Capillary Photoionization: A High Sensitivity Ionization Method for

Article pubs.acs.org/ac

Capillary Photoionization: A High Sensitivity Ionization Method for Mass Spectrometry Markus Haapala,* Tina Suominen, and Risto Kostiainen* Division of Pharmaceutical Chemistry, Faculty of Pharmacy, University of Helsinki, P.O. Box 56, FIN-00014 Viikinkaari 5 E, Finland S Supporting Information *

ABSTRACT: We present a capillary photoionization (CPI) method for mass spectrometric (MS) analysis of liquid and gaseous samples. CPI utilizes a heated transfer capillary with a vacuum ultraviolet transparent MgF2 window, through which vacuum UV light (10 eV) from an external source enters the capillary. The liquid or gaseous sample, together with dopant, is introduced directly into the heated transfer capillary between the atmosphere and the vacuum of the MS. Since the sample is vaporized and photoionized inside the capillary, ion transmission is maximized, resulting in good overall sensitivity for nonpolar and polar compounds. As in atmospheric pressure photoionization, ionization in CPI occurs either by proton transfer or by charge exchange reactions. The feasibility of CPI was demonstrated with selected nonpolar and polar compounds. A particular advantage of CPI is that it enables the analysis of nonvolatile and nonpolar compounds in liquid samples with high ionization efficiency. This is not possible with existing capillary ionization methods. The performance of CPI as an interface between GC and MS and its applicability for the analysis of steroids in biological samples are also demonstrated. The GC-CPI-MS method shows good chromatographic resolution, linearity (R2 > 0.993), limits of detection (LOD) in the range of 2−6 pg/mL and repeatability of injection with relative standard deviations of 4−15%.

I

between the atmosphere and the vacuum. Their system consists of a custom miniature spark discharge VUV lamp embedded in a transfer capillary with a windowless aperture. Sensitivity is very good for volatile compounds, but since the device is not heated and uses a windowless aperture, it is not suitable for nonvolatile compounds or liquid samples. We present a high sensitivity ionization method, capillary photoionization (CPI), which enables vaporization of a liquid sample and ionization of both polar and nonpolar molecules. In CPI, a sample is injected into the transfer capillary and vaporized by heat, and VUV radiation entering the capillary through a VUV transparent MgF2 window ionizes the analytes by photoionization processes. We also demonstrate CPI as an interface for gas chromatography−mass spectrometry (GC-MS) by showing its applicability for the analysis of steroids in biological samples.

onization efficiency and ion transmission are the key factors affecting the sensitivity of mass spectrometric analysis by atmospheric pressure ionization (API) methods, such as electrospray ionization (ESI), atmospheric pressure chemical ionization (APCI), and atmospheric pressure photoionization (APPI). When analytes are ionized at atmospheric pressure, however, most of the ions are lost when transferred to the vacuum of the mass spectrometer.1,2 It has been shown that the area from which ions are efficiently collected is small compared to sample vapor jets produced by API sources.3 Ionization of a liquid sample does not need to take place in an open atmosphere, even when modern liquid chromatography− mass spectrometry (LC-MS) instruments are used. Trimpin and coworkers4,5 recently presented a method called laserspray ionization (LSI) for easy generation of ions from a solid matrix. In LSI, ions are not generated at atmospheric pressure but in the heated transfer capillary between the atmosphere and the vacuum.6,7 Pagnotti et al.8−10 have reported a similar method, called solvent assisted inlet ionization (SAII), for liquid samples and demonstrated its feasibility as an interface between LC and MS. In SAII, ionization occurs without high voltage, electric discharge, or radiation. All that is required is heat. SAII has recently been modified into electrosprayed inlet ionization11 (ESII), which relies on voltage to enhance ionization, and nano-solvent assisted inlet ionization12 (nSAII), which utilizes very low flow rates. All these methods operate on ionization principles similar to those of ESI13 and thus, like ESI, they are well suited for polar analytes such as peptides and proteins; but they are not suitable for nonpolar compounds. Kersten et al.14 were the first to describe a photoionization method (cAPPI) for generating ions inside a transfer capillary © 2013 American Chemical Society

