Thin-Layer Chromatography and Mass Spectrometry Coupled Using

Solazzo , Matthew J. Collins , Jane Thomas-Oates , Edmund T. Bergström ..... Christina Y. Hampton , Rose McGready , Sivong Sengaloundeth , Nichol...
2 downloads 0 Views 429KB Size
Download PDF
Anal. Chem. 2005, 77, 1207-1215

Accelerated Articles

Thin-Layer Chromatography and Mass Spectrometry Coupled Using Desorption Electrospray Ionization Gary J. Van Berkel,* Michael J. Ford, and Michael A. Deibel†

Organic and Biological Mass Spectrometry Group, Chemical Sciences Division, Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831-6131

Desorption electrospray ionization (DESI) was demonstrated as a means to couple thin-layer chromatography (TLC) with mass spectrometry. The experimental setup and its optimization are described. Development lanes were scanned by moving the TLC plate under computer control while directing the stationary DESI emitter charged droplet plume at the TLC plate surface. Mass spectral data were recorded in either selected reaction monitoring mode or in full scan ion trap mode using a hybrid triple quadrupole linear ion trap mass spectrometer. Fundamentals and practical applications of the technique were demonstrated in positive ion mode using selected reaction monitoring detection of rhodamine dyes separated on hydrophobic reversed-phase C8 plates and reversedphase C2 plates, in negative ion full scan mode using a selection of FD&C dyes separated on a wettable reversedphase C18 plate, and in positive ion full scan mode using a mixture of aspirin, acetaminophen, and caffeine from an over-the-counter pain medication separated on a normal-phase silica gel plate.

“Desorption ionization” is a phrase that describes a diverse set of ionization methods for mass spectrometry (MS) in which the rapid addition of energy into a condensed-phase sample results in the production of gas-phase ions.1 Desorption electrospray ionization (DESI) is a new atmospheric pressure desorption * Corresponding author. Phone: 865-574-1922. Fax: 865-576-8559. E-mail: [email protected]. † Permanent address: Department of Chemistry, Earlham College, Richmond, IN 47374. (1) Busch, K. L. J. Mass Spectrom. 1995, 30, 233-240. 10.1021/ac048217p CCC: $30.25 Published on Web 01/29/2005

© 2005 American Chemical Society

ionization method introduced by Cooks and co-workers2 for the analysis of analytes on surfaces. This group described direct analysis of compounds on a number of different surface types using DESI-MS, including leather and nitrile gloves, a tomato skin, a medicine tablet, and even a blood drop on a finger, for a variety of analytes, from small pharmaceutical molecules to large biopolymers. With DESI, the charged liquid droplets and the gas jet from a pneumatically assisted electrospray (ES) ion source are directed at the surface to be analyzed, desorbing analytes into the gas phase by electrostatic and pneumatic means. The gas-phase ions ultimately generated from these desorbed analytes are then directed to the atmospheric sampling orifice of the mass spectrometer. The precise mechanisms of desorption and ionization are yet to be elucidated. It appears that gas-phase ions can be generated from those compound types typically amenable to analysis by ES-MS (e.g., ionic and polar molecules and biopolymers).3 In addition, initial results indicate that more nonpolar analytes, such as carotenoids, that are not particularly amenable to analysis by ES-MS4 may be analyzed by DESI. These compounds appear to be ionized via electron-transfer processes during the desorption process or in the gas phase following desorption. Another analytical surface that might be directly analyzed with DESI-MS is the chromatographic phase of a thin-layer chromatography (TLC) plate. The direct coupling of TLC and MS and tandem MS (MS/MS and MSn) has been of interest for a number of years.5-8 Matrix-assisted laser desorption/ionization (MALDI) (2) Taka´ts, Z.; Wiseman, J. M.; Golagan, B.; Cooks, R. G. Science 2004, 306, 471-473. (3) Cech, N. B.; Enke, C. G. Mass Spec. Rev. 2001, 20, 362-387. (4) Van Berkel, G. J.; McLuckey, S. A.; Glish, G. L. Anal. Chem. 1992, 64, 1586-1593. (5) Busch, K. L. J. Chromatogr., A 1995, 692, 275-290. (6) Wilson, I. D.; Morden, W. J. Planar Chromatogr. 1996, 9, 84-91. (7) Wilson, I. D. J. Chromatogr., A 1999, 856, 429-442.

