor of 10.This result is as expected from chromatographic results, where higher temperature gives a smaller retention volume in GC and in HPLC, and so smaller k and K values. Since salt affects solute solubility, a further study to test ITSPME for extraction of phenols from aqueous solution showed that both saturated salt solution and lowering the solution pH increase the extent of extraction by up to 10–20 times. For example, from Table 3 data, the peak for 2,4,6-TCP increases by about 3.5-fold, and ptert- butylphenol by about 25-fold.Fig. 4 is a representative GC trace of the extraction of the saturated salt, pH 2 buffered, aqueous solution. Since pKa values of phenols are !7, there is Table 1 Estimated amount of toluene sorbed per 100 mL of aqueous solution for the 20 mL min21 sorption profile at 20 °C Fraction number 1 2 3 4 5 6 7 8 9 10 11 12 13 Total Toluene sorbed/mg 2 2 2 2 2 1.7 1.6 1.3 0.9 0.5 0.2 0 0 16.2 Table 2 Distribution coefficient, K, data determined by ITSPME in comparison with literature log Kow and log K (SPME) (for 100% methyl siloxane coating) BTEX compound Log K (ITSPME)a Log Kow Log K (SPME) b Benzene 1.77 (7.4) 2.1317, 2.1315 2.3011, 2.1012, 2.308 Toluene 2.47 (4.2) 2.6917, 2.6915 2.8811, 2.5312, 2.888 Ethylbenzene 2.75 (3.9) 2.8413, 3.1515 3.3311, 2.7212, 3.338 p- and m-Xylene 2.81 (4.8) 3.1514 3.3111, 3.318 o-Xylene 2.69 (5.0) 2.7714, 2.7715 3.2611, 2.8212, 3.268 a Experimental value, this work; triplicate determinations; %rsd values in parentheses. b Values quoted for SPME studies.Fig. 3 Chromatograms of the hexane strip with different temperatures used for the aqueous extraction of 20 mg L21 of toluene. Curve a, extraction capillary at 20 °C; curve b, extraction capillary at 30 °C. Analyst, 1999, 124, 651–655 653only a moderate increase in extracted amounts with reduction in pH. Linearity of ITSPME extraction of toluene from water Equilibrium extraction conditions are established between the aqueous and stationary phase, so concentration and recovered solute should be linearly correlated.11,16 According to chromatographic theory, the partition ratio, k, is [equation (3)] k C V C V N N K V V = = = s s m m s m s m (3) where Ns and Nm are the number of moles of toluene in the stationary phase and mobile phase, respectively, and Vs and Vm are the volumes of the respective phases (note that Vm/Vs is normally referred to as the phase ratio).Rearranging the expression (3), and substituting CmVm for Nm, gives Ns = KVsCm = ACm where A = KVs = constant at a given temperature. Back extraction of 10, 20 and 40 mg L21 aqueous solutions of toluene, with added internal standard, was conducted in triplicate with an %RSD of about 4% for each concentration. Fig. 5 shows an overlay of one representative GC trace for each extracted concentration. The calibration graph of concentration versus area ratio (toluene/IS) had an R2 of 0.999. The gradient of the line allows determination of the partition coefficient if the volume of the stationary phase is known.The slope A is approximately 0.83 mL, giving an estimated log K value of 2.48, which is close to the previous estimation of log K for toluene between water and BP1 stationary phase. Conclusions Initial studies have proven ITSPME to be an effective extraction technique. Due to the availability of different polarity stationary phases, optimised extractions for target analytes of interest in routine analysis should be possible. If the extraction column is directly connected to the analytical column, maximum mass conservation in transferring the sorbed analyte to the analytical step can be realised.The first experiments using this have been encouraging. A solvent vent or waste line is required, and rather than solvent stripping, thermal desorption of extracted analyte will be explored with refocussing of the analyte at the head of the analytical column to minimise band broadening. A simple solution to this problem will be to use a recently demonstrated cryogenic technique to focus the target analyte band on the analytical column prior to chromatographic analysis.18 This preconcentration effect makes ITSPME suitable for sample preparation for trace analysis.Results presented here were logical and in agreement with expectations. The K values determined by ITSPME are comparable to those obtained with SPME and literature Kow values.It is anticipated that ITSPME will be suitable for routine analysis. The extraction capillary column can be re-used; thus far, there have been no carry-over problems, nor any evidence of performance deterioration in the extracting column. Bonded phase capillary columns are stable to organic solvent flushing. Large particulate material may require filtration prior to extraction, and, if necessary, a water rinse can be included between sorption and back extraction.Acknowledgement The authors wish to thank SGE International, Ringwood, Australia for providing the GC capillary columns for performing the extractions and Shimadzu Scientific Instruments, Rydalmere, Australia for GC facilities. References 1 Z. Zhang, M. J. Yang and J. Pawliszyn, Anal. Chem., 1994, 66, 844A. 2 D. Noble, Anal. Chem., 1993, 65, 693A. 3 C. L. Arthur and J. Pawliszyn, Anal. Chem., 1990, 62, 2145. Table 3 Comparison of extracted amounts of phenols (10 mg L21 each) from different aqueous matrices, as area ratio per cent of phenol peak/ xylene internal standard Milli-Q water pH 2 buffer pH 2 buffer + salta p-Cresol 10.1 (4.0)b 11.3 (6.2) 27.0 (3.7) 2,4-DCP 8.2 (9.6) 11.7 (3.4) 55.1 (2.0) p-tert-butylphenol 21.3 (2.8) 25.2 (9.1) 72.3 (4.3) 2,4,6-TCP 1.9 (6.2) 5.8 (3.5) 47.1 (5.5) m-Nitrophenol 0.7 (14.3) 1.6 (12.5) 4.5 (17.8) a Refer to Fig. 4 for the chromatogram of this solution. b %RSD (n = 3) values in parentheses.Fig. 4 Chromatogram of phenols sorbed from solution saturated with salt and buffered to pH 2. BP20 polyethylene glycol capillary used for sorption, with phenols back extracted from the capillary with 190 mL methanol. o- Xylene (10 mL of 25 mg L21 concentration) was added to collected methanol as internal standard. Fig. 5 Chromatograms of the hexane strip for aqueous extractions of toluene aqueous solution concentrations of (a) 10 mg L21, (b) 20 mg L21 and (c) 40 mg L21. Ethylbenzene is added to each extracted solution as internal standard at 25 mg L21. 654 Analyst, 1999, 124, 651–6554 J. R Dean, W. R. Tomlinson, V. Makovskaya, R. Cumming, M. Hetheridge and M. Comber, Anal. Chem., 1996, 68, 130. 5 S. P. Thomas, R. Sri Ranjan, G. R. B. Webster and L. P. Sarna, Environ. Sci. Technol., 1996, 30, 1521. 6 J. Pawliszyn and R. Eisert, Anal. Chem., 1997, 69, 3140. 7 R. G. Belardi and J. Pawliszyn, Water Pollut. Res. J. Can., 1989, 24, 179. 8 C. L. Arthur, D. W. Potter, K. D. Buchholz, S. Motlagh and J. Pawliszyn, LC-GC, 1992, 10, 656. 9 K. Grob, Split and Splitless Injection in Capillary Gas Chromatography with some remarks on PTV Injection; Huthig Verlag, Heidelberg, Germany, 1993. 10 C. L. Arthur, L. M. Killam, K. D. Buchholz, J. Pawliszyn and J. P. Berg, Anal. Chem., 1992, 64, 1960. 11 D. W. Potter and J. Pawliszyn, J. Chromatogr., 1992, 625, 247. 12 C. L. Arthur, L. M. Killam, S. Motlagh, D. W. Potter and J. Pawliszyn, Environ. Sci. Technol., 1992, 26, 979. 13 C. T. Chiou, D. W. Schmedding and M. Manes, Environ. Sci. Technol., 1982, 16, 4. 14 K. Verschuren, Handbook of Environmental Data on Organic Chemicals, Van Nostrand Reinhold, New York, USA, 2nd edn., 1983. 15 C. T. Chiou, Environ. Sci. Technol., 1985, 19, 57. 16 L. P. Sarna, G. R. B. Webster, M. R. Friesen-Fischer and R. Sri Ranjan, J. Chromatogr. A, 1994, 677, 201. 17 T. Fujita, J. Iwasa and C. Hansch, J. Am. Chem. Soc., 1964, 86, 5175. 18 R. M. Kinghorn and P. J. Marriott, Anal. Chem., 1997, 69, 2582. Paper 9/02567G Analyst, 1999, 124, 651–655 655

