Anal. Chem. 2005, 77, 6947-6953
Microfluidic Platform for Liquid Chromatography-Tandem Mass Spectrometry Analyses of Complex Peptide Mixtures Jun Xie,† Yunan Miao,‡ Jason Shih,† Yu-Chong Tai,† and Terry D. Lee*,‡
Department of Electrical Engineering, California Institute of Technology, Pasadena, California 91125, and Division of Immunology, Beckman Research Institute of the City of Hope, Duarte, California 91010
A microfluidic chip that integrates all the fluidic components of a gradient liquid chromatography (LC) system is described. These chips were batch-fabricated on a silicon wafer using photolithographic processes and with Parylene as the main structural material. The fabricated chip includes three electrolysis-based electrochemical pumps, one for loading the sample and the other two for delivering the solvent gradient; platinum electrodes for delivering current to the pumps and establishing the electrospray potential; a low-volume static mixer; a column packed with silica-based reversed-phase support; integrated frits for bead capture; and an electrospray nozzle. The fabricated structures were able to withstand pressures in excess of 250 psi. The device was used to perform a liquid chromatography-tandem mass spectrometry (LC-MS/MS) analysis of a mixture of peptides from the trypsin digestion of bovine serum albumen (BSA). Gradient elution through the 1.2-cm column was performed at a flow rate of 80 nL/min. Compared to the analysis of the same sample using a commercial nanoflow LC system, the chromatographic resolution was nearly as good, and the total cycle time was significantly reduced because of the minimal volume between the pumps and the column. Results demonstrate the potential of mass-produced, low-cost microfluidic systems capable of performing LC separations for proteomics applications. The development of capillary LC, driven in part by the need for high-sensitivity LC-MS/MS analyses, has progressed to the point that 75-µm-diameter columns and 200 nL/min flow rates are the norm for many applications. The 3-4 orders of magnitude reduction in the volume of such columns over the standard analytical 4.6-mm-diameter format has not been accompanied by a corresponding reduction in the scale of the HPLC pumping systems. A number of commercial solvent delivery systems are available that can create accurate solvent gradients at submicroliter-per-minute flow rates, but the physical size is much the same as systems operating at milliliter-per-minute flow rates. Interconnections, such as fittings and tubing, between different components in those systems often have a several microliter swept or * Corresponding author. Phone: 626 301-8301. E-mail:
[email protected]. † California Institute of Technology. ‡ Beckman Research Institute of the City of Hope. 10.1021/ac0510888 CCC: $30.25 Published on Web 10/07/2005
© 2005 American Chemical Society
dead volume. This can lead to long analysis times, peak broadening, and an increased number of places for leaks to develop. In addition to reducing these effects, there are a number of potential benefits to having an HPLC system that more closely matches the scale of the chromatography. These include improved coupling to other processes, greatly reduced cost of the system, and the possibility of even further reductions in the scale of the analysis. In recent years, there has been considerable research devoted to the development of microfluidic platforms capable of performing small-scale separations.1 Nearly every type of electrophoretically or electrokinetically driven separation has been demonstrated in a chip-based platform, including capillary electrophoresis (CE),2-4 isoelectric focusing (IEF),5-9 micellar electrokinetic chromatography (MEKC),10-13 and capillary electrochemical chromatography (CEC).14-16 This reflects a strong preference among researchers in the field for approaches