■

EXPERIMENTAL SECTION Chemicals and samples. Water was purified with a Milli-Q system (Millipore, Molsheim, France). LC-MS-grade methanol and acetonitrile, toluene (≥99.9% pure), HPLC-grade chlorobenzene, N-methyl-N-(trimethylsilyl)trifluoroacetamide (MSTFA), ammonium iodide, and dithioerythritol (DTE) were purchased from Sigma-Aldrich Chemie GmbH (Steinheim, Germany). The standard compounds benzo[a]pyrene (B[a]P), acetaminophen, verapamil, testosterone (T), β-estradiol (E2), 5-α-THDOC

Received: January 25, 2013 Accepted: May 15, 2013 Published: May 28, 2013 5715

dx.doi.org/10.1021/ac4002673 | Anal. Chem. 2013, 85, 5715−5719

Analytical Chemistry

Article

Figure 1. Structures and molecular masses of the standard compounds.

samples in the same way as the spiked artificial urine samples. MT was used as the internal standard (10 ng/mL) in all samples. Instrumentation. Direct Injection CPI-MS. The CPI device consists of a 1.5 mm i.d. stainless steel (SS) capillary (hereafter called CPI capillary) with a 1 mm wide and 15 mm long opening, a flat SS plate with a similar opening hard-soldered on the capillary, a 3 mm thick MgF2 window (Thorlabs Sweden AB, Goteborg, Sweden), a top plate with an 18 mm circular opening, and Teflon rings used as seals to hold the MgF2 window in place (Figure 2). The capillary (bottom) side of the setup was painted

(THDOC), progesterone (PROG), and methyltestosterone (MT) (Figure 1), diethyl ether, sodium dihydrogen phosphate, disodium hydrogen phosphate, and anhydrous sodium sulfate were also from Sigma-Aldrich. The sodium hydrogen carbonate−potassium carbonate mixture was prepared by mixing two parts of NaHCO3 from VWR (Leuven, Belgium) and one part of K2CO3 from J.T. Baker (Deventer, The Netherlands). Artificial urine was a self-made aqueous solution at pH 6, containing urea (Sigma-Aldrich) 14 g/L, diammonium hydrogen phosphate (Fluka) 1.7 g/L, creatinine (Sigma-Aldrich) 700 mg/L, glycine (Fabriques de Laire, Issy, France) 700 mg/L, DL-α-alanine (The British Drug Houses LTD, Poole, England) 700 mg/L, oxalic acid (Oriola Oy, Espoo, Finland) 700 mg/L, bovine albumine 350 mg/L (Sigma), glucose (BDH Limited, Poole, England) 350 mg/L, and sodium chloride (Akzo Nobel, Arnhem, The Netherlands) 250 mg/L. Stock solutions (1 mg/mL) of the analytes were prepared in methanol, except for B[a]P, which was prepared in toluene. Mixtures of B[a]P, testosterone, acetaminophen, and verapamil in acetonitrile/water (80/20, v/v) and acetonitrile/water/toluene (72/18/10, v/v/v) were used in initial optimization and evaluation of the CPI method. The stock solutions of steroids were diluted further with methanol to appropriate concentrations for GC-MS analysis. For GC experiments, the steroids were derivatized with trimethylsilyl (TMS). The steroids in methanol were evaporated to dryness, 50 μL of a mixture of MSTFA, ammonium iodide, and DTE (1000/2/4, v/w/w) was added, the vial was closed, the mixture was briefly vortexed, and the solution was incubated at 60 °C for 15 min. Samples were injected into the GC as such. Artificial urine samples spiked with steroids (T, PROG, E2 and THDOC) were used for the validation of the GC-CPI-MS/ MS method. The samples were prepared by adding 125 mg of NaHCO3/K2CO3 (2/1, w/w), 4 mL of diethyl ether and 1.5 g of anhydrous sodium sulfate to 2.5 mL of the artificial urine samples. The samples were then centrifuged at 2000 G for 10 min, and finally the organic layer was separated and evaporated to dryness (assisted with N2). Finally, 50 μL of derivatization reagent (MSTFA/NH4I/DTE as above) was added, and the samples were incubated at 60 °C for 15 min. The validated parameters were linearity, limit of detection (LOD), and repeatability of injection. The performance of the GC-CPI-MS/MS method was demonstrated by analyzing human urine samples from two healthy males. The steroids were analyzed by treating the urine