ISSN:0003-2654    

DOI:10.1039/a902567g

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

Determination of Cu, Fe, Mn, and Zn in blood fractions by SEC-HPLC-ICP-AES coupling Katerina Pomazal, Christian Prohaska, Ilse Steffan,* Gregor Reich and Josef F. K. Huber Institute of Analytical Chemistry, University of Vienna, Waehringerstr. 38, A-1090 Vienna, Austria Received 11th December 1998, Accepted 10th March 1999 The binding of Cu, Fe, Mn, and Zn to proteins in blood and in blood fractions was investigated, since their interactions in free radical metabolism in humans is of great interest.An HPLC-ICP-AES technique was developed allowing adequate separation of metalloproteins and of inorganic and organic metal species. For the separation of metalloproteins in erythrocytes and blood plasma a Merck Superformance Fractogel EMD BioSEC 650 (S) column was used. Size exclusion chromatography (SEC)-HPLC was hyphenated to ICP-AES both on-line and off-line for the detection of trace elements in the fractions resulting from HPLC separations. HPLC parameters, pH, temperature, flow rate and salt concentration were optimized for the protein separation and the optimal conditions were applied for the hyphenation to the ICP-AES detector.The separation column was calibrated with five standard proteins. For the element determination by ICP-AES a line selection with respect to the sensitivity was performed. Three different methods were used for the determination of trace elements in blood: direct determinations, on-line and off-line SEC-HPLC-ICP-AES measurements. For the optimizing experiments blood samples of one female subject were used. The direct determination by ICP-AES of the elements was performed in blood and blood fractions of ten different subjects to obtain the average concentration ranges.From the results the identification of the protein Cu/Zn superoxide dismutase in erythrocytes was possible. The LOD were 0.03 mg mL21 for Cu, 0.026 mg mL21 for Fe, 0.8 ng mL21 for Mn, and 0.09 mg mL21 for Zn in a synthetic blood matrix.Introduction Under normal physiological conditions cells of aerobic organisms generate potentially deleterious, reactive oxygen metabolites. A permanent generation of reactive oxygen radicals is system immanent. An imbalance of pro-oxidants and antioxidants in blood and blood cells is very dangerous and may cause damage.1–3 The phenomenon of oxygen toxicity is inherent in the atomic structure of oxygen. Molecular oxygen is a biradical that upon single electron addition is transformed into the partially reduced molecules O22·, H2O2, OH·, which may generate an array of additional reactive oxygen metabolites by further reactions, and cause extensive oxidative damage to biological macromolecules. 4–6 Free radicals are formed in vivo as organic and inorganic superoxides, peroxides, hydroxyl radicals or lipid peroxides.Breakdown of these species is controlled by antioxidant reagents. Increases in concentrations of oxygen and prooxidants in blood cause an increased production of reactive oxygen radicals leading to the disturbance of the equilibrium of the oxygen metabolism and finally to oxidative stress.DNA strands can break, and cell membranes and lipids can be oxidized leading to complete cell destruction. Decreased concentrations or defects of the antioxidant reagents result in oxidative stress as well.7–11 Proteins working as antioxidative reagents are involved in chronic diseases like diabetes mellitus, arteriosclerosis or cancers.They can influence the resulting after-effects of radical reactions in the cells. In relation to the oxidative stress defense system the concentrations of trace elements vary due to the changes in human metabolism. This may be regarded as an indication for disorders.12–15 Antioxidant enzymes regulating the metabolism of oxygen free radicals, superoxide dismutases (SODs), and glutathion peroxidase (GPX), are known to contain the essential trace elements Cu, Zn, Mn, and Se.Element species of Fe and Cu or other trace metals like Cr may also affect the balance of prooxidants and antioxidants in blood.16,17 Since blood fractions (blood plasma, erythrocytes, lymphocytes) contain a large variety of substances ranging from high molecular mass proteins to small metallic ions, there is a need for effective separation methods. In erythrocytes, hemoglobin, which represents up to 94% of the total amount of proteins, complicates the identification of other proteins, such as albumin in blood plasma.A separation of the proteins according to size is useful. The size is approximately directly proportional to the mass of the particles. Merck developed a solid phase tentacle system for the size exclusion chromatography (SEC) technique, which enables the preservation of the native form of the proteins in the separation systems. The column fill consists basically of a silica gel matrix carrying different polymeric chains with active ends.18–20 SEC enables a separation according to size and therefore to mass of the substances of interest thus giving the first important information as a result of the chromatographic procedure itself.Combination with other techniques enables an even more effective separation.21,22 Other techniques such as ion, affinity, and reversed-phase chromatography can lead to protein denaturation. The identification of the metal–protein complexes of interest is not possible according to size only, because blood fractions contain many proteins similar in size.Metals bound to proteins can be determined by element specific analytical methods and thereby the proteins can be identified. Atomic spectroscopic methods are frequently used for the specific determination of the mostly very low element concentrations in biological samples. In most of the matrices concerned inductively coupled plasma-atomic emission spectrometry (ICP-AES) and electrothermal atomic absorption spectrometry Analyst, 1999, 124, 657–663 657(ETAAS) are methods which enable a precise element specific determination of the elements in question after a high performance liquid chromatographic separation (HPLC) of the proteins.23–32 In this work ICP-AES was used for determination of the metals of interest.Three different methods were used: direct ICP-AES determination of the metals in whole blood and blood fractions after wet digestion; off-line coupling of SEC-HPLC to ICP-AES for the evaluation of the metal concentrations after the protein separation; and finally on-line hyphenation of SECHPLC and ICP-AES.The aim of this work was to optimize the hyphenation of ICPAES and HPLC for isolation and identification of Cu, Fe, Mn, and Zn particularly as metal–protein complexes in blood plasma and erythrocytes. The optimization of the coupled system was performed using the iron–hemoglobin complex.Experimental Chemicals Ficoll-Paque was provided by Pharmacia Biotech (Uppsala, Sweden); Heparin Immuno (5.000 I.E. mL21) by Immuno (Vienna, Austria), nitric acid suprapur, hydrochloric acid suprapur, hydrogen peroxide (30% H2O2) pro Analysi (p.A.), NaH2PO4 p.A., NaCl p.A. and single element standards p.A. (Cu, Fe, Mn, and Zn) by Merck (Darmstadt, Germany). Sample preparation and separation of proteins All experiments for the optimization of the methods were performed using blood samples of one female subject (29 years old, 170 cm tall, weight 55 kg).In whole blood, erythrocytes, blood plasma, and in the collected fractions after SEC-HPLC separation the Fe, Cu, Zn, and Mn concentrations were determined by ICP-AES. In addition these elements were determined in whole blood of ten different subjects for comparison. Blood samples were collected by Venflon to avoid metal contamination. The first 3 mL were discarded. Plasma and erythrocytes were separated by centrifugation (1200 g for ten minutes).After washing the cells with a physiological sodium chloride solution they were stored at 220 °C for further use. For HPLC analysis the erythrocytes were thawed, homogenized by a vortex mixer, centrifuged at 1200 g for ten minutes and filtered by a Micro Prep-Disc Filter [Bio-Rad (Richmond, CA, USA) PTFE membrane, pore size 5 mm] prior to injection. The plasma was thawed, homogenized and centrifuged as described for erythrocytes.Afterwards the proteins of blood plasma and erythrocytes were separated by SEC-HPLC. The SEC-HPLC procedure was carried out using an isocratic system from Hewlett-Packard (Avondale, PA, USA) (HP 1090 Liquid Chromatograph, PC HP 9000, Chem Station HP 9153C, printer HP Think Jet). A 250 mL injection loop was used (Hamilton, Reno, NV, USA, 250 mL Microliter Syringe). The chromatographic column used was Merck Superformance, Fractogel EMD BioSEC 650 (S), 600 3 16 mm id, particle size 20–40 mm, produced by Merck, Darmstadt, Germany.The column was thermostatted by a water cooling system. For the separation of the proteins an optimization of the parameters pH (6.5–7.2), temperature (25, 30 and 35 °C), flow rate (0.8–1.2 mL min21), and ion strength in the mobile phase (0.1–0.5 M NaCl) was performed. The mobile phase eluent solution (0.02 M NaH2PO4 + 0.3 M NaCl, at pH = 6.8 and t = 30 °C) was used, injection volumes varied between 100 and 250 mL. The proteins were detected by a UV/VIS diode array detector (DAD).The pressure on column was 38–42 bar (1 bar = 105 Pa). The optimized column conditions were used both for the off-line and the on-line coupling with ICP-AES. The calibration of the SEC column was performed with five standard proteins under optimized conditions (30 °C; pH = 6.8; 0.3 M NaCl, flow rate 1 mL min21). Myoglobin (18 kDa), ovalbumin (45 kDa), BSA (67 kDa), conalbumin (78 kDa), and g-globulin (150 kDa) were injected separately (0.5 mg mL21 of each protein in 100 mL) and as a mixture (0.1 mg mL21 of each protein in 100 mL).For the identification of the Cu/Zn–SOD, GPX, and hemoglobin, single standards and mixtures were injected (250 mL). The samples used are listed in Table 1. After separation by HPLC the amount of Cu, Fe, Mn, and Zn was determined by ICP-AES. ICP-AES determination For the element detection and determination an ARL 3520 ICP spectrometer (ARL, Ecublens, Switzerland) was used.All instrument parameters are listed in Table 2. (a) Line selection and determination of the limits of detection (LOD). Prior to analysis line selections were performed. Two spectral lines for Cu (324.694 and 327.396 nm), three spectral lines for Fe (259.940, 271.440 and 258.588 nm), one spectral line for Mn (257.61 nm), and one spectral line for Zn (213.856) were investigated. For the direct determination of the elements in whole blood, erythrocytes and blood plasma samples, scans using a synthetic blood matrix (containing 122.7 mg mL21 of Mg, 371.6 mg mL21 of P, 489.9 mg mL21 of Na, 406.5 mg mL21 of Ca, 3503.7 mg mL21 of K, 547.4 mg mL21 of Fe, and 2404.7 mg mL21 of S = ‘matrix’) and single element standards (10 mg ml21 each) were performed.For the line selection the following solutions were scanned: (a) matrix; (b) matrix + 10 mg element standard mL21; (c) 5 mg element standard mL21 in 1.4 M HNO3, and (d) 1.4 M HNO3. For the line selection for the determination of the elements after the SEC-HPLC separation blood plasma and erythrocytes were diluted with the HPLC eluent solution.The resulting solutions were scanned. Scans using (a) eluent; (b) 1 mL blood plasma in eluent (1 + 24); (c) 1 mL erythrocytes hemolyzate (1 + 9 in H2O) in eluent (1 + 24); (d) the solution (b) + 5 mg mL21 Table 1 Identification of the proteins Single proteins Mixture 1 Mixture 2 Mixture 3 Conc./ Volume/ Conc./ Volume/ Conc./ Volume/ Conc./ Volume/ Protein mg mL21 mL mg mL21 mL mg mL21 mL mg mL21 mL Erythrocytes —* 200 —* 200 — — — — Hemoglobin 5 200 5 200 5 75 5 20 SOD 5 200 5 200 5 75 5 20 GPX 1 100 — — 1 100 1 160 * Erythrocyte hemolyzate diluted 1 + 9 with doubly distilled water. 658 Analyst, 1999, 124, 657–663of Cu, Mn, and Zn; (e) the solution (c) + 5 mg mL21 of Cu, Mn, and Zn; (f) 5 mg element standard mL21 in 1.4 M HNO3; and (g) 1.4 M HNO3 were produced. The limits of detection (LOD) were calculated according to Boumans30 using 3s.LOD were determined in pure element standards and in the synthetic matrix. (b) Direct determination by ICP-AES. For the direct determination of the elements in whole blood, erythrocytes, and blood plasma the samples were digested in glass vessels with a mixture of nitric acid and hydrogen peroxide. For each series of digestions a reagent blank was prepared. The system was calibrated using mixed aqueous standards. The concentration ranges for the elements were: 0.5 to 5 mg mL21 of Cu, Mn and Zn for all samples, and 0.5 to 10 mg mL21 of Fe for blood plasma samples, and 10 to 100 mg mL21 for 1 + 9 dilutions of erythrocytes and whole blood samples, respectively.(c) Off-line coupling of SEC-HPLC-ICP-AES. After the SEC-HPLC separation of proteins, fractions of 2 mL volume were collected and analyzed by ICP-AES for Cu, Fe, Mn, and Zn. The fractions were collected by the fraction collector L 5200 Merck Hitachi. Calibration for ICP-AES was performed using adequate concentrations of the elements as described in (b) dissolved in the eluent solution.(d) On-line coupling of SEC-HPLC-ICP-AES. The determination of the Fe–hemoglobin complex in erythrocytes was performed using SEC-HPLC-ICP-AES coupling on-line. As a separation system a Fractogel EMD BioSEC 650 (S) column, a Merck Hitachi L-6200A Intelligent Pump, and a PTFE tube (1.5 id, 1 m long) were applied. For these experiments an erythrocyte hemolyzate diluted with doubly distilled water (1 + 9) was used.Results and discussion In order to understand the mechanisms and processes by which trace elements are absorbed, transported and incorporated into proteins it is important to monitor most of the essential trace elements because of their various interactions. The development of an analytical method for studying the protein binding and speciation of metals needs several prerequisites to be fulfilled: contamination has to be avoided, the separation system has to be optimized, interferences during spectroscopic detection have to be eliminated and the sensitivity for the metals of interest has to be optimized in the matrix given.For reducing the risk of contamination several precautions were applied. Glass-ware was purified by steaming with acid. The tendency for adsorption of metal species on glass was tested using standard solutions, e.g., 0.5 to 5 ng mL21 for manganese, since the lowest concentrations were expected for Mn.The recovery was 98 to 101%. The contamination caused by the reagents was corrected by a blank for iron, since the iron contamination of the prepared blank solutions was in the range of 0.5 mg mL21. For copper, zinc, and manganese the contamination was under the LOD, so a blank correction was not applicable. These data refer to measurements of pure aqueous standards. (see Table 2) The efficiency of the SEC-HPLC protein separation is influenced by the parameters pH, temperature, salt concentration, and flow rate.For both protein separation and coupling with ICP-AES the HPLC parameters had to be optimized with respect to the physical properties of the native proteins. For the separation of erythrocyte and blood plasma proteins the conditions described in the experimental section were selected. A pH of 6.8 was used, presenting a value in-between the known isoelectric values of the proteins of interest (hemoglobin 6.8, albumin 4.9, SOD 4.7–4.9, g-globulin 5.8–7.3, coeruloplasmin 4.4, catalase 5.8–6.5, cytochrome-c oxidase 10.6).Physiological processes take place under neutral conditions. Large amounts of erythrocyte proteins cannot be separated by the SEC column because of the similar size of the proteins and because of the overload of the column caused by hemoglobin. For blood plasma proteins the SEC method was not satisfactory either, but the albumin fraction could be separated from the other moieties.As a compromise a pH of 6.8 was chosen for all experiments. No significant changes in the resolution could be observed for temperatures between 25 °C and 35 °C. The results are shown in Fig. 1. Two replicates of each temperature were compared. All following experiments were performed at 30 °C, because of the best reproducibility of the retention time at this temperature. The concentration of NaCl in the eluent buffer solution was varied between 0.1 M and 0.5 M.The optimum concentration Table 2 Operating conditions for ICP-AES and line selection Operating conditions— Instrument: ARL 3520 ICP HF-generator Henry, 27.12 MHz RF power supply 1200 W Torch: Fassel type Outer gas flow 12 L min21 Intermediate gas flow 0.8 L min21 Aerosol carrier gas flow 1 L min21 Observation height 15 mm above coil Spectrometer: Paschen-Runge, sequential Grating 1080 lines mm21 Computer: DEC 316 sx Nebulizer: ARL MDSN, Babington-type Line selection— Element Wavelength/ nm LOD in matrix/ mg mL21 LOD in standards/ mg mL21 Background correction/nm Cu/1 324.694 0.03 0.01 ±0.04 Cu/2 327.396 0.05 0.03 ±0.04 Fe/1 259.940 0.026 0.024 ±0.04 Fe/2 271.440 0.384 0.358 ±0.04 Fe/3 258.588 0.072 0.068 ±0.05 Mn/1 257.610 0.0008 0.0003 ±0.05 Zn/1 213.856 0.086 0.044 ±0.04 Fig. 1 Optimization of the temperature, two replicates at each temperature: ..... 25 °C; — 30 °C; –– 35 °C. Analyst, 1999, 124, 657–663 659was found to be 0.3 M NaCl leading to a satisfactory peak resolution for the proteins in blood plasma and in erythrocytes.Despite the long separation time a flow rate of 1 mL min21 was chosen to enable hyphenation of the separation column to the ICP-AES. This flow rate is recommended by Merck as optimal for the separation column. This is in good agreement with the experiments performed using 0.8 mL min21 up to 1.2 mL min21 in 0.1 mL min21 steps. For 1.0 mL min21 the highest signal was obtained. After optimization of the HPLC system for the separation of the proteins, the column was calibrated with five proteins of different size.Fig. 2 shows the chromatograms of the single standards and of their mixture. The retention times of the single protein standards are in very good agreement with the peaks obtained by the separation of their mixture. The resulting calibration curve corresponds with the figure published in ref. 33 (selectivity curve). For the detection of the proteins only one wavelength is required, because for the given problem there is a need for the detection of the native proteins only.Two dimensional plots of erythrocytes and blood plasma chromatograms at 230 nm and 405 nm are presented in Fig. 3. The chromatograms of erythrocytes show a heme peak at 405 nm. In blood plasma only the signals of the amino acids could be observed (see dotted line at 230 nm). These chromatograms were compared to the profiles obtained by ICP-AES measurements (see below).The atomic spectrometric methods had to be optimized for the metal detection in the given matrix, as described below. ICP-AES determination (a) Line selection and determination of the limits of detection (LOD). Line selection was performed to choose the most sensitive analytical wavelength for the elements of interest. Spectral lines for analysis were selected with respect to two important criteria. The most sensitive line without spectral interferences in the sample matrix was used for analysis.For the given matrix three interference free lines for Fe, two for Cu, one for Zn, and one for Mn, respectively, were tested for their LOD in the matrix and in the eluent solution. In blood plasma and erythrocyte hemolyzate both diluted with eluent [solutions (b) and (c)] and the Fe concentration was high enough to be registered in the scans. For the other elements Cu and Zn and in particular for Mn single element standards (5 mg mL21 final concentration) were added [solutions (d) and (e)], because otherwise the signals obtained were too small for the line selection routine.For the determination of Cu the line at 324.694 nm, for Fe the line at 259.940 nm, for Mn the line at 257.610 nm, and for Zn the line at 213.856 nm were selected, because of their lowest LOD in the matrix given. (Table 2) (b) Direct determination by ICP-AES. The concentration of the elements in question were determined by direct measurement in whole blood, erythrocytes, and blood plasma samples of ten subjects (n = 10).The results are listed in Table 3. The second column of Table 3 gives the range of the measured concentration in digested samples (whole blood, erythrocytes, blood plasma) of 10 different individuals and the RSD range for three replicates. The RSD range from 0.1 to 9.8%. The highest RSD is caused by measurement of Cu in erythrocytes since the Cu content is very low in this case (approximately 0.3 mg mL21 in a complex matrix).The fourth column gives the mean concentration found for all persons tested (10) and the biological standard deviation (the Fig. 2 Separation of the standard protein mixture: 1 = BSA; 2 = conalbumin; 3 = g-globulin; 4 = myoglobin; 5 = ovalbumin; 6 = mixture. Fig. 3 2D chromatogram of erythrocytes (230 nm, 405 nm) —; and of blood plasma (230 nm) ·····. Table 3 Determination of Cu, Fe, Mn and Zn by ICP-AES Sample Measured conc./mg mL21 RSD (%) Conc.± SD/ mg mL21 RSD (%)* Whole blood— Cu 0.73–1.02 2.0–5.1 0.82 ± 0.12 14.6 Fe 183.49–267.01 0.1–1.2 223.18 ± 24.14 10.8 Mn 0.0039–0.0095 0.9–5.9 0.0063 ± 0.0016 25.4 Zn 4.28–6.44 0.4–3.1 5.34 ± 0.7 12.7 Erythrocytes— Cu 0.34–0.62 1.2–9.8 0.45 ± 0.13 28.9 Fe 431.81–527.07 0.9–1.4 484.04 ± 34.08 7.0 Mn 0.0086–0.0177 1.9–5.8 0.0134 ± 0.0029 21.3 Zn 9.42–12.12 1.2–3.7 10.5 ± 1.1 10.3 Blood plasma— Cu 0.84–1.45 5.7–8.1 1.07 ± 0.21 19.6 Fe 2.47–13.10 0.1–1.3 5.64 ± 3.56 63.0 Mn 0.0004–0.0027 0.7–5.0 0.0012 ± 0.0006 54.8 Zn 0.46–1.67 0.1–4.3 0.96 ± 0.36 37.5 * RSD gives the variation of the concentration according to biological differences, n = 10. 660 Analyst, 1999, 124, 657–663fifth column RSD%). It is evident that the biological range is rather high for the element investigated. It has to be stressed that the erythrocytes represent approximately 45% of the whole blood (hematocrit). In Table 3 the concentrations of elements in erythrocytes are not calculated according to their number per mL of whole blood, but are given for 1 mL of concentrate obtained after centrifugation.The element concentrations in whole blood, erythrocytes, and in blood plasma of the female subject used for the optimizing experiments were determined direct by ICP-AES as well. The measured concentration of Cu was 2.0 mg mL21 in whole blood, 1.0 mg mL21 in erythrocytes and 3.3 mg mL21 in blood plasma. The measured concentration of Fe was 213.7 mg mL21 in whole blood, 459.9 mg mL21 in erythrocytes and 6.2 mg mL21 in blood plasma. The measured concentration of Mn was 8.8 ng mL21 in whole blood, 17.0 ng mL21 in erythrocytes and 1.0 ng mL21 in blood plasma.The measured concentration of Zn was 6.1 mg mL21 in whole blood, 9.5 mg mL21 in erythrocytes and 1.3 mg mL21 in blood plasma. (c) Off-line coupling of SEC-HPLC-ICP-AES. The collected 2 mL fractions after the SEC-HPLC separation of the proteins were analyzed for Cu, Fe, Mn, and Zn by ICP-AES.Fig. 4a shows the determination of Fe in erythrocytes and in blood plasma. The determination of Fe in erythrocytes was easily performed in the collected fractions, because of the high Fe concentration (upper diagram). The main signal is caused by the Fe–hemoglobin complex. For the determination of Fe in blood plasma it was necessary to collect the fractions of three column separations to register sufficient intensities. Fig. 4b shows the signal intensities of Cu, Mn and Zn in erythrocytes after collection of the fractions of ten column operations.It is shown that Cu can be separated from Mn. The zinc peak is very small and is situated approximately at the same retention time as Cu. We conclude that the Zn signal belongs to Cu/Zn–SOD, which is also proved by the results of the protein separation and UV detection of the SOD standard on the same column (see also Fig. 6). Fig. 4c deals with Cu and Mn in blood plasma. As mentioned for Fig. 4b due to the low amounts of the elements present in blood plasma a ten-fold collection of the fractions was necessary to get measurable intensities. Cu in blood plasma seems to correspond to Cu bound to albumin, (the same retention time as the main signal in the chromatogram of blood plasma) since it is the main protein in this blood fraction. The second Cu peak like the Mn peaks could not be correlated to proteins investigated in plasma. We speculate that these element signals are caused by free metal ions present (see also Fig. 6). The profiles obtained by ICP-AES measurements (Fig. 4a–c) were compared to the two dimensional plots recorded at 230 nm (Fig. 3). (The comparison is shown in Fig. 6.) (d) On-line coupling of SEC-HPLC-ICP-AES. In order to identify and to quantify the metal–protein complexes, the HPLC separation system was coupled directly to the ICP-AES sample introduction device. The concentrations of Cu, Mn, and Zn protein complexes in blood are too low for the optimization of the hyphenated HPLC-ICP-AES method described. Therefore the Fe–heme complex of hemoglobin in erythrocytes was chosen to prove the reliability of the method.Hemoglobin represents up to 94% of the total protein in erythrocytes. In blood plasma the Fe present is primarily bound to transferrin and albumin. By coupling of the SEC-HPLC separation column and the ICP-AES detector the following chromatogram was obtained: Fe in erythrocytes at pH 6.8 in Fig. 5. The Fe intensity signal/V was plotted vs. separation time/min. Because of the relatively high separation volume of the SEC column (119 mL) it took 120 min to separate the proteins. Optimal protein separation was obtained using a sample flow rate of 1 mL min21, which is also a resonable flow rate for ICP-AES detection. The greatest challenge was to quantify the Fe intensity changing during the separation (peak) using a transient measuring mode. The Fe intensity was registered at time intervals of 0.5 s by the PC program ‘Nextview’ (BMC Puchheim-M�unchen, Germany).The statistical evaluation of the results was performed using the program ‘origin’ (MicroCal Software, Inc.). To summarize (c) and (d), Fig. 6 describes the applicability of the method developed. In part a the results of the protein separations and in part b the results of the off-line and on-line Fig. 4 Plot of the SEC-HPLC–ICP-AES off-line coupling: (a) Iron in erythrocytes and in blood plasma; (b) copper, manganese and zinc in erythrocytes; (c) copper and manganese in blood plasma.Analyst, 1999, 124, 657–663 661element detection are shown. The symbols mark (a) the retention times of the peak height maximum and (b) the fractions with the highest intensity. Conclusion The ICP-AES detection of Cu, Fe, Mn, and Zn in blood and blood fractions in combination with the separation of metalloproteins is a promising and suitable tool for the identification of proteins carrying these metals.The hyphenation of SEC-HPLC and ICP-AES was successfully applied off-line for the identification of the enzyme Cu/Zn–SOD in the erythrocytes. On-line only the determination of iron bound to hemoglobin in erythrocytes was performed in order to test the coupling system. For SOD-proteins containing copper and zinc, or manganese, the system has to be modified. The most important step will be the preconcentration of the protein complexes by column switching using different separation systems (hydrophobic-interaction chromatography or ionexchange chromatography).A limitation of the method described above is the multifold dilution of the sample during HPLC elution, therefore the dimensions of the separation system should be optimized as well. The method worked out was used for identification of the metal–protein complexes, but not for quantification. Especially for the on-line hyphenated system the calibration has to be performed in another study.The transient signals obtained have to be analyzed statistically to allow calibration. The calibration of the off-line coupling method could be performed easily. Iron could be detected in the 2 mL fractions after one separation run. The measured intensities were too low for manganese and not satisfying for copper and zinc after one separation run, since the proteins of interest are present in a very small concentration in blood and only 250 mL of the samples could be injected.A suitable separation system has to be found to overcome the collection of fractions of several separation runs. For the ICP-AES sample introduction system different nebulizers should be tested because of the high salt concentrations of the eluent used and other performances of different types of nebulizers. The flow rate has to be optimized according to the separation system used and to the requirements of the argon-plasma as discharge unit.Acknowledgement The authors wish to thank Merck, Darmstadt for supporting this study. The SEC-HPLC separation column was a present to Prof.-Dr. Dr. hc mult J.F.K. Huber. The authors wish to thank Doz. Dr. D.I.G. Vujicic for his kind support. References 1 K. R. Westerterp, G. A. L. Meijer, E. M. E. Janssen, W. H. M. Saris and F. Ten Hoor, Br. J. Nutr., 1992, 68, 21. 2 R. S. Sohal and R. Weindruch, Science, 1996, 273, 59. 3 B. Halliwell and J. M. Gutteridge, Arch. Biochem.Biophys., 1990, 280, 1. 4 M. Lopez-Torres, R. Perez-Campo, C. Rojas, S. Cadenas and G. Barja, Mech. Aging Dev., 1993, 70, 177. 5 F. Van Lente, Anal. Chem., 1993, 65, 374R. 6 J. M. Gutteridge, Chem. Biol. Interact., 1994, 91, 133. 7 R. Meneghini, M. S. Benfato, C. R. Bertoncini, H. Carvalho, S. A. Gurgueira, R. L. Robalinho, H. D. Teixeira, C. M. A. Wendel and A. L. T. O. Nascimento, Cancer J., 1995, 8, 3. 8 P. I. Oteiza, K. L. Olin, C. G. Fraga and C. L. Keen, J. Nutr., 1995, 125, 823. 9 J. M. Gutteridge, Med. Biol., 1985, 63, 41. 10 B. Halliwell, Neurotoxicology, 1984, 5, 113. 11 O. Lux and D. Naidoo, J. Nutr. Biochem., 1995, 6, 43. 12 B. Halliwell and J. M. Gutteridge, Methods Enzymol., 1990, 186, 1. 13 R. G. Cutler, Age, 1995, 18, 91. 14 M. Iskra, J. Patelski and W. Majewski, J. Trace Elem. Electrolytes Health Dis., 1993, 7, 185. 15 M. Rautalahti and J. Huttunen, Ann. Med., 1994, 26, 435. 16 M. Grootveld and B. Halliwell, Free Radical Res. Commun., 1986, 1, 243. 17 D. J. Anderson, Clin. Chem., 1993, 65, 434R. 18 L. Hagel, J. Chromatogr., 1992, 591, 47. 19 M. Favarato, C. A. Mizzen and D. R. McLachlan, J. Chromatogr., 1992, 576, 271. Fig. 5 Plot of the SEC-HPLC-ICP-AES on-line coupling: iron in erythrocytes. Fig. 6 Summary of the results: (a) signals of the proteins obtained by the SEC-HPLC separation; (b) signals of the metals obtained by atomic spectrometric detection. 662 Analyst, 1999, 124, 657–66320 D. J. Anderson, Clin. Chem., 1993, 65, 434R. 21 K. Vorauer, M. Skias, A. Trkola, P. Schulz and A. Jungbauer, J. Chromatogr., 1992, 625, 33. 22 B. Gercken and R. M. Barnes, Anal. Chem., 1991, 63, 283. 23 W. Mertz, Nutr. Rev., 1995, 53, 179. 24 G. F. Van Landeghem, P. C. D’Haese, L. V. Lamberts and M. E. De Broe, Anal. Chem., 1994, 66, 216. 25 L. A. Melton, M. L. Tracy and G. M�oller, Clin. Chem., 1990, 36, 247. 26 W. Flapper, A. G. M. Theeuwes and J. T. G. Kierkels, J. Chromatogr., 1990, 533, 47. 27 D. Beauchemin, J. C. Y. Le Blanc, G. R. Peters and A. T. Persaud, Anal. Chem., 1994, 66, 462R. 28 A. Mazzucotelli, A. Viarengo, L. Canesi, E. Ponzano and P. Rivaro, Analyst, 1991, 116, 605. 29 Inductively Coupled Plasmas in Analytical Atomic Spectrometry, A. Montaser and D. W. Golightly, VCH, Weinheim, 1987. 30 P. W. J. M. Boumans, in Basic Concepts and Characteristics of ICPAES, Inductively Coupled Plasma Emission Spectroscopy, Part 1, Methodology, Instrumentation, and Performance, ed. P. W. J. M. Boumans, Wiley, New York, 1987, ch. 4. 31 G. Vujicic and I. Steffan, ed. P. Hoffmann, 2nd Russian-Ukrainian- Austrian-German Analytical Symposium, Hirschegg/Kleinwalsertal, Germany, 1993, pp. 305–320. 32 I. Steffan and G. Vujicic, ed. P. Hoffmann, 2nd Russian-Ukrainian- Austrian-German Analytical Symposium, Hirschegg/Kleisertal, Germany, 1993, pp. 287–304. 33 ChromBook Merck, Germany, Biochromatography, 2nd edn., 1994. Paper 8/09688K Analyst, 1999, 124, 657–663 663

