Comparison of Direct and Alternating Current Vacuum Ultraviolet

Jan 9, 2012 - *Phone: +358 9 191 59169. ... The emission distributions along the diameter of the lamp exit window were measured, and they showed that ...
1 downloads 0 Views 620KB Size
Download PDF
Article pubs.acs.org/ac

Comparison of Direct and Alternating Current Vacuum Ultraviolet Lamps in Atmospheric Pressure Photoionization Anu Vaikkinen,† Markus Haapala,† Hendrik Kersten,‡ Thorsten Benter,‡ Risto Kostiainen,† and Tiina J. Kauppila*,† †

Division of Pharmaceutical Chemistry, Faculty of Pharmacy, University of Helsinki, P.O. Box 56, 00014 Helsinki, Finland Department of Physical and Theoretical Chemistry, University of Wuppertal, Gausstrasse 20, 42119 Wuppertal, Germany

‡

S Supporting Information *

ABSTRACT: A direct current induced vacuum ultraviolet (dc-VUV) krypton discharge lamp and an alternating current, radio frequency (rf) induced VUV lamp that are essentially similar to lamps in commercial atmospheric pressure photoionization (APPI) ion sources were compared. The emission distributions along the diameter of the lamp exit window were measured, and they showed that the beam of the rf lamp is much wider than that of the dc lamp. Thus, the rf lamp has larger efficient ionization area, and it also emits more photons than the dc lamp. The ionization efficiencies of the lamps were compared using identical spray geometries with both lamps in microchip APPI mass spectrometry (μAPPI-MS) and desorption atmospheric pressure photoionization-mass spectrometry (DAPPI-MS). A comprehensive view on the ionization was gained by studying six different μAPPI solvent compositions, five DAPPI spray solvents, and completely solvent-free DAPPI. The observed reactant ions for each solvent composition were very similar with both lamps except for toluene, which showed a higher amount of solvent originating oxidation products with the rf lamp than with the dc lamp in μAPPI. Moreover, the same analyte ions were detected with both lamps, and thus, the ionization mechanisms with both lamps are similar. The rf lamp showed a higher ionization efficiency than the dc lamp in all experiments. The difference between the lamp ionization efficiencies was greatest when high ionization energy (IE) solvent compositions (IEs above 10 eV), i.e., hexane, methanol, and methanol/water, (1:1 v:v) were used. The higher ionization efficiency of the rf lamp is likely due to the larger area of high intensity light emission, and the resulting larger efficient ionization area and higher amount of photons emitted. These result in higher solvent reactant ion production, which in turn enables more efficient analyte ion production.

Atmospheric pressure photoionization (APPI) was introduced for liquid chromatography−mass spectrometry (LC-MS) in 2000 by Robb et al.1 and Syage et al.2 These two independent contributions made possible the LC-MS analysis of various neutral and nonpolar compounds that could not be ionized efficiently by either electrospray (ESI) or atmospheric pressure chemical ionization (APCI). Both articles introduced the ionization of soluted analytes by vaporizing liquid sample and ionizing the analytes in the gas-phase by 10.0 (and 10.6) eV photons emitted by a krypton discharge lamp. Robb et al. also introduced the use of agents that enhance the ionization, i.e., dopants, such as toluene and acetone.1 The APPI technique has gathered much attention ever since, as presented comprehensively in current reviews,3,4 and has become an essential technique in areas such as the analysis of polyaromatics, drugs, and petroleum. APPI has also been successfully miniaturized for combination with capillary LC,5,6 and the same method of ionization has been applied to the direct analysis of solid samples in desorption atmospheric pressure photoionization (DAPPI).7 Despite a decade of intense study on the ionization mechanism of APPI, some discrepancies between the results © 2012 American Chemical Society

