High-performance liquid chromatography with packed microchips

This work is explaining the importance of generating a densely packed chromatographic bed to achieve optimum separation efficiency in miniaturized liquid chromatography. The reduction of peak dispersion and enhancement of separation efficiency is the important problem in micro- and nano-separation that is to be solved to generate highly efficient chromatographic systems that can exploit all advantages of miniaturization. Therefore, the influence of the geometrical wall effect in cylindrical nanobore columns and its influence to the obtained packing porosities were determined, by varying the column diameter from 30 µm to 250 µm and constant particle diameter (5 µm) (Chapter 2). In addition, two independent porosity data generating chromatographic approaches based on inverse size exclusion and Donnan exclusion chromatography were investigated, to prove the reliability of the porosity data (Chapter 3). Chapter 4 is explaining the optimization of the Agilent chromatographic chip system packing process by varying the packing conditions (pressure and implementation of ultrasound) and provides an experimental analysis and consistent interrelation of the packing procedure, resulting packing density (interparticle porosity), pressure drop over the packed inherent noncylindrical microchannel, and separation efficiency. The data revealed a significant improvement in separation efficiency with improved bed porosity received at different packing conditions. This was evaluated by use of isocratic measure of plate height curves. Chapter 5 and 6 investigated the enhanced separation efficiencies based on the improved bed porosities under isocratic and gradient elution conditions for small pharmaceutical molecules and more complex biological applications (BSA and Human plasma protein fraction; Cohn fraction IV-4) with the Agilent high performance liquid chromatography mass spectrometry chip system. The data showed that the separation efficiency and identification of peptides is significantly improved, even under gradient elution with continuously improved bed porosities. The improvement in peak width and resolution lead to better identifications and quantification of the more complex biological chromatographic problem. In particular, this work provides evidence for the operation of a geometrical wall effect in slurry-packed capillaries in a range of capillary diameter (dc) to particle diameter (dp) ratios of 5 < dc/dp < 50 (Chapter 2 - Packing density of slurry-packed capillaries at low aspect ratios). Packing densities are assessed by a polystyrene standard which is size-excluded from the intraparticle pore space of the packings [1, 2]. It is noted first that, as expected, the values for the intraparticle porosity (intra  0.29) remain independent of dc/dp because the available intraparticle pore space with respect to the overall particle volume for rigid particles should remain unaffected by the actual density of a packed bed. In contrast, the packed beds external porosity shows a steady decrease from inter  0.47 at dc/dp = 5 towards inter = 0.36-0.37 at dc/dp = 40-50. This systematic increase in inter (interparticle porosity) and total (total porosity) at decreasing capillary diameter (or in other words decreasing dc/dp) lends support to the operation of a geometrical wall effect which affects and even limits the achievable packing density due to the inherent oscillations of interparticle voidage in a transition region between the hard inner surface of the fused-silica capillaries and the bulk, random-close packing of particles, if the latter can be reached at all which depends on the actual dc/dp-ratio. When the volumetric contribution of this critical wall region to the overall volume of the packed bed becomes significant, the interparticle porosity is expected to increase. This interplay between a more loosely packed wall region and a more tightly packed core region forms the basis for explaining the improved performance of fused-silica capillaries packed with 5 µm-sized porous C18-silica particles in the work of Jorgenson and co-workers [3, 4] as the capillary inner diameter is decreased from 50 to 12 µm. With such a decrease in aspect ratio the core region ultimately disappears and the packing structure is dominated by the loosely packed wall region; the packing structure becomes effectively more homogeneous. This thesis also presents a very simple, fast, and reliable approach for the analysis of packing densities by using the interparticle Donnan (electrostatic) exclusion method (Chapter 3 - Determination of the interparticle void volume in packed beds via intraparticle Donnan exclusion). Therefore, interparticle void volumes (Vinter) and porosities (inter) of particle-packed beds within 75 µm i.d. fused-silica capillaries have been determined. The theory of electrostatic exclusion provides a clear, physically sharp definition of the boundary conditions between (charge-selective) mesopore space and (charge-nonselective) macropore space in porous media like packed beds. Hence, a complete intraparticle Donnan exclusion of an unretained co-ionic tracer (nitrate ions) was established and used to determine the porosity of the investigated packed beds with different pore and particle sizes, and with different surface modifications (bare silica, reversed-phase, and strong cation-exchange materials) or surface charge densities of the particles, in dependence of the mobile phase ionic strength (Tris-HCl buffer). This approach allowed the investigation of donnan exclusion of the charged analyte based on the electrical double layer (EDL) overlap under the aforementioned conditions. The determined interparticle porosities agreed well with those analyzed by inverse size-exclusion chromatography (ISEC) [2]. Limitations to the use of Donnan exclusion (electrostatic exclusion) and ISEC (mechanical exclusion) arose from the same principle. Exclusion becomes noticeable in the cusp regions between particles if the particles became very small (3 µm or less) in conjunction with a low bed porosity, and when the intraparticle pore diameter is so large that complete electrostatic and size exclusion is difficult to realize. Nevertheless, this simple and fast technique provides reliable porosity data and can be used for, e.g., quality control of packed conduits, if the particle packed beds are investigated due to the applicability of this approach. Based on the findings in chapter 2 and 3, this work also provides an experimental analysis and consistent interrelation of the packing procedure, resulting packing density (interparticle porosity), pressure drop over the packed inherent noncylindrical microchannel, and separation efficiency under both isocratic and gradient elution conditions for packed beds employed in microchip-HPLC (High Performance Liquid Chromatography) (Chapter 4 Chapter 4 - Separation Efficiency of Particle-Packed HPLC-Microchips). First, a new prototype HPLC/UV-microchip design was developed based on the commercial HPLC/MS (mass spectrometry) microchips, to ensure a suitable porosity measurement of the established beds. This became necessary due to the used ISEC-method [2] which needed an UV-detection mode and was not applicable with MS. The size excluded polystyrene standards reflecting the bed porosity in a chromatographic run, were not able to be ionized with the electrospray ionization (ESI) mode used with the commercially available microchips. Therefore an almost dead volume free on-chip UV-detection was implemented to overcome this issue. Both chip types had a high, comparable ratio of the separation channel volume (packed conduit) to the chip dead volume (on chip external volume) of about 110 for the HPLC/UV chip design and 150 for the HPLC/MS chip design. This allowed investigating the two different chip systems more in detail without remarkable aberrations in external contributions (external band broadening on-chip). In particular, it is demonstrated that the separation channels of suitable microfluidic analysis systems, which often cannot tolerate the high packing pressures used in conventional column packing and the application of ultrasound, can be packed as densely as the cylindrical fused-silica capillaries commonly used in nano-HPLC (inter = 0.42, with 5 µm-size packing materials) (Chapter 4 - Separation Efficiency of Particle-Packed HPLC-Microchips). The achieved packing densities were comparable to those in nano-HPLC for capillary columns characterized by similar column-to-particle size ratios as the trapezoidal microchip separation channels when packed with the same particles. A consistent decrease in inter of the microchips with increasing separation efficiency was found. inter decreased in the following series of packing modes: 150 bar > 150 bar & ultrasound > 300 bar > 300 bar & ultrasound being accompanied by a decrease in plate height by a factor of about three (with 80/20 acetonirtile/water (v/v)) from the microchip packed at 150 bar (inter = 0.475) to the one packed at 300 bar and with ultrasound assistance (inter = 0.42) for the particles with 5 µm size. These complementary data confirm that the improvement in separation efficiency can be explained by higher packing densities achieved by the increased packing pressure and simultaneous application of ultrasound [5]. For a final comparison of the data determined with the prototype HPLC/UV chips, the HPLC/MS chips were packed in analogy to the developed packing procedure of the HPLC/UV chips (Chapter 5 - Performance of HPLC/MS microchips in isocratic and gradient elution modes). The data clearly showed the same trend, as expected. The packing porosity decreased for the MS chips from inter ≈ 0.46 (150 bar) to inter ≈ 0.41 (300 bar & ultrasound) for the 5 µm particles. For the investigated 3.5 µm particles, bed porosities of inter = 0.40 (UV chips) and inter ≈ 0.39 (MS chips) were found. The slightly different porosities determined could be explained by the difference in separation channel lengths (UV chips 73 mm and MS chips 43 mm) which generated a different pressure drop over the length during the packing procedure favoring the MS chips. The separation efficiency was investigated by using octanophenone at 50/50 acetonitrile/water (v/v) at a k’ = 28 (HPLC/MS) and k’ = 35 (HPLC/UV). The decreased retention on the HPLC/MS chips with the same stationary and mobile phases is a consequence of the elevated temperature inside the MS-chip-cube compartment (313 ± 1 K) [6], while the separations with the prototype HPLC/UV chips were carried out at 298 ± 1 K. Both ultrasound and high pressure are crucial during packing to obtain improved separation efficiencies, where the minimum of the plate height curves (Hmin) is shifted to lower plate heights and higher velocities. For chips packed with 3.5 µm particles at 300 bar and ultrasonication a good chromatographic performance was obtained as reflected by the reduced minimum plate heights of hmin = 2.5 (HPLC/UV) and hmin = 2.1 (HPLC/MS). Even more important, the slope of the plate height curves at higher velocities is smaller for the chips packed under optimal conditions, enabling shorter analysis times without significant reduction of resolution. Surprisingly, the separation efficiency for the HPLC/MS chips at low k’-values was significantly reduced compared to the HPLC/UV chips. Both compartments showed a k’ dependent separation efficiency, but the HPLC/MS chips revealed a much stronger one at low k’ than the HPLC/UV chips. Assuming that the injection process and the separation channel to on chip dead volumes were comparable, the diminished efficiency of the HPLC/MS is explained by extra-column band broadening that occurs after the chromatographic separation. While external contributions to peak dispersion have a constant value, their relative contribution to peak width became less pronounced at high k’-values where the peak is more dispersed due to its longer residence time inside the separation conduit [6]. The trend of increased separation efficiencies with the packing quality is also clearly confirmed under gradient elution conditions. For the investigated pharmaceutical test mixtures the resolution and peak width is remarkably improved with decreased bed porosities. Even for steep gradients, the peak width is decreased by ~15% and the resolution is increased by ~20% after changing the packing conditions from 150 bar without ultrasound to 300 bar and ultrasound for the 5 µm particles. Reduction of particle size from 5 to 3.5 µm decreased the peak width further by ~15% and increased the resolution by ~20%. These findings correspond to the results from the isocratic elution experiments and demonstrate that the analytical performance of the HPLC/MS chip is critically affected by the quality of the chromatographic separation even under gradient elution conditions. Last but not least, the separation efficiency with a low complexity (BSA tryptic digest) and a highly complex (Cohn4-IV fraction digest) biological samples were investigated (Chapter 6 - Improved particle-packed HPLC/MS microchips for proteomic analysis). The improved packing quality led to four trends important for the field application. First, the reproducibility of independently accomplished separation improved with the packing quality, which is favorable for the longtime investigations and process control. Secondly, it was possible to reduce the lower limit of quantification (LOQ) for the BSA digest from 10 fmol for the 5 µm particles packed with 150 bar to 1 fmol for the packings packed at 300 bar and with ultrasound (3.5 µm). The impressive improvement in peak width and resolution led to a significantly reduced co-elution of peaks and therefore to a better quantification due to the detection of discrete (independent) peaks. Thirdly, because the higher chromatographic resolution yields a larger number of discrete peaks it was also possible to identify a higher number of peptides with the complex biological probe (Cohn4-IV fraction digest). However, the optimized packing conditions for the 3.5 µm particles enabled an average number of 175 identified peptides from the tryptic digest, an improvement by 22% and 39% compared with the chips packed with 5 µm and 150 bar and those with 5 µm, 300 bar and ultrasound, respectively. Fourth, the peak capacities improve significantly with the optimized packing conditions, similar to the sequence coverage and peptide identification data. With respect to the 5 µm particles packed at 150 bar without ultrasound it was found a relative improvement in peak capacity of about 37-41% and 76-94% for the different BSA concentrations and nearly 58% and 89% for the Cohn fraction, when the 5 and 3.5 µm particles are packed at 300 bar with ultrasound, respectively. To conclude, this work demonstrates the fundamental importance to generate a densely packed chromatographic bed to achieve an excellent, reliable and reproducible chromatographic performance. A lower porosity (denser bed) leads to significantly improved separation efficiency especially in inherently noncylindrical conduits. Basically, under gradient elution conditions, widely used in the pharmaceutical industry, the improved peak width and resolution lead to better identifications, peak capacities and quantification mainly for complex matrices. However, it is possible to generate packing densities comparable to widely used cylindrical nanobore columns, the only restriction is made by the stability of the microfabricated conduit, whether it can withstand high packing pressures. The application of ultrasound particularly for low aspect ratio conduits is crucial for the generation of a densely packed separation channel. Taking this as a basis, the effect of miniaturization and integration of chromatography onto single devices (micro total analyzing systems; µTAS) and all its positive contributions due to reduction of band broadening and traveled distances for the analytes, is more pronounced and promises the best separation on nanoscale chromatography platforms. References [1] Ehlert, S.; Rösler, T.; Tallarek U. J. Sep. Sci. 2008, 31, 1719-1728. [2] Halász, I.; Martin, K. Angew Chem. Int. Ed. 1978, 17, 901-908. [3] Kennedy, R. T.; Jorgenson, J. W. Anal. Chem. 1989, 61, 1128-1135. [4] Hsieh, S.; Jorgenson, J.W. Anal. Chem. 1996, 68, 1212-1217. [5] Ehlert, S.; Kraiczek, K.; Mora, J.-A.; Dittmann, M.; Rozing, G. P.; Tallarek, U. Anal. Chem. 2008, 80, 5945-5950. [6] Guiochon, G. J. Chromatogr. A 2006, 1126, 6-49.