A Fully Integrated Monolithic Microchip Electrospray Device for Mass

Anal. Chem. 2000, 72, 4058-4063

A Fully Integrated Monolithic Microchip Electrospray Device for Mass Spectrometry Gary A. Schultz,* Thomas N. Corso, Simon J. Prosser, and Sheng Zhang

Advanced BioAnalytical Services, Inc., 30 Brown Road, Ithaca, New York 14850

A novel microfabricated nozzle has been developed for the electrospray of liquids from microfluidic devices for analysis by mass spectrometry. The electrospray device was fabricated from a monolithic silicon substrate using deep reactive ion etching and other standard semiconductor techniques to etch nozzles from the planar surface of a silicon wafer. A channel extends through the wafer from the tip of the nozzle to a reservoir etched into the opposite planar surface of the wafer. Nozzle diameters as small as 15 µm have been fabricated using this method. The microfabricated electrospray device provides a reproducible, controllable, and robust means of producing nanoelectrospray of a liquid sample. The electrospray device was interfaced to an atmospheric pressure ionization timeof-flight mass spectrometer using continuous infusion of test compounds at low nanoliter-per-minute flow rates. Nozzle-to-nozzle signal intensity reproducibility using 10 nozzles was demonstrated to be 12% with single-nozzle signal stability routinely less than 4% relative standard deviation (RSD). Solvent compositions have been electrosprayed ranging from 100% organic to 100% aqueous. The signal-to-noise ratio from the infusion of a 10 nM cytochrome c solution in 100% water at 100 nL/min was 450:1. Microchip electrospray nozzles were compared with pulled capillaries for overall sensitivity and signal stability for small and large molecules. The microchip electrospray nozzles showed a 1.5-3-times increase in sensitivity compared with that from a pulled capillary, and signal stability with the microchip was 2-4% RSD compared with 4-10% with a pulled capillary. Electrospray device lifetimes achieved thus far have exceeded 8 h of continuous operation and should be sufficient for typical microfluidic applications. The total volume of the electrospray device is less than 25 pL, making it suitable for combination with microfluidic separation devices. The potential benefits of microfluidic devices for chemical analysis are increased sample throughput, better reproducibility, higher sensitivity, and significantly lower cost per analysis over conventional, higher flow rate analyses. Microchip separation devices are being developed for rapid analysis of large numbers of samples. Examples of microchip separation devices include * Correspondence: (phone) 607-257-0183; (fax) 607-257-0359; (e-mail) [email protected].