obtained with the two available commercial APPI sources remain. While the APPI source developed by Robb et al., and commercialized by Sciex, requires the use of dopant that enhances the ionization performance,1,8 the APPI source developed by Syage can be operated without a dopant,9 although dopants have been reported to enhance also the ionization efficiency of the Syagen source.8 Other less pronounced differences between the ion sources have also been found, such as the somewhat different signal dependence on solvent flow rate discussed by Kauppila et al.8 These inconsistencies have been thought to arise from the geometries of the ion sources and the vacuum ultraviolet (VUV) lamps utilized. The VUV lamp used in the Sciex source excites the krypton gas with direct current (dc), whereas the lamp used in the Syagen source uses alternating current at radio frequency (rf). The rf lamp has been previously suggested to emit more photons than the dc lamp.2 Received: September 20, 2011 Accepted: January 9, 2012 Published: January 9, 2012 1408

dx.doi.org/10.1021/ac2024574 | Anal. Chem. 2012, 84, 1408−1415

Analytical Chemistry

Article

Figure 1. (a) The μAPPI and (b) the DAPPI setups.

spectrometer (Acton Research Corporation, Acton, MA). The lamps were positioned on the entrance flange of the spectrometer, which has an aperture of 0.25 mm width and 15 mm length centered on the flange. The spectrometer chamber was continuously flushed with helium held slightly above atmospheric pressure, so that a low gas flow exited the aperture and encompassed the exit window of the lamp. The VUV radiation further passed a slit of 0.25 mm and was then reflected via an MgF2 coated parabolic and movable grating (1200 G mm−1) onto a 0.15 mm wide exit slit. Subsequently, a scintillator (sodium salicylate) coated lens was used to convert the dispersed VUV radiation into visible light peaking at around 416 nm.14 The fluorescent light was focused onto a photomultiplier tube (IP28, Hamamatsu Photonics K.K., Shimokanzo, Japan) with a matching spectral response. The electrical signal was amplified 100-fold, digitized, and transferred at a rate of 10 Hz to a personal computer. A custom Visual Basic (Visual Basic 2010 Express, Microsoft Corporation, Redmond, WA) program provided the synchronization of the spectrum scanning and data recording. The scanning speed of the grating was set to 1 nm/s resulting in one data point per 0.1 nm. Additionally, the emission distributions along the diameter of the lamp exit window were measured for both lamp types. For this purpose the dc and rf lamps were mounted on a translational stage and moved along the entrance aperture of the spectrometer in 0.25 mm and 0.50 mm steps, respectively. For each position a full scan between 110 and 140 nm was recorded, and the total integrated emission of each spectrum (peak area of the 10.0 and 10.6 eV emission) was plotted as a function of the window position. MS Instrumentation. The MS spectra were acquired using an Agilent 6330 ion trap mass spectrometer (Agilent Technologies, Santa Clara, CA). The capillary voltage was set to −1650 V in positive ion mode, and 1750 V in negative ion mode. The spray shield of the mass spectrometer was replaced by a capillary extension (Agilent Technologies). Nitrogen drying gas was heated to 285 °C, and the drying gas flow rate was 4 L/min in DAPPI and 3 L/min in μAPPI. The MS data of analytes were collected by scanning m/z range 50−500 in negative ion mode and 50−600 in positive ion mode. The reactant ion data were recorded using m/z range 20−400. The data were processed using DataAnalysis software (DataAnalysis for 6300 series ion trap LC/MS version 3.4 (Build 192); Agilent Technologies). μAPPI Setup. The μAPPI (Figure 1 a) setup,15 the microchips used, and the chip holder16 have been described elsewhere previously. Microchip heating power was set to 3.0 W (ISO-TECH programmable power supply 603, RS Components, Northants, U.K.), which heated the microchip

In this contribution we have measured the emission spectra of a dc krypton discharge lamp similar to the one used by Robb et al.,1 and an rf excited krypton discharge lamp similar to the lamp used by Syage,2 and used two noncommercial, open ionization sources, namely microchip APPI (μAPPI) and DAPPI, to compare the performances of the lamps in APPIMS. Our open source designs of μAPPI and DAPPI enable the use of identical ion source geometries so that the effect of the lamps on the ionization process can be compared. Since solvents have a significant role in APPI ionization processes,9−13 six different solvent compositions in μAPPI, five spray solvents in DAPPI, and completely solvent-free DAPPI were studied to gain a comprehensive view on the performance of the two lamps.

