Anal. Chem. 2009, 81, 7281–7287
Microstructured Photonic Fibers as Multichannel Electrospray Emitters Shuqin Su, Graham T. T. Gibson, Samuel M. Mugo, Dale M. Marecak, and Richard D. Oleschuk* Department of Chemistry, Queen’s University, Kingston, Ontario, Canada K7L 3N6 Novel multichannel electrospray emitters are presented that use silica-based microstructured fibers (MSFs) to split the flow allowing efficient desolvation during electrospray. The MSFs investigated in this study possess 30-168 individual fluidic channels (each channel being 5 µm in diameter) that form a 2D emitting array. Multiple flow paths afford stable electrospray at flow rates ranging from the microspray (e.g., 1000 nL/min) to the nanoelectrospray (e.g., 10 nL/min) regime with moderate to negligible flow-induced backpressures. The electrospray stability of highly aqueous solutions (up to 99.9% water with 0.1% acetic acid) is enhanced through modification of the emitting surface with a hydrophobic silylation reagent (chlorotrimethylsilane). Furthermore, by successfully spraying highly concentrated salt solutions, this study demonstrates that multichannel MSF emitters provide enhanced robustness to clogging, leading to increased operational throughput. Electrospray ionization (ESI) is capable of transferring analytes from the solution phase to ions in the gas phase in an efficient fashion without inducing significant fragmentation and, as a result, has become the method of choice for coupling liquid separation techniques to mass spectrometry (MS) in laboratories around the world. Desolvation is critical to the efficient formation of gas phase ions by electrospray ionization. A number of methods (e.g., using sheath gas and/or heated curtain gas) have been used to aid in the desolvation process and improve ionization efficiency, thus increasing sensitivity. Further sensitivity enhancements can be recognized by reducing the flow rate to the nanoelectrospray regime, typically 18 MΩ · cm) was obtained from a Milli-Q Gradient water purification system (Millipore, Bedford, MA). The microstructured fibers (containing 30, 54, 84, and 168 holes) were purchased from Crystal Fibre (Denmark), with product numbers LMA-PM-16 (30 holes, no longer available), LMA-PM-16 (54 holes, no longer available), LMA-15 (84 holes), and LMA-20 (168 holes). Holes are ∼5 µm in diameter and spaced ∼7 µm apart, typically in a hexagonal pattern. The pulled-tip emitters (uncoated SilicaTips, 360 µm o.d. with tip diameters of 5, 15, and 30 µm) were from New Objective Inc. (Woburn, MA). Backpressure Measurements. The flow-induced backpressure of the MSFs was measured using an Eksigent NanoLC pump (Livermore, CA) with a 1:1 acetonitrile/water solvent composition (with 0.1% formic acid) using a technique modified from one reported earlier.24 Briefly, each MCN emitter was flushed for several minutes at 1 µL/min before measuring backpressure, and measurements were taken at each of three flow rates (50, 500, and 1000 nL/min). Each measurement is the average of five pressure readings taken every 10 s, and the pressure was allowed to stabilize between measurements (after changing the flow rate, usually 1-5 min). Since there is a linear relationship between backpressure and flow rate (R2 > 0.999), the slope of the line of the pressure vs flow rate plot gives the best indication of the backpressure associated with the emitter. Such a slope was measured three times for two emitters of each type. Measurements without an emitter attached were taken in the same way, and the backpressure associated with the system was found to be negligible. Each of the emitters was 4 cm in length and was connected to the pump outlet by coupling it via a PEEK union to a 200 µm i.d. transfer capillary. Chemical Modification of Microstructured Fibers and Tapered Emitters. The MSF segments were cut to 4 cm in length using a fiber cleaver (FiTel, Furukawa Electric, Japan). The acrylate coating was facilely removed from the entire emitter length after softening by immersion in toluene. One end of the cleaved fiber was inserted in a solution of 20% (v/v) CTMS in toluene and left overnight to render the emitter face hydrophobic. For comparison, the commercial tapered emitters were also coated with CTMS in the same manner. Following silylation, the MCN emitters were flushed for 10 min each with an acetonitrile/water solution (80/20 v/v) using a syringe pump (Harvard Apparatus, Holliston, MA) set at 500 nL/min. The tapered emitters were flushed with the same solvent mixture using a NanoLC-1D pump from Eksigent (Dublin, CA) directly prior to spraying to reduce the chances of clogging. Mass Spectrometry. Mass spectra were obtained using an API 3000 triple-quadrupole mass spectrometer (MDS Sciex/ Applied Biosystems, Streetsville, ON, Canada) coupled with a nanospray interface (Proxeon, Odense, Denmark). The MCN emitters and tapered emitters were affixed with a MicroTee (24) Gibson, G. T. T.; Koerner, T. B.; Xie, R.; Shah, K.; de Korompay, N.; Oleschuk, R. D. J. Colloid Interface Sci. 2008, 320, 82–90.