Analytical Chemistry, Vol. 77, No. 5, March 1, 2005 1207

has become a popular approach to coupling TLC and MS. However, extensive post separation preparation of the plates prior to analysis, the need for specialized plates, low mass chemical noise from the MALDI matrix, and the requirement that the analysis be carried out in the vacuum chamber of the mass spectrometer have limited the technique. We recently reported a coupled TLC/ES-MS system for the analysis of TLC plates that makes use of a combined surface sampling probe/ES emitter as the coupling interface.9-11 That system exploits a sampling probe-to-TLC plate liquid microjunction and a self-aspirating ES emitter for the direct readout of developed TLC plates by ES-MS. The basic operation and analytical utility of this atmospheric pressure approach to TLC/MS have been illustrated using a single quadrupole mass spectrometer, a threedimensional quadrupole ion trap, and a hybrid triple quadrupole linear ion trap in the analysis of a variety of ES active components separated on commercially available hydrophobic reversed-phase (RP) C8 and C18 TLC plates. DESI provides a new alternative to either the MALDI or the surface sampling probe ES approach to coupling TLC and MS. From our perspective, DESI provides many of the same advantages of the surface sampling probe ES approach to coupled TLC/ MS when compared to MALDI. Furthermore, because the surface to be analyzed is not touched with a continuous liquid stream, DESI should circumvent surface wetting issues that have limited the current sampling probe ES approach.9-11 The surface sampling ES approach has been limited to hydrophobic RP-TLC phases and to eluting solvent systems containing at least 40% water by volume to avoid rapid wetting of the TLC phase. Rapid wetting of the chromatographic phase by the eluting solvent develops the analyte radially out from the vicinity of the surface sampling probe, inhibiting efficient sampling. In this paper, we report the successful coupling of TLC and MS using DESI for a variety of hydrophobic and wettable TLC stationary phases. The basic experimental setup and the optimization of TLC/DESI-MS conditions are discussed, including solvent flow rate, solvent composition, nebulizing gas flow rate, DESI emitter-to-surface distance, and the effect of surface scan rate on signal levels and chromatographic readout resolution. TLC/DESIMS fundamentals and applications are demonstrated using rhodamine dyes separated on hydrophobic RP C8 and RP C2 plates; FD&C dyes separated on a wettable C18 plate incorporating a preconcentration zone; and a mixture of caffeine, acetaminophen, and aspirin from a medicinal formulation separated on a normalphase silica gel plate. EXPERIMENTAL SECTION Chemicals. HPLC grade methanol was purchased from Burdick and Jackson (Muskegon, MI), ethanol was from Fisher Scientific (Gibbstown, NJ), and ACS grade acetone and ethyl acetate were from EM Science (Gibbstown, NJ). Ammonium acetate (99.999%) and acetic acid (PPB/Teflon Grade) were obtained from Sigma Aldrich (Milwaukee, WI). Rhodamine 6G (8) Gusev, A. I. Fresenius’ J. Anal. Chem. 2000, 366, 691-700. (9) Van Berkel, G. J.; Sanchez, A. D.; Quirke, J. M. E. Anal. Chem. 2002, 74, 6216-6223. (10) Ford, M. J.; Van Berkel, G. J. Rapid. Commun. Mass Spectrom. 2004, 18, 1303-1309. (11) Ford, M. J.; Kertesz, V.; Van Berkel, G. J. J. Mass Spectrom., in press.