■

EXPERIMENTAL SECTION Chemicals. Reagents were obtained as follows: toluene (HPLC-grade), benzo[a]pyrene, and testosterone from SigmaAldrich (Steinheim, Germany); methanol (HPLC-grade) and anthracene from Fluka Chemie GmbH (Buchs, Switzerland); hexane (HPLC-grade) from VWR international (Espoo, Finland); verapamil hydrochloride, 2-naphthoic acid, and 1,4dinitrobenzene from Aldrich (Milwaukee, WI); acetone (HPLC-grade) and acetaminophen from Merck (Darmstadt, Germany); midazolam from Hoffman-La Roche (Basel, Switzerland); and nicotine from Alfa Aesar (Karlsruhe, Germany). A Milli-Q Plus water purification system (Millipore, Molsheim, France) was used to purify water. Samples. Stock solutions (10 mM) of the studied analytes were prepared in toluene (anthracene, benzo[a]pyrene, 1,4dinitrobenzene), water (nicotine), or methanol (acetaminophen, midazolam, testosterone, verapamil, and 2-naphthoic acid). A 1 μM working mixture of acetaminophen, anthracene, benzo[a]pyrene, nicotine, midazolam, testosterone, and verapamil was prepared in each of the studied solvents for studies in positive ion μAPPI and 10 μM mixtures of acetaminophen, 1,4-dinitrobenzene, and 2-naphthoic acid for studies in negative ion μAPPI. In DAPPI the working mixture consisted of 10 μM acetaminophen, anthracene, benzo[a]pyrene, nicotine, midazolam, verapamil, and 1 μM or 10 μM testosterone in methanol/water (1:1, v/v) in positive ion mode and 10 μM of acetaminophen, 1,4-dinitrobenzene, and 2naphthoic acid in methanol/water (1:1, v/v) in negative ion mode. In DAPPI 1 μL droplets (corresponding to 10 pmol of each analyte per spot except for 1 pmol or 10 pmol of testosterone per spot) of the sample were applied on a PMMA plate (Vink Finland, Kerava, Finland). Determination of UV Emission. Emission spectra of both lamp types were recorded with a modified ARC VM-502 VUV 1409

dx.doi.org/10.1021/ac2024574 | Anal. Chem. 2012, 84, 1408−1415

Analytical Chemistry

Article

to approximately 300 °C. The sample was introduced by direct infusion, using flow rate 3 μL/min (PHD 2000 syringe pump; Harvard Apparatus, MA). The nebulizer gas (nitrogen) flow rate was 80 mL/min (mass flow controller GFC17, Aalborg, Orangeburg, NY). The microchip was aligned parallel to the MS inlet at the same horizontal plane, and the microchip nozzle was set to approximately 13 mm from the tip of the capillary extension (a in Figure 1a). The rf VUV lamp was a PKR 100 from Heraeus Noblelight (Cambridge, U.K.). The dc VUV lamp was a PKS 100 from Heraeus Noblelight, and it was operated at 0.8 mA current. The dc lamp was housed in a cover with a well for the lamp, which set the lamp window approximately 3 mm from the lamp surface. All given distances were measured from the lamp windows. Both VUV lamps were aligned orthogonally with the microchip, at 5−22 mm distance from the sample plume (b in Figure 1a), and were aimed at the spray exiting from the nozzle of the microchip. DAPPI Setup. The DAPPI setup (Figure 1 b) has been described elsewhere in detail.7 Microchip heating power was set to 4.5 W (the ISO-TECH programmable power supply 603), corresponding to DAPPI spray temperature of approximately 220 °C at 10 mm distance from the microchip nozzle. The DAPPI spray solvent flow rate was 10 μL/min (the PHD 2000 syringe pump), and nebulizer gas (nitrogen) flow rate was 180 mL/min (the GFC17 controller). In solvent-free DAPPI 260 mL/min gas flow rate was used. The microchip was aligned parallel to the MS inlet, and the microchip nozzle was set to approximately 3 mm above the capillary extension. The spray angle was approximately 45°. The plume was aimed at the sampling surface at approximately 3 mm from the tip of the capillary extension. The VUV lamps were aligned orthogonally with the capillary extension, at 7−30 mm distance (a in Figure 1 b) from the microchip spray nozzle.