Figure 1. (A) Experimental online setup for the MCN emiter. (B) Apparatus for imaging offline nanoelectrospray from an MCN emitter. (C) Setup for MCN emitter anticlogging evaluation.
(Upchurch Scientific, Oak Harbor, WA), which was positioned with an x-y-z stage. The translation stage and two CCD cameras were used for final alignment of the emitter end at a distance of 2-15 mm from the MS orifice. No additional desolvation aids were employed. Delivery of samples to the MS was accomplished by direct infusion from a 30 µL fused-silica capillary custom loop connected to a 6 port ChemInert valve (VICI/Valco, Brockville, ON, Canada) coupled with the nanopump. The liquid junction technique was used to supply the electrospray potential, which was applied to a Pt electrode fitted into the MicroTee union upstream of the emitter, with the MS orifice plate acting as the ground (Figure 1A). Electrospray potentials were optimized for total ion current and spray stability and generally ranged from 2000 to 3500 V for the MCN emitters. Q1/positive mode was set to unit resolution. Total ion current was monitored from 400 to 800 m/z, while extracted ion current for leucine enkephalin was monitored from 555 to 557 m/z. Emitter performance was evaluated by the intensity and stability of the extracted ion current (XIC) signal in addition to the intensity of the analyte ion peak of the mass spectrum per mole of analyte consumed using a leucine enkephalin (LE) solution (1 µM). Solvent compositions ranging from highly organic solutions (90% methanol) to highly aqueous solutions (99.9% H2O) with 0.1% formic or acetic acid were investigated to assess the electrospray stability using solvent compositions typically encountered over the course of a reverse phase separation. Offline Electrospray Imaging. Aqueous electrospray samples (99.9% water, 0.1% formic acid) were delivered through a MCN emitter using a 0.5 mL Hamilton syringe (Gastight #1750) driven by an 11Plus syringe pump (Harvard Apparatus, Holliston, MA). Electrospray was generated offline via an Upchurch MicroTee using a liquid junction with an electrode having a voltage applied with a Trisep 2100 high-voltage module (Unimicro Technologies Inc., Pleasanton, CA). A grounded metal plate was placed ap-
proximately 5 mm from the emitter tip (Figure 1B). The resulting electrospray plumes were imaged using a Nikon Eclipse TE 2000-U (Nikon Canada Inc., Mississauga, ON, CAN) microscope equipped with a direct visualization system from Q-Imaging, QICAM with Simple PCI software (Compix Inc. Imaging Systems, PA). Emitter Anticlogging Evaluation. Hanks’ solution was used to evaluate emitter robustness in terms of resistance to clogging. The 100 mL aqueous Hanks’ solution contained 0.8 g of sodium chloride, 0.02 g of calcium chloride, 0.02 g of magnesium sulfate, 0.04 g of potassium chloride, 0.01 g of monobasic potassium phosphate, 0.127 g of sodium bicarbonate, 0.01 g of dibasic sodium phosphate, and 0.2 g of glucose.23 As illustrated in Figure 1C, the emitter was connected first to a 63 cm long 50 µm i.d. fusedsilica capillary filled with a 5 µM leucine enkephalin solution in 50/50 (v/v) H2O/ACN with 0.1% acetic acid to establish a stable MS signal. This capillary was then connected through a microunion to a 50 cm long 150 µm i.d. capillary filled with Hanks’ solution, which was then connected to a nanopump through a six-port valve and sample loop. The solutions were sequentially infused through the emitters at 300 nL/min and electrosprayed into the MS. The backpressure (i.e., time to clog) and total ion current were used to ascertain the relative robustness of each emitter. Clogged emitters were inspected using an optical microscope (Nikon Eclipse ME600, Nikon Canada Inc., Mississauga, ON, CAN) to verify blockage at the tip. No increase in backpressure resulting from particulate buildup at the head of the MCN emitter was observed during anticlogging evaluation. Emitter Spray Longevity Evaluation. MCN emitters were allowed to continuously spray at 250 nL/min a solution of verapamil (0.6 µM) and leucine enkephalin (0.7 µM) in 50% aqueous CH3OH with 0.2% acetic acid for over 5 h. The stability and intensity of the total ion current (TIC) were used to evaluate emitter longevity. Scanning Electron Microscopy (SEM). The MCN emitters were mounted vertically on an aluminum stub using tape to