1208

Analytical Chemistry, Vol. 77, No. 5, March 1, 2005

(1, basic red 1, CAS No. 989-38-8) and rhodamine B (2, basic violet 10, CAS No. 81-88-9) were obtained from Eastman Kodak Company (Rochester, NY). FD&C Red no. 3 (4, erythrosine B, CAS No. 16423-68-0) was from J. T. Baker (Phillipsburg, NJ); and rhodamine 123 (3, CAS No. 123333-76-6), FD&C Blue no. 1 (6, brilliant blue FCF, CAS No. 2650-18-2), FD&C Green no. 3 (5, fast green FCF, CAS No. 2353-45-9), FD&C Yellow no. 6 (7, sunset yellow FCF, CAS No. 2783-94-0), aspirin (8, CAS No. 50-78-2), acetaminophen (9, CAS No. 103-90-2), and caffeine (10, CAS No. 58-08-2) were from Sigma Aldrich. Extra Strength Excedrin (Bristol-Myers Squibb, New York, NY) containing aspirin, acetaminophen, and caffeine was purchased over the counter. Structures of all compounds investigated are shown in Figure 1, drawn as the free acid forms to simplify the mass spectrometry nomenclature for the gas-phase ions observed. Standard stock solutions of the rhodamines were prepared for TLC and ES-MS detection optimization by dissolving the dyes in methanol at a concentration of 200 ng/µL (0.45 nmol/µL for rhodamines 6G and B and 0.57 nmol/µL for rhodamine 123, respectively). Analytical standards were prepared from this stock by dilution with methanol for TLC spotting and by 60/40 (v/v) methanol/water with 0.1% acetic acid by volume for ES-MS optimization. Standard stock solutions of the FD&C dyes were prepared for TLC and ES-MS detection optimization by dissolving the compounds in either methanol or 80/20 (v/v) methanol/water at a concentration of 680 ng/µL (1.5 nmol/µL) for FD&C Yellow no. 6 (7), 1000 ng/µL (1.2 nmol/µL) for FD&C Green no. 3 (5), 1300 ng/µL (1.6 nmol/µL) for FD&C Blue no. 1 (6), and 3000 ng/µL (3.4 nmol/µL) for FD&C Red no. 3 (4). Analytical standards for TLC spotting were prepared by mixing aliquots of these stocks, and for ES-MS optimization, by dilution of this mixture in 50/50 (v/v) methanol/water. Standard stock solutions of aspirin, acetaminophen, and caffeine were prepared for ES-MS detection optimization by dissolving the compounds in methanol (aspirin and caffeine) or 50/50 (v/v) methanol/water (acetaminophen) at a concentration of 250 ng/µL (1.4 nmol/µL) for aspirin, 1000 ng/µL (6.6 nmol/µL) for acetaminophen, and 1000 ng/µL (5.2 nmol/µL) for caffeine. Analytical standards for ES-MS optimization were prepared from these stocks by dilution with methanol containing 0.1 vol % formic acid. TLC. For the rhodamines, TLC was carried out using hydrophobic Merck RP C8 plates (P/N 13725-5, EM Science, Gibbstown, NJ) and Merck RP C2 plates (P/N 5746-7, EM Science). Dye standards were spotted on the plates in 0.5-µL aliquots. Plates were developed in 80/20 (v/v) methanol/water containing ∼200 mM ammonium acetate. Developed plates were dried in an oven (110 °C) for 15 min just prior to analysis. Photographs of the developed plates were taken with a Coolpix 990 digital camera (Nikon, Tokyo, Japan) using white light illumination. Fluorescence images of the plates were taken using a Versadoc fluorescence imaging system (BioRad Laboratories, Hercules, CA) operating in UV transillumination mode with a 50-s exposure time, 0.5× gain (lowest sensitivity), and 1 × 1 binning (highest resolution).

Figure 1. Structure and mass-to-charge ratio observed for the compounds investigated.

For the FD&C dyes, TLC was carried out with a procedure adapted from Milojkovic-Opsenica et al.12 using wettable RP C18 plates (Whatman LKC-18, P/N 4800-600, Whatman, Middlesex, U.K.). Standards were spotted on the plates in 1.0-µL aliquots. Plates were developed in 70/30 (v/v) water/acetone containing ∼500 mM ammonium acetate. Developed plates were dried in an oven (110 °C) for 15 min just prior to analysis. Photographs of the developed plates were taken with a Coolpix 990 digital camera (Nikon, Tokyo, Japan) using white light illumination. The TLC separation of aspirin, acetaminophen, and caffeine extracted from an Excedrin tablet was carried out with a procedure adapted from Williamson13 using normal-phase silica gel plates with organic binder and UV254 indicator (P/N 59077, Alltech, Deerfield, IL). Prior to use, the plates were completely developed with 99/1 (v/v) ethyl acetate/acetic acid and dried in an oven (110 °C) for 30 min. The three components were extracted from (12) Milojkovic-Opsenica, D. M.; Lazarevic, K.; Ivackovic, V.; Tesic, Z. L. J. Planar Chromatogr. 2003, 16, 276-279. (13) Williamson, K. L. Macroscale and Microscale Organic Experiments, 4th ed.; Houghton Mifflin: Boston, MA, 2003, pp 160-162.

a weighed portion of a powdered Excedrin tablet using 50/50 (v/ v) ethanol/ethyl acetate (13.9 mg tablet powder/mL of solution). The extract was centrifuged and filtered through a 0.20-µm filter. This solution was spotted onto the plates in 1.0-µL aliquots. Plates were developed in 99/1 (v/v) ethyl acetate/acetic acid. Developed plates were dried in an oven (110 °C) for 15 min just prior to analysis. Photographs of the developed plates were taken with a Coolpix 990 digital camera (Nikon, Tokyo, Japan) using shortwavelength UV illumination. TLC/DESI-MS System. Figure 2 shows a schematic and photograph of the TLC/DESI-MS experimental setup. The mass spectrometer was a 4000 QTrap (MDS SCIEX, Concord, Ontario, Canada) hybrid triple quadrupole linear ion trap fitted with the nanospray interface orifice and operated using Analyst Software, version 1.4. The pneumatically assisted ES emitter was fashioned from a prototype “microionspray” head supplied by MDS SCIEX. The spray emitter was a 5.2-cm-long taper-tip fused-silica capillary (100-µm i.d., 360-µm o.d., New Objective, Woburn, MA). The inner diameter of the nebulizing tube was 500 µm, providing a nebulizing Analytical Chemistry, Vol. 77, No. 5, March 1, 2005