4058 Analytical Chemistry, Vol. 72, No. 17, September 1, 2000

those for capillary electrophoresis1 (CE), capillary electrochromatography2 (CEC), and high-performance liquid chromatography (HPLC).3,4 Coupling microfluidic devices with electrospray mass spectrometry is important to fully achieve sensitive and selective detection of analytes for qualitative and quantitative analysis with these devices. Electrospray provides for the atmospheric pressure ionization of a liquid sample by creating highly charged droplets at atmospheric pressure that, under evaporation, create gas-phase ions representative of the species contained in the solution. An ion-sampling orifice of a mass spectrometer is used to sample these ions for mass analysis. Several mathematical models have been generated to explain the principals governing electrospray.5,6,7,8,9 The three main parameters that determine if an electrospray will be formed for a fluid from a capillary tip at a given applied fluid voltage are the fluid surface tension, the capillary tip diameter, and the distance of the capillary tip from a counter electrode. According to these models and in practice, using smaller capillary tips requires lower fluid voltages for the formation of a stable electrospray. Also, decreasing the capillary tip diameter allows for stable electrospray at lower flow rates.10,11 It would be desirable to define an electrospray device that could form a stable electrospray of all fluids commonly used in CE, CEC, and HPLC. The surface tension of solvents commonly used as the mobile phase for these separations range from 100% aqueous (γ ) 0.073 N/m) to 100% methanol (γ ) 0.023 N/m). As the surface tension of the electrospray fluid increases, a higher onset voltage is required to initiate an electrospray for a fixed capillary diameter.12 As an example, a capillary with a tip diameter of 14 µm is required to electrospray 100% aqueous solutions with an onset voltage of 1000 V. The work of Wilm et al. first demonstrated (1) Harrison, D. J.; Fluri, K.; Seiler, K.; Fan, Z.; Effenhauser, C. S.; Manz, A. Science (Washington, D.C.) 1993, 261, 859-897. (2) Jacobson, S. C.; Hergenro ¨der, R.; Koutny, L. B.; Ramsey, J. M. Anal. Chem. 1994, 66, 1114-1118. (3) Jacobson, S. C.; Hergenro ¨der, R.; Koutny, L. B.; Ramsey, J. M. Anal. Chem. 1994, 66, 2369-2373. (4) He, B.; Tait, N.; Regnier, F. Anal. Chem. 1998, 70, 3790-3797. (5) Dole, M.; Mack, L. L.; Hines, R. L.; Mobley, R. C.; Ferguson, L. D.; Alice, M. B. J. Chem Phys. 1968, 49, 2240-2249. (6) Iribarne, J. V.; Thomson, B. A. J. Chem. Phys. 1976, 64, 2287-2294. (7) Thomson, B. A.; Iribarne, J. V. J. Chem. Phys. 1979, 71, 4451-4463. (8) Yamashita, M.; Fenn, J. J. Phys. Chem. 1984, 88: 4451-4459. (9) Smith, D. P. H. IEEE Trans. 1nd. Appl. 1986, IA-22, 527. (10) Mann, M.; Wilm, M. S. Int. J. Mass Spectrom. Ion Processes 1994, 136, 167180. (11) Mann, M.; Wilm, M. S. Anal. Chem. 1996, 68, 1-8. (12) Kebarle, P.; Ho, Y. Electrospray Ionization Mass Spectrometry; Cole, R. B., Ed.; ISBN 0-471-145, John Wiley & Sons: New York, 1997; Chapter 1. 10.1021/ac000325y CCC: $19.00

© 2000 American Chemical Society Published on Web 08/08/2000

nanoelectrospray from a fused-silica capillary pulled to an outer diameter of 5 µm at a flow rate of 25 nL/min.10 Attempts have been made to manufacture an electrospray device for microchip separations.13,14 Xue et al. described an electrospray from the edge of a planar glass microchip by applying 4 200 V to the fluid exiting the separation channel on the edge of the microchip at a distance of 3-8 mm from the ion-sampling orifice of an API mass spectrometer. Ramsey et al.14 also described an electrospray from the planar edge of a microchip separation device coupled with MS. This work demonstrated electrospray at 90 nL/min from the edge of a planar glass microchip using electroosmotic flow. Approximately 12 nL of the sample fluid collects at the edge of the microchip before the formation of a Taylor cone and electrospray from the edge of the microchip. A channel volume of 19.8 nL makes electrospray directly from the edge of a microchip impractical since this system has a dead volume approaching 60% of the channel (column) volume. Furthermore, because these devices provided a flat surface for the formation of the electrospray, these devices required an impractically high voltage to overcome the fluid surface tension to initiate an electrospray. Techniques designed to limit the spreading of the fluid on the edge of a microchip, such as incorporation of a hydrophobic coating or an on-chip nebulizer, met with limited success.13,15 More recently, fused-silica or glass capillaries have been glued or fitted into the end of the microchip channel to provide a surface from which an electrospray can be more easily formed.16,17,18 The extra-column volume associated with coupling a relatively long fused-silica capillary to a microfluidic device and the associated band-broadening due to the dead volume at the chip-to-capillary interface places some limitations on this approach. These approaches would not be practical for high-volume production of electrospray devices to meet the needs of this emerging field. Lee and co-workers19 described a process to fabricate a parylene nozzle on the edge of a silicon microchip. In this design, a nozzle was formed to provide a surface from which an electrospray can more easily be generated. This device required a relatively high voltage to form an electrospray, with the nozzle positioned in very close proximity to the mass spectrometer ionsampling orifice. This device also has a relatively high device volume (with nozzle lengths in the millimeters range), thus limiting its usefulness with coupling to microchip separation devices. To generate a useful electrospray from a microchip, a high aspect ratio nozzle structure of small dimensions is required. Deep reactive ion etching (DRIE) technologies allow high aspect ratio structures to be fabricated and are widely available for the etching (13) Xue, Q.; Foret, F.; Dunayevskiy, Y. M.; Zavracky, P. M.; McGruer, N. E.; Karger, B. L. Anal. Chem. 1997, 69, 426-430. (14) Ramsey, R. S.; Ramsey, J. M. Anal. Chem. 1997, 69, 1174-1178. (15) Zhang, B.; Liu, H.; Karger, B. L.; Foret, F. Anal. Chem. 1999, 71, 32583264. (16) Hsieh, F.; Baronas, E.; Muir, C.; Martin, S. A. Rapid Commun. Mass Spectrom. 1999, 13, 67-72. (17) Li, J.; Thibault, P.; Bings, N. H.; Skinner, D. D.; Wang, C.; Colyer, C. L.; Harrison, D. J. Anal. Chem. 1999, 71, 3036-3045. (18) Li, J.; Kelly, J. F.; Chernushevich, I.; Harrison, D. J.; Thibault, P. Anal. Chem. 2000, 72, 599-609. (19) Licklider, L.; Wang, X.; Desai, A.; Tai, Y.; Lee, T. D. Anal. Chem. 2000, 72, 367-375.