facilitate imaging of the fiber cross section and subsequently sputter coated with gold to render the sample conductive. SEM images were obtained using either a Jeol JSM-840 (Tokyo, Japan) or a Leo 1530 field emission (Oberkochen, Germany) scanning electron microscope. RESULTS AND DISCUSSION In this study, we have investigated several microstructured fibers (MSFs) as multichannel emitters for electrospray. As optical fibers, light is confined to the core of the fiber by a periodic refractive index introduced by the regularly spaced holes. The MSFs are manufactured commercially by stacking silica capillaries to a desired preform followed by heating and pulling (a technique referred to as “stack and draw”).22 The result is a remarkable material with a homogeneous hole structure that runs the length of the fiber. The fibers are typically coated with an acrylate polymer, but for the fluidic applications presented here, the coating is removed because of its poor resistance to organic solvents (e.g., methanol and acetonitrile). The SEM images in Figure 2 show cross sections of the MSFs examined in this study. Characteristics of each fiber are listed in Table 1. The hole diameter shown for each emitter was obtained using SEM micrographs by measuring Analytical Chemistry, Vol. 81, No. 17, September 1, 2009
7283
Figure 2. SEM images of the distal ends of four multichannel nanoelectrospray emitters and a tapered emitter: (A) 30 orifice emitter; (B) 54 orifice emitter; (C) 84 orifice emitter; (D) 168 orifice emitter; (E) tapered emitter with 15 µm aperture. Scale bars in A, B, and C represent 50 µm, while in D and E the scale bar represents 100 µm. The emitter in panel D is shown with its polymeric coating removed.
Table 1. Characteristics of the Multichannel Emitters Shown in Figure 2 number of hole diameter, diameter of o.d. without o.d. with holes µm hole region, µm coating, µm coating, µm 30 54 84 168
4.9 3.8 4.3 5.6
61 75 92 185
230 ± 3 230 ± 3 230 ± 5 230 ± 3
330 ± 10 330 ± 10 405 ± 10 350 ± 10
several holes across several cross sections and reporting the median value. The diameter of the hole region was also measured from SEM images, taken as the distance between the furthest points of the hole pattern. The outer diameters listed are those provided by the manufacturer. Each MCN emitter was used to spray from a low nanoflow rate (i.e., 500 nL/min). All MCN emitters investigated show negligible backpressures at low nanoflow rates and only moderate backpressures at flow rates as high as 1000 nL/min as outlined in Figure 3. As expected, an increase in the number of fluidic channels decreased the backpressure at a given flow rate. The flow-induced backpressures from the multichannel emitters are in the range of 1-15 bar for the 4 cm emitters at a flow rate of 1000 nL/min. Clearly, shorter emitters will cause less backpressure, but the 4 cm design was chosen because it fits the majority of MS ion sources. The ability of these emitters to spray with stability over a large range of flow rates is also advantageous for LC-ESI-MS operations. The performance of the modified MCN emitters was tested by electrospraying a solution of the peptide leucine enkephalin (LE) by constant infusion, using a range of organic solvent compositions (methanol or acetonitrile in water with 0.1% acetic or formic acid) and flow rates (50-1000 nL/min). Relatively similar performance was observed across all conditions (data not shown), but as a representative set of conditions, the presented data was obtained using a solution of 1 µM leucine enkephalin in 1:1 CH3OH/H2O with 0.1% acetic acid. Shown in Figure 4 is a series of extracted ion current (XIC) traces over a range of flow rates depicting electrospray performance for the various MCN emitters compared to tapered emitters under the typical electrospray conditions. Tapered emitters with different tip sizes (30, 15, 7284
Analytical Chemistry, Vol. 81, No. 17, September 1, 2009
Figure 3. Flow-induced backpressures at different flow rates from various 4 cm long multichannel emitters. The lines shown are representative experiments for each emitter type where the slope is closest to the average determined for that type, as described in the Experimental Section. Error bars represent the standard deviation in the backpressure readings at that flow rate for that experiment and were generally less than 10%.