1209

from a 500-µL glass syringe connected to the emitter with ∼30 cm of 254-µm-i.d. (1/16-in. o.d.) Teflon tubing. The MS2000 x-y-z robotic platform (Applied Scientific Instrumentation Inc., Eugene, OR) and control software used to manipulate the TLC plate relative to the stationary DESI emitter have been described previously.11 The platform was operated manually or by computer control in the x-y plane for development lane scanning. A sidearm extension fabricated from plexiglass was attached to the TLC plate-holding platform to reach closer to the curtain plate of the mass spectrometer (Figure 2a). The TLC plate was held on this extension in the vertical x-y plane using doublesided tape, at an ∼50° angle to the DESI emitter and ∼10° angle from the axis of the sampling orifice of the mass spectrometer. For initial positioning and adjustment, the platform was controlled with the joystick (x-y) and jog wheel (z-surface-to-DESI emitter axis) of the manual control unit. The position of the TLC plate relative to the stationary DESI emitter was monitored with a Panasonic GP-KR222 closed circuit camera (Panasonic Matsushita Electric Corporation of America, Secaucus, NJ) with an Optem 70 XL zoom lens (Thales Optem Inc., Fairport, NY). The camera image was output to the mass spectrometer PC, and the image was monitored and captured using VidCap32 software (Microsoft, Redmond, CA) (Figure 2b).

Figure 2. (a) Schematic illustration of the TLC/DESI-MS experimental setup and (b) a color photograph of the DESI emitter and the TLC plate as viewed through the camera monitor during a TLC/DESIMS experiment. Separated bands of rhodamine 6G (1, orange band), rhodamine B (2, pink band), and rhodamine 123 (3, yellow band) are observed on the RP C8 TLC plate.

gas (nitrogen) jet annulus area of about 1.5 × 10-7 m2. The ES emitter was mounted ∼4 mm from the curtain plate of the mass spectrometer at an ∼50° angle relative to the TLC plate surface. Optimum spacing of the nebulizer tip was 0.5-1.0 mm from the TLC surface using a nebulizer gas flow rate of ∼2.4 L/s (275 m/s nebulizing gas jet linear velocity). The TLC plate was mounted so that the edge nearest the mass spectrometer was in line with the far edge of the heater orifice and about 10° off-axis from the line of sight down the sampling orifice. Plates were cut or samples separated in development lanes near the plate edge to align the sample bands with the DESI plume. The ES high voltage ((4.0 to 4.5 kV) was applied to the stainless steel body of the microionspray head. The ES solvent was delivered to the emitter by a syringe driver, controlled by the mass spectrometer software, 1210

Analytical Chemistry, Vol. 77, No. 5, March 1, 2005

RESULTS AND DISCUSSION Optimization and General Performance Metrics. Proper positioning of the ES emitter and the TLC plate relative to the atmospheric sampling orifice of the mass spectrometer was critical to the success of the TLC/DESI-MS experiments. To find the proper positioning, a glass microscope slide was used as the analysis surface, representative analytes were infused through the ES emitter, and the charged droplet ES plume was essentially bounced or reflected from the slide surface toward the mass spectrometer orifice. A set of rhodamine dyes whose performance metrics in our prior surface sampling probe TLC/ES-MS work was well-established were chosen for this task.11 The relative positions and angles of the glass slide and DESI emitter versus the sampling orifice were adjusted until the analyte signal was maximized. The schematic in Figure 2a shows the optimized TLC/DESIMS experimental setup. The angle of incidence of the DESI droplet jet onto the surface was steeper than the sampling angle into the sampling orifice of the mass spectrometer. The sputtered plume had a shallow angle off the surface with a portion of the droplet jet “skidding” along or just above the surface. Because of this, the edge of the TLC plate or sample holder closest to the sampling orifice served as a sputtered plume/ion guide into the atmospheric sampling region of the mass spectrometer. When the edge of the plate was pulled back from the curtain plate by more than ∼2 mm without changing the position of the DESI emitter, the signal levels were dramatically reduced. This observation pointed to the need for the sampling portion of the mass spectrometer to be located very close to the desorption region, consistent with the findings of Cooks and co-workers2. They accomplished that close positioning by extending the sampling capillary of the mass spectrometer out from the instrument by 30 cm. In our case, the region of the surface to be analyzed was positioned as near the sampling orifice as possible within the constraints of the current instrument interface design. For the

