Anal. Chem. 1998, 70, 3790-3797
Fabrication of Nanocolumns for Liquid Chromatography Bing He,† Niall Tait,‡ and Fred Regnier*,§
Department of Chemistry, Purdue University, Lafayette, Indiana 47907, Alberta Microelectronics Corporation, Edmonton, Canada T6G 2T9, and PerSeptive Biosystems, Framingham, Massachusetts 01701
This paper shows that in situ micromachining can be used to simultaneously position and define (i) support particles, (ii) convective transport channels, (iii) an inlet distribution network of channels, and (iv) outlet channels in multiple chromatography columns on a single quartz wafer to the level of a few tenths of a micrometer. Stationary phases were bonded to 5 × 5 × 10 µm collocated monolith support structures separated by rectangular channels 1.5 µm wide and 10 µm deep with a low degree of deviation of channel width between the top and bottom of channels. High aspect ratio microfabrication can only be achieved with deep reactive ion etching. The volume of a 150 µm × 4.5 cm column was 18 nL. Column efficiency was evaluated in the capillary electrochromatography (CEC) mode using rhodamine 123 and a hydrocarbon stationary phase. Plate heights in these columns were typically 0.6 µm in the nonretained and 1.3 µm in the retained modes of operation. Columns were designed to have identical mobile-phase velocity in all channels in an effort to minimize outgassing during operation. When the total lateral cross-sectional area of channels at all points along the separation axis is identical, linear velocity of the mobile phase in a CEC column should be the same. Columns were operated at atmospheric pressure. There is great interest today in miniaturized analytical systems for life science research, the clinical environment, drug discovery, biotechnology, quality control, and environmental monitoring. The dream is that through micro- and nanofabrication it will be possible to prefabricate either single or multiple, integrated, self-contained laboratories of a size equivalent to a computer chip. It is possible that these systems will be capable of executing analyses in which chemical reactions, separations, and various forms of detection are integrated into a single method. For this dream to become reality, it will be necessary to miniaturize all the analytical methods and devices currently used in conventional laboratories. This paper addresses the problem of miniaturizing liquid chromatography columns in such a way that (i) separations can be tightly coupled to other analytical operations in an analytical method, (ii) columns can be easily produced in large numbers for parallel processing, and (iii) the †
Purdue University. Alberta Microelectronic Corp. § PerSeptive Biosystems. ‡
3790 Analytical Chemistry, Vol. 70, No. 18, September 15, 1998
trend toward increasingly smaller channels and support structures in liquid chromatography can be accommodated. Liquid chromatography has been achieved in columns packed with particles for more than half a century. The function of these particles is to (i) support the stationary phase, (ii) provide a large interfacial boundary between the stationary and mobile phase for analyte partitioning, (iii) form a homogeneous channel network for the convective transport of mobile phase through the column, and (iv) initiate interchannel mixing to minimize peak dispersion. This is most easily achieved in porous particles. The problem with porous particles is that stagnant pools of mobile phase accumulate in the pore network and limit mass transfer.1-3 Remarkably, liquid chromatography has been scaled in column volume over 8 orders of magnitude using packed particles. At the lower end, microliter-volume packed columns can be produced with fused-silica capillary tubing and 3-5-µm particles. It is when large numbers of columns with even smaller volumes are need that the universality of the packed column approach begins to break down, particularly when the objective is to make multiple columns on a wafer. Rectangular capillaries have been micromachined in silica wafers and packed with particles,4 but fabricating frits, nonuniformity of packing at the walls and corners of the rectangular channels, and the difficulty of packing columns through the tortuous channel network on a wafer are substantial problems. An alternative that circumvents the use of particles might be to fabricate nanocolumns from membranes5 or the structurally similar continuous polymer rods.6 Unfortunately, the problems are still the same. Open tubular columns with the stationary phase supported on the tubing walls is another way to circumvent the use of particles and the accompanying packing problems.7 Although open tubular columns were first described nearly three decades ago,8 they have never become popular in liquid chromatography. This is probably because channel widths of 2 µm or less are required to deal with the limited rates of (1) Giddings, J. C. Unified Separation Science; John Wiley & Sons: New York, 1991. (2) Horvath, C.; Lin, H.-J. J. Chromatogr. 1976, 126, 401-409. (3) Horvath, C.; Lin, H.-J. J. Chromatogr. 1978, 149, 43-70. (4) Ocvirk, G.; Verpoote E.; Manz, A.; Grasserbauer, M.; Widmer, H. M. Anal. Methods Instrum. 1995, 22, 1-9. (5) Tennikova, T. B.; Svec, F. J. Chromatogr., A 1993, 646, 279-288. (6) Wang, Q. C.; Svec, F.; Frechet, J. M. J. Chromatogr., A 1994, 669, 230235. (7) Yun, H.; Markides, K. E.; Lee. M. L. J. Microcolumn Sep. 1995, 7, 153158. (8) Nota, G.; Marrino, G.; Buonocore, V.; Ballio, A. J. Chromatogr. 1970, 46, 103-106. S0003-2700(98)00028-6 CCC: $15.00
© 1998 American Chemical Society Published on Web 08/15/1998
mobile-phase mass transfer in liquid chromatography.9 Columns this small are also easily plugged, the loading capacity is extremely small, gradient elution with positive displacement pumps is difficult, and the optical path length used for detection is very short. The equivalent of open tubular columns in a microfabricated format is a semirectangular open channel that has been etched into a silicon or quartz wafer.10 Results are encouraging with columns having channels of roughly 10 µm depth and 100 µm width, but again the requisite channel width of 2 µm or less9 is not met. Micellar electrokinetic chromatography (MEKC) in microfabricated channels is still another approach.10 Very nice separation in the MEKC mode have been achieved on chips, but unfortunately, it does not allow the equivalent of gradient elution separations which are so widely used in the resolution of complex mixtures. This leads to the question of whether there are other, yet unexplored ways of producing liquid chromatography columns with channel widths of 2 µm or less that would have the millions of support surfaces found in a packed column. The fact that microlithography is routinely used to produce millions of submicrometer-size transistors in modern integrated circuits draws attention to this mode of fabrication. First, fabrication of liquid chromatography columns by microlithography would allow the creation of structures and channels in situ. This would circumvent the need for packing particles in a column. Second, objects could be positioned and arranged in columns by design. Particles and channels would be located where the designer placed them in the column design. Third, microlithography would allow simultaneous creation of all the structures and channels in a column. And finally, it should be possible to create multiple columns on a single silicon or quartz wafer. This paper addresses the basic design and fabrication issues associated with producing liquid chromatography columns in situ on silicon and quartz wafers. EXPERIMENTAL SECTION Materials. Photolithography masks, SL-4006-2C-AR3-AZ1350, were purchased from Hoya Corp. (Shelton, CT). This mask is composed of three layers: (i) fused-silica substrate SL (SiO2 70%, Na2O 8%, K2O 9%, and RO 13%; 1.5 mm), (ii) chromfilm AR3, 1050 Å, and (iii) positive photoresist AZ-1350, 4800 Å. Quartz wafers, QZ-3W40-225-UP, were also purchased from Hoya Corp. This quartz wafer, 3 in. in diameter and 400 µm thick, has an extremely low thermal expansion coefficient (