and 5 µm) were used for electrospray at different flow rates (1000, 500, and 50 nL/min, respectively) according to the manufacturer’s recommendations. Using the LE solution, all emitters show similar performance at a moderate flow rate of 500 nL/min, as shown in Figure 4B. At a low nanoflow rate (e.g., 50 nL/min, shown in Figure 4C), the 30-hole MCN emitter provided increased ion current and enhanced stability relative to other emitters tested, while the other MCN emitters showed similar performance to that of the tapered emitter. It should be noted that the optimal working distance (a combination of sensitivity and ease of operation) is somewhat different for the MCN emitters than for the tapered emitters. The conventional tapered emitters have an optimal tip-to-orifice distance of approximately 1-5 mm, whereas the MCN emitters were found to be optimal at 5-15 mm from the MS orifice. While signal intensity continues to increase with shorter working distances (i.e., 2 mm), producing a consistent signal becomes more problematic. This increased distance can be seen as an advantage since greater operational distance reduces the risk of electrical arcing to the curtain plate, as well as enabling the integration of online sample preparation techniques such as desalting and preconcentration with reduced risk of ion source contamination. The ability to electrospray highly aqueous solutions is important for employing reverse phase LC gradients, which involve the transition from a highly aqueous to highly organic mobile phase. Furthermore, in structural proteomics, some samples cannot tolerate significant organic solvent content without denaturation, hence necessitating working in aqueous environments. However, highly aqueous solutions are notoriously difficult to spray with hydrophilic silica emitters owing to both the high surface tension and low volatility of water. To improve the emitter’s ability to spray aqueous solutions, the silica exit surface of the cleaved fiber can be modified with a silylation reagent to alter its wetting characteristics. In particular, chlorotrimethylsilane (CTMS) is effective at making silica relatively hydrophobic, thus preventing the surface wetting of the capillary exit by polar solvents. Experiments conducted in our laboratory have shown that the water contact angle is increased from 50° to 127° for CTMS-derivatized fused
Figure 5. Total ion current produced by derivatized 30-hole MCN emitters while spraying 1 µM LE in 99% H2O with 0.1% acetic acid at flow rates from 50 to 1000 nL/min (squares, right axis); signal intensity of the LE peak of the mass spectrum relative to the amount of analyte delivered at different flow rates using the 30-hole MCN emitter (circles, left axis).
silica.25 With the surface treatment, the MCN emitters can easily spray highly aqueous solutions as well as highly organic solutions. For example, the 30-hole MCN emitter was able to generate efficient electrospray with 99.9% aqueous solutions (with 0.1%
acetic acid) from 50 to 1000 nL/min, as shown in Figure 5. Fully aqueous solutions could also be sprayed at different flow rates switching to tapered emitters with varying tip i.d.’s, but in our hands, we found that “spitting” was more pronounced and the emitter was optimal over a narrower range of electrospray conditions. Commensurate with nanoelectrospray behavior, the MS signal at the m/z of an analyte of interest does not decrease linearly with flow rate.3 Indeed, if the signal is plotted relative to the number of moles of analyte delivered by the system as a function of flow rate, as in Figure 5, the sensitivity rises sharply as flow rate decreases. Such a plot demonstrates the sample utilization efficiency of the system as one can use much less material without greatly compromising sensitivity. For conventional tapered nanoelectrospray emitters, the Taylor cone can be used to monitor the stability of the electrospray process; however, the presence of a traditional Taylor cone is dependent on the flow rate while spraying from a MCN emitter. At relatively low flow rates (e.g., 25 nL/min water with 0.1% formic acid, Figure 6B), the typical Taylor cone is absent; at higher flow rates, a more traditional Taylor cone is observed (e.g., 300 nL/ min, Figure 6A). Similar electrospray behavior was reported for PPM-assisted electrospray emitters17,26 and entrapped microsphere emitters.18 The diameter of the emitting region of the MCN emitter is much larger than the diameter of the aperture of a conventional emitter (Table 1), which results in a larger electrospray plume that may not be as well sampled by the mass spectrometer. The fact that the 30-hole MCN emitter consistently provides better sensitivity than the 168-hole emitter (Figure 4) supports this idea. The multiple-electrospray emitters reported by Kelly et al. also suffered from poor MS sampling, and subsequently it was determined that a specially designed MS inlet incorporating an orifice aligned with each individual sprayer and an ion funnel provided the best ion current, presumably due to optimized ion transmission.14–16 It is, therefore, envisioned that MCN emitters used in conjunction with such an ion funnel would allow the collection of more analyte ions from the electrospray plume and, hence, increase sensitivity.
(25) Gibson, G. T. T.; Mugo, S. M.; Oleschuk, R. D. Polymer 2008, 49, 3084– 3090.
(26) Lee, S. S. H.; Douma, M.; Koerner, T.; Oleschuk, R. D. Rapid Commun. Mass Spectrom. 2005, 19, 2671–2680.
Figure 4. Extracted ion current produced by MCN emitters and tapered emitters while infusing 1 µM LE in 1:1 methanol/water solution (with 0.1% acetic acid) at (A) 1000, (B) 500, and (C) 50 nL/min. Tapered emitters with different aperture sizes were utilized at different flow rates commensurate with manufacturer specifications.