Figure 3. (a) Color photograph of the desorption trail left by the DESI emitter during a surface scan (44 µm/s) longitudinally along a developed lane of rhodamine B (2) spotted on a hydrophobic RP C8 plate using DESI solvent (methanol) flow rates of 2.5, 5.0, and 10 µL/min.

analysis of the TLC plates, this meant development lanes were run near a plate edge, or the developed plates were scored and cut to provide the required band locations on the plate for analysis. A modification to the sampling portion of the instrument similar to that described by Cooks and co-workers2 would be one means to eliminate this constraint on the TLC. Optimization of analyte signal levels in actual TLC plate analysis was accomplished by analyzing several centimeter-long lanes of selected analytes spotted onto the TLC plate surface. The spotted samples were developed a few millimeters up the plate to provide a uniform analyte band. The analyte band was then analyzed by moving the plate surface relative to the stationary DESI emitter under computer control, which provided a relatively constant analyte signal for a given set of conditions. In this fashion, the optimum DESI emitter-to-surface distance was found to be between 0.5 and 1.0 mm with a nebulizer gas linear velocity of 275 m/s out from the spray capillary/nebulizer annulus. These same experiments showed that as the DESI solvent flow rate was increased, at a fixed DESI emitter-to-surface distance, the width of the desorption region traced through a spotted band increased, as illustrated by the photograph in Figure 3. At solvent flow rates of 2.5, 5.0, and 10 µL/min, the width of the desorption trace was ∼300, 600, and 1000 µm, respectively. The mass spectral signal levels increased as the flow rate increased, probably reflecting the larger amount of analyte desorbed into the gas phase from the wider desorption region (see below). However, beyond a flow rate of 10 µL/min, there was not a major enhancement in the signal. This may be related to the substantial wetting of the plate observed at these higher flow rates, which appeared to elute a substantial fraction of the analyte to the edges of the desorption region, rather than sputter the material into the gas phase. This “elution versus desorption” phenomenon was indicated by the deep color ridge along the edges of the desorption trace in Figure 3 at both 5 and 10 µL/min. The sprayed solvent impinging on the chromatographic phase elutes a portion of the analyte out from the desorption region, even at these lower flow rates. Methanol was the exclusive DESI solvent used in the experiments reported here. Ethanol, 2-propanol, butanol, and 50/50 (v/ v) methanol/water were investigated with all the RP plates in addition to methanol. These other organic solvents provided about equal responses for the analytes when compared among themselves, but the mass spectral signal level was over an order of magnitude less than the signal recorded when using methanol. The 50/50 (v/v) methanol/water solvent physically removed stationary phase from the plate in the desorption region, resulting in a trench in the stationary phase along the path of the surface scan. Moreover, no signal from the analyte was observed. Mixtures containing smaller amounts of water were not investi-

gated. Methanol proved to be the best solvent with the normalphase plates, as well. The methanol/water mixture did not damage these plates; however, this solvent provided no better signal than methanol alone. For the most part, no signal advantage was observed when using solvent additives, such as formic acid, to enhance protonation of basic analytes. This result was somewhat surprising because solvent composition is typically an important factor in the ES response of an analyte.3 However, this observation may be a result of the analytes investigated rather than a general trend or indication of ionization mechanism in DESI. Investigation of other solvents and solvent combinations with a variety of analytes is warranted. The enhancement of analyte signal observed with increasing DESI emitter solvent flow rate is illustrated by the data in Figure 4, showing the positive ion mode selected reaction monitoring (SRM) ion current traces for rhodamines 6G (1), B (2), and 123 (3) separated on a RP C2 plate. These data were obtained from a surface scan of replicate development lanes using DESI solvent flow rates of 2.5, 5.0, and 10 µL/min. Signal levels were a factor of ∼3 larger at 10 µL/min, as compared to 2.5 µL/min. These same data showed that the higher DESI flow rate had minimal effect on chromatographic readout resolution. The resolution, R, of two chromatographic bands on the TLC plate was calculated from the fluorescence image of the plate, and the mass spectral data using eq 1 where d was the distance between the band centers and W1 and W2 was the width of the two bands.