Figure 2. Schematic of the capillary photoionization device connected to a mass spectrometer.

with heat-resistant black paint and heated with IR radiation from a halogen lamp driven by a DC power supply (ISO-TECH IPS2010, RS Components, Northants, U.K.). The injection end of the CPI capillary was heated with a resistance wire heater, also driven by a DC power supply (ISO-TECH IPS603, RS Components). An rf excited 10 eV (124 nm) krypton VUV lamp (Heraeus Noblelight Analytics Ltd., Cambridge, U.K.) was used to initiate photoionization. The mass spectrometer for CPI experiments was an Agilent 6330 ion trap MS (Agilent Technologies, Santa Clara, CA) with a commercial capillary extension (KR Analytical Ltd., Sandbach, U.K.). The CPI setup was connected to the capillary extension by inserting the extension a few millimeters into the CPI capillary. Samples of 2 μL were injected manually, with a 10 μL syringe, about 5 mm inside the heated capillary. Injection speed was about 1 μL/s. GC-CPI-MS. An HP 5890 GC (Hewlett-Packard, Waldbronn, Germany) equipped with a CTC-A200S autosampler (CTC Analytics, 5716

dx.doi.org/10.1021/ac4002673 | Anal. Chem. 2013, 85, 5715−5719

Analytical Chemistry

Article

Zwingen, Switzerland) was utilized for the gas chromatography experiments. Splitless injection was done with 2 μL injection volume and 1 min splitless time. The temperature of the injector was 280 °C. The carrier gas was helium (≥99.996%, AGA, Espoo, Finland) with 100 kPa constant pressure. The GC temperature program was started with a 1 min isothermal period (190 °C), after which the temperature was increased to 240 °C at 20 °C/min, from 240 to 255 °C at 3 °C/min, and from 255 to 330 °C at 12 °C/min. The temperature was kept at 330 °C for 2 min. Figure 3 shows a schematic of the GC-CPI-MS connection. A methyl-deactivated retention gap (length 1 m, i.d. 0.25 mm)

Figure 3. Schematic of the GC-CPI-MS interface. Figure 4. Typical mass spectra recorded in CPI-MS experiments with 2 μL injections of 1 μM mixtures of the test compounds in (a) methanol/water with VUV lamp off, (b) methanol/water with VUV lamp on, and (c) methanol/water/toluene with VUV lamp on.

was connected in front of the analytical BXP5 column (length 15 m, i.d. 0.25 mm) and a deactivated fused silica capillary (length 1 m, i.d. 0.15 mm) after the column (all three from SGE Europe Ltd., Milton Keynes, U.K.) with pressfit connectors (BGB Analytik AG, Boeckten, Switzerland). The fused silica transfer capillary was fed through an SS t-piece placed in the GC oven and then through 0.5 mm i.d. SS transfer tubing connected to the t-piece. The SS transfer tubing, with the fused silica inside, was fed through a heated transfer line, and the tip of the SS tubing was inserted a few millimeters into the CPI capillary. The transfer line was heated by an in-house made resistance wire heater and an MSD transfer line, both set at 300 °C. Nitrogen was used as an auxiliary gas at 80 mL/min. Dopant (chlorobenzene) was pumped with a syringe pump at 2 μL/min, vaporized, and mixed with the nitrogen. The nitrogen-dopant gas mixture was led in 1/16 in. o.d. SS tubing into the t-piece and passed coaxially through the 0.5 mm i.d. SS transfer tubing into the CPI capillary.