ISSN:0003-2654    

DOI:10.1039/a809688k

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

Gas chromatographic determination of some alkoxysilanes for use in occupational exposure assessment Jukka Mäittälä,a Sirpa Pennanena and Jyrki Liesivuori*b a Finnish Institute of Occupational Health, P.O.B. 93, FIN-70701 Kuopio, Finland b Department of Pharmacology and Toxicology, University of Kuopio, P.O.B. 1627, FIN-70211 Kuopio, Finland. E-mail: jyrki.liesivuori@uku.fi Received 1st February 1999, Accepted 23rd March 1999 The manufacture and application of organosilicon compounds, especially silanes, have increased dramatically during recent decades.This has led to an increase in the number of exposed workers in different areas of industry. Therefore, there is an urgent need for an analytical method which can assess exposure to these compounds. A capillary column gas chromatographic (GC) method was developed for detecting 3-methacryloxypropyltrimethoxysilane, 3-glycidoxypropyltrimethoxysilane and 3-aminopropyltriethoxysilane. The silanes diluted in heptane were analysed by GC using flame ionisation detection.Gas chromatography–mass spectrometry was used to confirm the identity of the GC peaks. The analytical range of the method varied from 1 or 5 mg ml21 to 500 mg ml21 depending on the silane being studied. The detection limits were 1 mg ml 21 for 3-methacryloxypropyltrimethoxysilane and 3-glycidoxypropyltrimethoxysilane and 5 mg ml21 for 3-aminopropyltriethoxysilane. The mean recovery of silanes tested with patch samples was > 95% for all of the silanes.The repeatability of the patch sample method for silanes varied from 6.5 to 10.1%. This new GC method allows the simultaneous determination of three organosilicon compounds for occupational exposure assessment. The most commonly used organic silicon compounds can be divided into siloxanes and silanes. Siloxanes are compounds which have a general silicon–oxygen–silicon backbone. The most important siloxane polymers are polydimethylsiloxanes, commonly known as silicones.These can be linear polymers with varying degrees of polymerisation or cyclic chains or rings. Organosilicon monomers (silanes) are composed of one or more silicon atom linked to hydrogen, nitrogen, halogen or some organic group. The variety and diversity of mono- and polysilanes are due to the great number of possibilities for combining substituents to the silicon. The organofunctional substituents often used with silanes are vinyl, methacryloxy, epoxy, sulfur, aminopropyl, ureic or isocyanato groups.Although they have been known since the 1950s, the use of organosilicon compounds such as silanes has increased dramatically during recent decades. Organofunctional trialkoxysilanes of the general formula RSi(OR1)3 are widely used as coupling agents for surface coatings and adhesion promoters of polymers to glass and metal surfaces. In glass-fibre production, propyltrialkoxysilanes such as 3-glycidoxypropyltrimethoxysilane, 3-methacryloxypropyltrimethoxysilane and 3-aminopropyltriethoxysilane (Fig. 1) are used as agents binding the glass filaments and various resins in the primary production.1 In hydrolysis, these silanes are partly transformed into analogous silanols which can readily form hydrogen bonds to water molecules adsorbed on an alloy (glass) surface. Side hydroxide groups become condensed, forming a siloxane network. The organofunctional end of the molecules, the methacrylate, epoxy or amino group, reacts with the resin and the organosilicon compounds effectively glue the resin to the glass filament.It has been previously estimated that the number of workers exposed to these three silanes increased from about 2000 in 1974 to over 130 000 in 1983 in the USA alone.2 Organic silicon polymers (silicones) have commonly been assayed using atomic absorption (AAS) or atomic emission spectrometry (AES).3 This is based on the determination of elemental silicon but it gives no information on the structure of the siloxane(s) present.Instead, infrared (IR) and nuclear magnetic resonance (NMR) spectrometric methods are generally used for both qualitative and quantitative analyses of siloxanes.3,4 In order to separate and identify the many different organosilicon compounds present in the same sample, it is necessary to use chromatographic methods as reviewed by Steinmeyer and Becker.5 However, only a few chromatographic methods, especially for trialkoxysilanes, have been published.Some of the recent studies on alkoxysilanes have dealt with NMR and IR spectrometric or gel permeation chromatographic methods.6,7 According to Shatz et al.,8 several gas chromatographic (GC) methods with different kinds of packing materials in packed columns have been developed for alkoxysilanes, but these are no longer acceptable. Mlejnek et al.6 introduced a method for vinyltri(2-methoxyethoxy)silane using GC with a fused-silica wall-coated open-tubular (WCOT) capillary column.A liquid chromatographic method was recently published9 for 3-glycidoxypropyltrimethoxysilane used as a fixing additive in silicone and polyurethane resins. According to the current literature, there is no method available for the assessment of occupational exposure to Fig. 1 Structures of (a) 3-glycidoxypropyltrimethoxysilane, (b) 3-methacryloxypropyltrimethoxysilane and (c) 3-aminopropyltriethoxysilane. Analyst, 1999, 124, 665–668 665commonly used trialkoxysilanes either for biological monitoring of workers or for occupational hygiene measurements at the work site.The need for this kind of exposure assay has arisen from several recent case reports describing contact allergy in the glass filament and plastics industries.9–11 The source of the contact dermatitis and sensitisation in the glass filament production was an amine-functional methoxysilane in two reported cases.10,11 A worker in the plastics industry was found to be allergic to an epoxysilane, 3-glycidoxypropyltrimethoxysilane, which was contaminated with a reactive epoxy diluent.9 In occupational hygiene measurements, several kinds of sampling methods are available for determining the inhalation or dermal exposure to a certain chemical agent.For the assessment of exposure through inhalation, air samples are collected on adsorbents of different material or filters or in absorbent liquids. More emphasis has been placed on dermal exposure recently, since respiratory exposure has decreased owing to improvements in preventative measures, e.g., better ventilation systems.12 Collection of patch samples and handwash samples is a commonly used method for assessing dermal exposure, especially in pesticide research.13,14 The aim of this study was to develop a sensitive and reliable GC method for the three above-mentioned trialkoxysilanes commonly used in glass-fibre production.We hoped to develop an efficient analytical tool for exposure assessment of workers handling several coating chemicals, especially silanes, in primary glass filament production.Experimental Chemicals The three trialkoxysilanes studied, 3-glycidoxypropyltrimethoxysilane (purity > 98%), 3-methacryloxypropyltrimethoxysilane ( > 98%) and 3-aminopropyltriethoxysilane ( > 98 %), manufactured by (Witco Europe S.A. Geneva, Switzerland), were supplied by a local glass filament factory which uses these silanes as detachment agents in the coating of glass filaments.The solvent, heptane, was obtained from Merck (Darmstadt, Germany) and was of analytical-reagent grade. Samples The validity of the method was tested with samples prepared under laboratory conditions where the three silanes were diluted with heptane in order to prepare the desired dilution series for GC analysis. Different concentrations of silane solutions were spread on a-cellulose patches (10 3 10 cm) which were extracted with 25 ml of heptane.The samples were then analysed by GC. The mean recovery of samples with the relative standard deviation (RSD) was calculated from five different concentrations in the linear analytical range. During validation of the method, we also tested methanol as the solvent but the recovery and sensitivity were better with heptane. Gas chromatography The gas chromatograph was a Hewlett-Packard (Avondale, PA, USA) Model 5880-A with a flame ionisation detector (FID).A DB-5 capillary column was used (J&W Scientific, Folsom, CA, USA) (30 m 3 0.32 mm id, film thickness 0.25 mm). The injection volume was 1 ml, the injector temperature was 220 °C and the splitless time was 0.5 min. Helium was used as the carrier gas at a flow rate of 1.5 ml min21 and nitrogen as the make-up gas at a flow rate of 20 ml min21. The initial temperature of the oven was 80 °C, which was raised at a rate of 3 °C min21 to 150 °C then at 20 °C min 21 to the final temperature of 250 °C, which was held for 3 min.The detector temperature was 250 °C. Gas chromatography–mass spectrometry A Hewlett-Packard Model 6890 gas chromatograph equipped with a Hewlett-Packard Model 5973 mass spectrometer and a Hewlett-Packard Model 6890 autosampler was used. The oven temperature programming was identical with that for GC-FID. The column was the same DB-5 and the carrier gas (helium) flow rate was 1.0 ml min21. The transfer line heater temperature was set at 280 °C.The ion source temperature was 200 °C and the electron energy was 70 eV. The injector temperature was 220 °C, the splitless time was 0.5 min and the injection volume was 1 ml. Results and discussion A method for the simultaneous GC determination of 3-methacryloxypropyltrimethoxysilane, 3-glycidoxypropyltrimethoxysilane and 3-aminopropyltriethoxysilane was developed (Fig. 2). The analytical range of the method for both 3-methacryloxypropyltrimethoxysilane and 3-glycidoxypropyltrimethoxysilane was 1–500 mg ml21 with a detection limit of 1 mg ml21. The corresponding values for 3-aminopropyltriethoxysilane were 5–500 and 5 mg ml21.The mean recovery for occupational hygiene samples (patch samples) was 102% with an RSD of 2% for 3-methacryloxypropyltrimethoxysilane. For 3-glycidoxypropyltrimethoxysilane, the mean recovery for samples was 103% with an RSD of 3% and for 3-aminopropyltriethoxysilane the values were 100 and 5%, respectively. The GC peaks of the three silanes studied were verified by mass spectrometry. The identification of these silanes was based on the interpretation of the electron ionization (EI) mass spectra (Fig. 3). The solvent experiments revealed that the use of heptane in the GC analysis improved the sensitivity compared with methanol (from 5 to 1 mg ml 21 and 20 to 5 mg ml 21) depending on the silane tested. The Dermal Exposure Network funded by the European Union is striving to develop a validated generic model of dermal Fig. 2 Gas chromatogram of 3-aminopropyltriethoxysilane, 3-methacryloxypropyltrimethoxysilane and 3-glycidoxypropyltrimethoxysilane. The retention times were 8.3, 14.3 and 15.2 min, respectively. The concentration of each silane was 50 mg ml21. 666 Analyst, 1999, 124, 665–668exposure for use in risk assessment. A number of models have been developed for specialist areas, e.g., the occupational use of pesticides and the occupational use of industrial chemical compounds.15 Validation of the models requires detailed measurements of all aspects of exposure, including skin surface contamination surveys.The patch sampling method has been applied in assessing dermal exposure to pesticides and wood preservative agents studied in our laboratory.14,16 Warm and Fig. 3 EI mass spectra of 3-aminopropyltriethoxysilane (top), 3-methacryloxypropyltrimethoxysilane (middle) and 3-glycidoxypropyltrimethoxysilane (bottom).Analyst, 1999, 124, 665–668 667humid working conditions, the mode of exposure (splashing, contact) and the physical state of the chemical substance (aerosol) found in greenhouses and glass-fibre production are surprisingly similar, as seen in surveys concerning chemical and microbiological exposures and general working conditions. 14,17,18 Our results showed that the GC-FID method developed for 3-methacryloxypropyltrimethoxysilane, 3-glycidoxypropyltrimethoxysilane and 3-aminopropyltriethoxysilane is sufficiently reliable for occupational hygiene measurements and evaluation of dermal exposure to these silanes from patch samples.With this method, one achieves equal sensitivities and linear analytical ranges for 3-methacryloxypropyltrimethoxysilane and 3-glycidoxypropyltrimethoxysilane and approximately the same reproducibility of parallel samples. The sensitivity for 3-aminopropyltriethoxysilane is slightly lower but nevertheless at the same level as for the other silanes.A lower detection limit could be achieved by concentrating the silane solutions via evaporation (to dryness) before the GC analysis, but we wanted to avoid any possible detachment of silanes from solid (glass) surfaces in case of unwanted hydrolysis. The volatilisation of the silanes during sample preparation is presumably negligible because of their low vapour pressure ( << 1 mmHg). The chemical heterogeneity of silanes poses challenges to analytical chemistry.In this work, the alkoxysilanes studied are similar in their chemical composition, all having the same propyltriethoxy or propyltrimethoxy structure. Nonetheless, these silanes show considerable differences in their physicochemical features, i.e., water solubilities and boiling-points. Special attention needs to be paid to avoid hydrolysis as the alkoxysilanes react readily with acids, bases or water (humidity). From a practical point of view, it will be a challenge to find an organic solvent suitable for the extraction of the organosilicon compounds, especially alkoxysilanes, and their metabolites from biological material.19 Only a few studies have been published on the toxicological and environmental effects of organic silicon compounds, probably because of analytical difficulties in the determination of the dose.In conclusion, a reliable and sensitive GC method was developed for the simultaneous determination of some alkoxysilanes commonly applied in glass-fibre production. This method will also prove useful in surveys to determine the amounts of organosilicon residues in the environment and to assess consumer exposure to personal care products such as shampoos and hair conditioners containing organic silicon compounds.The authors thank Mrs. Kirsi Immonen and Mr. Jarmo Hartikainen for their excellent technical assistance. Financial support from the Finnish Work Environment Fund is gratefully acknowledged.References 1 The Plastic Composites, ed. I. Airasmaa, J. Kokko, V. Komppa and O. Saarela, Gummerus, Jyväskylä, 1991, pp. 102–110 (in Finnish). 2 RTECS: Registry of Toxic Effects of Chemical Substances, National Institute for Occupational Safety and Health, Cincinnati, OH, 1999. 3 B. A. Cavic-Vlasak, M. Thompson and D. C. Smith, Analyst, 1996, 121, 53. 4 P. Fux, Analyst, 1989, 114, 445. 5 R. D. Steinmeyer and M. A. Becker, in The Analytical Chemistry of Silicones, ed. A. L. Smith, Wiley, New York, 1991, pp. 255–303. 6 O. Mlejnek, M. Lacuska and P. Liptak, J. Chromatogr., 1984, 286, 301. 7 T. Anagtopoulos, G. Eliades and G. Palaghias, Dent. Mater., 1993, 9, 182. 8 V. D. Shatz, R. Ya. Sturkovich and E. Lukevics, J. Chromatogr., 1979, 165, 257. 9 A. Dooms-Goossens, M. Bruce, L. Buysse, S. Fregert, B. Gruvberger and H. Stals, Contact Dermatitis, 1995, 33, 17. 10 F. Toffoletto, G. Cortona, G. Feltrin, A. Baj, E. Goggi and R. Cecchetti, Contact Dermatitis, 1994, 31, 320. 11 T. Heino, K. Haapa and F. Manelius, Contact Dermatitis, 1996, 34, 294. 12 V. Fiserova-Bergerova, Ann. Occup. Hyg., 1993, 37, 673. 13 R. A. Fenske, Ann. Occup. Hyg., 1993, 37, 687. 14 J. Kangas, S. Laitinen, A. Jauhiainen and K. Savolainen, Am. Ind. Hyg. Assoc. J., 1993, 54, 150. 15 J. J. Van Hemmen, Ann. Occup. Hyg., 1997, 41, 729. 16 S. Kröger, J. Liesivuori and A. Manninen, Int. Arch. Occup. Environ. Health, 1990, 62, 213. 17 A. Manninen, J. Kangas, A. Tuomainen and R. Tahvonen, Toxicol. Environ. Chem., 1996, 57, 213. 18 D. K. Milton, J. Amsel, C. E. Reed, P. L. Enright, L. R. Brown, G. L. Aughenbaugh and P. R. Morey, Am. J. Ind. Med., 1995, 28, 469. 19 S. Varaprath, K. L. Salyers, K. P. Plotzke and S. Nanavati, Anal. Biochem., 1998, 256, 14. Paper 9/00859D 668 Analyst, 1999, 124, 665–668