■

Figure 2. (a) UV emission spectra of the dc and rf lamps. (b) The total integrated intensity of 10.0 and 10.6 eV photons emitted by the dc and rf lamps plotted as a function of the lamp window position.

RESULTS AND DISCUSSION

The UV emission spectra of the two lamps were essentially identical: the relative intensities of 10.0 and 10.6 eV photons were on the same level with both lamps (approximately 7:1, Figure 2a). However, measurement of the emission distributions along the diameter of the lamp exit window showed that the beam of the dc lamp is much narrower than the beam of the rf lamp (Figure 2b). This leads to larger efficient ionization area and a significantly higher amount of photons with the rf lamp. In addition an approximately 5-fold higher total photon flux of the rf lamp was measured. The effects of the different lamps on ionization in mass spectrometry were compared using the solvent compositions listed in Table 1, and μAPPI and DAPPI ion source geometries presented in Figure 1. The effects of the lamps on the photoionization process and ionization efficiency were studied using compounds presented in Figure 3.

from the sample plume without a noticeable change in the analyte signal, but decrease in the rf lamp distance from the sample plume led to increase in background signals, especially oxidized toluene species, which will be discussed in more detail further below. Corresponding results were obtained with acetone and methanol/toluene (9:1) as solvents. With methanol as the solvent, the analyte signals decreased over the studied range of approximately 5−22 mm with both lamps when they were placed further from the plume, but the absolute abundances recorded with the rf lamp were approximately 6 times higher than those with the dc lamp. The differences between the results with different solvents are discussed in more detail in the following sections. The lamp window to microchip nozzle distance was set to 8 mm for the rf lamp and 11 mm for the dc lamp for further comparisons with all solvents, when not stated otherwise.

■

■

LAMP PLACEMENT In μAPPI the position of the lamp was optimized by varying the lamp to sample plume distance (b in Figure 1a). Figure 4 presents the signal abundance of benzo[a]pyrene molecular ion (M+•) at different lamp to sample plume distances with toluene as the solvent. With the dc lamp the analyte signal intensity decreased over the studied range when the lamp was placed further from the sample plume, which was thought to be due to the spreading of the light beam and absorbance of the UV light in the ion source. The rf lamp could be placed at 8−15 mm

POSITIVE ION MODE The positive reactant ions of different solvents produced by the rf and dc lamps were recorded in μAPPI and DAPPI, and the observed ions are listed in Table 1 and S1 in Supporting Information, respectively. With both lamps and all solvents except toluene, intense background ion signals were detected at m/z ≥ 149, most likely due to phthalate esters and their fragments. Therefore, only ions below m/z 149 are discussed, except for toluene. In general, the reactant ion compositions 1410

dx.doi.org/10.1021/ac2024574 | Anal. Chem. 2012, 84, 1408−1415

Analytical Chemistry

Article

Table 1. Studied Solvent Compositions, Gas-Phase Properties of the Solvents, and Observed Reactant Ions in Positive Ion μAPPIa gas-phase properties of the solvent(s)24 solvent composition

IE (eV)

PA (kJ/mol)

toluene acetone hexane

8.8 9.7 10.1

784.0 812.0

methanol

10.8

754.3

methanol (S1)/ water (S2) (1:1) methanol (S1)/ toluene (S2) (9:1)

10.8/12.6

754.3/691.0

10.8/8.8

754.3/784.0

m/z of the observed reactant ions (suggested identity and relative abundance in parentheses) dc lamp

rf lamp

92 (S+•, 100), 106 (25) 43 (10), 59 ([S + H]+, 68), 99 (100), 117 ([2S + H]+, 86) 43 (23), 55 (22), 57 (21), 83 (47), 85 ([S − H]+, 100), 97 (18), 99 (38), 101 (55), 115 (83), 117 (24) 33 ([S + H]+, 64), 63 (29), 65 ([2S + H]+, 20), 73 (19), 87 (34), 103 (18), 121 (47), 129 (37), 140 (100), 141 (22) 33 ([S1 + H]+, 100), 47 (82), 59 (14), 65 ([2S1 + H]+, 91), 73 (41), 87 (42), 101 (21), 117 (13), 133 (19), 135 (12)