molecule of verapamil was detected (Figure 4a). No significant change was seen when the lamp was turned on (Figure 4b). In these two cases, the system evidently behaves like solvent assisted inlet ionization (SAII), and the analytes are ionized inside the CPI capillary by ESI-like ionization of polar compounds based on highly charged droplets.13 This mechanism is capable of ionizing polar and ionic compounds, such as verapamil, but the ionization efficiency for neutral compounds is poor. When 10% of toluene was added to the methanol/water mixture and the VUV lamp was turned on, ionization of the standard compounds changed drastically. The intensity of the protonated molecule of verapamil increased, and acetaminophen, testosterone, and B[a]P were efficiently ionized via proton transfer reactions (Figure 4c). Here, toluene is easily ionized by 10 eV photons and efficiently produces reactant ions. The ionization mechanism is similar to that presented earlier for dopant-assisted APPI.15 Radical cations of the dopant react with solvent molecules or clusters to produce protonated solvent clusters, which can ionize the analytes via proton transfer. If the ionization energy of the analyte is lower than that of the dopant and competing reactions are energetically unfavorable, the dopant radicals can also react with analytes by charge exchange and produce molecular ions. Charge exchange was observed with trimethylsilylated steroids in GC−CPI-MS experiments (see below). The sensitivity of direct injection CPI-MS was tested in MS/ MS mode. The limit of detection (LOD) for testosterone was at low fmol level (2 μL injection), which is highly promising considering the nonoptimized construction of our prototype system and the age of the mass spectrometer. Repeatability of the manual injection with a syringe was relatively poor, on the other hand, owing to variations in injection speed and the

■

RESULTS AND DISCUSSION The performance of CPI was tested with a 1 μM mixture of B[a]P, testosterone, acetaminophen, and verapamil in methanol/water/toluene (72/18/10, v/v). The effect of temperature on the signal intensity was tested by increasing the power of the halogen lamp from 60 to 110 W with 10 W increments, which corresponds to an estimated temperature range of 200 to 300 °C in the CPI capillary. The signals of testosterone and verapamil increased with heating power up to 110 W, and 110 W was chosen for subsequent measurements. The effect of VUV light and dopant on the ionization process was studied with the same 1 μM mixture of the standards in methanol/water (80/20, v/v) and methanol/water/toluene (72/18/10, v/v/v). When methanol/water was used as solvent and the VUV lamp was turned off, only the protonated 5717

dx.doi.org/10.1021/ac4002673 | Anal. Chem. 2013, 85, 5715−5719

Analytical Chemistry

Article

The performance of GC−CPI-MS/MS was tested with the TMS-derivatized steroids. The extracted ion chromatograms of the TMS-steroids (Figure 6) show that the CPI interface does not cause peak broadening or tailing. This is attributable to the high temperature of the setup. The peak widths at half height were typically 2.7−3.8 s. The splitting of the peak of progesterone is due to the formation of two isomeric enol

position of the syringe needle in the inlet capillary. Achieving good repeatability will require development of an automated liquid injection system. Sample carryover from 1 μM sample to blank was minimal as was demonstrated with two injections of 2 μL of 1 μM testosterone sample followed by three injections of pure solvent (Figure 5). The peak area of the first blank peak

Figure 5. Test of sample carryover. Extracted product ion chromatrogram (MS/MS) of m/z 97 for two injections of 1 μM testosterone and three injections of solvent (note the scale difference).