of silicon.20,21 DRIE methods, originally developed for making microelectromechanical systems (MEMS), are increasingly being applied for the fabrication of microfluidic devices. DRIE is used to form high aspect ratio structures from the surface of a substrate with a high degree of anisotropy. Since DRIE methods etch features from the surface of a substrate, multiple devices can be formed simultaneously, creating low-cost, highly reproducible devices. Many approaches and substrate materials are available for microchip fabrication (e.g., silicon, glass, plastics, etc.). Silicon has many advantages as a substrate because relatively small high aspect ratio features may be formed. This makes silicon a good match for the development of a nanoelectrospray device. In addition, the processing techniques developed by the semiconductor industry are amenable to high-volume mass production. A wellknown process for DRIE of silicon is called the Bosch process.21 Currently the Bosch process provides etch rates up to 10 µm/ min, depending on the feature size. The process also provides selectivity to etching silicon versus silicon dioxide or photoresist of greater than 300:1 which is important when deep silicon structures are desired. These features of DRIE are necessary in order to form a high aspect ratio, anisotropic etch for the formation of small-diameter nozzles from the surface of a silicon substrate. Here we describe the fabrication and operation of an electrospray device formed from the planar surface of a monolithic silicon substrate for the electrospray MS analysis of liquids at low nanoliter-per-minute flow rates. The volume of this electrospray device is less than 25 pL, making it suitable for combination with microchip separation devices. The design of this electrospray device is robust, such that devices can be readily mass-produced in a cost-effective, high-yielding process. EXPERIMENTAL SECTION Microchip Fabrication. A schematic of the electrospray device fabrication steps is shown in Figure 1. The substrate consisted of a 4-in. double-side polished silicon wafer. A 2-µm layer of silicon dioxide was thermally grown on both sides of the silicon wafer to serve as an etch mask. A positive photoresist was spun onto the topside of the wafer (Figure 1A), and a 10-µm nozzle channel feature was patterned in the photoresist using UV light. The UV exposed photoresist regions were then removed by development in an alkaline base, for clearing the photoresist and exposing the silicon dioxide mask. The silicon dioxide was then selectively dry plasma etched from the region of this circular channel, exposing the silicon substrate. A highly anisotropic, deep reactive ion etch (DRIE) was then conducted to etch the silicon forming the nozzle channel in the substrate (Figure 1B). Following the silicon nozzle channel etch, all remaining photoresist on the topside of the wafer was removed using an oxygen plasma. To form the nozzle feature, a new layer of photoresist was spun onto the topside surface of the wafer (over the 10-µm nozzle channel). A pattern in the shape of a torus was aligned and patterned in the photoresist around the 10-µm hole (Figure 1C). The torus defines the nozzle outer diameter (for this experiment, 20 µm) and the outer diameter of a recessed annular region surrounding the nozzle. Development of the photoresist and subsequent (20) Laermer, F.; Schilp, A. U.S. Patent 5, 501, 893, 1996. (21) Laermer, F.; Schilp, A.; Funk, K.; Offenberg, M. IEEE Int. MEMS Conference, 1999, 17-21.