in which processes can be controlled simply by adjusting voltage potentials at different points in the microfluidic system. Despite this considerable advantage and the (1) Lion, N.; Rohner, T. C.; Dayon, L.; Arnaud, I. L.; Damoc, E.; Youhnovski, N.; Wu, Z. Y.; Roussel, C.; Josserand, J.; Jensen, H.; Rossier, J. S.; Przybylski, M.; Girault, H. H. Electrophoresis 2003, 24, 3533-3562. (2) Jacobson, S. C.; Hergenroder, R.; Koutny, L. B.; Ramsey, J. M. Anal. Chem. 1994, 66, 1114-1118. (3) Lacher, N. A.; de Rooij, N. F.; Verpoorte, E.; Lunte, S. M. J. Chromatogr., A 2003, 1004, 225-235. (4) Liu, Y.; Foote, R. S.; Jacobson, S. C.; Ramsey, R. S.; Ramsey, J. M. Anal. Chem. 2000, 72, 4608-4613. (5) Hofmann, O.; Che, D.; Cruickshank, K. A.; Muller, U. R. Anal. Chem. 1999, 71, 678-686. (6) Tan, W.; Fan, Z. H.; Qiu, C. X.; Ricco, A. J.; Gibbons, I. Electrophoresis 2002, 23, 3638-3645. (7) Li, Y.; DeVoe, D. L.; Lee, C. S. Electrophoresis 2003, 24, 193-199. (8) Tsai, S. W.; Loughran, M.; Hiratsuka, A.; Yano, K.; Karube, I. Analyst 2003, 128, 237-244. (9) Herr, A. E.; Molho, J. I.; Drouvalakis, K. A.; Mikkelsen, J. C.; Utz, P. J.; Santiago, J. G.; Kenny, T. W. Anal. Chem. 2003, 75, 1180-1187. (10) Kutter, J. P.; Jacobson, S. D.; Ramsey, J. M. Anal. Chem. 1997, 69, 51655171. (11) Culbertson, C. T.; Jacobson, S. C.; Ramsey, J. M. Anal. Chem. 2000, 72, 5814-5819. (12) Ramsey, J. D.; Jacobson, S. C.; Culbertson, C. T.; Ramsey, J. M. Anal. Chem. 2003, 75, 3758-3764. (13) Rocklin, R. D.; Ramsey, R. S.; Ramsey, J. M. Anal. Chem. 2000, 72, 52445249. (14) Slentz, B. E.; Penner, N. A.; Regnier, F. E. J. Chromatogr., A 2002, 948, 225-233. (15) Svec, F.; Peters, E. C.; Sykora, D.; Frechet, J. M. J. Chromatogr., A 2000, 887, 3-29. (16) Throckmorton, D. J.; Shepodd, T. J.; Singh, A. K. Anal. Chem. 2002, 74, 784-789.
Analytical Chemistry, Vol. 77, No. 21, November 1, 2005 6947
progress that has been made, there are a number of drawbacks, including the need to work with very high voltages, a strong dependence of performance on the nature of the solvent system, limits to the range of sample types that can be analyzed, and difficulties with interfacing to MS. LC is still the method of choice for many applications, including many analytical methods in the field of proteomics. Some progress has been made toward performing HPLC separations in chip-based platforms.17 It is now apparent that standard beaded chromatography supports can be readily packed in channels on a chip.18,19 One company (Agilent) now markets a microfluidic chip that integrates a trapping column, separation column, and electrospray source within a single structure;20,21 however, the device was still connected to a conventional LC system to deliver the gradient. Recently, we reported on work to develop a chip-based electrochemical pumping system capable of forming solvent gradients at flow rates and pressures consistent with many microscale LC separations.22 In this article, we describe integration of that pumping system with other components to create an HPLC on a chip and show that it can be used for the LC-MS/MS analysis of complex peptide mixtures. EXPERIMENTAL SECTION Materials. The Al was obtained from Williams Advanced Materials (Brewter, NY). Al etchant and buffered HF were purchased from Transene Inc. (Danvers, MA). Acids used to etch the Pt were obtained from VWR Scientific Products (West Chester, PA). All photoresist materials (AZ1518, AZ4400, and AZ4620) were purchased from AZ Electronic Materials (Sommerville, NJ). Parylene C was purchased from Uniglobe Kisco Inc. (San Jose, CA). A-174 adhesion promoter for parylene deposition was supplied by Specialty Coating Systems (Indianapolis, IN). XeF2 was purchased from Pelchem (Pretoria, South Africa). SU-8 photoresist and its developer were supplied by Microchem (Newton, MA). Microfabrication. The chips were fabricated using our