ISSN:0003-2654    

DOI:10.1039/a900859d

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

Characterization of within-tree variation of lignin components in Eucalyptus camaldulensis by pyrolysis–gas chromatography Hiroaki Yokoi,a Yasuyuki Ishida,a Hajime Ohtani,a Shin Tsuge,*a Tetsuya Sonodab and Toshihiro Onab a Department of Applied Chemistry, Graduate School of Engineering, Nagoya University, Nagoya 464-8603, Japan. E-mail: shin@apchem.nagoya-u.ac.jp b Kameyama Research Center, Forestry Research Institute, Oji Paper Co., Ltd., Kameyama, 519-0212, Japan Received 25th November 1998, Accepted 22nd March 1999 Pyrolysis–gas chromatography (Py-GC) using a vertical microfurnace pyrolyzer was applied to the precise determination of the ‘within-tree variation’ of the ratio of syringyl and guaiacyl units (S/G ratio) for lignin in Eucalyptus camaldulensis.On the pyrogram of the Eucalyptus obtained at 450 °C, many characteristic peaks derived from syringyl and guaiacyl units of lignin were reproducibly observed, together with those from cellulose and hemicellulose.On the basis of the relative intensities of those characteristic peaks relating to the syringyl and guaiacyl units, the ‘within-tree variation’ of the S/G ratio was determined precisely in about 1 h using only about 0.1 mg of the powdered wood sample with a 2% relative standard deviation. The results obtained were also compared with those obtained by the thioacidolysis method. Introduction Lignin is a three-dimensional cross-linked polymer consisting of various phenylpropane units, which constitutes a tree trunk together with cellulose and hemicellulose.It is known that the lignin in hardwoods such as eucalyptus and beech consists of guaiacylpropane units (G) and syringylpropane units (S) containing one and two methoxy groups, respectively, whereas the lignin in softwoods such as pine and spruce mostly consists of guaiacylpropane units (G).1 Moreover, the lignin in herbaceous plants, such as wheat and rice, comprises p-hydroxyphenylpropane units (H) without any methoxy groups in addition to G and S.It is well known that the lignin content and its chemical structure have a significant influence on the pulping process of the woods. In particular, the ratio of syringyl and guaiacyl units (S/G ratio) in hardwood lignin affects not only the pulping efficiency2 but also the strength of manufactured pulp and papers.3 Therefore, a simple and reliable method to determine the S/G ratio has been desirable in the pulp and paper industry.Recently, ‘quality breeding’, in which trees suitable for pulp production are selectively cultivated, has been carried out to increase pulp production efficiently. Generally, the results of ‘quality breeding’ have been judged on the basis of various pulp properties measured after pulping the wood. However, in order to carry out ‘quality breeding’ efficiently, the pulp properties of the wood should be determined on the basis of the lignin content and its S/G ratio in several milligrams of wood sample obtained from part of trunk without cutting down the whole tree.Furthermore, it has been generally observed that there are some variations in the S/G ratio even in the same trunk.1 Therefore, it is essential for ‘quality breeding’ to determine the within-tree variations in lignin content and its S/G ratio using a small amount of wood sample taken from any desired part of a trunk. So far, various methods utilizing chemical degradation followed by chromatography have mainly been employed for determination of lignin content and its S/G ratio in woods.Permanganate oxidation and alkaline nitrobenzene oxidation have often been used for this purpose.4 However, these degradation methods require not only tedious and timeconsuming procedures but also fairly large amounts of wood sample, up to 100 mg. Furthermore, because the strong acidity of the reagents used in the degradation process often causes the denaturation of the lignin structures, the results obtained do not always reflect the original structure of the lignin.The thioacidolysis method (TAM), in which lignin is solvolyzed with dioxane and ethanethiol, has also been utilized for the determination of the S/G ratio.5 This method is expected to yield more representative results for the original chemical structure of lignin than other chemical degradation methods because lignin is decomposed under relatively mild acidic conditions.By using this method, Ona et al.6 investigated within-tree variations of the S/G ratio for Eucalyptus camaldulensis and Eucalyptus globulus. However, it usually takes several hours for the thioacidolysis reaction and derivatization for gas chromatographic analysis. Moreover, because only the b-O-4 bonds in lignin are selectively cleaved among the various intermonomeric linkages during thioacidolysis, the resulting S/G ratio might be different to that obtained if all kinds of intermonomeric linkages in lignin were to be cleaved evenly.Spectroscopic analyses, such as Fourier transform infrared (FTIR) spectroscopy7 and nuclear magnetic resonance (NMR) spectroscopy,8 have also been utilized for qualitative and/or quantitative analysis of the functional groups in lignin. Faix7 classified lignins from different botanical origins by FTIR spectroscopy without using authentic lignin samples well characterized by other methods. Manders8 utilized solid state 13C NMR spectroscopy to determine the S/G ratio in various hardwoods. However, it is difficult to determine the precise S/G ratio in lignin by these spectroscopic methods owing to both insufficient sensitivity and poor resolution of the spectra.Pyrolysis–gas chromatography (Py-GC) has been found to be a rapid and highly sensitive method for characterizing the structure of lignin.9 So far, classification of tree species based on the distribution of lignin-derived pyrolyzates in trace wood samples10,11 and the characterization of lignin treated chemically during the pulping process12,13 using Py-GC have been Analyst, 1999, 124, 669–674 669reported.Genuit et al.14 determined the S/G ratio in chemically isolated milled wood lignin without using any tedious pretreatments. Furthermore, Faix et al.15 compared the S/G ratio determined by Py-GC with those obtained both by nitrobenzene oxidation and by FTIR spectroscopy. However, the above Py-GC studies were mostly carried out using a Curie-point or a filament pyrolyzer, which might cause undesirable denaturation and/or degradation of thermally labile samples before final pyrolysis since prior to the final pyrolysis the samples had to be exposed to the interface heating zone in the pyrolysis chamber at around the maximum temperature of the separation column, typically between 200 and 300 °C.16 Therefore, these pyrolysis methods are not always suitable to discriminate subtle differences in lignin structures. On the other hand, in the case of the vertical microfurnace pyrolyzer used in this work, the sample taken into a sample cup is first placed at the waiting position of the pyrolyzer kept at around room temperature, and then it is dropped into the heated center of the pyrolyzer, controlled at a prefixed pyrolysis temperature (400–600 °C), just before the final pyrolysis.Therefore, preferable instantaneous pyrolysis at the desired pyrolysis temperature is expected to be achieved without causing any thermal denaturation and/or degradation of thermally unstable samples such as lignin at intermediate temperatures, otherwise the samples should be exposed to the interface temperature for a while before the final pyrolysis.In this work, Py-GC using a vertical microfurnace pyrolyzer was applied to the precise determination of the S/G ratio in various local samples taken from a given Eucalyptus. The observed within-tree variations of the S/G ratio were compared with those obtained by TAM.Experimental Materials A Eucalyptus camaldulensis tree grown in Western Australia was investigated in this work. Fig. 1 shows the procedures for sample preparation from the trunk of the tree. First, debarked wood disks of 6 cm thickness were obtained by cutting the trunk from 0.3 m above the ground at 1 m intervals [Fig. 1(a)]. Then, 2 cm wide wood bars sawn from the center of each disk [Fig. 1(b)] were divided at each 2 cm depth into blocks (ca. 2 3 2 3 6 cm) [Fig. 1(c)]. Each block was cryo-milled into a finely powdered sample using a freezer mill [Spex (Metuchen, NJ, USA) 6700] in order to homogenize and improve the efficiency of the subsequent solvent extraction. Fig. 2 shows the sampling position of the block samples, among which the shadowed ones were used in this work. Here, the alphabetical letters show the axial position order from the bottom of the trunk.Likewise, the numbers show the radial position order from the pith of the trunk. Among these, A-1–A-7 and E-1–E-4 were utilized for the determination of the radial variations, and A-1–K-1 and A- 4–G-4 were utilized for the determination of the axial variations of the S/G ratio. According to a previous report,6 in order to remove potential interference from the so-called extractives included in the wood samples with the determination of the S/G ratio, each powdered sample of the selected blocks was treated in advance with a sequence of Soxhlet extractions using ethanol–toluene (1 + 2, v/v) for 6 h, ethanol for 4 h and finally distilled water for 2 h.The thus treated powdered samples were subjected to Py-GC measurement after being vacuum dried at room temperature for 24 h. Py-GC and Py-GC-MS Fig. 3 shows a schematic diagram of the vertical microfurnace pyrolyzer [Frontier Lab (Koriyama, Japan) PY-2010D] used in this work, which is basically the same as that described previously.16 In this vertical microfurnace pyrolyzer, the sample taken in a small platinum sample holder is set at around room temperature (position I in Fig. 3) until it leaves the sample holder for the heated center of the quartz tube for the final pyrolysis (position II in Fig. 3). The pyrolyzer was directly attached to a gas chromatograph [Hewlett-Packard (Avondale, PA, USA) HP6890] with a flame ionization detector (FID). About 100 mg of the cryo-milled wood sample were pyrolyzed under a flow of helium carrier gas.The optimum pyrolysis temperature of 450 °C was determined empirically. A metal capillary column (Frontier Lab Ultra ALLOY+ 21, 30 m 3 0.15 mm id, coated with 0.25 mm of polydimethylsiloxane immobilized through chemical cross-linking) was used. The 50 ml min21 helium carrier gas flow rate at the pyrolyzer was reduced to 1.0 ml min21 at the capillary column by means of a splitter. The column temperature was programmed from 50 to 280 °C at 5 °C min21.Identification of the peaks on the pyrograms was carried out mainly by using a gas chromatograph –mass spectrometer [Jeol (Tokyo, Japan) Automass 150 system II] with an electron ionization source (70 eV), to which the pyrolyzer was also directly attached. The mass spectra of the Fig. 1 Procedures for sample preparation. (a) Cutting at 1 m intervals in height; (b) sawing out from the center of the disk; (c) dividing into 2 cm widths. Fig. 2 The position of the block samples. 670 Analyst, 1999, 124, 669–674thermal degradation products of wood reported previously17,18 were also considered for peak identification. Thioacidolysis method Reference data for the lignin S/G ratio were obtained by TAM as reported previously.6 First, 10 mg of the extractives-free powdered sample were mixed with 5 ml of dioxane–ethanthiol (9 + 1, v/v) containing 0.2 m BF3 etherate. Then the mixture was heated at 100 °C for 5 h in a nitrogen atmosphere to complete the thioacidolysis reactions.6 The reactant was extracted with CH2Cl2 and finally trimethylsilylated with N,O-bis(trimethylsilyl) trifluoroacetamide (BSTFA) at 100 °C for 10 min.After removing the excess BSTFA, the resulting trimethylsilyl derivative was analyzed by GC. The lignin S/G ratio was determined based on the peak intensity ratio reflecting the syringyl- and the guaiacylpropane units. Results and discussion Within-tree radial variation of the S/G ratio Fig. 4 shows a typical pyrogram of a Eucalyptus sample at 450 °C observed by FID. Many pyrolyzates derived from lignin were observed as peaks 1–26 on the pyrogram along with those derived from cellulose and hemicellulose such as levoglucosan. Table 1 shows the assignment of these characteristic peaks from lignin together with their effective carbon number (ECN), corresponding to the relative molar sensitivity of the FID. The ECN for molar sensitivity corrections to FID was determined empirically on the basis of the structure of each pyrolyzate.19 Here, the symbols S and G mean the pyrolyzates derived from syringyl- and guaiacylpropane units, respectively, and the 26 peaks were assigned to 13 S and 13 G pyrolyzates with the corresponding side-chain structures.On the basis of the relative molar yields of the 26 characteristic peaks listed in Table 1, the S/G ratio for each sample can be evaluated as follows. First, the relative molar yield (Yi) of each pyrolyzate of lignin was determined by use of the following equation: Y I I i i i i i i ( %) mol ECN ECN = � = å1 26 100 (1) where Ii and ECNi are the peak intensity obtained by FID and the ECN for the ith peak, respectively.Then, the S/G ratio was evaluated from the ratio of the sum of the molar yields of characteristic peaks of S and G units by S/G ratio S G = = = å å Y Y m m n n 1 13 1 13 (2) where YSm and YGn are the molar yields for the mth and nth pyrolyzate derived from S and G units, respectively.The relative standard deviation (RSD) for the determination of the S/G ratio by Py-GC was 1.5% for four repeated runs with sample A-1, suggesting adequate reproducibility to characterize the within-tree variation of the S/G ratio. First, samples A-1–A-7 taken at 0.3 m above the ground and samples E-1–E-4 at 4.3 m above the ground were measured by Py-GC to determine the variation of the S/G ratio in the radial Fig. 3 Schematic diagram of a vertical microfurnace pyrolyzer (Frontier Lab Py 2010D).Position I, ambient temperature; position II, pyrolysis temperature. Fig. 4 A typical pyrogram for Eucalyptus obtained at 450 °C (sample A-1). The peak assignments are given in Table 1. Analyst, 1999, 124, 669–674 671direction. Table 2 summarizes the relative molar yields among the 26 pyrolyzates from lignin listed in Table 2 for samples A- 1–A-7. The distribution of each pyrolyzate in samples A-1–A-7 showed a similar tendency, vinylsyringol (peak 13), vinylguaiacol (3), syringol (4), methylsyringol (8) and trans-propenylsyringol (19) being the main pyrolyzates.However, the yields of vinylguaiacol (3) and vinylsyringol (13) having the same aldehyde side-chain derived from G and S units, respectively, decreased from the pith to the bark, whereas those of transconiferyl alcohol (22) and trans-sinapyl alcohol (26), having a 1-propenyl alcohol side-chain, showed the reverse tendency. Similarly, the distributions of these pyrolyzates for samples E- 1–E-4 showed the same tendency as those for samples A-1–A-7.These observations suggest that the within-tree variation of the S/G ratio for the radial direction would show a similar tendency at any height of the tree. Fig. 5 illustrates the within-tree variation of the S/G ratio in the radial direction at 0.3 m above the ground obtained on the basis of the molar yields of the pyrolyzates together with those obtained by TAM.As for the radial variation of the S/G ratio at this height, sample A-1 at the extreme pith side showed the highest S/G ratio value of 2.13. The S/G ratio then decreased gradually towards the bark, with the lowest value of 1.57 for sample A-7. Here, the average value of the S/G ratio in the radial direction was determined as 1.80, which was approximately equal to that for sample A-4. Although the S/G ratios obtained by Py-GC were always higher than those obtained by TAM, a similar tendency was observed for the within-tree variation of the S/G ratio obtained by both methods.Fig. 6 illustrates the within-tree variation of the S/G ratio in the radial direction at 4.3 m above the ground obtained by Py- GC together with those obtained by TAM. The withinree radial variation of the S/G ratio obtained by Py-GC at this level showed a gradually decreasing tendency toward the bark. However, the S/G ratio for E-3 obtained by TAM showed an unexpectedly higher value, probably related to an abnormal formation where the S/G ratio of lignin structures linked by b- O-4 bonds was different to that at other positions. Within-tree axial variation of the S/G ratio The within-tree axial variation of the S/G ratio was determined by using samples A-1–K-1 taken at the extreme pith side and samples A-4–G-4 taken 6 cm apart from the pith.Table 3 Table 1 Peak assignment in the pyrogram of a Eucalyptus camaldulensis sample Peak No. Name Origina Molecular mass Effective carbon numberb 1 Guaiacol G 124 5.45 2 4-Methylguaiacol G 138 6.45 3 Vinylguaiacol G 150 7.35 4 Syringol S 154 5.65 5 Eugenol G 164 8.35 6 Vanillin G 152 6.65 7 cis-Isoeugenol G 164 8.35 8 Methylsyringol S 168 6.65 9 Homovanillin G 166 6.65 10 trans-Isoeugenol G 164 8.35 11 Acetoguaiacone G 166 6.65 12 Guaiacylacetone G 180 7.65 13 Vinylsyringol S 180 7.55 14 Allylsyringol S 194 8.55 15 Syringaldehyde S 182 5.65 16 cis-Propenylsyringol S 194 8.55 17 cis-Coniferyl alcohol G 180 7.75 18 Homosyringaldehyde S 196 6.85 19 trans-Propenylsyringol S 194 8.55 20 trans-Coniferaldehyde G 178 7.55 21 Acetosyringone S 196 6.85 22 trans-Coniferyl alcohol G 180 7.75 23 Syringylacetone S 210 7.85 24 cis-Sinapyl alcohol S 210 7.95 25 trans-Sinapaldehyde S 208 7.75 26 trans-Sinapyl alcohol S 210 7.95 a S = syringyl unit, G = guaiacyl unit.b Mole sensitivity corrections to FID. Table 2 Radial variation of pyrolyzates for the samples at 0.3 m above the ground Molar yield (%) Peak No.Compound Origin A-1 A-2 A-3 A-4 A-5 A-6 A-7 1 Guaiacol G 4.90 4.96 4.96 4.41 4.42 4.12 5.38 2 4-Methylguaiacol G 5.55 4.95 4.55 4.64 5.04 5.01 4.47 3 Vinylguaiacol G 6.93 6.75 6.28 6.18 6.42 7.01 5.93 4 Syringol S 9.24 8.49 9.45 8.17 9.94 7.55 8.06 5 Eugenol G 0.94 0.85 0.76 0.63 0.94 1.07 1.32 6 Vanillin G 1.49 1.36 1.49 1.60 1.79 1.88 2.21 7 cis-Isoeugenol G 0.45 0.43 0.44 0.46 0.57 0.50 0.38 8 Methylsyringol S 9.29 8.75 8.18 7.78 8.60 8.03 6.54 9 Homovanillin G 1.86 1.63 1.41 1.63 1.40 1.77 1.44 10 trans-Isoeugenol G 3.80 3.48 3.32 3.35 4.17 4.03 3.16 11 Acetoguaiacone G 1.00 0.93 1.00 1.00 0.95 0.94 1.01 12 Guaiacylacetone G 0.61 0.48 0.58 0.21 0.61 0.65 0.40 13 Vinylsyringol S 14.01 14.37 12.68 13.72 11.29 11.80 11.99 14 Allylsyringol S 2.37 2.43 1.98 1.77 2.04 2.04 1.58 15 Syringaldehyde S 5.16 5.23 5.22 5.48 6.04 6.32 6.51 16 cis-Propenylsyringol S 1.46 1.29 1.22 1.18 1.43 1.35 1.06 17 cis-Coniferyl alcohol G 0.32 0.53 0.64 0.66 0.64 0.60 0.71 18 Homosyringaldehyde S 3.71 2.92 2.79 2.82 2.71 2.85 1.92 19 trans-Propenylsyringol S 9.70 8.40 7.91 7.42 8.22 8.19 6.61 20 trans-Coniferaldehyde G 1.25 1.41 1.49 1.54 1.63 1.55 1.79 21 Acetosyringone S 2.75 3.11 3.29 3.39 3.59 3.42 3.94 22 trans-Coniferyl alcohol G 2.46 5.17 6.88 7.69 6.18 6.42 9.25 23 Syringylacetone S 1.75 1.66 1.59 1.54 1.43 1.45 1.26 24 cis-Sinapyl alcohol S 0.72 1.27 1.57 1.53 1.24 1.19 1.77 25 trans-Sinapaldehyde S 6.67 6.71 6.63 6.62 6.69 7.39 7.69 26 trans-Sinapyl alcohol S 1.61 2.44 3.69 4.58 2.02 2.87 3.62 Total 100 100 100 100 100 100 100 672 Analyst, 1999, 124, 669–674summarizes the relative molar yield of each pyrolyzate of the lignin for samples A-4–G-4.The distributions of the pyrolyzates from the samples taken at different axial positions proved to be similar to each other. Similarly, no specific variations in the observed pyrolyzates were recognized in the axial direction among the extreme pith side samples A-1–K-1.Fig. 7 shows the within-tree variation of the S/G ratio in the axial direction at 6 cm apart from the pith obtained by Py-GC and by TAM. As for the axial variation of the S/G ratio, there was a slightly increasing tendency in the S/G ratios for the samples at higher parts. As with the radial variation, although the S/G ratios obtained by Py-GC were always higher than those obtained by TAM, the axial variation of the S/G ratio obtained by both methods showed almost the same tendency.Basically the same tendency was also observed for the extreme pith side samples A-1–K-1. Conclusion A highly sensitive method to determine the lignin S/G ratio reflecting the whole lignin structures in a given tree sample was developed by means of Py-GC using a vertical microfurnace pyrolyzer. By using this technique, the within-tree variation of the S/G ratio was characterized precisely within about 1 h using only about 0.1 mg of wood sample.The S/G ratio can be Fig. 5 Radial variation of S/G ratio at 0.3 m above the ground. 5, Py-GC; 8, TAM. Fig. 6 Radial variation of S/G ratio at 4.3 m above the ground. 5, Py-GC; 8, TAM. Table 3 Axial variation of pyrolyzates for the samples at a 6 cm distance from the pith Molar yield (%) Peak No. Compound Origin A-4 B-4 C-4 D-4 E-4 F-4 G-4 1 Guaiacol G 4.41 5.00 4.38 4.35 4.67 4.90 4.75 2 4-Methylguaiacol G 4.64 4.56 4.15 3.71 3.25 4.42 3.76 3 Vinylguaiacol G 6.18 6.13 5.26 5.24 5.21 5.59 5.52 4 Syringol S 8.17 9.89 9.87 9.80 10.05 11.17 10.76 5 Eugenol G 0.63 0.82 0.90 0.75 0.74 0.87 0.80 6 Vanillin G 1.60 1.92 1.92 1.83 2.16 1.79 1.92 7 cis-Isoeugenol G 0.46 0.52 0.49 0.51 0.46 0.56 0.56 8 Methylsyringol S 7.78 8.08 8.16 7.25 5.64 8.45 6.89 9 Homovanillin G 1.63 1.38 1.34 1.23 1.01 1.22 1.11 10 trans-Isoeugenol G 3.35 3.58 3.19 3.12 2.73 3.37 3.16 11 Acetoguaiacone G 1.00 0.79 0.89 1.03 0.83 0.91 0.83 12 Guaiacylacetone G 0.21 0.16 0.35 0.44 0.47 0.51 0.50 13 Vinylsyringol S 13.72 12.43 11.92 11.83 11.73 12.37 12.67 14 Allylsyringol S 1.77 1.94 2.00 1.97 1.68 2.21 2.05 15 Syringaldehyde S 5.48 5.68 6.44 6.69 6.83 7.12 6.34 16 cis-Propenylsyringol S 1.18 1.25 1.29 1.33 1.06 1.44 1.40 17 cis-Coniferyl alcohol G 0.66 0.62 0.64 0.62 0.85 0.59 0.79 18 Homosyringaldehyde S 2.82 2.29 2.64 2.17 1.50 2.78 2.10 19 trans-Propenylsyringol S 7.43 7.66 7.87 7.74 6.43 8.45 7.89 20 trans-Coniferaldehyde G 1.54 1.51 1.52 1.59 1.64 1.64 1.58 21 Acetosyringone S 3.39 3.32 3.35 3.50 3.61 3.61 3.49 22 trans-Coniferyl alcohol G 7.69 6.85 7.33 7.65 10.44 4.78 7.63 23 Syringylacetone S 1.54 1.38 1.44 1.37 1.36 1.48 1.56 24 cis-Sinapyl alcohol S 1.53 1.36 1.46 1.50 1.55 1.25 1.56 25 trans-Sinapaldehyde S 6.61 6.99 7.13 7.82 7.16 7.31 6.92 26 trans-Sinapyl alcohol S 4.58 3.89 4.07 4.96 6.94 1.21 3.46 Total 100 100 100 100 100 100 100 Fig. 7 Axial variation of S/G ratio at 6 cm apart from the pith. 5, Py-GC; 8, TAM. Analyst, 1999, 124, 669–674 673measured by this technique with very high reproducibility, with RSD < 2%. The observed results indicated that (1) the S/G ratio generally showed a higher value at the pith side and decreased gradually towards the bark in the radial direction; (2) there was a slightly increasing tendency in the S/G ratios for the samples at higher parts in the axial direction; and (3) in order to obtain a representative S/G ratio for a given tree of this kind, the sample taken from half of the radius at a moderate height could be used.Acknowledgement Financial support by the Grant-in-Aid for Scientific Research (A) (09305056) and (B) (09555262) of the Ministry of Education, Science, Sports and Culture, Japan, and by a grant from the ‘Research for the Future’ Program of the Japan Society for the Promotion of Science (JSPS-RFTF, 96 R11601) is gratefully acknowledged. References 1 K. V. Sarkanen and H. L. Hergert, in Lignins—Occurrence, Formation, Structure and Reaction, ed. K. V. Sarkanen and C. H. Ludwig, Wiley, New York, 1971, pp. 43–94. 2 D. Collins, C. Pilotti and A. Wallis, Appita J., 1990, 43, 193. 3 T. Ona, T. Sonoda, K. Ito, M. Shibata, Y. Tamai and Y. Kojima, Appita J., 1996, 49, 325. 4 M. Tanahashi and T. Higuchi, Methods Enzymol., 1992, 161, 101. 5 C. Lapierre and B. Monties, Holzforschung, 1986, 40, 113. 6 T. Ona, T. Sonoda, K. Ito and M. Shibata, Holzforschung, 1997, 51, 396. 7 O. Faix, Holzforschung, 1991, 45, 21. 8 W. F. Manders, Holzforschung, 1987, 41, 13. 9 D. Meier and O. Faix, in Methods in Lignin Chemistry, ed. S. Y. Lin and C. W. Dence, Springer, Berlin, 1992, pp. 177–199. 10 K. Kuroda and A. Yamaguchi, J. Anal. Appl. Pyrolysis, 1995, 33, 51. 11 J. M. Challinor, J. Anal. Appl. Pyrolysis, 1996, 37, 1. 12 M. Kleen and G. Gellerstedt, J. Anal. Appl. Pyrolysis, 1991, 19, 139. 13 M. C. Terrón, M. L. Fidalgo, A. E. González, G. Almendros and G. C. Galletti, J. Anal. Appl. Pyrolysis, 1993, 27, 57. 14 W. Genuit, J. J. Boon and O. Faix, Anal. Chem., 1987, 59, 508. 15 O. Faix, D. Meier and I. Grobe, J. Anal. Appl. Pyrolysis, 1987, 11, 403. 16 S. Tsuge, H. Ohtani, H. Matsubara and M. Ohsawa, J. Anal. Appl. Pyrolysis, 1987, 11, 181. 17 O. Faix, D. Meier and I. Fortmann, Holz Roh- Werkst., 1990, 48, 281. 18 O. Faix, D. Meier and I. Fortmann, Holz Roh- Werkst., 1990, 48, 351. 19 A. D. Jorgensen, K. C. Picel and V. C. Stamoudis, Anal. Chem., 1990, 62, 683. Paper 8/09217F 674 Analyst, 1999, 124, 669–674