Analytical Chemistry, Vol. 72, No. 17, September 1, 2000

4059

Figure 1. Microfabrication process used to fabricate a microchip electrospray device (not to scale).

etching of the silicon dioxide and silicon formed a silicon nozzle coplanar with the surface of the substrate. Upon forming the nozzle on the topside of the wafer, a reservoir was patterned on the backside of the silicon wafer for receiving samples (Figure 1D). To form the reservoir, photoresist was spun on the backside of the silicon wafer followed by UV light patterning of a circular feature, aligned with the nozzle channel previously formed on the topside of the substrate. Following UV patterning of the reservoir feature, the exposed regions of the resist were removed in an alkaline base followed by removal of the exposed silicon dioxide via a dry plasma etch. A DRIE was then conducted to remove the exposed silicon until the nozzle channel was reached, thus creating a through wafer channel. The remaining photoresist and oxide mask was removed to complete the device (Figure 1E). Thin films may be further applied to control the surface chemistry on the device. Scanning electron micrographs (SEMs) showing a completed device are shown in Figure 2. Apparatus. A cross-sectional view of the chip handling and fluid-delivery system is shown in Figure 3. A custom-machined stainless mounting bracket was used to hold the microchip and for mounting to the translational stages. Manual stages (NewPort, Irvine, CA) were used for accurate positioning of the microchip in front of the mass spectrometer ion-sampling orifice. Fluid solutions were delivered to the electrospray chip using a 1/16-in. stainless steel tube (0.004-in. inner diameter) fitted with an O-ring on the end of the tube. A fused-silica capillary was coupled to the inlet end of the stainless steel tube, while the inlet end of the fused-capillary was placed in a sample fluid reservoir. A gas regulator was used to adjust the fluid flow to the electrospray microchip to control the head pressure of the reservoir. During operation, the fluid-delivery probe was prealigned to the ionsampling orifice. The fluid-delivery probe is retractable in the z plane in order to make contact with the microchip. A nozzle was then positioned in front of the ion-sampling orifice, and the liquid probe was engaged on the backside of the chip to the reservoir. The fluid voltage was applied through the stainless steel tube. Pressure was applied to the sample-delivery reservoir to deliver 4060 Analytical Chemistry, Vol. 72, No. 17, September 1, 2000

Figure 2. SEM images of a microfabricated silicon electrospray device. (A) shows an electrospray nozzle tilted at a 30° angle. The nozzle has dimensions of 10 µm, inner diameter, and 20 µm, outer diameter. Also shown is the recessed annular region surrounding the nozzle etched to a depth of 50 µm and diameter of 600 µm. (B) Shows the reservoir etched 100 µm into the backside of the substrate. The 10-µm channel extending from the tip of the nozzle to the reservoir is visible as a dark circle in the middle of the reservoir.

fluid to the microchip. To increment to the next nozzle, the probe was retracted and the process repeated. The nanoelectrospray started immediately upon contact of the probe with the microchip. Mass Spectrometry. A Micromass (Manchester, UK) LCT time-of-flight mass spectrometer (TOFMS) was used for the work described herein. The experiments were performed without any modification to the instrument. The nanoelectrospray microchip was positioned approximately 5 mm in front of the ion-sampling orifice of the LCT. The ion source temperature was maintained at 70 °C. The nanoelectrospray MS cone nozzle was used with 50 L/h nitrogen to provide a counter current gas flow to assist in droplet evaporation prior to the ions entering the ion-sampling orifice. Mass spectral data were acquired using one-second ion integration times. The Micromass nanospray ion source was used for both microchip and the pulled capillary experiments. For the pulled capillary experiments, commercially available 15-µm i.d. fused-silica electrospray tips (part no. FS360-75-15-N-20) were obtained from New Objective (Cambridge, MA). Sample Preparation. A 1 mg/mL standard of horse heart cytochrome c in water (Catalog No. C-7752, Sigma, St. Louis, MO) and a 0.1 mg/mL standard of a nine-peptide mixture in water (Catalog No. 148-2012, Bio-Rad, Hercules, CA) were used as stock solutions. Further dilutions are listed in the figure descriptions. All samples were filtered using 0.45-µm filters (Millipore, Bedford,

Figure 3. Schematic representation of a complete chip handling and fluid delivery system positioned in front of the ion-sampling orifice of a mass spectrometer.