Analytical Chemistry, Vol. 81, No. 17, September 1, 2009
7285
Figure 6. Photomicrographs of electrospray from of a CTMSmodified 168-hole MCN emitter at different flow rates. (A) A traditional Taylor cone observed from offline electrospray of water with 0.1% formic acid at 300 nL/min; (B) a mist observed from offline electrospray at 25 nL/min.
A major advantage of the MCN emitters is their resistance to clogging, owing to the lack of tapered channels and the improbability that enough channels will clog to compromise electrospray ability. The clogging resistance of the emitters was ascertained by infusing Hanks’ solution, a concentrated nonvolatile salt mixture used for cell culturing. This method is used by some manufacturers of mass spectrometers to assess electrospray ionization source robustness. A commercial tapered emitter with a 5 µm tip aperture was used for a comparison with a 30-hole MCN emitter. A solution of leucine enkephalin in 50/50 (v/v) ACN/H2O with 0.1% acetic acid and Hanks’ solution were sequentially infused through each of the emitters at 200 nL/min using the setup shown schematically in Figure 1C. The resulting TIC (Figure 7A) and mass spectra were monitored to gauge the relative robustness of each emitter type. With the continuous infusion of Hanks’ solution, the tapered emitter experienced a sharp rise in flow-induced backpressure and a sharp decline in MS signal, completely clogging almost immediately following the arrival of the Hanks’ solution at the emitter. In contrast, the MCN emitters do not completely clog during the experiment. When the Hanks’ solution arrives at the emitter, there is an immediate drop in total ion current resulting from ion suppression; however, the emitter continues to produce ions during the entire Hanks’ solution infusion (Figure 7B). Once the Hanks’ solution is cleared, the total ion current levels again increase, albeit not to original levels. Although this is an extreme case, it demonstrates the relative robustness of the MCN emitter to clogging. The MCN emitter was also tested for longevity by monitoring its electrospray stability over 5 h. The 30-hole emitter continuously sprayed a solution of verapamil (0.6 µM) and leucine enkephalin (0.7 µM) in 50% CH3OH with 0.2% acetic acid at 250 nL/min without loss of TIC signal for the entire 5 h with a RSD of the TIC of 11% (data not shown). The sensitivity of the analyte ions detected was also maintained throughout the experiment. CONCLUSION Microstructured fibers have been demonstrated as efficient multichannel electrospray emitters. MCN emitters are easily 7286
Analytical Chemistry, Vol. 81, No. 17, September 1, 2009
Figure 7. Emitter robustness to clogging: (A) TIC traces for a tapered emitter (red trace) and a MCN emitter when a plug of Hanks’ solution is infused for 25 min followed by a 50/50 (v/v) ACN/H2O solution of 1 µM leucine enkephalin. Traces have been offset for easier viewing. Inset: Optical microscope images of the two emitters, (I) tapered emitter (scale bar is 100 µm) and (II) MCN emitter (scale bar is 200 µm) following Hanks’ solution infusion. (B) Total ion current trace during Hanks’ solution infusion showing that ion current is still produced by the MCN.
adapted to conventional nano-ESI instrumentation due to their dimensional similarity with commonly used capillary tubing diameters. With the hydrophobic treatment of the exit surface, MCN emitters are capable of electrospraying highly aqueous solutions as well as highly organic solutions, making them ideal for gradient LC-MS. Because of the lower flow-induced backpressure owing to the multiple channels, the MCN emitter can spray efficiently from a wide range of flow rates (e.g., from 10 to >1000 nL/min). MCN emitters can be directly used with standard nanoESI-MS interfaces; however, further gains in sensitivity may be possible with improved ion focusing. Moreover, the microstructured fibers utilized in this study were designed and optimized to offer reduced optical loss for photonic applications and not for the delivery of fluids and electrospray purposes. Consequently, further sensitivity enhancements would likely result from fabricating an optimized fiber based on modeling the size, pattern, and number of the orifices at the exit aperture according to specific applications. The merits of the improved robustness, increased sample throughput, low backpressure, large flow rate range, and good sensitivity offered by MCN emitters will provide researchers in any field with a useful tool to further explore the potential
advantages of multichannel electrospray in cases where they could benefit from moving to the nanoelectrospray regime but were previously reluctant to do so because of robustness concerns.
Innovation, and the Natural Sciences and Engineering Research Council of Canada.
ACKNOWLEDGMENT This research has been financially supported by Queen’s University, Ontario Innovation Trust, Canadian Foundation for
Received for review May 12, 2009. Accepted July 27, 2009. AC901026T
Analytical Chemistry, Vol. 81, No. 17, September 1, 2009
7287