ISSN:0003-2654    

DOI:10.1039/a809217f

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

Determination of testosterone : epitestosterone ratio after pentafluorophenyldimethylsilyl-trimethylsilyl derivatisation using gas chromatography-mass spectrometry in equine urine Man Ho Choi,a Jin Young Kimb and Bong Chul Chung*a a Doping Control Center, Korea Institute of Science and Technology, P.O. Box 131, Cheongryang, Seoul, Korea. E-mail: bcc0319@kist.re.kr b Racing Laboratory, Korea Racing Association, Kyonggi 427-070, Korea Received 15th February 1999, Accepted 9th March 1999 A highly specific method is described for measuring the testosterone : epitestosterone ratio in equine urine by gas chromatography-mass spectrometry (GC-MS) with stable isotope internal standards.The procedure was based on Serdolit Pad-1 resin extraction, enzymatic hydrolysis, and chemical derivatisation prior to instrumental analysis. The mixed derivatives, 3-trimethylsilyl-17-pentafluorophenyldimethylsilyl ether (3-TMS-17-flophemesyl) testosterone and epitestosterone, were found to have excellent analytical properties. The specificity of the derivatisation method exploits a unique feature of steroids: the selective exchange of the alcoholic flophemesyl ether for the trimethylsilyl ether.The sensitivity and specificity of the mixed 3-TMS-17-flophemesyl derivatives allow adequate determinations of testosterone and epitestosterone, even in urine from mares, in 5 ml samples. The repeatability of testosterone and epitestosterone was 6.2 and 5.7%, respectively, and their reproducibility was in the range of 6.4–8.7%. 1. Introduction Testosterone is the principal endogenous androgenic-anabolic steroid in humans and equines. In human athletes, testosterone is the substance most frequently reported in steroid misuse, and the accepted test for testosterone administration has been the urinary testosterone to epitestosterone (T : E) ratio, a value of > 6 being taken as the hallmark of drug abuse.1 In contrast to humans, low concentrations of epitestosterone have been detected in normal equine urine.High resolution mass spectrometry (HRMS) or tandem mass spectrometry (MS-MS) coupled with gas chromatography (GC) or high-performance liquid chromatography (HPLC) has been used to identify and determine the T :E ratio by several groups.2,3 Many steroids are thermally labile and must be derivatised prior to analysis to avoid decomposition of the compound and to improve its chromatographic performance. Generally, the derivatisation of testosterone and epitestosterone with 3,17-bistrimethylsilyl ether (TMS) has been the most common approach. 2,4 Possible methods for improving the selective and sensitive detection of steroids in human and equine urine include modifications to: the GC temperature program; the derivatisation method; the type of GC column; and the sample purification.5–8 In general, alkyl or aryl compounds with closely bound fluorine atoms are remarkable in that they show little increase in boiling point compared to hydrocarbons containing a similar number of carbon atoms.However, the pentafluorophenyl ring is a strong electron attracting group which is able to influence the mode of fragmentation of steroid derivatives under electron impact in a way which leads to diagnostic mass spectra. The spectra often show marked differences from those of the TMS ethers. The pentafluorophenyldimethylsilyl (abbreviated to flophemesyl for convenience) derivatives generally show a strong molecular ion and provide much more detailed diagnostic information.9,10 The objective of this work was to improve the determination of testosterone : epitestosterone ratios in equine urine by using a novel mixed flophemesyl-trimethylsilyl ether derivatisation method.The new method was assessed by GC-MS analyses and comparisons with deuterium labelled testosterone and epitestosterone internal standards were made. 2. Experimental 2.1. Chemicals Testosterone (4-androsten-17b-ol-3-one) and epitestosterone (4-androsten-17a-ol-3-one) were purchased from Sigma Co.(St. Louis, MO, USA). The 1:1 (v/v) mixture of 16,16,17-2H3- testosterone (90 ng ml21) and 16,16,17-2H3-epitestosterone (15 ng ml21), an internal standard, was obtained from Cologne Laboratory (Institute of Biochemistry, German Sports University, Germany). N-methyl-N-trimethylsilyltrifluoroacetamide (MSTFA), ammonium iodide (NH4I), pentafluorophenyldimethyl chlorosilane (flophemesyl chloride) and dithioerythritol (DTE) were purchased from Sigma Co.(St. Louis, MO, USA). Serdolit Pad-1 resin (particle size 0.1–0.2 mm) was supplied by Serva Co. (Heidelberg, Germany) and washed with acetone, methanol and distilled water before use. b-Glucuronidase–arylsulfatase from Helix pomatia (aqueous solution stabilized with thiomerosal) was purchased from Boehringer Mannheim Co. (Mannheim, Germany). 2.2. Apparatus The GC-MS system (Model 5973MSD combined with a Model 6890 plus gas chromatograph, Hewlett-Packard; Avondale, PA, USA) was used in both scan and selected ion monitoring (SIM) modes.The electron energy was 70 eV and the ion source temperature was 230 °C. The gas chromatograph was equipped with a 17 m 3 0.2 mm id 3 0.11 mm film thickness capillary column coated by cross-linked 5% phenyl methyl silicon fluid Analyst, 1999, 124, 675–678 675(Hewlett-Packard). The carrier gas was helium at a column head pressure of 121 kPa.The split (1 : 10) method of injection was used. The temperature program was as follows: initial temperature 200 °C (2 min); program rate 10 °C min21 to 250 °C (5 min); 10 °C min21 up to a final temperature of 320 °C, where it was held for 3 min. 2.3. Extraction and derivatisation procedure An aliquot of urine (5 ml) was taken and internal standard solution (20 ml) was added. An aqueous Serdolit Pad-1 slurry was filled into a Pasteur pipette until a bed height of 1.5 cm was achieved.The column was washed with 3 ml of distilled water, then the mixture was loaded onto a Serdolit Pad-1 resin cartridge. The column was washed with water (5 ml) and nhexane (5 ml), then eluted with methanol (2 31.5 ml) into a test tube; the eluate was evaporated to dryness (40 °C under nitrogen). The residues were redissolved in acetate buffer of pH 5.4 (0.2 mol l21; 1 ml). In order to hydrolyze the conjugated form, 0.1 ml of enzyme solution was added to the acetate buffer, and the solution was heated at 80 °C for 3 h, then cooled to room temperature and the pH adjusted by adding 20 mg potassium carbonate along with 5 ml n-pentane.The mixture was mechanically shaken (10 min) and centrifuged (2400 rpm, 5 min) and the organic phase was transferred to a test tube. The organic layer was evaporated to dryness in a rotary evaporator. The hydrolyzed buffer solution was extracted two additional times with 5 ml n-pentane to enhance the recovery of this extraction method. The n-pentane fraction was evaporated to dryness.The residue was dried in a vacuum desiccator over P2O5–KOH for at least 30 min before the derivatisation procedure. The flophemesyl chloride solution (50 ml) was added to the residue, and the mixture was allowed to stand at room temperature for 15 min. After the excess reagent had been evaporated under a stream of nitrogen at 70 °C, trimethylsilylating reagent (50 ml, MSTFA–NH4I–DTE, 1000 : 4 : 2, v/w/w) was added to the residue, and the mixture was heated at 60 °C for 15 min.An aliquot of the flophemesyl-TMS derivatised sample solution was injected into the GC-MS. 2.4. Evaluation of repeatability and reproducibility The repeatability of the chromatographic analysis was determined by ten replicate 2 ml injections of a mixture of derivatised standards. The reproducibility for urine samples was examined by several injections of a 2 ml portion of ten derivatised extracts obtained from fortified water samples at 10 and 50 ppb levels. 3. Results 3.1. Establishment of a suitable derivatisation method Testosterone and epitestosterone as 17-epimers have two ionizable hydrogen atoms. In order to stabilize the compounds and improve the GC properties, an initial effort was made to examine the derivatisation method using flophemesyl chloride. Indeed, the injection of the flophemesyl derivatives of both testosterone and epitestosterone, by GC-MS, showed the presence of two peaks with identical mass spectra in EI mode. Because of mono- and bis-flophemesyl derivatives, we suspected that the reaction at the 3-enol keto position had taken place during the GC injection in the injection port.In contrast, the efficiency of 3-TMS-17-flophemesyl derivatives was tested by a full-scan spectra of pure steroid standards. The chromatograms did not show any peaks of an unexpected nature, corresponding to derivatised or partially derivatised Fig. 1 Total ion chromatogram of epitestosterone after flophemesyl derivatisation procedure (A) and 3-TMS-17-flophemesyl derivatisation procedure (B). Fig. 2 Scan spectra of 17-flophemesyl-epitestosterone (A), 3,17-bisflophemesyl- epitestosterone (B) and 3-TMS-17-flophemesyl-epitestosterone (C). 676 Analyst, 1999, 124, 675–678steroids indicating that the derivatisation reaction was complete (Fig. 1). 3.2. Mass spectral analysis Fig. 2 shows the principal ions detected in the present study.In all the spectra, the compound loss of a methyl group led to M215 mass units of abundance, less than that of the molecular ion. The flophemesyl derivatives are characterized by 77 u.10 Peaks at mass units of M2167 [M2C6F5]+, M2225 [M2C6F5Si(CH3)2]+ and M2241 [M2C6F5Si(CH3)2O]+ also occurred to varying extents in all the spectra. Likewise, occurrence of trimethylsilylation is identified by mass units of M272 [M2Si(CH3)3]+, M290 [M2Si(CH3)3OH]+ and M2105 [M2Si(CH3)3OH2CH3]+, even though their abundances are quite low.In all mass spectra for the flophemesyl derivatisation method, there is a significant peak at 81 u which corresponds to the [Si(CH3)F2]+ mass units. 3.3. Stability of the mixed flophemesyl-TMS derivatisation method The repeatability of the method was evaluated using 1 ppm derivatised mixed standards of testosterone and epitestosterone, and the reproducibility was assessed using extracts fortified with standards of both testosterone and epitestosterone at 10 and 50 ppb levels in water.The peak areas of selected ions (molecular ions) were obtained for testosterone and epitestosterone. They were quantitated by the ratio of the peak area from the spiked sample to that from the corresponding internal standard (16,16,17-2H3-testosterone) and the absolute values were calculated. The repeatability for testosterone and epitestosterone was 6.2 and 5.7%, respectively, and their reproducibility was in the range of 6.4%–8.7% (Table 1). 3.4. Determination of testosterone : epitestosterone ratio d3-Testosterone and d3-epitestosterone were introduced as internal standards because calibrating the GC-MS system using external standards is difficult due to matrix problems and linearity problems in the GC-MS system.11 To reduce some of the problems arising from these matrices and instruments, d3- testosterone and d3-epitestosterone were introduced as internal standards to obtain standardized conditions in the human doping groups.12 Two same-mare urine samples were analyzed by GC-MS after derivatisation using two different methods.In order to determine the T : E ratio, a derivatisation method using a silylating agent was carried out for the formation of 3,17-bis- TMS derivatives in most of the doping groups. Especially in the case of epitestosterone, these derivatives produce peak broadening or are not detected because of very low concentrations in mares and geldings, but 3-TMS-17-flophemesyl derivatives reduce matrix interference and show excellent peak shape (Fig. 3). Three different urine samples (OC-218, 219 and 220) were analyzed using two different derivatisation methods. All samples were analyzed three times by GC-MS, and the results of the T : E ratio were similar to each other. The results are listed in Table 2. Fig. 3 SIM chromatograms for detection of epitestosterone (A), testosterone (B), d3-epitestosterone (C) and d3-testosterone (D) after mixed TMSflophemesyl (left) and trimethylsilyl (right) derivatisation procedures. Table 1 The repeatability and reproducibility of 3-TMS-17-flophemesyl derivatives (n = 10) Reproducibility (RSD,%) Repeatability Substance (RSD,%) 10 ppb 50 ppb Testosterone 6.2 7.3 8.7 Epitestosterone 5.7 7.1 6.4 Analyst, 1999, 124, 675–678 6774.Discussion The reactants uncatalyzed by flophemesyl chloride rapidly reacted with unhindered secondary hydroxyl groups to produce silyl ethers.This reagent did not cause the formation of enol ethers and hindered the hydroxyl groups from reacting.13 Thus we used the silylating agent and flophemesyl chloride for 3-enol keto and 17-hydroxyl groups, respectively. The derivatisation method of testosterone and epitestosterone to 3,17-bis-TMS derivatives has been the mass (432 u) common approach for determination of the T : E ratio, but the mass increment provided by trimethylsilylation is rather low. Therefore, 3-TMS-17-flophemesyl derivatives have been advocated as a better choice for SIM at the higher mass (584 u) of the molecular ion.The advantages of using the mixed flophemesyl-TMS derivatisation method include: the ease of reagent removal, without loss of products, by excessive nitrogen gas at high temperature; rapid reaction time; good GC properties; and the formation of intense molecular ions under electron impact mass spectrometry (EI-MS). Moreover, on a theoretical basis, specificity should be improved, because the selective exchange of an alcoholic flophemesyl ether for TMS ether is an exclusive feature of testosterone and epitestosterone.This study may be the starting point of further studies, which could screen and confirm unambiguously the structure of steroids and diverse applications in chromatographic research. References 1 D. H. Catlin, D. A. Cowan, R. de la Torre, M. Donike, D. Fraisse, H. Oftegro, K. Hatton, B. Starcevie, M.Becchi, X. de la Torre, H. Norli, H. Geyer and C. J. Walker, Int. J. Mass Spectrom. Ion Processes, 1996, 31, 397. 2 S. Horning, W. Schänzer, G. Sigmund and M. Donike, in Proceedings of the 10th International Conference of Racing Analysts and Veterinarians, Stockholm, Sweden, ed. D. E. Auer and E. Houghton, R and W Publications, Newmarket, 1994, pp. 127–131. 3 P. W. Tang, W. C. Law and D. L. Crone, in Proceedings of the 11th International Conference of Racing Analysts and Veterinarians, Queensland, Australia, ed. D.E. Auer and E. Houghton, R and W Publications, Newmarket, 1996, pp. 73–76. 4 Y. Bonnaire, L. Dehennin, P. Plou, M. A. Popot and L. Tcutain, in Proceedings of the 10th International Conference of Racing Analysts and Veterinarians, Stockholm, Sweden, ed. D. E. Auer and E. Houghton, R and W Publications, Newmarket, 1994, pp. 187–194. 5 A. Leinonen, L. Savonen and K. Kuoppasalmi, in Recent Advances in Doping Analysis: Proceedings of the 11th Cologne Workshop on Dope Analysis, ed.W. Schänzer, H. Geyer, A. Gotzman, U. Mareck- Engelke and S. Rauth, Sport und Buch Strauss, Köln, 1994, pp. 25–31. 6 H. Geyer, U. M. Engelke, W. Schänzer and M. Donike, in Recent Advances in Doping Analysis: Proceedings of the 11th Cologne Workshop on Dope Analysis, ed. W. Schänzer, H. Geyer, A. Gotzman, U. Mareck-Engelke and S. Rauth, Sport und Buch Strauss, Köln, 1994, pp. 97–103. 7 P. W. Tang, W. C. Law and D. L. Crone, in Proceedings of the 11th International Conference of Racing Analysts and Veterinarians, Queensland, Australia, ed.D. E. Auer and E. Houghton, R and W Publications, Newmarket, 1996, pp. 68–72. 8 M. H. Choi, J. Y. Kim and B. C. Chung, Anal. Lett., 1999, in the press. 9 C. F. Poole, A. W. Zlatkis, F. Sye and S. Singhawangcha, Lipids, 1980, 15, 734. 10 L. Y. Jayasinghe, P. J. Marriott, P. D. Carpenter and P. D. Nichols, J. Chromatogr., 1998, 809, 109. 11 E. Nolteernsting, W. Schänzer and M. Donike, in Recent Advances in Doping Analysis (3): Proceedings of the 13th Cologne Workshop on Dope Analysis, ed. W. Schänzer, H. Geyer, A. Gotzman and U. Mareck-Engelke, Sport und Buch Strauss, Köln, 1996, pp. 191–199. 12 E. Nolteernsting, H. Geyer, U. Mareck-Engelke, W. Schänzer and M. Donike, in Recent Advances in Doping Analysis (2): Proceedings of the 12th Cologne Workshop on Dope Analysis, ed. W. Schänzer, H. Geyer, A. Gotzman and U. Mareck-Engelke, Sport und Buch Strauss, Köln, 1995, pp. 113–120. 13 E. D. Morgan and C. F. Poole, J. Chromatogr., 1975, 104, 351. Paper 9/01242G Table 2 Comparison of T : E ratio using two different derivatisation methods T: E ratioa (n = 3) Sample Sex 3,17-bis-TMS 3-TMS-17-flophemesyl OC-218 Mare 12.4 ± 0.35 11.6 ± 0.33 OC-219 Gelding 8.1 ± 0.28 9.7 ± 0.31 OC-220 Stallion 10.6 ± 0.21 9.9 ± 0.41 a T: E ratio calculated by peak area. 678 Analyst, 1999, 124, 675–678