was only about 0.4% of the sample peaks, and the signals of the second and third blanks were at background level. The ionization process in GC-CPI-MS was tested with and without dopant with trimethylsilyl (TMS)-derivatized steroids (T, PROG, E2, THDOC, and MT). In GC-CPI-MS with chlorobenzene as dopant, the TMS-steroids produced intense molecular ions (M+•) via charge exchange reaction, without detectable formation of protonated molecules. The fragmentation was minimal (Figure S-1 in Supporting Information), unlike in conventional GC-EI-MS where analytes are usually strongly fragmented. The chlorobenzene radical cation (m/z 112) can be seen in the spectra together with unknown chlorine-containing reactant ions (m/z 128 and 188). The GC-CPI setup was also tested without dopant, which should permit direct photoionization of the analytes. The same radical cations of the analytes were observed, but the intensity was only 1/100 of that obtained with dopant. The narrow slit in the capillary and the rather thick MgF2 window decrease the amount of VUV radiation entering the capillary, and we conclude that the system would benefit from a thinner window and higher intensity VUV lamp.

Figure 7. Extracted ion GC-CPI-MS/MS chromatograms of (A) testosterone and (B) progesterone showing spiked artificial urine samples (10 pg/mL), reagent blank samples (RBL), and two authentic urine samples. (*Correct peak of T identified by MS/MS spectrum).

Figure 6. Extracted ion GC-CPI-MS/MS chromatograms of the TMS derivatives T, PROG, E2, and THDOC and the internal standard MT spiked in artificial urine at 10 ng/mL.

Table 1. Validation Results for the TMS-Derivatized Steroids Analyzed with the GC-CPI-MS Method m/z

a

+

analyte

M*

production

T E2 PROG THDOC

432 416 458 550

417 285 443 265

linearity range 10 10 10 10

pg/mL−100 pg/mL−100 pg/mL−100 pg/mL−100

ng/mL ng/mL ng/mL ng/mL

R2

LOD pg/mL (S/N = 3)

repeatability of injection, rsd % (5 ng/mL, N = 6)

0.9956 0.9987 0.9988 0.9931

6a 2a 5a 5a

6.61 4.91 7.02 4.42

Calculated based on S/N values of lowest tested concentration and RBL. 5718

dx.doi.org/10.1021/ac4002673 | Anal. Chem. 2013, 85, 5715−5719

Analytical Chemistry

Article

(4) Trimpin, S.; Inutan, E. D.; Herath, T. N.; McEwen, C. N. Anal. Chem. 2010, 82, 11−15. (5) Trimpin, S.; Inutan, E. D.; Herath, T. N.; McEwen, C. N. Mol. Cell. Proteomics 2010, 9, 362−367. (6) McEwen, C. N.; Pagnotti, V. S.; Inutan, E. D.; Trimpin, S. Anal. Chem. 2010, 82, 9164−9168. (7) McEwen, C. N.; Trimpin, S. Int. J. Mass Spectrom. 2011, 300, 167−172. (8) Pagnotti, V. S.; Chubatyi, N. D.; McEwen, C. N. Anal. Chem. 2011, 83, 3981−3985. (9) Pagnotti, V. S.; Inutan, E. D.; Marshall, D. D.; McEwen, C. N.; Trimpin, S. Anal. Chem. 2011, 83, 7591−7594. (10) Chubatyi, N. D.; Pagnotti, V. S.; Bentzley, C. M.; McEwen, C. N. Rapid Commun. Mass Spectrom. 2012, 26, 887−892. (11) Pagnotti, V. S.; Chakrabarty, S.; Harron, A. F.; McEwen, C. N. Anal. Chem. 2012, 84, 6828−6832. (12) Wang, B.; Inutan, E.; Trimpin, S. J. Am. Soc. Mass Spectrom. 2012, 23, 442−445. (13) Trimpin, S.; Wang, B.; Inutan, E.; Li, J.; Lietz, C.; Harron, A.; Pagnotti, V.; Sardelis, D.; McEwen, C. J. Am. Soc. Mass Spectrom. 2012, 23, 1644−1660. (14) Kersten, H.; Derpmann, V.; Barnes, I.; Brockmann, K.; O’Brien, R.; Benter, T. J. Am. Soc. Mass Spectrom. 2011, 22, 2070−2081. (15) Kauppila, T. J.; Kuuranne, T.; Meurer, E. C.; Eberlin, M. N.; Kotiaho, T.; Kostiainen, R. Anal. Chem. 2002, 74, 5470−5479. (16) Magnisali, P.; Dracopoulou, M.; Mataragas, M.; DacouVoutetakis, A.; Moutsatsou, P. J. Chromatogr.,. A 2008, 1206, 166−177. (17) Hansen, M.; Jacobsen, N. W.; Nielsen, F. K.; Björklund, E.; Styrishave, B.; Halling-Sørensen, B. Anal. Bioanal. Chem. 2011, 400, 3409−3417.