developed Parylene surface micromachining technology.22-25 In this process, alternating layers of Parylene and photoresist were deposited, with the photoresist ultimately being dissolved away, leaving only the desired Parylene structures. The process used to fabricate the chip (Figure 1) began with a 4-in. silicon wafer with 1.5 µm of thermally grown oxide on the surface. Electrolysis electrodes were deposited by e-beam evaporation of platinum/titanium (Pt/Ti) (200 nm/30 nm). This metal deposition was performed using the Integrated Micromachines Inc. (Monrovia, CA) foundry service. Heated (∼80 °C) aqua regia (HNO3/HCl ) 1:6) was used to pattern the Pt/Ti. After patterning (17) He, B.; Regnier, F. J. Pharm. Biomed. Anal. 1998, 17, 925-932. (18) Li, J.; LeRiche, T.; Tremblay, T. L.; Wang, C.; Bonneil, E.; Harrison, D. J.; Thibault, P. Mol. Cell Proteomics 2002, 1, 157-168. (19) Ekstrom, S.; Malmstrom, J.; Wallman, L.; Lofgren, M.; Nilsson, J.; Laurell, T.; Marko-Varga, G. Proteomics 2002, 2, 413-421. (20) Yin, H.; Killeen, K.; Brennen, R.; Sobek, D.; Werlich, M.; van de Goor, T. Anal. Chem. 2005, 77, 527-533. (21) Fortier, M. H.; Bonneil, E.; Goodley, P.; Thibault, P. Anal. Chem. 2005, 77, 1631-1640. (22) Xie, J.; Miao, Y.; Shih, J.; He, Q.; Liu, J.; Tai, Y. C.; Lee, T. D. Anal. Chem. 2004, 76, 3756-3763. (23) Xie, J.; Shih, J.; Lin, Q.; Yang, B.; Tai, Y. C. Lab Chip 2004, 4, 495-501. (24) Licklider, L.; Wang, X.-Q.; Desai, A.; Tai, Y.-C.; Lee, T. D. Anal. Chem. 2000, 72, 367-375. (25) Ho, C.-M.; Tai, Y.-C. Annu. Rev. Fluid Mech. 1998, 30, 579-612.
6948 Analytical Chemistry, Vol. 77, No. 21, November 1, 2005
Figure 1. Fabrication process for the LC chip. For clarity, the thicknesses of the various layers deposited on top of the Si wafer are not drawn to scale.
the electrodes, the oxide was etched using buffered HF with AZ1518 photoresist as the mask. Aqua regia is strongly acidic, and HF is an extremely toxic substance. Personnel working with both aqua regia and HF must take adequate precautions to protect against exposure to either the liquid solution or vapors. This step was necessary to create anchors for the channels and also to provide a way to make the electrospray nozzle free-standing at the end, both of which were done using XeF2 (Pelchem, Pretoria, South Africa), a gas-phase isotropic silicon etchant. After A-174 adhesion promoter (Specialty Coating Systems, Indianapolis, IN) was applied to the substrate, the first Parylene (Uniglobe Kisco Inc., San Jose, CA) layer (4.5 µm) was deposited and patterned by oxygen plasma, using AZ4400 photoresist as a mask. Then, a 20-µm AZ4620 sacrificial photoresist layer was spun onto the wafer and lithographically patterned to define the fluid channels. For the mixer, nozzle, and filter regions, a partial lithographic exposure was performed to reduce the height to 5 µm, producing a bilevel (20 µm/5 µm) photoresist sacrificial layer. Before deposition of the second Parylene layer, a short XeF2 etch was performed to roughen the silicon surface exposed during the oxide patterning step. This roughened silicon served to promote adhesion between the Parylene and the substrate, allowing our structures to sustain higher pressures. XeF2 is readily hydrolyzed to HF and must be handled using the same precautions. The second Parylene layer (4.5 µm) was then deposited and patterned by oxygen plasma using a 150-nm sputtered or thermally evaporated aluminum layer as a mask. This patterning step also defines the shape of the electrospray nozzle. Wafers then went through a 5% HF dip and oxygen plasma cleaning before a 50-µm SU-8 (Microchem, Newton, MA) layer was spin-coated on. The SU-8 layer was lithographically patterned and served to provide a flat overall chip surface to support the reservoir structure. The SU-8 also helped strengthen the channels
Figure 2. Photograph of the fabricated LC chip.