R ) d/[(W1 + W2)/2]

(1)

The calculated chromatographic readout resolutions for the separation of rhodamine 6G (1) and rhodamine B (2) (R6G/B) and separation of rhodamine B (2) and rhodamine 123 (3) (RB/123) are shown in the respective panels in Figure 4. The mass spectral readout resolution was ∼20% poorer than the chromatographic resolution determined from visual examination of the fluorescence image of the separated bands on the plate (Figure 4b), but these readout resolutions were consistent for each of the three solvent flow rates. However, close visual inspection of the fluorescence image (Figure 4b) shows there was some small degree of overlap between compounds 1 and 2 (white haze between the spots) and that compounds 2 and 3 were just completely separated. We posit that the mass spectral readout resolution may be closer to the true resolution. Degradation in readout resolution, if any, was probably caused by the finite width of the DESI plume and desorption region (>0.3 mm). While scanning over closely spaced bands, it may be possible to begin desorption of ions from the beginning of the second band while still desorbing ions from the trailing edge of the first band. Using the same separation system, it was further found that chromatographic readout resolution was not affected by a change in surface scan rate up to a rate of 190 µm/s. This is illustrated by the SRM ion current profiles in Figure 5. Using a DESI emitter solvent flow rate of 5.0 µL/min, the mass spectral readout resolution at scan rates of 19, 44, and 190 µm/s was about the same in each case and nearly equal to that determined from the data shown in Figure 4. Faster surface scan rates were attempted. At 500 µm/s, the SRM signal was quite severely reduced (data not shown). This observation might have been related to the time Analytical Chemistry, Vol. 77, No. 5, March 1, 2005

1211

Figure 5. Positive ion SRM ion current profiles for 1 (m/z 443 f 415, black trace), 2 (m/z 443 f 415, red trace), and 3 (m/z 345 f 285, blue trace) obtained during development lane scans of replicate development lanes of the RP C2 TLC separation of a mixture (50 ng each) of rhodamines 6G (1), B (2), and 123 (3) at surface scan rates of (a) 19, (b) 44, and (c) 190 µm/s using a DESI solvent (methanol) flow rate of 5.0 µL/min. Dwell time was 100 ms for each transition. Signal levels were normalized to signal in panel (c). Chromatographic resolution, R, calculated from the data using eq 1 is shown in each respective panel.

Figure 4. (a) Color photograph and (b) fluorescence image of RP C2 TLC separation of a mixture (50 ng each) of rhodamines 6G (1), B (2), and 123 (3) and the positive ion SRM ion current profiles for 1 (m/z 443 f 415, black trace), 2 (m/z 443 f 415, red trace), and 3 (m/z 345 f 285, blue trace) obtained during development lane scans (44 µm/s) of replicate development lanes using a DESI solvent (methanol) flow rate of (c) 2.5, (d) 5.0, and (e) 10 µL/min. Dwell time was 100 ms dwell for each transition. Signal levels were normalized to maximum signal level in panel (e). Chromatographic resolution, R, calculated from the data using eq 1 is shown in each respective panel.

necessary to minimally wet the surface and sputter gas-phase ions or related to other desorption ionization phenomena not yet determined. Optimization of the DESI emitter position and solvent flow rate for higher surface scan rates was not attempted. In any case, the surface scan rate of 190 µm/s translated to an analysis time of 4 min or less for development lanes no longer than 5 cm. Low-level detection of rhodamine dyes was also examined in SRM mode. Plots of mass spectral peak areas versus the amount 1212 Analytical Chemistry, Vol. 77, No. 5, March 1, 2005

of dye spotted on the plates are shown in the graphs of Figure 6. These data were obtained during a surface scan at a fixed RF value across a dilution series of bands from a mixture of rhodamine 6G (1, Figure 6a) and rhodamine 123 (3, Figure 6b) separated on a RP C8 plate. The lowest levels spotted and detected were 250 pg for each dye (0.56 pmol compound 1, 0.72 pmol for compound 3). These detection levels are within a factor of 2 of those we recently reported for these dyes using our surface-sampling TLC/ ES-MS system11 and nearly equal to the detection levels reported by Cooks and co-workers2 in their DESI-MS paper for analytes of comparable mass. The signals plotted in Figure 6 were near linear up to about 10 ng (22.6 pmol) and 5.0 ng (14.5 pmol) spotted for compounds 1 and 3, respectively. Beyond these amounts spotted, the concentration in the desorbed ES plume appears high enough to exceed the dynamic range of the ES process (i.e., typically a few tens of micromolar or less),3 resulting in a signal level plateau. Applications. Beyond the basic optimization and detection figures of merit described above using rhodamine dyes separated on hydrophobic RP C8 and RP C2 plates, we examined two

Figure 6. Plots of mass spectral peak areas versus amount spotted obtained from the positive ion SRM ion current profiles during a surface scan (44 µm/s) at a fixed RF value across a dilution series of (a) rhodamine 6G (1, m/z 443 f 415) and (b) rhodamine 123 (3, m/z 345 f 285) bands separated on a RP C8 plate. The DESI solvent (methanol) flow rate was 5.0 µL/min, and the dwell time was 100 ms for each transition. Signal levels are normalized with respect to the maximum signal for the respective dye.