ethers at the C17 position of the side chain. The LODs (based on S/N of 3) for the steroids spiked in artificial urine were between 2 and 6 pg/mL being lower or equal to those presented for GC-EI-MS/MS.16,17 Repeatability of injection with GC-CPIMS/MS was good: RSDs of the peak areas were 7% or less at a concentration of 5 ng/mL (Table 1). The linearity of the method was measured in the range of 10 pg/mL to 100 ng/mL, and the coefficient of determination (R2) was higher than 0.993 indicating good linearity of the method. The feasibility of the method was demonstrated by the analysis of urine samples of two healthy men. Although most of the steroids are excreted into urine in conjugated form, the high sensitivity of the method allows the detection of minor amounts of free testosterone and progesterone excreted in urine in unconjugated form (Figure 7).

■

CONCLUSIONS The presented prototype capillary photoionization is a rapid, sensitive, and versatile ionization method. CPI shows good potential as a simple and effective ion source for the analysis of both polar and nonpolar compounds unlike previous inlet ionization methods, which behave similarly to ESI. Compared to the previously presented cAPPI method, our device is heated and can also handle liquid samples and nonvolatile compounds. Thus it can be used both for direct liquid injection and for connecting LC to MS. CPI also enables GC instruments to be used with API mass spectrometers intended for LC-MS. The initial results indicate high sensitivity, thanks to high ion transmission efficiency. Quantitative performance is promising, and the results from the analysis of selected steroids in urine samples show that the method has good potential for being used in applications of analyzing less polar compound such as steroids in biological samples. Further experiments with GC, LC, and other types of mass spectrometers are needed to demonstrate the full potential of CPI. Improvement of the construction should enhance the performance and facilitate operation. Heating with a halogen lamp is effective but impractical, and another means of heating the capillary, such as enclosing the heat source in the structure, needs to be pursued.

■

ASSOCIATED CONTENT

S Supporting Information *

CPI-MS spectra of the steroids used in the study. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: markus.haapala@helsinki.fi (M.H.); risto.kostiainen@ helsinki.fi (R.K.). Phone: +358 (0)9 191 59171 (M.H.); +358 (0)9 191 59134 (R.K.). Notes

The authors declare no competing financial interest.

■ ■

ACKNOWLEDGMENTS Financial support from the Academy of Finland (grants 257316 and 251575) is gratefully acknowledged. REFERENCES

(1) Page, J.; Kelly, R.; Tang, K.; Smith, R. J. Am. Soc. Mass Spectrom. 2007, 18, 1582−1590. (2) Cech, N. B.; Enke, C. G. Mass Spectrom. Rev. 2001, 20, 362−387. (3) Lorenz, M.; Schiewek, R.; Brockmann, K. J.; Schmitz, O. J.; Gäb, S.; Benter, T. J. Am. Soc. Spectrom. 2008, 19, 400−410. 5719

dx.doi.org/10.1021/ac4002673 | Anal. Chem. 2013, 85, 5715−5719