Figure 3. Diagram of the LC chip showing the placement of the solvent reservoir and cover plate on top of the main chip.
for the necessary high-pressure operation. After SU-8 developing, wafers were left inside the SU-8 developer (propylene glycol monoether acetate, PGMEA, Microchem) to dissolve the sacrificial photoresist. Finally, XeF2 was used to etch away the exposed silicon underneath the nozzle and make it free-standing. Wafers were then diced into 1 cm × 2 cm chips (Figure 2). Column Packing and Device Assembly. The column on each chip was individually packed with Polaris 3-µm C18-A silicabased support (Varian, Palo Alto, CA). To do this, the chip was mounted into a polyetherimide (Ultem) jig (not shown) that coupled a port at the front of the column to Teflon tubing via a poly(dimethylsiloxane) (PDMS) gasket. A slurry of the beaded support in 2-propanol from a pressurized (250 psi) reservoir was forced into the column through an access port between the mixer and the column until the entire length of the column was filled. The nominal 5-µm channel height at the entrance of the mixer and the end of the column, along with an array of closely spaced posts at the end of the column was sufficient to retain even 3-µm beads. The side channels between the pump chambers and the mixer were similarly packed with the same 3-µm beaded support. Once packed, the chip was mounted in a different holder that utilized a 5-mm-thick Ultem cover (Figure 3). Chambers, which matched up with the pumps on the chip, were machined into this cover piece to form reservoirs for the sample and solvent. A PDMS gasket provided the seal between the Ultem cover and the chip. Each of the solvent chambers was 10 mm2 in area and 2 mm tall, creating a reservoir 20 µL in total volume. The sample chamber was only 3.9 mm2 in area and 1.3 mm tall for a total volume of ∼5 µL. These chambers could be filled using a syringe via two access ports at either end of each chamber. For our experiments, the solvent A reservoir was filled with a 95/5/0.1 (water/methanol/ formic acid) solution, and the solvent B chamber was filled with a solution of the same components in a ratio of 40/60/0.1. The sample chamber was filled with a 1 pmol/µL solution of tryptically
Figure 4. Diagram of the chip holder assembly showing the orientation to the inlet of the Agilent MSD ion trap mass spectrometer.