mixtures of significantly different analytes using yet different stationary phases for each to illustrate the potential breadth of application of TLC/DESI-MS. FD&C Dyes. A wettable RP C18 plate with a preconcentration zone was used to separate a mixture of four FD&C dyes, namely, Red no. 3 (4), Green no. 3 (5), Blue no. 1 (6), and Yellow no. 6 (7). Panel a in Figure 7 shows a picture of the development lane on the plate, where 21 mm on the distance scale corresponds to the C18/preconcentration zone transition. With a major analyte band very near this transition point (low RF), the development lane was scanned from high to low RF (left to right in the picture) to avoid destabilization of the ES-MS signal as this transition region was passed. Signal levels across the m/z range were severely attenuated when analyzing from the preconcentration zone, as compared to the C18 surface. The desorption trail through bands on the development lane was clearly visible in the photograph of the plate, which was taken following the analysis. Panel b in Figure 7 shows the base peak chromatogram acquired with the mass spectrometer set to collect negative ion full scan mass spectra using the Q3 ion trap mode (EMS mode, m/z 350-1000 range) during a surface scan of the TLC plate. Panels c, d, and e are the mass spectra recorded at the location of each band on the plate. These data show that each dye was detected as either the singly or doubly charged molecular anion or both. From the base peak chromatogram it appeared that Red no. 3 (4), the least polar of the compounds, had a desorption ionization efficiency that was significantly higher than the other three dyes. However, the band for compound 4 was the narrowest band on the plate, and the greater signal level may have reflected the higher amount of the dye per linear area on the plate, as compared to the other dye bands. In addition, the difference in signal level among the dyes was less dramatic in the summed ion current chromatogram for the molecular species of each dye

Figure 7. (a) Picture of wettable RP C18 TLC plate development lane showing the separated bands of a four-component spotted (1.0µL) mixture of FD&C dyes containing ∼320 ng of Yellow no. 6 (7), 250 ng of Green no. 3 (5), 260 ng of Blue no. 1 (6), and 240 ng Red no. 3 (4). (b) Base peak chromatogram from full scan negative ion EMS mode data (m/z 300-1000) acquired scanning the development lane shown in panel a at 190 µm/s from high to low RF. The background-subtracted, averaged mass spectra at the distance in the chromatogram corresponding to the respective band positions on the plate are (c) Yellow no. 6 (7, m/z 407 and 429), (d) Green no. 3 (5, m/z 381 and 763), Blue no. 1 (6, m/z 373 and 747), and (e) Red no. 3 (4, m/z 834). The DESI solvent was methanol sprayed at 10 µL/ min.

(data not shown) because compounds 5 (m/z 381 and 763), 6 (m/z 373 and 747), and 7 (m/z 407 and 429) each had two major molecular ions in their spectra, whereas compound 4 was observed only as the deprotonated molecule (m/z 834). Analytical Chemistry, Vol. 77, No. 5, March 1, 2005

1213

an Excedrin tablet extract containing these three compounds. Panel a in Figure 8 shows a picture of the development lane taken before the DESI-MS analysis. Panel b in Figure 8 is the base peak chromatogram acquired with the mass spectrometer set to collect positive ion full scan mass spectra in EMS mode during the surface scan of the TLC plate. Four peaks were observed in the base-peak chromatogram. The first three peaks correspond in location to the three analyte bands observed on the plate. The fourth peak, observed at 40 mm, corresponds to the location of the solvent front. The mass spectra for each of the three analyte peaks in the base peak chromatogram are shown in panels c, d, and e. Each compound was detected as both the singly charged protonated and sodiated molecule. In comparison to the other analyses shown here, it was necessary to spot significantly larger amounts (microgram quantities) of these compounds on the plate to obtain the high-quality mass spectra shown. The poorer desorption ionization efficiency of these compounds appeared to be related at least in part to the normal-phase stationary phase. In other experiments (data not shown), we did note lower signal levels for these three compounds, and others, from this phase, as compared to the other plates investigated, particularly in comparison to the hydrophobic C8 plates. The fundamental and mechanistic aspects of this observation are under investigation.

Figure 8. (a) Picture of normal-phase silica gel TLC plate development lane showing the separated components of a spotted Excedrin tablet extract (2.0 µL) containing approximately 2.5 µg of caffeine (8), 10 µg of acetaminophen (9), and 10 µg of aspirin (10). (b) Base peak chromatogram from full scan positive ion EMS mode data (m/z 60300) acquired scanning the development lane shown in panel a at 190 µm/s from low to high RF. The background-subtracted, averaged mass spectra at the distance in the chromatogram corresponding to the respective band positions on the plate correspond to (c) caffeine (8, m/z 195 and 217), (d) acetaminophen (8, m/z 152 and 174), and (e) aspirin (10, m/z 203 and 225). The DESI solvent was methanol sprayed at 10 µL/min.