MA). Solvents were from JT Baker, Phillipsburg, NJ. RESULTS AND DISCUSSION Figure 2A shows a SEM image of a single silicon nozzle at a 30° angle. The nozzle has an inner diameter of 10 µm, an outer diameter of 20 µm, and a length of 50 µm from the base of the annulus. The recessed annular region around the nozzle defines the electrospray surface (nozzle tip) with respect to the substrate surface. By etching the nozzle from the surface of the substrate in this way, the nozzle tip is coplanar with the surface. One advantage of this coplanar design is that the nozzle is protected against accidental breakage during device handling. Nozzles ranging from 100 to 15 µm in diameter have been fabricated. This report will show results obtained from a 20-µm-diameter nozzle. Use of this size nozzle for infusion experiments provides a robust device with few clogging problems while still allowing for application of fluid voltages less than 1500 V for electrospray operation at flow rates less than 100 nL min-1. However, this fabrication method allows for nozzle dimensions to be tailored for specific flow-rate requirements. Figure 2B displays an SEM image (30° angle) of a reservoir on the backside of the wafer. The small dark spot in the center of the reservoir is the 10-µm-diameter through-wafer nozzle channel. The reservoir has a diameter of 370 µm and depth of 100 µm, corresponding to a volume of approximately 10 nL. The reservoir volume may be varied dependending on the use of the device. A 10-nL sample reservoir was used for this device, to evaluate the deposition of a discrete sample volume within the reservoirs followed by sample reconstitution and ESI-MS analysis. Removal of the reservoir and etching of the 10-µm-diameter nozzle channel from the nozzle tip to the backside of the wafer would result in a 25-pL volume for the device. A low dead volume ensures almost instantaneous onset of electrospray when the flow is initiated, even at nanoliter per minute flow rates, making rapid sample throughput possible. The low nozzle dead volume is also important when

Figure 4. A digital image of a 20-µm-diameter nozzle showing an actual electrospray exiting the nozzle tip, resulting from the infusion of water at a flow rate of 100 nL/min using a syringe pump. The applied fluid voltage was 1250 V.

coupling with microchip separation devices where postcolumn peak broadening would preclude conventional nanoelectrospray by use of a fused-silica capillary. Figure 4 shows a digital image of a nozzle positioned approximately 5 mm from the ion-sampling orifice of the mass spectrometer. The image was captured through a microscope mounted above the microchip. The outer diameter of the nozzle is 20 µm, and the diameter of the annulus is 600 µm. This image shows an electrospray of water from the nozzle at a flow rate of 100 nL min-1. The lighting for the image was adjusted to enhance the visibility of the liquid jet and the plume of droplets which emanate from the nozzle tip toward the right-hand side of the image. A reflection of this spray on the highly polished silicon substrate can be seen to the left of the nozzle; there is also a reflection of the nozzle in the annulus, making the nozzle appear slightly longer. Etching a nozzle from the surface of a substrate has several advantages. The nozzle tip is flat and uniform in size and shape, Analytical Chemistry, Vol. 72, No. 17, September 1, 2000

4061

Figure 5. The TIC (m/z range 500-2000 Da) from the infusion of a 100 nM cytochrome c solution in 50% methanol/50% water with 0.1% acetic acid. (A) Shows the TIC from a microchip electrospray device with a 2.1% RSD over the 5-min acquisition. (B) Shows the TIC from a pulled capillary with a 9.8% RSD over the 5-min acquisition.