digested BSA. The access ports were later sealed using an acrylic cover piece and another PDMS gasket. Electrical contact to the chip electrodes was accomplished using a printed circuit board clamped to the metal pads on the chips with a conductive elastomeric connector (Fujipoly, Carteret, NJ) between the two pieces. Mass Spectrometry. All MS analyses were performed using an Agilent 1100 series LC-MSD SL ion trap mass spectrometer (Palo Alto, CA) equipped with an Agilent orthogonal nanospray ion source operating in the positive ion mode with software version 4.2. The voltage setting was -1900 V for the MS inlet and -1400 V for the counter electrode. For the on-line capillary LC separations used for comparison, an Agilent 1100 series nanoflow LC equipped with a 75-µm-i.d. 15-cm-long column packed with 3.5µm Zorbax 300SB-C18 reversed-phase support was used. The column was coupled to a fused-silica needle with an 8-µm diameter at the tip pulled from 20 µm i.d. × 360 µm o.d. tubing (New Objective Inc., Woburn, MA). Samples (0.6 µL of a 1 pmol/µL solution) were first loaded from the autosampler onto a 300-µmi.d. 5-mm-long trap column packed with 5-µm Zorbax 300SB-C18 support at a flow rate of 15 µL/min using solvent A (0.1% formic acid in water). After 2 min, the trap column was then switched on-line, and the peptides were eluted with a 60-min gradient of 5-55% solvent B (0.1% formic acid in acetonitrile) at a flow rate of 300 nL/min. Both full range mass spectra and MS/MS spectra were collected in an automated fashion using programs built into the Agilent data system software. For chip LC analyses, the completed chip assembly was mounted on the Agilent electrospray ion source housing using a modified probe that incorporated a three-dimensional positioner. The chip electrospray needle was mounted in the same position as the standard electrospray needle, orthogonal to axis of the MS inlet (Figure 4). The MS inlet voltage was typically in the range of -1900 to -2200 V with -1400 to -1700 V on the counter electrode. Electrolysis pumps on the chip were galvanostatically controlled using output currents from custom-built voltage-tocurrent converters. Control of these current sources was achieved using LabView and a DAQPad-6020E interface board from National Instruments (Austin, TX). Separations were performed at a flow rate of ∼80 nL/min with gradients going from 0 to 100% Solvent B (40/60/0.1 water/methanol/formic acid) over 40 min. In these experiments, ∼600 nL of a 1 pmol/µL BSA digest was loaded onto the column from the 5-µL sample chamber. This was Analytical Chemistry, Vol. 77, No. 21, November 1, 2005
6949
done by directing a predetermined current to the sample pump electrode for a calibrated period of time. RESULTS AND DISCUSSION Device Fabrication and Assembly. One of the guiding principles of our work to develop microfluidic sample preparation systems for mass spectrometry is to utilize fabrication techniques that can be adapted to mass production. The Parylene surface micromachining technology allowed us to batch-fabricate (Figure 1) the microscale LC system using standard lithographic processes on a silicon wafer. Parylene exhibits high strength, chemical inertness, and optical transparency and can be deposited as a conformal film at room temperature. These properties make Parylene an ideal candidate from both application and fabrication standpoints. A sacrificial layer of photoresist between two layers of Parylene is used to define the channels within the structure. Subsequent removal of this layer creates the network of open spaces that become the fluidic network of the device. By using a bilevel sacrificial layer, channels of different heights (in this case, 20 and 5 µm) are created. The lower channel height can function as a filter or frit to confine particles to particular regions of the chip, for example, when chromatography beads are packed into an onchip column. Each fabricated chip contains two solvent pumps and a sample pump. Each of these pumps consists of a pair of electrodes that are used for electrochemical pumping. The chips also integrate a mixer, an LC column, and a free-standing electrospray needle extending 1 mm past the edge of the chip (Figure 2).24 Electrochemical Pumps. The electrolysis-based pumps have been previously described.24 These pumps are capable of producing the flow rates and pressures required for an LC separation. The pumps are controlled galvanostatically, and typical currents used during operation of the chip range from 0 to 400 µA. Ideally, 100% of the current passing between the two electrodes of an electrochemical pump is converted into a corresponding amount of gas. In practice, there are losses due to chemical reactions other than hydrolysis, recombination of the hydrogen and oxygen, and diffusion through the chip structure. Even with these losses, pumping efficiencies of up to 10% can be achieved, and the total power consumption at any given time of all three pumps together was generally around 2 mW. When the decision was made to utilize an electrochemical pump, there was a concern that dissolved gas in the solvents would result in gas bubble formation at the end of the column. We have carefully monitored the solvent stream between the end of the column and the electrospray needle and have not observed any bubble formation for pressure differentials as high as 140 psi. We attribute the lack of gas formation to two factors. First, the solvent chambers are large enough that the gas generated over the electrodes (in the rear of the chamber) does not have time to diffuse to the front of the chamber (near the outlet to the mixer and the column) within the duration of a typical separation (