ISSN:0003-2654    

DOI:10.1039/a901242g

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

Simultaneous determination of guanine and adenine contents in DNA, RNA and synthetic oligonucleotides using a chemically modified electrode Jyh-Myng Zen,* Ming-Ren Chang and Govindasamy Ilangovan Department of Chemistry, National Chung-Hsing University, Taichung 402, Taiwan. E-mail: jmzen@mail.nchu.edu.tw Received 19th November 1999, Accepted 15th March 1999 Simultaneous determination of guanine and adenine contents in DNA, RNA and synthetic oligonucleotides on a Nafion–ruthenium oxide pyrochlore modified electrode is described.This chemically modified electrode shows very pronounced electrocatalytic effects towards the oxidation of guanine and adenine. The experimental parameters for individual determinations of guanine and adenine on the modified electrode were first optimised. Excellent detection limits (S/N = 3) of 0.86 ng ml21 (5.7 nM) and 2.7 ng ml21 (20 nM) for guanine and adenine, respectively, were obtained. Simultaneous determination of guanine and adenine in a mixture was also optimised. The denatured calf thymus DNA and yeast RNA showed oxidation peaks corresponding to guanine and adenine with a linear calibration line up to 5 mg ml21 in pH 4.0 phosphate buffer.The detection limit estimated for these nucleic acids were 11.5 ng ml21 and 37.7 ng ml21 for guanine and adenine, respectively. The new method provides an alternative to estimate the guanine and adenine contents more selectively and sensitively than currently applied techniques.Many methods for the detection and quantification of purine bases, such as guanine and adenine, in nucleic acids have been developed. Among them, the most used are the spectroscopic methods coupled with chromatography or electrophoresis.1–6 Conversely, voltammetric methods are practically suited for the analysis of these purines in nucleic acids.7 To date, the electrochemical detection protocols for nucleic acids are based on the electrochemical activity of the purine bases.8–13 One exception employs sugar oxidation on a Cu electrode to determine the DNA.14 The direct estimation of these purine bases in nucleic acids is highly cumbersome, especially in the case of double stranded DNA (ds-DNA), in which the purines are paired up with pyrimidine bases (Watson–Crick base pairs) by inter-strand hydrogen bonding.This double helix structure makes the bases embed well inside as stacks and exist in a highly hydrophobic environment. Electrochemical methods for the determination of DNA after denaturing seem particularly well suited for generally limited sample sizes for DNA analysis since they can be determined in nano-liter or pico-liter volumes without sacrificing the selectivity and sensitivity.The voltammetric analysis of nucleic acids has been extensively attempted on hanging drop mercury electrodes (HMDE) by many authors using adsorptive square-wave voltammetry (AdSWV),15–17 adsorptive transfer stripping voltammetry (AdTSV)18–23 and cathodic stripping voltammetry (CSV).In both AdSWV and AdTSV, nucleic acid is physically adsorbed (no faradaic process) on the HMDE surface by dipping into a very small quantity of sample, say 5 to 20 ml, depending upon the dimension of the working HMDE. In CSV, the nucleic acids are accumulated through cathodic polarisation (faradaic process involved). These extensive studies showed very good detection limits of ng ml21 or even pg ml21 of nucleic acids.Unfortunately, these reports do not deal with the question of direct quantification of the purine bases present in nucleic acids.23 Measuring individual concentrations of adenine and guanine or their ratio in DNA or RNA is equally important to the determination of the nucleic acid concentration itself. Nucleic acids exist in physiological fluids, such as tissues and cells from catabolism of nucleic acids, enzymatic degradation of tissues, dietary habits, and various salvage pathways.The importance of such a determination stems from the fact that changes in the concentration of these purines may reflect alterations in the activity of catabolic, anabolic, and interconvertion enzymes and may indicate the presence of various diseases. Compared to spectroscopic methods, electrochemical methods reported for guanine and adenine determination suffer severe setbacks, as these purine bases irreversibly adsorb on the electrode surface thus hampering the estimations.Despite the early pioneering studies by Palecek and Dryhurst,24–26 Hart and co-workers reported the determination of guanine in nucleic acids at a glassy carbon electrode (GCE) and carbon paste electrode by differential-pulse voltammetry.27 Later, complexation of guanine with Hg in differential-pulse adsorption stripping voltammetry to detect guanine at nanomolar levels in the presence of 10-fold excess of guanosine was reported.28 A similar method using Cu(ii) was also reported with a detection limit of 0.5 nM.29–31 Use of chemically modified electrodes (CME) against conventional mercury electrodes is always beneficial since the CME render certain advantages like lowering of peak potential (Ep) due to catalytic activity.32 Moreover, HMDE is limited to negative potentials owing to the dissolution of Hg at more positive potentials.We report here a sensitive voltammetric method for the simultaneous determination of guanine and adenine contents in ds-DNA, RNA and other synthetic oligonucleotides (ss-DNA) on the Nafion–ruthenium oxide pyrochlore CME.We have already demonstrated the catalytic activity of this CME for various applications.33–38 In this report, we describe that nucleic acid concentrations can be estimated from the quantitative oxidation of the purine bases of nucleic acids using SWV. The redox behaviour of these purines and the optimum conditions for their estimation on the CME are first evaluated.Next, since guanine and adenine yield two distinct oxidation peaks separated by more than 300 mV at the CME, simultaneous estimation of adenine and guanine in a mixture is described. Finally, the protocol is extended to determine purine base concentrations in nucleic acid solutions. Analyst, 1999, 124, 679–684 679Experimental Chemical and reagents Nafion perfluorinated ion-exchange powder and 5 wt% solution in a mixture of lower aliphatic alcohols and 10% water were obtained from Aldrich (Milwaukee, WI, USA). Ruthenium chloride and lead nitrate were also obtained from Aldrich.Guanine and adenine (Sigma, St. Louis, MO, USA), and all other compounds (ACS certified reagent grade) were used as received without further purification. Aqueous solutions were prepared with doubly distilled and deionized water. Apparatus All the electrochemical measurements were performed on a Bioanalytical Systems (West Lafayette, IN, USA) BAS 50 W electrochemical analyser using a BAS VC-2 electrochemical cell.The three-electrode system consisted of the CME, an Ag/ AgCl reference electrode (Model RE-5, BAS), and a platinum wire auxiliary electrode. Since dissolved oxygen did not interfere with the anodic voltammetry, no deaeration was performed. General procedure The procedure for the preparation of the CME has been described in detail elsewhere.3,34 The formation of ruthenium oxide pyrochlore was confirmed by XRD as the diffraction pattern matched with the reported values.33 The GCE (3 mm diameter, BAS) was polished to mirror finish using the BAS polishing kit sequentially with diamond paste of decreasing particle size (15, 3, 1, and 0.05 mm).It was then rinsed with distilled water and further subjected to ultrasonic cleaning in the buffer and deionized water successively. For studies concerned with Ep shift with pH, the cleaning was performed in the respective pH buffer solutions.The Nafion coating was obtained by carefully spreading 6 ml of 4 wt% solution on the GCE and the electrode was spun at 2000 rev min21 to get a uniform film. The CME was equilibrated in test buffer solution before measurement. In general, the measurement regression cycles to prove the reproducibility of the electrode surface was carried out as follows. As soon as the measurement was carried out in test solutions, the electrode was transferred to a blank solution and SWV was carried out until the original background current was regained.Preparation of DNA and RNA solutions The ds-DNA and RNA were hydrolysed as follows for quantification of guanine and adenine. A gentle treatment of DNA with 1 mM HCl leads to the selective removal of its purine bases by cleavage of purine glycoside bonds.39 The calf thymus ds-DNA (sodium salt type I, Sigma) and RNA (Bakers yeast Type III, Sigma) were digested by mixing samples of Bakers yeast RNA (20 mg) and 1 M HCl (5 ml) in a sealed glass ampoule.After heating in a boiling water bath for 80 min, the solution was adjusted with 1 M NaOH to pH 9.2. The solution was diluted to 25 ml as the stock solution. The calf thymus ds- DNA was digested using 5 ml of 1 M HCl solution. Concentrations of DNA were determined spectrophotometrically with e260 = 6600 M21 cm21.40 The base sequences of synthetic oligonucleotides are as below: 5AGCGGTACAAAATGGGCGC3A (D55) 5AGTGCAATGCAATGCAAC3A (BR2) Spectrophotometric estimations of nucleotide base concentration in synthetic nucleotides were performed using the reported e values for all the nucleotides used.Results and discussion Electrochemical behaviour of guanine and adenine The SW voltammetric responses of 10 mM guanine in 0.1 M, pH 3.0 phosphate solution on three different electrodes are Fig. 1 SW voltammograms for (A) 10 mM guanine in 0.1 M, pH 3.0 phosphate solution and (C) 10 mM adenine in 0.1 M, pH 4.5 phosphate solution at a bare GCE (a), the Nafion–GCE (b), and the Nafion–ruthenium oxide pyrochlore CME (c).SW amplitude, 60 mV; SW frequency, 70 Hz; step height, 5 mV. Pp is 20.4 V for (A) and 20.2 V for (C); tp = 14 s. (B) Cyclic voltammograms for 10 mM guanine and adenine in 0.1 M, pH 3.0 phosphate solution with a scan rate of 300 mV s21. 680 Analyst, 1999, 124, 679–684illustrated in Fig. 1A. As can be seen, the response at the bare GCE is very poor with a small hump at +1.05 V. The Nafioncoated GCE gives increased signal at the same potential indicating that the Nafion coating improves the accumulation of guanine.The response of guanine on the CME is very pronounced resulting in a huge and well-defined peak at +0.92 V. The enhancement in current response and the lowering of oxidation potential are clear evidences of the catalytic effect of the CME toward guanine oxidation. We have observed a similar lowering of SWV Eps for various biologically important compounds.33–38 The inherent catalytic activity of the CME results from the fact that the outermost region of the Nafion bound oxide surface is hydrated; the catalytically active centres are the hydrated surface bound oxy–metal groups which act as binding centres for substrates.Previous studies indicated that guanine can be oxidised and undergoes adsorption on the carbon surface at pH 7.0 or above.41–44 This adsorption behaviour can hamper the determination of guanine by causing non-linear calibration curves.Whereas, it becomes perfectly diffusion-controlled at lower pH solutions.44 In this study, the extent of adsorption on the CME also depends on pH as described later. As adsorption seems to occur for almost all purine bases on solid electrodes, it is important to investigate the mechanism of the charge transfer of guanine on the CME first. The results from linear scan voltammetry (LSV) showed that the peak current (ip) was directly proportional to the scan rate (v) and the Ep shifted to more positive values with increasing v.The current function (ip/ Cv) remained unchanged with the v indicating an adsorption process. Meanwhile, the process was irreversible as confirmed from the fact that there was no cathodic peak observed in cyclic voltammetry (Fig. 1B) and from the observation of the Ep shift with scan rate in LSV. The number of electrons involved in the charge transfer step was then evaluated. Based on the slope of 65 mV decade21 from the linear variation of Ep with log(v) in LSV, the ana is calculated to be 0.98.45 Assuming a is equal to 0.5, the value for na is 2.It is known that, on carbon electrodes, guanine is oxidised by a (2e2, 2H+) removal and yields 8-oxyguanine which is capable of getting oxidised at less anodic potentials to yield various products.42 It indicates that the formation of 8-oxyguanine is the controlling step at the CME as in the case of a graphite electrode.To infer more about the adsorption of guanine on the CME, chronocoulometry was studied. As shown in Table 1, from the intercepts of the Anson’s plots,46 the excess charge corresponding to the adsorbed component (after background charge correction from the backward step) is found to be 7.787 mC at pH 3.0. Using geometric area and approximating the roughness factor as one—the observed charge corresponds to a surface concentration of 5.76 3 10210 mol cm22. The SW voltammograms of 10 mM adenine on three different electrodes are illustrated in Fig. 1C. On a bare GCE, the adenine oxidation peak is very weak yielding a small hump at +1.25 V. The CME yields a well-defined peak at +1.1 V, a potential of 150 mV less than Nafion-coated GCE. Unlike the very illdefined peak for adenine oxidation reported on a carbon fibre microelectrode,47 the peak here is well defined with a nearly perfect shape. The lower oxidation Ep again indicates the electrocatalytic effect of this CME.It seems that the electrocatalytic activity of this CME is common for both guanine and adenine but to a different magnitude (Fig. 1A and Fig. 1C). The charge transfer process for adenine is also found to be adsorption-controlled and irreversible. The n value in the charge transfer step is calculated as follows. The kinetic parameter ana was found to be 0.98 from the slope of 67 mV decade21 for Ep vs. log(v) plots in LSV. The value of na is equal to 2 indicating that the formation of 2-hydroxyadenine is the controlling step at the CME.Previous electrochemical oxidation studies on carbon electrodes showed evidence for the oxidation of adenine by an overall 6e2 in two steps.48 Removal of 6e2 from adenine suggests that the process proceeds initially by a (2e2, 2H+) oxidation to 2-hydroxyadenine and then rapidly to 2,8-dihydroxyadenine by a 4e2 removal and coupled chemical steps.48 The above results for both adenine and guanine indicate that the rate determining charge transfer steps are pretty similar to the results on graphite electrodes and the sequence of coupled chemical reactions and the final products formed are assumed to be the same.No further attempts were made to characterise the final products formed, as it has been very well established in the literature.42–44,47,48 Individual determination of guanine and adenine To arrive at the optimum conditions, the major factors that should be considered are the preconcentration time (tp), the preconcentration potential (Pp) and the SW parameters.The effect of Pp on the SW response for guanine is shown in Fig. 2A. As can be seen, the ip increases with the increase of Pp in the anodic direction and reaches a maximum at 20.4 V. Similarly, the tp also has tremendous influence in guanine determination as shown in Fig. 2B. The ip increases with the increase in tp and attains saturation from 5 s onward. The SW parameters were then optimised at Pp = 20.4 V and tp = 14 s.Increase in the SW amplitude increases the ip and shows saturation around 50 mV. Note that the peak widths also increase with increase in the SW amplitude. When the SW amplitude is higher than 60 mV, the resolution between these two purines is affected. As to the SW frequency, the ip increases initially and attains a maximum around 60 Hz and decreases on further increase in the Table 1 Chronocoulometrya of 10 mM guanine in 0.1 M, pH 3.0 phosphate solution at the CME pH CME/ intercept/mC Nafion–GCE intercept/mC GCE intercept/mC 2.0 5.224 0.754 1.344 3.0 7.787 0.793 1.408 4.0 7.381 0.736 1.780 5.0 6.981 0.837 2.019 6.0 5.482 0.743 2.513 a Experimental conditions: Ei = +0.67 V; Ef = +1.09 V; pulse, 350 ms. Intercepts were obtained after background charge correction from the backward step from Anson plots.Fig. 2 The effects of preconcentration potential (A) and preconcentration time (B) on the SWV response for guanine at the CME.Conditions are as in Fig. 1. Analyst, 1999, 124, 679–684 681frequency. We studied the effect of step width and it shows little influence on the ip. The best parameters selected for the determination of guanine at the CME were Pp = 20.4 V; tp = 14 s; SW amplitude 60 mV; SW frequency 70 Hz; step width 5 mV. The calibration curve for guanine in pH 3.0 phosphate solution showed two linear segments: the initial linear increase up to 1.1 mM with a higher slope of 8.06 mA mM21 and the second linear segment up to 10 mM with a lower slope of 2.02 mA mM21.The initial linear portion is understandably due to the strong adsorption of guanine on the CME. However, if the pure adsorption is the only controlling factor, the ip should linearly increase initially and eventually saturate at higher concentrations. The fact that there is a second linear segment indicates the onset of a diffusion process on the monolayer-covered surface. Nevertheless, these two linear regions can be used for the quantification of guanine at different concentration domains.The detection limit (S/N = 3) was 5.7 nM. To characterise the reproducibility of the CME, repetitive measurement–regeneration cycles were carried out in 10 mM and 0.3 mM guanine solutions. The results of 20 successive measurements showed 1.41% and 1.73% relative standard deviation for 10 mM and 0.3 mM guanine, respectively, indicating good renewals of the CME. As to adenine, similar studies were also done as in the case of guanine. The best parameters to determine adenine on the CME are Pp = 20.2 V; tp = 30 s; SW amplitude = 60 mV; SW frequency = 70 Hz; step width = 5 mV.Using these parameters, a calibration curve was constructed in 0.1 M, pH 4.5 phosphate solution. The calibration curve yields a linear range up to 5 mM and is saturated at higher concentrations. The detection limit (S/N = 3) is 20 nM. Similarly, to infer the reproducibility of the electrode, 10 continuous detections of 10 mM and 0.3 mM adenine show excellent reproducibility with a 2.54% and 2.78% relative standard deviation, respectively.Simultaneous determination of guanine and adenine Simultaneous determination of guanine and adenine on the CME shows strong pH dependence as shown in Fig. 3. The same trend was also observed when they were studied individually. The ip increases initially with pH and the highest ip was observed at pH 3.5 for guanine and pH 4.5 for adenine.The decreasing portion of ip with pH in the range of pH 1.0 to 3.0 can not be explained as the characteristic of guanine. Because guanine exists as a cation (pKb = 3.2)49 in this region, very good accumulation in the Nafion film is expected leading to the opposite trend. It seems that either the lower activity of Nafion in more acidic solutions or the effect internally associated with the CME can cause the result. Since the decrease in peak magnitude does not exceed 50% even at higher pHs, the sensitive estimation of these purine bases is still practically possible.Another important advantage is noticed in the Ep shift with pH. The Ep, for both guanine and adenine, changes linearly with pH with a slope of 60 mV pH21 indicating an even number of electrons and protons involved in the charge transfer step. Note that the variation of pH significantly affects the quality of peak separation between adenine and guanine, and subsequently quantitative estimation becomes virtually impossible in acidic pH values due to the inseparable conditions in reversed-phase capillary zone electrophoresis with end column amperometric detection.47 Such a problem does not appear in the present method, because both peaks shifted with an equal magnitude, and thus, the peaks are always well-separated in any of the pHs studied.Chronocoulometry was used to evaluate the adsorption characteristics of both guanine and adenine in various pH solutions.The variation of faradaic charge with pH is qualitatively similar to the variation of ip with pH. The surface coverage, calculated from an Anson plot, is relatively higher at pH 3.0 to 5.0 for both guanine and adenine. It is interesting to observe that the adsorption of both guanine and adenine depends on the pH of the solution. Similar trends for variation of ip with pH and surface coverage with pH confirm that the oxidation current observed is due to the adsorbed species on the electrode surface.The quantitative comparison made use of the charge calculated at all pHs being approximately equal substantiating the adsorption control of the charge transfer step for both adenine and guanine. The effect of SW parameters on the oxidation of guanine and adenine mixtures remains the same as they were studied separately. The SW parameters used in the analysis of mixtures are 60 mV and 80 Hz. The effect of Pp on the SW response for the simultaneous determination of guanine and adenine is shown in Fig. 4. As can be seen, the ip increases with the increase of Pp in the anodic direction and reaches a maximum at 20.2 V for both guanine and adenine. Thus, Pp = 20.2 V was selected for the simultaneous determination of guanine and adenine. As mentioned earlier, the tp needed for guanine is 14 s but it is 30 s for adenine. Thus, in analysing the mixtures, a longer tp of 30 s was chosen. As mentioned earlier, the calibration curve for guanine in pH 3.0 phosphate solution showed a linear increase up to 1.1 mM.A linear variation of ip up to 1.1 mM of guanine was thus observed in the presence of 1.1 mM adenine as shown in Fig. 5A. As can be seen, the ip of 1.1 mM adenine was found to remain a constant. It is therefore possible to determine quantitatively without any considerable influence on each other at this low concentration range. Meanwhile, a calibration curve was observed in pH 4.5 phosphate solution with a linear range up to 5 mM for adenine.The influence of ip for adenine was studied in Fig. 3 The pH effect in the simultaneous determination of guanine and adenine at the CME. Fig. 4 The effect of preconcentration potential on the SWV response for the simultaneous determination of guanine and adenine at the CME. 682 Analyst, 1999, 124, 679–684the presence of 5 mM guanine as shown in Fig. 5B. As can be seen, the ip of guanine was observed to remain constant when the concentration of adenine was increased close to 2 mM.On the other hand, a similar increase of ip with concentration was observed for adenine up to around 2 mM. This observation clearly demonstrates that these two purine bases can be estimated from a mixture at the concentration range studied. A set of calibration graphs could be constructed in the simultaneous determination of guanine and adenine and the linear range is up to 1.1 mM. Analytical applications It is envisaged that the method developed here can be used to detect the guanine and adenine concentrations in plasmid DNA samples.All detection strategies for the purine bases in ds-DNA face a severe decrease in signal as compared to that for ss-DNA since the bases are embedded in the interior of the double helix and therefore their detection is sterically hindered due to the crowded phosphate group on the exterior of the helix.50 The denaturation results in an unwinding of duplex DNA and renders the guanine and adenine residues more accessible to the electrode surface owing to the increased flexibility in its structure.In the present work, we investigated the possibility of using the proposed method to determine the purine bases in DNA and RNA after denaturation. The anodic SW sweep in BR2 or D55 DNA solution yielded two oxidation peaks at +0.84 V and +1.08 V, which correspond to the oxidation of guanine and adenine, respectively. To confirm the above assignment, guanine and adenine were spiked to the DNA solution.Spiking with a known amount of guanine proportionately increased the peak at +0.84 V, while the peak at +1.08 V corresponding to the adenine content of DNA remained constant. Similarly, when known aliquots of adenine were spiked, the more anodic peak at +1.08 V increased proportionately while the guanine peak of DNA remained unaffected. The concentrations of adenine and guanine in BR2 and D55 synthetic DNA were first measured from ip using the calibration graph constructed previously for simultaneous detection of a guanine and adenine mixture and the results are summarised in Table 2.As can be seen, the measured concentrations of guanine and adenine were in good agreement with the concentrations obtained by a spectrophotometric method. Thus, according to the present method, the guanine and adenine in nucleic acid samples can be accurately measured by simply spiking the test DNA solution to the known amount of guanine and adenine mixture.In the case of calf thymus DNA, Fig. 6 shows typical SWV responses and calibration curves for the simultaneous determination of guanine and adenine with increasing DNA concentrations of 0.33 ppm to 3.96 ppm at 0.33 ppm intervals on the CME. The resulting peak heights are more prominent than the AdTSV signal recently reported on HMDE for denatured calf thymus DNA.23 The following comparison can be made.The AdTSV from 5 ppm (mg ml21) denatured DNA yielded a signal (corresponding to adenine and cytosine reduction) of 7.5 nA on 0.4 mm2 HMDE, i.e., 1.3 nA cm22. However, on the CME used in the present work 500 nA cm22 was yielded for the adenine or guanine oxidation, for the same 5 ppm (Fig. 6). This shows that the SW voltammetry on the present CME is ca. 4003 more sensitive than the recent report on HMDE.23 The lowest point that could be detected on the CME was 0.33 ppm of DNA.The detection limit (S/N = 3) of DNA calculated are 11.5 ng ml21 and 37.8 ng ml21 for guanine and adenine, respectively. Furthermore, the [guanine] : [adenine] ratio also testifies to the above contention. According to the Chargaff rule, the ratio of Fig. 5 The influence of ip for (A) guanine in the presence of 1.1 mM adenine and (B) adenine in the presence of 5 mM guanine. SW amplitude, 60 mV; SW frequency, 70 Hz; step height, 5 mV; Pp = 20.2; tp = 30 s. The supporting electrolyte is 0.1 M, pH 4.0 phosphate solution.Table 2 Simultaneous determination of guanine and adenine in synthetic oligonucleotides with the CME BR2/guanine BR2/adenine D55/guanine D55/adenine Detected value after dilution/mM 0.151 ± 0.003 0.226 ± 0.018 0.336 ± 0.006 0.240 ± 0.008 Spike/mM 0.3 0.3 0.3 0.3 Detected value after spike/(mM) 0.450 ± 0.016 0.506 ± 0.012 0.637 ± 0.007 0.533 ± 0.018 Recovery (%) 99.67 93.33 99.93 97.67 Total valuea/mM 50.45 75.48 67.30 48.08 Reference valueb/mM 50.47 75.71 67.49 48.21 a Total value was obtained by multiplying the detected value with the dilution factor of 333.3 and 200 for BR2 and D55, respectively.Number of samples assayed was three. b Reference value was obtained by the spectrophotometric method. Analyst, 1999, 124, 679–684 683[guanine] : [adenine] from our data is calculated as 0.770. The estimated value is fairly close to the reported value of 0.786.47 Similarly, in the case of yeast RNA, the estimated molar percentage of the guanine and adenine result is 1.021, which is also fairly close to the reported value of 0.969.47 This result proves the accuracy of the present protocol.On the other hand, the ip for adenine and guanine are directly related to the quantity spiked revealing that any amount of DNA or RNA can be directly spiked and the contents can be inferred. The above observation indicates the advantage in using the CME in the simultaneous determination of guanine and adenine in DNA and RNA.However, it should be mentioned that the volume requirement in the present approach is relatively higher (minimum 2 ml used), since the analysis is made in the DNA solution itself. To address this, another advantage of the present method can be realised in the form of a disposable screenprinted electrode for routine DNA analysis, which is currently being worked out in our laboratory. Conclusion We have demonstrated that nucleic acid samples can be analysed on the Nafion–ruthenium oxide pyrochlore CME for their quantitative estimations and the sensitivity is comparable with the conventional HMDE.The main advantage of CME is that the guanine and adenine oxidation can be monitored unlike HMDE where the mercury dissolution will occur so that the oxidation can not be studied. The oxidation of these purines occurs at well-separated Eps and thus the concentration of nucleic acids can be directly measured from the magnitudes of these peaks.Unlike the overlap in electropherograms in chromatographic detection posing problems in separating the crowded peaks, the Ep separation for guanine and adenine remain the same at all pH values on the CME. Meanwhile, the CME was previously applied to the detection of codeine in human plasma with excellent selectivity.51 Thus, the present method can be advantageously used at any desired pH and should bring electroanalysis closer to the needs of contemporary molecular biology research.The authors gratefully acknowledge financial support from the National Science Council of the Republic of China under Grant NSC 88-2113-M-005-019. The authors thank Dr. S.-H. Chou for providing the synthetic nucleotides. References 1 R. Sheng, F. Ni and T. M. Cotton, Anal. Chem., 1991, 63, 437. 2 M. Kai, Y. Ohkura, S. Yonekura and M. Iwasaki, Anal. Chim. Acta, 1994, 287, 75. 3 B. Todd, J. Zhao and G. Fleet, J. Microbiol. Methods, 1995, 22, 1. 4 T.Torda and J. M. Saavedra, Neuroendocrinology, 1990, 52, 361. 5 N. Kuroda, K. Nakashima and S. Akiyama, Anal. Chim. Acta, 1993, 278, 275. 6 H. C. Tseng, C. Dadoo and R. N. Zare, Anal. Biochem., 1994, 222, 55. 7 J. P. Hart, Electroanalysis of Biologically Important Compounds, Ellis Horwood, Chichester, 1990, p. 51. 8 D.-W. Pang, Y.-P. Qi, Z.-L. Wang, J.-K. Cheng and J.-W. Wang, Electroanalysis, 1995, 7, 774. 9 T. Nakahara, M. Okuzawa, H. Maeda, M. Hirano, T. Matsumoto and H. Uchimura, J.Chromatogr., 1992, 15, 1785. 10 H. Lin, D. K. Xu, and H. Y. Chen, J. Chromatogr. A, 1997, 760, 227. 11 J. Wang, L. Chen and M. Chicharro, Anal. Chim. Acta, 1996, 319, 349. 12 D. K. Xu, L. Hua and H. Y. Chen, Anal. Chim. Acta, 1996, 335, 95. 13 P. Singhal and W. G. Kuhr, Anal. Chem., 1997, 69, 3552. 14 P. Singhal and W. G. Kuhr, Anal. Chem., 1997, 69, 4828. 15 J. Wang, in Electroanalytical Chemistry, ed. A. J. Bard, Marcel Dekker, New York, 1989, vol. 16, p. 1. 16 E.Palecek, P. Boullikova and F. Jelen, Anal. Chim. Acta, 1986, 187, 99. 17 P. Boulikova, F. Jelen and E. Palecek, Stud. Biophys., 1986, 114, 83. 18 E. Palecek, I. Post Bieglova, J. Electroanal. Chem., 1986, 214, 359. 19 E. Palecek, Anal. Biochem., 1988, 170, 171. 20 E. Palecek, Bioelectrochem. Bioenerg., 1988, 20, 171, 21 E. Palecek, Bioelectrochem. Bioenerg., 1992, 28, 71. 22 E. Palecek, F. Jelen, C. Teijeiro, V. Fucik and T. M. Jovin, Anal. Chim. Acta, 1993, 273, 175. 23 E. Palecek and M. Fojta, Anal.Chem., 1994, 66, 1566. 24 E. Palecek, Anal. Lett., 1980, 13(B5), 331. 25 E. Palecek, Anal. Biochem., 1980, 108, 129, 137. 26 G. Dryhurst, Anal. Chim. Acta, 1971, 57, 137. 27 M. A. T. Gilmartin and J. P. Hart, Analyst, 1992, 117, 1613. 28 X. Sun, Y. Chi and H. Chen, Microbiol. J., 1993, 47, 287. 29 R. M. Shubietah, A. Z. Abu Zuhri and A. G. Fogg, Fresenius’ J. Anal. Chem., 1994, 348, 754. 30 G. Dryhurst, Talanta, 1972, 19, 769. 31 X. Cai, B. Ogarevc and K. Kalcher, Electroanalysis, 1995, 7, 1126. 32 R. L. McCreery, in Electroanalytical Chemistry, ed. A. J. Bard, Marcel Dekker, New York, 1991, vol. 13, p. 221. 33 J.-M. Zen and C.-B. Wang, J. Electroanal. Chem., 1994, 368, 251. 34 J.-M. Zen and J.-S. Tang, Anal. Chem., 1995, 67, 1892. 35 J.-M. Zen and Y.-S. Ting, Anal. Chim. Acta, 1997, 342,, 175. 36 J.-M. Zen and J.-S. Tang, Anal. Chem., 1995, 67, 208. 37 J.-M. Zen and I.-L. Chen, Electroanalysis, 1997, 9, 537. 38 J.-M. Zen, I.-L. Chen and Y. Shih, Anal. Chim. Acta, 1998, 369, 103. 39 R. H. Garrett and C. M. Crishan, Biochemistry, Saunders College Publishing, Orlando, 1995. 40 D. H. Johnson, K. C. Glasgow and H. H. Thorp, J. Am. Chem. Soc., 1995, 117, 8933. 41 K. Albert and E. Bayer, Trends Anal. Chem., 1988, 7, 288. 42 R. Y. Goyal and G. Dryhurst, J. Electroanal. Chem., 1982, 135, 75. 43 Y. Yao, T. Wasa and S. Musha, Bull. Chem. Soc. Jpn., 1977, 50, 2919. 44 J. Wang and B. Freika, Bioelectrochem. Bioenerg., 1984, 12, 225. 45 A. J. Bard and L. R. Faulkner, Electrochemical Methods, Fundamental and Applications, Wiley, New York, 1980, ch. 6. 46 A. J. Bard and L. R. Faulkner, Electrochemical Methods, Fundamental and Applications, Wiley, New York 1980, ch. 5. 47 W. Jin, H. Wei and X. Zhao, Electroanalysis, 1997, 9, 770. 48 G. Dryhurst and P. J. Elving, J. Electrochem. Soc., 1968, 115, 1014. 49 Merck index, 11th edition. 50 E. Palecek, Electroanalysis, 1996, 8, 7. 51 J.-M. Zen, M.-R. Chang, H.-H. Chung and Y. Shih, Electroanalysis, 1998, 10, 517. Paper 9/00532C Fig. 6 Typical SWV responses and calibration curves for the simultaneous determination of guanine and adenine in DNA with increasing DNA concentrations of 0.33 ppm to 3.96 ppm at 0.33 ppm intervals at the CME. The supporting electrolyte is 0.1 M, pH 4.0 phosphate solution. 684 Analyst, 1999, 124, 679–684