since it is part of the original, highly polished surface of the substrate. As a result of the fabrication process, individual nozzles provide very reproducible electrospray results. In addition, using this parallel etching technology allows high-density nozzle arrays to be fabricated on a single wafer, resulting in a reduced cost per nozzle. Also, fabrication costs are mostly independent of wafer diameter, so more nozzles may be formed simply by increasing the wafer size. The microchip electrospray devices have been found to be reliable and robust as long as the fabrication of the nozzles was carried out correctly. At present, the yield of correctly functioning nozzles is 80-90% using research-grade fabrication tools. Yields should approach 100 percent by moving the fabrication out of a development laboratory and to production facilities. The nozzles provide a stable electrospray for many hours and may be reused, although they are most likely to serve as one-time-use disposable devices with applications in high-throughput, fast analyses. The nozzles perform best at flow rates of 50-300 nL min-1 and fluid electrospray voltages of 800-1500 V. The back pressure generated by a nozzle operated at 100 nL min-1 was less than 2 psi and does not add significant back pressure to a chromatographic system. Sealing the fluid probe at the back of the microchip was straightforward, with the O-ring seal capable of holding pressures greater than 50 psi. Comparison of Microchip Electrospray Device to a Pulled Capillary. To evaluate the performance of the microchip device, standard solutions were analyzed using a chip and a commercially available pulled capillary nanoelectrospray device with a 15-µm i.d. tapered tip. To make the comparison as meaningful as possible, most of the instrumental parameters were kept constant, except for adjustments to the head pressure applied to the fluid reservoir for each experimental setup in order to maintain a flow rate of approximately 100 nL min-1 for all experiments. In addition, the position of the microchip and pulled capillary and the electrospray voltage were optimized for signal intensity. A 100 nM cytochrome c solution in 50% methanol/50% water with 0.1% acetic acid was infused through the microchip electrospray device with a nozzle diameter of 20 µm and through a 15µm i.d. pulled capillary. Figure 5A shows the total ion current (TIC) from the microchip electrospray device over a period of 5 min. The TIC signal stability was calculated by measuring the 4062 Analytical Chemistry, Vol. 72, No. 17, September 1, 2000

Figure 6. Mass spectrum of cytochrome c resulting from summing the first minute of data shown in Figure 5 for (A) the microchip electrospray device and for (B) the pulled capillary.

relative standard deviation of the signal intensity for the 5-min experiments. For the microchip electrospray device, the signal stability was 2.1% RSD. Figure 5B shows the TIC from the pulled capillary, which displayed a signal stability of 9.8% RSD. Note that the experiments were optimized for maximum signal intensity. In previous pulled-capillary experiments, we have observed signal stabilities as low as 4% RSD. Over a 30 min data collection period, the stability of the ion current from the microchip device was still 4.0%, demonstrating good long-term stability as well as an improved ratio of signal-to-noise. The improvement in stability of the microchip over that from a pulled capillary has been consistently observed on three different types of mass spectrometers. The mechanism behind this is not yet fully understood, and we are currently conducting experiments to investigate this phenomenon. Figure 6A shows the resulting mass spectrum for cytochrome c from the summation of the first 60 s of data acquired in Figure 5A. Figure 6B shows the resulting mass spectrum for cytochrome c, from the summation of the first 60 s of data shown in Figure 5B. Overall, the mass spectra are the same except that the signal intensity using the microchip electrospray device was 1.5 times higher than that using the pulled capillary. A mixture of nine peptides (BioRad, Hercules, CA) was used to compare the performance of the microchip electrospray device for the analysis of small-molecule mixtures to that of the pulled capillary. The mixture was diluted to a concentration of 160 pg µL-1 in water for each peptide. Figure 7A shows a mass spectrum resulting from the summation of a one-minute period acquired from the microchip nanoelectrospray device. Figure 7B shows a mass spectrum resulting from the summation of a one-minute period obtained using a pulled capillary. The mass spectral peaks are labeled with the compound names. Overall, the mass spectral content was the same, with a 1.7 times increase in sensitivity from the microchip electrospray device. Signal stability measured from the TIC for the microchip nanoelectrospray device over a 5-min acquisition was 2.1% RSD compared with a value of 7.0% RSD for the pulled capillary. Solvent Composition. Various solvent compositions were tested using the microchip, ranging from 100% organic to 100% aqueous. Water can be problematic in electrospray because of its high surface tension and low volatility; however, the microchip nozzles operated well with water when the electrospray voltage was increased approximately 150 V over that used for the 50%

Table 1. Comparison of Nozzle-to-Nozzle Electrospray Reproducibilitya nozzle 1 2 3 4 5 6 7 8 9 10 avg intranozzle %RSD avg signal intensity standard deviation % RSD Figure 7. Mass spectrum of a nine-peptide mixture (160 pg µL-1 each component) in water, resulting from summing of the first minute of data from (A) a microchip electrospray device and (B) a pulled capillary.