Aspirin, Acetaminophen, and Caffeine in Excedrin Tablets. One Extra Strength Excedrin tablet contains 250 mg each of aspirin (8) and acetaminophen (9) as well as 65 mg of caffeine (10). A normal-phase silica gel plate was used to separate an aliquot of 1214 Analytical Chemistry, Vol. 77, No. 5, March 1, 2005

CONCLUSIONS In this paper, we demonstrated the use of DESI for coupling TLC and MS. Analytes separated on commercially available hydrophobic RP C8 and C2 plates, wettable RP C18, and normalphase plates were amenable to analysis in positive or negative ion mode. Positioning of the DESI emitter, TLC plate surface, and the atmospheric sampling orifice of the mass spectrometer were found to be crucial for obtaining maximum analyte signal levels. Close positioning of the desorption region to the sampling orifice was somewhat constrained by the configuration of our particular mass spectrometer. This meant that TLC development lanes needed to be run near the plate edge or the plates needed to be cut following development to provide the proper positioning for DESI-MS. With a change in the sampling orifice similar to that shown by Cooks and co-workers,2 this constraint on the TLC would be eliminated. These same changes might be expected to improve the sampling efficiency and, thus, absolute detection levels. Detection levels in atmospheric pressure MALDI (APMALDI), another recently developed desorption method, rapidly improved as methods for sampling into the mass spectrometer advanced.14,15 Nonetheless, detection levels for the rhodamine dyes investigated were within a factor of 2 of those detection levels we previously determined using a surface sampling probe/ES emitter interface for TLC/MS.11 The results presented point to other further studies. Desorption ionization from all TLC phases was not equivalent. The normalphase surface did not provide the lower detection levels garnered when using the RP plates. The mechanistic aspects and practical implications of this observation will need to be addressed through analysis of a wide range of analytes on these and other TLC phases (14) Miller, C. A.; Yi, D. H.; Perkins, P. D. Rapid Commun. Mass Spectrom. 2003, 17, 860-868. (15) Tan, P. V.; Laiko, V. V.; Doroshenko, V. M. Anal. Chem. 2004, 76, 24622469.

using a variety of DESI solvents and conditions. Means to enhance desorption ionization efficiency while minimizing damage to the TLC plate surface will also need to be addressed. Physical damage to the TLC plates occurred because of the use of aqueous solvents or, less frequently, because of the mechanical forces of the pneumatic DESI gas jet. Operation under conditions that damaged the chromatographic phase hindered the generation of analyte gas-phase ions and necessitated relatively frequent cleaning of the mass spectrometer interface to remove the sputtered stationary phase particles and restore optimum instrument performance. ACKNOWLEDGMENT M.J.F. acknowledges an Oak Ridge National Laboratory (ORNL) appointment through the ORNL Postdoctoral Research Associates Program. M.A.D. acknowledges an appointment to the U.S. Department of Energy (DOE) Higher Education Research Experiences (HERE) Program for Faculty at the Oak Ridge National Laboratory administered by the Oak Ridge Institute for Science and Education with support from the Oak Ridge Science Semester Program administered by Denison University. Becky R. Maggard (ORNL) is thanked for creation of Figure 2. Dr. Fred Rabel (EDM Chemicals, Inc., Gibbstown, NJ) is thanked for

samples of the RP C2 plates. The DESI-MS research was sponsored by the Division of Chemical Sciences, Geosciences, and Biosciences, Office of Basic Energy Sciences, U.S. Department of Energy. The TLC/MS plate scanning platform was developed with funding from the Laboratory Directed Research and Development Program of ORNL and support from ORNL Technology Transfer and Economic Development (TTED) Royalty Funds. ORNL is managed by UT-Battelle, LLC, for the U.S. Department of Energy under contract DE-AC05-00OR22725. This manuscript has been authored by a contractor of the U.S. Government under contract DE-AC05-00OR22725. Accordingly, the U.S. Government retains a paid-up, nonexclusive, irrevocable, worldwide license to publish or reproduce the published form of this contribution, prepare derivative works, distribute copies to the public, and perform publicly and display publicly, or allow others to do so, for U.S. Government purposes.

Received for review December 29, 2005.

December

1,

2004.

Accepted

AC048217P

Analytical Chemistry, Vol. 77, No. 5, March 1, 2005

1215