ISSN:0003-2654    

DOI:10.1039/a900532c

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

Development of novel thermochromic plastic films for optical temperature sensing Andrew Mills* and Anne Lepre Department of Chemistry, University of Wales Swansea, Singleton Park, Swansea, UK SA2 8PP. E-mail: a.mills@swansea.ac.uk Received 19th January 1999, Accepted 19th March 1999 The preparation of plastic film optical ‘CO2-based’ temperature-sensing films that utilise the temperature-dependent acid–base equilibria of indicator dyes is described. In film formulations a suitable dye, such as phenolphthalein, in a hydrophobic base, tetraoctylammonium hydroxide, solubilised within a plasticised, hydrophobic polymer matrix, creates a system which is sensitive to ambient CO2 levels.The resultant solution, when cast on to glass supports, yields ‘CO2-based’ temperature-sensitive films which change colour in response to a highly temperature-dependent reaction between the deprotonated form of the dye and CO2 dissolved in the film. The absorbance characteristics of these films display a fully reversible response to temperature over a temperature range which is largely determined by the pKa of the dye and the ambient CO2 concentration.The magnitude of the response is dependent on the dye concentration. The response time towards changes in temperature is typically @2.2 min and the films show good stability under operational conditions. A simple mechanism of the reaction is suggested and an associated working equation has been derived and fitted to data obtained for a typical sensor functioning over the range 278–333 K.A ‘CO2-based’ temperature-sensing film is used successfully alongside a standard CO2 sensing film. This combination not only provides temperature information but also ensures that the response of the CO2 sensor is corrected for any changes in temperature. In addition, both sensors use the same interrogating light and light intensity monitoring system because they contain the same phenolphthalein dye. The latter two features represent an improvement on the existing optical systems used to measure CO2 and temperature. 1. Introduction Considerable work has focused on the development of optical sensors for the quantitative detection of a wide range of analytes, including gases, ionic species and biomolecules.1,2 Another parameter which is of considerable biological and commercial importance and which often needs to be measured concurrently with the above species is temperature, and this also lends itself to measurement using optical sensors. Where temperature is to be measured at the same time as another analyte, it is advantageous to use the same method of detection for both as this reduces the complexity of the instrumentation required.3 A range of optical sensors4–6 recently developed for analyte detection in areas such as blood gas analysis, bioreactor monitoring and modified atmosphere packaging utilise a pHsensitive absorbance- or fluorescence-based indicator which detects the analyte by interacting with acidic or basic species in the sample.The use of the same measurement technique (i.e., absorbance or fluorescence) and, if possible, the same sensor unit for temperature measurements would therefore be useful, especially where temperature is known to affect the sensor’s response to the other analyte of interest and so needs to be continuously monitored. Several groups7–9 have utilised the temperature sensitivity of pH indicator–buffer systems to develop optical thermometers.Most of these rely on the temperature sensitivity of the pKa of the buffer component, usually tris(hydroxymethyl)aminomethane (TRIS), which has a particularly high DpH/DT coefficient (20.015 °C21).8 This is coupled to the pH sensitivity of a dye to produce a measurable optical response. An early example is the work by Bowie et al.,8 who used TRIS and the pH indicator cresol red to prepare an aqueous temperature-indicating solution for monitoring intra-cuvette temperatures in clinical analysers.Straub and Seitz3 developed this idea further in their design of a fibre optic-based temperature sensor which also utilised the temperature sensitivity of the pKa of TRIS. In this sensor, the acid–base indicator dye phenol red was immobilised in a crosslinked polyacrylamide matrix which was soaked in an aqueous solution of TRIS buffer. This was used as the basis of a semisolid state sensor which was capable of making remote temperature measurements.Both of these approaches exploited the temperature sensitivity of TRIS buffer coupled with a pH-sensitive indicator dye to produce an optical response. However, many dyes themselves display marked temperature sensitivity and the thermochromism of pH-sensitive indicators has been well documented.10,11 This thermochromic behaviour results from the temperature dependence of the acid–base equilibrium DH D H a [| K T , - + + (1) where DH and D2 are the protonated and deprotonated forms of the dye, respectively, and Ka,T is the acid dissociation constant for the dye at temperature T.We have found that dyes of this type, when ion-paired to a suitable phase transfer agent, may be readily incorporated into hydrophobic polymers to yield thin, effectively solid state films.4,5 Such films have been successfully used, in their pHsensing capacity, to produce optical sensors for the detection of acid gases such as carbon dioxide, which is the principle acid gas found in air and an important analyte in medical monitoring.These pH indicator-based, optical CO2 sensors change colour in response to a reversible change in the local pH induced by CO2 as it interacts with the film. As in solution, the pH equilibria set up in these polymer films, and hence their response to CO2, are highly temperature dependent. In general, an increase in temperature decreases the sensitivity of such sensors toward Analyst, 1999, 124, 685–689 685CO2. In their role as CO2 sensors this cross-sensitivity is a problem and such film sensors are generally thermostated at a constant temperature to overcome it. It may be envisaged, however, that if, instead of keeping the temperature constant, the level of CO2 in the environment surrounding the sensor is kept constant, the sensor can be transformed into a novel temperature sensor or optical thermometer.In this paper, the development and characterisation of solid state thin film optical thermometers based on the temperature sensitivity of the reaction of pH indicator dyes with ambient CO2 is described.A model is presented which allows the behaviour of such sensors to be predicted quantitatively and the possibility of the simultaneous measurement of both an analyte gas (in this case CO2) and temperature using essentially the same sensing unit is investigated. 2. Experimental 2.1 Materials Ethylcellulose (ethoxyl content 46%), tributyl phosphate, tetraoctylammonium bromide and the dyes phenolphthalein, thymol blue and m-cresol purple were obtained from Aldrich Chemicals (Gillingham, Dorset, UK).A 0.5 mol dm23 tetraoctylammonium hydroxide solution in methanol was prepared from the corresponding bromide solution using wet silver oxide to effect the ion exchange.12 The gases used, N2 and a 5% CO2–N2 blend, were of high purity ( > 99%) and were purchased from BOC, Guildford, Surrey, UK.The solvents used to make up the various solutions were of high purity and were purchased from Aldrich Chemicals. 2.2 Preparation of sensing films The temperature-sensing polymer film sensors had the general composition: indicator dye–phase transfer agent–polymer– plasticiser–glass support. The phase transfer agent, tetraoctylammonium hydroxide (TOAOH), serves both to solvate the hydrophilic indicator dye anion in the hydrophobic polymer matrix and to provide the trace water required to allow dissolution of CO2 in the film.The plasticiser, tributyl phosphate, is present to improve the gas permeability of the film. The standard temperature-sensitive plastic film sensor used in this work contained phenolphthalein as the dye. Film solutions were prepared by adding 5 mg of phenolphthalein to 1.25 cm3 of methanolic TOAOH (0.5 mol dm23) and then adding a further 0.5 cm3 of methanol. This was then added to 5 cm3 of a 10% (m/v) solution of ethylcellulose in toluene– ethanol (80 + 20).Finally, 0.5 cm3 of the plasticiser, tributyl phosphate, was added and the film solution was stirred thoroughly. This gave a final dye concentration in the film solution of 2.2 3 1023 mol dm23. Where other dyes were used, the quantity added was adjusted to give the same molarity. The final solvent-free films were prepared by casting the film solutions on to glass microscope cover-slips through a 100 mm thick brass template with a rectangular hole (0.8 3 1.5 cm).Films were left to dry overnight at ambient temperature and were subsequently stored in a desiccator. Calculations based on the known area and mass of the films gave film thickness estimates of 15–20 mm. 2.3 Instrumentation UV/VIS absorption spectra and single wavelength absorption measurements were recorded using a Lambda 3 double beam scanning spectrometer (Perkin-Elmer, Norwalk, CT, USA) and a Model 8625 single beam spectrometer (Unicam, Cambridge, UK).The glass cover-slips on to which the films had been cast were taped on to a 1 cm thick cylindrical glass vessel equipped with water inlet and outlet tubes. Temperature control was achieved by pumping water from a thermostated water-bath through this vessel using a Grant SU6 thermostat unit (Grant Instruments, Cambridge, UK). A Grant Type CC25 cooling unit was used where temperatures below room temperature were required. Where concentrations of CO2 other than atmospheric were required, the appropriate blend was generated using a gas blender (Model 852VI-B, Signal Instruments, Camberley, Surrey, UK).Computer modelling was performed using a Jandel Scientific (San Rafael, CA, USA) curve fitting program. 3. Theory It is useful first to consider the simple case of a pH-sensitive dye dissolved in water and buffered at a fixed proton concentration, [H+]fixd. In aqueous solution the dye can exist in both a protonated and a deprotonated form (DH and D2) and the relationship between the relative concentrations of these two species is determined by the equilibrium reaction (1).Ka,T, the acid dissociation constant for the dye at the temperature of the experiment, is then defined as follows: Ka,T = [D2][H+]fixd/[DH] (2) The value of Ka,T will vary with temperature Ka,T = Ka,T=°. exp(2DH°/RT) (3) where DH° is the change in standard enthalpy for equilibrium reaction (1) and is invariably > 0 and Ka,T=° is the value of Ka,T at infinite temperature [ = exp(DS°/R)].Often the UV/VIS absorption spectra of DH and D2 are completely different; typically the wavelength of maximum absorption of DH is much less than that of D2, i.e., lmax(DH) < lmax(D2). In many cases this difference in UV/VIS absorption spectra is such that the molar absorptivity of DH at lmax(D2) is approximately zero and, under these conditions, any absorbance measurements made at lmax(D2), i.e., Abs(D2)T, reflect solely the concentration of D2 in the aqueous solution at temperature T and [H+]fixd.If we assume that such a situation holds for the dye in reaction (1), it follows from eqns. (2) and (3) and Beer’s law that Ka,T/[H+]fixd = (Ka,T=°/[H+]fixd)exp(2DH°/RT) = Abs(D2)T/[Abs(D2)T=° 2 Abs(D2)T] (4) where Abs(D2)T=° is the absorbance of the aqueous solution at infinite temperature, at which imaginary point all the dye will be fully deprotonated, i.e., Abs(D2)T=° is a direct measure of the total dye concentration, where [dye]total = [D2] + [DH], assuming the UV/VIS absorption spectrum of D2 does not change with temperature. A value of Abs(D2)T=° can be gleaned through a knowledge of the original total dye concentration, [dye]total, and the molar absorptivity of D2 at lmax(D2), e(D2), since Abs(D2)T=° = [dye]total·e(D2) 3 pathlength.It follows from eqn. (4) that the experimentally measurable parameter Abs(D2)T is related to the temperature of the aqueous solution as follows [Abs(D2)T]21 = exp(DH°/RT)[H+]fxd/[Ka,T=° Abs(D2)T=°] + [Abs(D2)T=°]21 (5) In earlier work,10,11 a pH buffer was used to achieve a ‘fixed’ pH.However, not surprisingly, the pH of the buffer is also temperature dependent, {i.e., [H+]fxd,T = [H+] fxd,T=° exp- (2DH°buff/RT), where DH°buff is usually > 0}, and as a result, the observed value for DH, DHobs, as determined using a suitably modified version of eqn. (5), will be the difference in DH, for the indicator and the buffer, i.e., DH°obs = DH° 2 DH°buff. 686 Analyst, 1999, 124, 685–689In our work using thin plastic films for sensing carbon dioxide, the pH-sensitive dye is solubilised using a phase transfer agent, Q+OH2, into the hydrophobic medium of the encapsulating plastic medium. However, the pH-sensitive dye is still able to reflect a change in the pH of the surrounding medium and change colour when exposed to CO2. The key equilibrium reaction is as follows Q HCO H O DH Q D H O CO Colour B Colour A + - + - × - ( ) × × + 3 2 2 2 1 x x KT [| * (6) where Q+HCO32·(x 2 1)H2O·DH and Q+D2·xH2O are the lipophilic forms of the protonated and deprotonated forms of the dye, with very similar UV/VIS absorption spectral characteristics to DH and D2, respectively.KT* is the equilibrium constant for the overall process at temperature T and is directly related to the acid dissociation constant for the dye, i.e., Ka,T. The overall process, reaction (6), is in fact a combination of the equilibrium reaction involving dissolved CO2 and the deprotonated dye and the equilibrium reaction between dissolved CO2 and the CO2 in the gas phase.In the past, when using plastic films as optical sensors for CO2 at a fixed temperature, T, any variation in the ambient partial pressure of CO2, PCO2, can be detected by monitoring Abs(D2)T,film, the absorbance at lmax(D2) of the film at ambient temperature. In the present work, where these films are used as temperature sensors, it is the partial pressure of ambient CO2 which is fixed.Under these conditions, when the ambient temperature is varied, the films should behave in a manner similar to that for a pH-sensitive dye in aqueous solution, in which the [H+] is fixed, and the key equation relating the observed absorbance due to D2 encapsulated in the film, Abs(D2)T,film, and temperature will be similar to that of eqn. (5), i.e., [Abs(D2)T,film]21 = exp(DH°obs/RT)PCO2,fxd/[K*T=° Abs(D2)T,film*] + [Abs(D2)T,film*]21 (7) where K*T=° is the value of K*T at infinite temperature, DH°obs is the enthalpy change for the overall process and Abs(D2)T,film* is the absorbance of the film when all of the dye is in the form of D2.From eqn. (7), it follows that the sensitivity of the film at any temperature, i.e., dAbs(D2)T,film/dT, is given by the expression d D d D ilm obs obs obs film Abs T H RT H RT H RT Abs T f T ( ) / ( / ) exp( / ) { exp( / ) [ ( ) ] } , , - - - = × × + D D D o o o 2 1 2 b b (8) where b = PCO2,fxd/K*T=°·Abs(D2)T,film* (9) A useful rough guide to a temperature around which the particular thermochromic sensor, based on the above CO2- sensitive plastic films, will operate is T(S = 1 2), the temperature at which Abs(D2)T,film is 50% of its maximum value, i.e., when Abs(D2)T,film = Abs(D2)T=°,film/2 = Abs(D2)T,film*/2.Under these conditions, it can be shown from eqn. (7) that T(S = 1 2) is related to the experimental parameters PCO2,fxd, K*T=° and DH° as follows: T(S = 1 2) = DH°obs/Rln(K*T=°/PCO2,fxd) (10) From eqn.(10), it follows that this type of thermochromic sensor will operate at lower T(S = 1 2) values if (a) the dye is changed to one with a lower pKa (since K*T is related directly to the acid dissociation constant for the dye) or (b) if the initial, fixed partial pressure of carbon dioxide, PCO2,fxd, is decreased. From a combination of eqns. (8) and (10), the following expression for the film sensitivity, at T(S = 1 2), can be derived: [dAbs(D2)T,film/dT]T(S = 1 2) = R·Abs(D2)T,film*[ln(K*T=°/ PCO2,fxd)}2/4DH°obs (11) If this model applies to our system, various predictions can be made based on these equations.From eqn. (11), it follows that the sensitivity of a film at T(S = 1 2) can be increased by (i) changing to a dye with a larger Ka,T=°, i.e., one with a lower pKa, (ii) decreasing the initial, fixed partial pressure of carbon dioxide, PCO2,fxd, or (iii) increasing the initial dye concentration [which will increase Abs(D2)T,film*].In addition, eqn. (7) can be rearranged to the following form: Abs(D2)T,film = 1/[bexp (a/T) + g] (12) where a = DH°obs/R, b = PCO2,fxd/K*T=°Abs(D2)T,film* and g = [Abs(D2)T=°,film*]21; a is in units of K21 and b and g are both unitless. It follows from eqn. (12) that from experimental data of the form Abs(D2)T,film vs. temperature, values for a, b and g, which are characteristic of the film, can be extracted.In the following section the model and its associated equations are tested using experimental data for a series of different film temperature sensors. 4. Results and discussion 4.1 Temperature response of a phenolphthalein film in atmospheric [CO2] A thin plasticized ethylcellulose film containing the indicator dye phenolphthalein was prepared and its temperature sensitivity was investigated. This dye has a high pKa (9.6)13 and at room temperature and atmospheric CO2 levels (3 3 1024 atm) it is almost exclusively in its protonated, colourless form.As the temperature of the film was raised an absorption peak gradually developed at 573 nm. Increasing the temperature drives the equilibrium reaction (6) more to the right, thereby increasing the concentration of the deprotonated form of the dye, which is deep pink. These spectral changes are illustrated in Fig. 1. Further work showed that the response of the film was fully reversible over many cycles between 305 and 345 K (32–70 °C).Above a temperature of 343 K for a prolonged period ( > 30 min), the film became permanently coloured, probably owing to thermal destruction of the base, tetraoctylammonium hydroxide. Cycling the temperature between the physiologically important range 295 and 318 K (which experimentally is achieved in 15 s) gave a 90% response and recovery time of 2.2 and 1.0 min, respectively. In order to extract values for a, b and g, a set of absorbance values obtained at 573 nm for the film over the range 305–345 K were fitted to eqn.(12). Fig. 2 shows the experimentally determined data points and the line passing through them represents the least-squares line of best fit of the data to eqn. Fig. 1 Effect of temperature on the absorption spectrum of a phenolphthalein –ethyl cellulose film. Temperature (from top to bottom): 308, 311, 314, 317, 321 and 324 K. Analyst, 1999, 124, 685–689 687(12).This line of best fit yielded values for a, b and g of 6.65 3 103 K, 5.62 3 1029 and 1.93, respectively. From the optimised value for a obtained from this work, a DH° value of 55.3 kJ mol21 was calculated for this reaction, which falls into the DH° range 31–71 kJ mol21 and which has been reported previously for dye protonation reactions.4,14 In all subsequent work, pH-sensitive dyes of a similar nature were used and the key reaction (6) was the same. As a result, the value of DH obtained from the value of a, i.e., 55.3 kJ mol21, was used in all of the model calculations and optimisations as it is likely to be the same for all of the systems studied.To check that this assumption was reasonable, some optimisations were also re-run without inputting this value as a constant, thereby allowing the optimisation program to generate its own value for a. When this was done, the value for DH obtained was always very close to 55.3 kJ mol21. The value of g yields a value of Abs(D2)T,film* of 0.52, which agrees reasonably well with the value of 0.58 calculated from a knowledge of initial dye concentration, the film thickness and the molar absorptivity of D2 at its lmax.The slight discrepancy between Abs(D2)T,film* and the calculated film absorbance may arise because the molar absorptivity used in the latter was for the D2 in aqueous solution not the encapsulating medium. The latter probably differs slightly from that associated with the anionic form of the dye, now in the form of an ion pair, Q+D2·xH2O, in a plastic film environment.The optimised values of a, b and g in eqn. (12) allows a complete Abs(D2)T,film versus temperature plot over an extended temperature range to be constructed for the phenolphthalein –ethylcellulose film. The inset in Fig. 2 shows this calculated profile together with the data points from which it was derived. It should be stressed that this full curve is not intended to represent the operational capabilities of the sensing film, which is restricted to the relatively narrow range 305–345 K, but rather illustrates the overall shape predicted by the mathematical function in eqn.(12). 4.2 Effect of local PCO2, pKa of the dye and initial dye concentration In a series of additional experiments, the effects of changes in local PCO2, pKa of the dye and initial dye concentration on the absorbance versus temperature profiles of the temperature film sensors were investigated. Eqn.(12) predicts that altering the former two parameters will affect the value of the variable b, while altering the latter will affect the value of g. Fig. 3 shows the effect of four different PCO2 levels on the temperature sensitivity of a standard phenolphthalein film. The data points were experimentally determined and the lines through them represent the lines of best fit to these data points determined using eqn. (12), a = 6.65 3 103 K and optimised values of g and b.As expected from eqn. (12), an increase in PCO2 resulted in an increase in the value of b, and the Abs(D2)T,film versus temperature profiles were shifted to higher temperatures, i.e., T(S = 1/2) increased. As predicted by eqn. (9) of the model, a plot of the optimised values of b versus PCO2 (shown in the inset to Fig. 3) is a good straight line. The pKa of the indicator dye used was also varied and the results of this work are illustrated in Fig. 4. The choice of dye was limited to those with a suitable pKa for making measurements at atmospheric levels of CO2. Hence the two phthalein dyes phenolphthalein (pKa = 9.6)13 and o-cresolphthalein (pKa = 9.4)15 and the sulphonphthalein dye thymol blue (pKa = 9.0)16 were used. From the results in Fig. 4 it can be seen that as the pKa of the indicator was decreased the value of b decreased and, as a consequence, T(S = 1/2) decreased. Eqn. (12) predicts a linear inverse relationship between the Ka of the dye and b and Fig. 2 Absorbance versus temperature plot of a phenolphthalein– ethylcellulose film. The line through the experimentally determined data points represents the least-squares line of best fit calculated using eqn. (13). Optimised values of a, b and g were a = 6.65 3 103, b = 5.62 3 1029 and g = 1.94. The inset represents the theoretical curve over an extended temperature range calculated using eqn. (13) and values of a, b and g obtained from the main diagram.Fig. 3 Effect of PCO2 on experimentally determined absorbance versus temperature plots. PCO2 values were 2.00 3 1025 (5), 3.90 3 1024 (2), 1.37 3 1023 (½) and 2.76 3 1023 (8) atm. The lines through the experimentally determined data points represent the least-squares line of best fit calculated using eqn. (13) and taking a = 6.65 3103. The optimised b and g values obtained were 8.17 3 10210, 1.96 (5), 2.05 3 1029, 1.32 (2), 3.78 3 1029, 1.00 (½) and 7.47 3 1029, 1.26 (8).The inset shows a plot of b versus PCO2 using optimised b values obtained from plots in the main diagram. Fig. 4 Effect of dye pKa on experimentally determined absorbance versus temperature plots. The dyes used were phenolphthalein (5), o-cresolphthalein (8) and thymol blue (½). The lines through the experimentally determined data points represent the least-squares line of best fit calculated using eqn. (13) and taking a = 6.65 3 103. The optimised b and g values obtained were 5.62 31029, 1.94 (5), 1.84 31029, 2.08 (8), 6.45 310211, 1.93 (½).The inset shows a plot of 1/b versus dye Ka values using optimised b values obtained from plots in the main diagram. 688 Analyst, 1999, 124, 685–689this appears to be confirmed by the data plot of 1/b versus dye Ka (inset in Fig. 4). Fig. 5 illustrates the results of a study of the effect of the initial dye concentration on the absorbance versus temperature profiles of a series of films which used the same dye, phenolphthalein.From these data it can be seen that raising the initial dye concentration did not affect the position of the profile, i.e., T(S = 1/2) remained unchanged, but did increase the total absorbance change observed. As a result, the temperature sensitivity of such sensors, dAbs(D2)T,film/dT, increased with increasing amount of dye [i.e., increasing Abs(D2)T,film]. From eqn. (12) of the model, a linear relationship between 1/g and Abs(D2)T,film was expected and found, as illustrated by the plot of the data in the inset Fig. 5.From these results, it appears that the model applies well to this system and that it may be used to make both qualitative and quantitative predictions about the response characteristics of this type of temperature-sensing film. 4.3 Simultaneous measurement of temperature and CO2 As already stated, pH-sensitive films which have been designed for monitoring analytes such as CO2 are frequently thermostated to avoid interference from temperature sensitivity.This may not always be desirable, however, owing to factors such as cost, accessibility and ease of miniaturisation. As it is easy to correct the response of such sensors for temperature effects, thermostating might be avoided if a reliable temperaturesensing element could be incorporated into the unit. Experiments showed that the temperature-sensing film (in which a fixed level of CO2 was encapsulated by a gasimpermeable membrane) could be readily used next to a CO2- sensing film which utilised the same dye, i.e., m-cresol purple.Both films could be interrogated using the same optics. The temperature-sensing film provided the necessary information to correct any change in response in the CO2 sensing film due to a change in ambient temperature over the range tested, i.e., 298–318 K. The attractive feature of this system is the use of the same optical system to provide both temperature and PCO2 data using two simple, similar optical sensors. 5. Conclusions A ‘CO2-based’ temperature-sensing film has been developed and the parameters affecting its response characteristics have been quantified. It has been used successfully alongside a standard CO2-sensing film to monitor both temperature and PCO2 simultaneously using the same interrogating and monitoring system. The reliability of the sensor for CO2 detection has been demonstrated many times in previous papers.4,17 Most commonly, IR spectroscopy is used to confirm the reliability of the CO2 sensor.In the work described in this paper, mercury thermometers were used to record independently temperature and the typical absorbance versus temperature profiles illustrated were recorded several times in order to confirm their reliability and reproducibility. As noted earlier, although the films functioned well between 305 and 345 K, at temperatures above 345 K prolonged exposure caused some degree of hysteresis in the absorbance versus temperature profiles of the sensor.In the present study, experiments were performed using variable PCO2 levels provided by a gas blender and variable temperature achieved by thermostating the gas and the cell block containing the sensor elements inside the spectrometer cell compartment. However, in practice, for real environment measurements of PCO2 and temperature, it is envisaged that both optical sensors (i.e., the CO2 and the ‘CO2-based’ temperature sensors) would be used in combination with fibre optics as is common practice with other multi-analyte sensors.18 6.Acknowledgement We gratefully acknowledge support of this research by Johnson and Johnson Medical Ltd. 7. References 1 O. S. Wolfbeis, in Fiber Optic Chemical Sensors, ed. O. S. Wolfbeis, CRC Press, Boca Raton, FL, 1991, ch 1. 2 W. R. Seitz, CRC Crit. Rev. Anal. Chem., 1988, 19, 135. 3 A. E. Straub and W. R. Seitz, Anal. Chem., 1993, 65, 1491. 4 A. Mills, O.Chang and N. McMurray, Anal. Chem., 1992, 64, 1383. 5 A. Mills and Q. Chang, Anal. Chim. Acta, 1994, 285, 113. 6 B. H. Weigl and O. S. Wolfbeis, Anal. Chim. Acta, 1995, 302, 249. 7 P. Blume, in Enzymology in the Practise of Laboratory Medicine, ed. P. Blume and E. F. Freier, Academic Press, New York, 1974, pp. 246–249. 8 L. Bowie, F. Esters, J. Bolin and N. Gochman, Clin. Chem., 1976, 22, 449. 9 T. D. O’Leary, J. L. Badenoch and R. Bais, Ann. Clin. Biochem., 1983, 20, 153. 10 M. Tajima, H. Inoue and M. Hida, Dyes Pigm., 1987, 8, 119. 11 D. G. Hafeman, K. L. Crawford and L. J. J. Bousse, Phys. Chem., 1993, 97, 3058. 12 Sidgwick’s Organic Chemistry of Nitrogen, eds. I. T. Millar and H. D. Springall, Clarendon Press, Oxford, 3rd edn., 1966, p. 117. 13 I. M. Kolthof, Acid–Base Indicators, Macmillan, New York, 1937. 14 A. Mills and L. Monaf, Analyst, 1996, 121, 535. 15 F. J. Green, in The Sigma–Aldrich Handbook of Stains, Dyes and Indicators, Aldrich Chemicals, Milwaukee, WI, 1990. 16 L. Meites, Handbook of Analytical Chemistry, McGraw-Hill, New York, 1963. 17 A. Mills, A. Lepre and L. Wild, Sens. Actuators B, 1997, 38/39, 419. 18 T. A. Dickinson, J. White, J. S. Kauer and D. R. Walt, Nature (London), 1996, 382, 697. Paper 9/00531E Fig. 5 Effect of dye concentration on experimentally determined absorbance versus temperature plots. Dye concentrations: 1.05 (5), 2.10 (2), 3.15 (½) and 4.20 (8) mmol dm23. The lines through the experimentally determined data points represent the least-squares line of best fit calculated using eqn. (13) and taking a = 6.65 3 103. The optimised b and g values obtained were 6.86 3 1029, 4.02 (5), 5.62 3 1029, 1.94 (2), 5.54 3 1029, 1.48 (½) and 4.29 31029, 1.08 (8) The inset shows a plot of 1/g versus dye concentration using optimised g values obtained from plots in the main diagram. Analyst, 1999, 124, 685–689 689