Figure 8. (A) Mass spectrum of 100 nM cytochrome c in water from a microchip electrospray device resulting from summing one minute of data. (B) 10 nM cytochrome c in water from a microchip electrospray device resulting from summing of one minute of data.

methanol/50% water mixture. Figure 8A shows the electrospray mass spectrum for a 100 nM cytochrome c solution in water at 100 nL min-1 with an electrospray voltage of 1250 V. The signal stability measured from the TIC was 3 times larger than that from the 50% methanol/50% water mixture. Figure 8B shows the electrospray mass spectrum for one-tenth the concentration (10 nM cytochrome c in water) with a signal stability measured from the TIC of 8.7% RSD. The different charge-state distributions for cytochrome c in differing solvent composition observed in Figures 6A and 8A are consistent with previous findings of the acid-induced unfolding process of cytochrome c,22,23 in which protonation of the N2 atom of His18 (that coordinates to the heme iron) has been shown to be a key step of the unfolding process.24,25 In addition, the presence of 50% organic solvent in our experiment may facilitate this unfolding process as well. Internozzle Reproducibility. The internozzle reproducibility is an important feature of devices designed for one-time use. Such a device must operate robustly with adequate accuracy and precision so that the signal measured over the course of an experiment (which may be on the order of tens of seconds) from (22) Konermann, L.; Douglas, D. J. Biochemistry 1997, 36, 12296-12302. (23) Konermann, L.; Douglas, D. J. Rapid Commun. Mass Spectrom. 1998, 12, 435-442. (24) Knapp, J. A.; Pace, C. N. Biochemistry 1974, 13, 1289-1294. (25) Dyson, H. J.; Beattie, J. K. J. Biol. Chem. 1982, 257, 2267-2273.

signal intensity* 6216 6498 4947 6086 6780 7033 6134 6250 5433 4852

intranozzle % RSD 2.6 3.1 8.7 6.4 3.4 3.4 5.4 2.7 4.7 8.2 4.9

6023 730 12.1

a The data were obtained by infusion of a small-molecule test solution using 10 different electrospray nozzle devices from the same wafer. The average signal stability for individual nozzles was measured over 30 s from the TIC.

any given nozzle would be the same as that obtained using another. Table 1 shows intranozzle reproducibility for infusion results of a small-molecule test solution analysis between 10 different electrospray nozzle devices from the same wafer. The fluid probe was manually moved from the back of one reservoir to that of another, while positioning a new nozzle in front of the ion-sampling orifice of the mass spectrometer. The average signal stability for individual nozzles over 30 s was measured from the TIC at 4.9% RSD while the internozzle reproducibility for the 10 nozzles resulted in a RSD of 12.1%. It is anticipated that further improvements in the electrospray device fabrication and the addition of robotic positioning of the microchip will improve upon these results. CONCLUSIONS This microchip electrospray device provides a novel, reliable, and reproducible electrospray of fluid samples for MS analysis. The signal stability and intensity is comparable with that obtained using a pulled capillary of similar dimensions. The fabrication method allows for high-density nozzle arrays of any layout to be made. Future work will demonstrate the direct transfer of samples from 96, 384, and 1536 well plates to multiple nozzle arrays for electrospray mass spectrometry analysis of the transferred solutions. The volume associated with this electrospray device is less than 25 pL, making it suitable for combination with microchip separation devices. The design of this electrospray device is also robust, such that the device can be readily mass-produced in a cost-effective, high-yielding process. ACKNOWLEDGMENT Advanced BioAnalytical Services performed the microchip fabrication in part at the Cornell Nanofabrication Facility (CNF) (a member of the National Nanofabrication Users Network), which is supported by the National Science Foundation under Grant ECS9319005, Cornell University, and industrial affiliates. We acknowledge the many helpful discussions with Gary Bordonaro, Mike Skvarla, and Phil Infante of CNF. Received for review March 17, 2000. Accepted June 29, 2000. AC000325Y Analytical Chemistry, Vol. 72, No. 17, September 1, 2000

4063