ISSN:0003-2654    

DOI:10.1039/a900531e

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

The performance of oxygen sensing films with ruthenium-adsorbed fumed silica dispersed in silicone rubber Chun-Man Chan, Mee-Yee Chan, Minquan Zhang, Waihung Lo and Kwok-Yin Wong* Department of Applied Biology and Chemical Technology, The Hong Kong Polytechnic University, Hunghom, Kowloon, Hong Kong. E-mail: bckywong@polyu.edu.hk Received 12th January 1999, Accepted 8th March 1999 Oxygen sensing films were prepared by adsorbing the tris(4,7-diphenyl-1,10-phenanthroline)ruthenium dye onto high surface area hydrophilic fumed silica and dispersing the ruthenium-loaded silica in a silicone rubber support.These sensing films possess desirable properties including higher luminescence intensity, larger response towards oxygen and more linear Stern–Volmer calibration curves than films prepared by simple mixing of the ruthenium dye with silicone rubber. However, the response time of these silica-containing films in the recovery cycle is ca. 30–50% longer than the films without silica.Quantitative determination of oxygen concentration is important in chemical analysis related to environmental, clinical and industrial applications.1,2 In recent years, much effort has been devoted to the development of optical oxygen sensors based on luminescence quenching of transition metal dyes because of their high sensitivity and desirable optical properties. 3 Among the different metal complexes being investigated, ruthenium tris(4,7-diphenyl-1,10-phenanthroline), [Ru(Ph2- phen)3]2 4–7 and platinum/palladium porphyrins8–10 are the two classes of luminescent metal dyes that have received most attention.In the fabrication of an oxygen sensing film, the luminescent dye has to be immobilized in a polymer support which provides the mechanical strength and protects the dye from potential interferents in the measuring environment. Although the hydrophobic metalloporphyrins encounter relatively little ‘solubility’ problems in most polymer matrices, the ionic ruthenium dyes are immiscible with hydrophobic polymers such as silicone rubber.Uneven distribution of the ruthenium dye within the polymer matrix would result in the formation of aggregated ion pairs,7 leading to a large portion of the dye molecules not quenched by oxygen. Previous attempts to solve this ‘solubility’ problem include entrapping an aqueous emulsion of the ruthenium complex in silicone rubber,11 adding polar co-polymer cross-linkers to the nonpolar polymer12 and employing a ruthenium dye with a surfactant anion such as dodecyl sulfate.7 During our investigations on optical oxygen sensors, we found that a convenient way to prepare oxygen sensing films is to adsorb the ionic ruthenium dye onto high surface area fumed silica followed by dispersing the silica particles in silicone rubber.This technique has been used in the fabrication of sensing films in a prototype oxygen sensor for monitoring the toxicity effect of heavy metals on activated sludge microorganisms.13 A comparison of the performance of oxygen sensing films with and without fumed silica is reported in this article.Experimental Materials [Ru(Ph2phen)3](ClO4)2 was synthesized and purified according to a published method.14 The RTV silicone rubber purchased from Nice Top Quality Ltd. (West Palm Beach, FL, USA) is a one-part clear polymer containing no silica filler. The fumed silica employed in this study were Cab-O-Sil fumed silica LM- 130, HS-5 and EH-5 obtained from Cabot Corporation (Tuscola, IL, USA).The surface of these fumed silica contains hydroxyl groups (about 3.5–4.5 hydroxyl groups per square nanometers of silica surface15) and is therefore hydrophilic. These fumed silica particles are three-dimensional branched chain aggregates with a length of approximately 0.2 to 0.3 microns. The difference in the different grades of fumed silica used in this study lies in their surface areas which are 130, 325 and 380 m2 g21 for LM-130, HS-5 and EH-5 fumed silica, respectively.15 Other chemicals used in this study were analytical-grade reagents and were used without further purification. Oxygen and nitrogen gases (99.9%) were purchased from Hong Kong Oxygen Co.(Hong Kong). Preparation of the oxygen sensing films A 0.18 mM stock solution of [Ru(Ph2phen)3](ClO4)2 in ethanol was prepared. An aliquot of the stock solution (0.1–1 ml depending on the required loading of ruthenium on silica) was transferred into a test tube containing 0.05 g fumed silica.The mixture was allowed to stand at room temperature for 24 h, after which most of the ruthenium complex had adsorbed onto the silica and the ethanol solution became colorless or very pale in color. The silica was then separated from the liquid by centrifuge and dried under vacuum overnight. The amount of ruthenium adsorbed on the silica surface was estimated from the difference in ruthenium content of the ethanol solutions before and after treatment with silica by measuring its absorbance at 467 nm.The ruthenium-loaded silica was then mixed thoroughly with 0.5 g of uncurred silicone rubber and 3 ml toluene in an ultrasonic bath. A 0.3 ml portion of the slurry mixture was transferred to a glass slide of 11 mm diameter. For comparison purpose, films containing ruthenium complex but no silica were prepared by mixing appropriate aliquot of the ruthenium complex solutions in tetrahydrofuran with silicone rubber and toluene under sonication and transferring the mixture to the glass slide.The films were left undisturbed for 24 h to allow complete curing. The films prepared in this way have a thickness of about 0.2 mm. Variation of film thickness can be achieved by pipetting different amount of the slurry mixture to the glass slide. The concentration of ruthenium complex in the film was calculated from the amount of ruthenium complex Analyst, 1999, 124, 691–694 691added and the volume of the film which was determined from the density of the silicone rubber as previously described.16 Instrumentation Luminescence intensity measurements were performed using a Perkin-Elmer (Norwalk, CT, USA) LS50B Luminescence Spectrometer with a 20 kW xenon discharge lamp as light source.The excitation and emission wavelengths used in this work were 467 nm and 598 nm respectively. Two gas flowmeters calibrated individually by a volumetric method were used to monitor the flow rate of oxygen and nitrogen.A flow cell setup similar to that in our previous report16 was employed in this study. All measurements were conducted at room temperature (25 ± 2 °C) and atmospheric pressure. Results and discussion Cab-O-Sil LM-130, HS-5 and EH-5 are high surface area hydrophilic fumed silica with acidic hydroxyl groups on the silica surface.15 Based on the difference in ruthenium content of the ethanol solutions before and after treatment with silica, it was estimated that the maximum loading of [Ru(Ph2phen)3]2+ on these silica surfaces is approximately 3.0 and 3.5 mmol ruthenium complex per gram of silica for LM-130 and HS- 5/EH-5 respectively.It appears that the maximum loading does not depend much on the surface area of the fumed silica used in this study. If one takes the radius of [Ru(Ph2phen)3]2+ as 15 Å,17 the amount of adsorbed ruthenium complex only represents a 4–10% coverage of the surface area of the silica.However, it should be noted that Cab-O-Sil fumed silica are silica aggregates15 and its clustered nature may make it difficult for large molecules such as [Ru(Ph2phen)3]2+ to gain access to all surfaces which can be reached by the gas molecules in the surface area measuring procedure. Moreover, the cationic ruthenium complex may interact more favorably with the hydrophilic hydroxyl groups on the silica surface. Hence it is not unreasonable that the maximum loading does not represent a complete coverage of the silica surface.The emission spectra of [Ru(Ph2phen)3](ClO4)2 in silicone rubber with and without fumed silica are shown in Fig. 1. The emission wavelength of the ruthenium complex remains essentially unchanged in the presence of fumed silica. The optimum concentration of the ruthenium dye (which corresponds to a particular loading of ruthenium on silica) in the sensing film was determined from the luminescence intensity under 100% nitrogen atmosphere (I0) with different concentrations of the immobilized dye.The plots of luminescence intensity versus concentration of [Ru(Ph2phen)3](ClO4)2 in different polymer supports are shown in Fig. 2. The plots show an initial upward curve indicating an increase in luminescence intensity with the ruthenium concentration followed by a downward curve as the dye concentration further increases. The decrease in luminescence intensity at high ruthenium concentration may be attributed to self-quenching of the complexes on the silica surface.It was also noticed that the quenching ratio I0/I100 (I100 corresponds to the intensity under 100% oxygen) does not change much with the dye concentration. The optimum concentration of the ruthenium dye with its quenching ratio in different polymer supports are summarized in Table 1. These films with optimum ruthenium concentration were used in subsequent studies to evaluate the performance of the sensor.Fig. 2 indicates that the luminescence intensities of films with ruthenium-loaded fumed silica are higher than those films without silica. Immobilization of the luminescent ruthenium complex into a rigid medium such as fumed silica is known to lower the rate of non-radiative relaxation and will result in higher luminescence intensity.17 The higher luminescence intensity may also be a result of the more even distribution of the ruthenium complex in the polymer matrix which minimizes the formation of aggregates and hence decreases the self quenching effect. The Stern–Volmer plots of [Ru(Ph2phen)3]2+ in different supports are shown in Fig. 3. Similar to other luminophores in heterogeneous systems,8,12 the quenching curves are not linear. However, in silica-containing polymer systems, the Stern– Volmer plots are obviously more linear and the quenching ratio is higher. Good quenching occurs when the complexes are Fig. 1 Emission spectra of [Ru(Ph2phen)3](ClO4)2 in different supports. (1) Silicone rubber, (2) silica(HS-5)–silicone, (3) silica(LM-130)–silicone and (4) silica(EH-5)–silicone. Fig. 2 Plots of luminescence intensity (I0) versus concentration of [Ru(Ph2phen)3](ClO4)2 in (3) silica(LM-130)–silicone and (o) silicone rubber support. Table 1 Optimum concentration of [Ru(Ph2phen)3](ClO4)2 in different supports and its quenching ratio (I0/I100) Support I0/I100 Optimum Ru concentration/mM Silicone rubber 6.31 ± 0.6 0.20 Silicone rubber with silica (LM-130) 13.7 ± 1.0 0.22 Silicone rubber with silica (HS-5) 14.6 ± 1.2 0.25 Silicone rubber with silica (EH-5) 14.4 ± 1.1 0.27 692 Analyst, 1999, 124, 691–694easily accessible to oxygen.The silica causes a more even distribution of the complexes in the polymer (an improvement in homogeneity) and the amount of unquenchable complexes is greatly reduced. On the other hand, the presence of silica particles may favor the adsorption of oxygen molecules which would enhance the effective collision between the ruthenium dye and the oxygen quencher.Therefore, higher sensitivity and more linear plots are obtained. Apart from sensitivity, fast response is another important requirement for oxygen sensors. The effect of fumed silica on the response time of the sensor was also investigated in this study. The response times of the sensor in different polymer supports were measured by exposing the film to pure oxygen and pure nitrogen environment alternatively. A typical response curve of [Ru(Ph2phen)3]2+ in silica-containing silicone support is shown in Fig. 4. The response of the complex in these supports is completely reversible and highly reproducible. The quenching time (N2 to O2 cycle) and recovery time ( O2 to N2 cycle) taken as 95% of the full signal are given in Table 2. Our results show that while there is not much change in the quenching time, the recovery time is significantly lengthened in the presence of silica.When the amount of incorporated silica is decreased by 40%, a shortening of the recovery time from 64 s to 43 s was observed for the LM-130 films. Small molecules such as oxygen are well known to adsorb strongly on silica surfaces and the longer recovery time can be attributed to slow desorption of oxygen from the silica surface in the support. As expected, the response time of the film depends on its thickness and the response time for films of different thicknesses are shown in Fig. 5. [Ru(Ph2phen)3](ClO4)2 is known to be a relatively stable dye towards photochemical degradation.4 Under our experimental conditions with a 20 kW xenon lamp as light source and the excitation wavelength and bandwidth set at 467 nm and 15 nm respectively, no photo bleaching of the ruthenium dye was observed under several hours of illumination. A small decrease in luminescence intensity (less than 5%) could be observed, however, if the film were exposed under extensive illiumination for 2–3 d.This indicates that the photostability of the ruthenium dye is superior over the platinum porphyrins which show a photo bleaching effect under several hours of illumination.8 Conclusions It is well known that silicone rubber is a suitable matrix for fabrication of oxygen sensors due to its high permeability toward oxygen and chemical inertness. However, the solubility of an ionic dye such as [Ru(Ph2phen)3](ClO4)2 in silicone rubber is poor which is detrimental to the performance of the sensor.We have presented here a convenient method to incorporate ruthenium dye in silicone rubber by dispersing dyecontaining fumed silica particles in the polymer matrix. The luminescence intensity and quenching ratio of the new sensing films are much improved. However, these improvements are obtained with a sacrifice of the response time in the recovery cycle. Hence these silica-containing sensing films are best utilized in the continuous monitoring of oxygen concentration such as the prototype oxygen sensor we have constructed13 in which no switching between high and low oxygen environment is required.Fig. 3 Stern–Volmer plots of [Ru(Ph2phen)3](ClO4)2 in different supports. (-) silicone rubber, (5) silica(LM-130)–silicone, (×) silica(EH- 5)–silicone and (+) silica(HS-5)–silicone. Fig. 4 Response curve of [Ru(Ph2phen)3](ClO4)2 in LM-130 silica– silicone support when subjected to step changes in oxygen concentration. Table 2 Response time for [Ru(Ph2phen)3](ClO4)2 in different supports Support Quenching time/s Recovery time/s Silicone rubber 25 ± 6 48 ± 6 Silicone rubber with silica (LM-130) 27 ± 7 64 ± 8 Silicone rubber with silica (HS-5) 30 ± 5 74 ± 10 Silicone rubber with silica (EH-5) 24 ± 4 72 ± 8 Fig. 5 Effect of film thickness on the response time of [Ru(Ph2phen)3- (ClO4)2] in silicone rubber: (-) N2 to O2 transition, (5) O2 to N2 transition; and in silica(LM-130)–silicone rubber: (+) N2 to O2 transition, (×) O2 to N2 transition.Analyst, 1999, 124, 691–694 693Acknowledgement We acknowledge the financial support from The Hong Kong Polytechnic University, the Research Grants Council (grant no. HKP 124/93M) and the Croucher Foundation. References 1 M. R. Surgi, in Applied Biosensors, ed. D. L. Wise, Butterworths, Boston, 1989, ch. 9. 2 A. Gottlieb, S. Divers and H. K. Hui, in Biosensors with Fiberoptics, ed. D. L. Wise and L. B. Wingard, Jr., Humana Press, New Jersey, 1991, p.325. 3 J. N. Demas and B. A. DeGraff, Anal. Chem., 1991, 63, 829A. 4 J. R. Bacon and J. N. Demas, Anal. Chem.,1987, 59, 2780. 5 X. M. Li, F. C. Ruan and K. Y. Wong, Analyst, 1993, 118, 289. 6 P. Hartmann and M. J. P. Leiner, Anal. Chem., 1995, 67, 88. 7 I. Klimant and O. S. Wolfbeis, Anal. Chem., 1995, 67, 3160. 8 W. W. S. Lee, K. Y. Wong, X. M. Li, Y. B. Leung, C. S. Chan and K. S. Chan, J. Mater. Chem., 1993, 3, 1031. 9 D. B. Papkovsky, G. V. Ponomarev, W. Trettnak and P. O’Leary, Anal. Chem., 1995, 67, 4112. 10 A. Mills and A. Lepre, Anal. Chem., 1997, 69, 4653. 11 O. S. Wolfbeis, M. J. P. Leiner and H. E. Posch, Mikrochim. Acta, 1986, 3, 359. 12 W. Xu, R. C. McDonough III, B. Langsdorf, J. N. Demas and B. A. DeGraff, Anal. Chem., 1994, 66, 4133. 13 K. Y. Wong, M. Q. Zhang, X. M. Li and W. H. Lo, Biosens. Bioelectron., 1997, 12, 125 . 14 C. T. Lin, W. Bottcher, M. Chou, C. Creutz and N. Sutin, J. Am. Chem. Soc., 1976, 98, 6536. 15 Cab-O-Sil Untreated Fumed Silica Properties and Functions, Cabot Corporation, Illinois, 1993. 16 W. W. S. Lee, K. Y. Wong and X. M. Li, Anal. Chem., 1993, 65, 255. 17 E. R. Carraway, J. N. Demas and B. A. DeGraff, Langmuir, 1991, 7, 2991. Paper 9/00367C 694 Analyst, 1999, 124, 691–694

ISSN:0003-2654    

DOI:10.1039/a900367c

出版商:RSC    

年代:1999

数据来源: RSC