92 (S+•, 100), 106 (20), 108 ([S + O]+, 10), 199 (12) 43 (10), 59 ([S + H]+, 70), 99 (100), 117 ([2S + H]+, 86) 43 (20), 55 (17), 57 (20), 83 (38), 85 ([S − H]+, 100), 99 (21), 101 (33), 115 (10), 117 (10) 33 ([S + H]+, 73), 47 (55), 65 ([2S + H]+, 55), 73 (47), 87 (100), 101 (37), 121 (18), 129 (22), 136 (23), 140 (34) 33 ([S1 + H]+, 100), 47 (62), 65 ([2S1 + H]+, 75), 73 (30), 87 (37), 101 (16), 115 (12), 117 (11), 136 (23), 141 (13)

33 ([S1 + H]+, 89), 47 (56) 65 ([2S1 + H]+, 72), 73 (57), 87 (100), 93 ([S2 + H]+, 68), 101 (41), 107 (56), 139 (59), 141 (33)

33 ([S1 + H]+, 100), 47 (60), 65 ([2S1 + H]+, 81), 73 (57), 87 (99), 93 ([S2 + H]+, 73), 101 (44), 107 (46), 125 (47), 139 (36)

The 10 most abundant ions with relative intensity ≥10% and S/N ≥ 3 in m/z range 20−148 are reported for acetone, hexane, methanol, methanol/ water, and methanol/toluene. Toluene ions in the studied range (m/z 20−400) are reported. IE = ionization energy, PA = proton affinity, S = solvent. a

noise in the spectrum recorded with the dc lamp. These ions were thought to result from oxidation of toluene, as also reported in an earlier study by Tubaro et al. with the Agilent APPI ion source employing an rf lamp.17 The intensities of the oxidation products were observed to increase due to higher intensity of radiation as the rf lamp was placed closer to the sample spray. In μAPPI, acetone and methanol produced MH+ ions with both lamps. The MH+ ions were thought to have formed via formation of solvent or solvent cluster radical cations (Scheme 1, reaction 1) followed by a fast self-protonation reaction. The photoionization of acetone occurs since its ionization energy (IE) is lower than the 10.6 eV energy of the photons emitted by the VUV lamps. Methanol is ionized because the IEs of methanol clusters are below 10.6 eV,18,19 although methanol molecule has a higher IE of 10.8 eV. When water was added to methanol (1:1, v:v), very similar reactant ions as with pure methanol were produced, because the IE of water is too high (12.6 eV) for it to be ionized by the 10.6 eV photons. Addition of toluene to methanol (1:9, v/v) led to production of MH+ and [2M + H]+ of methanol, and MH+ of toluene. Methanol/ toluene was observed to produce reactant ions more efficiently than pure methanol: 80- and 600-fold intensities of methanol [2M + H]+ ions, and 74- and 240-fold intensities of methanol MH+ ions were obtained with the rf and dc lamps, respectively. The higher amount of reactants was thought to be formed because toluene has a low IE (8.8 eV) and is therefore ionized more efficiently by the 10.0 and 10.6 eV photons than methanol. In this case the M+• of toluene can also transfer protons to methanol clusters as discussed earlier.11,12 Hexane, which has IE of 10.1 eV, was efficiently ionized and produced [M − H]+ ions. With all solvent systems, the rf lamp showed a higher total amount of reactant ions than the dc lamp as presented in Figure 5a. With methanol and methanol/water (1:1) the effect was most drastic, and the rf lamp gave even an order of magnitude higher total reactant ion signal than the dc lamp. Because the emission spectra with both lamps were similar (Figure 2a), the higher amount of reactant ions formed by the rf lamp was thought to be due to the larger efficient ionization area caused by the broader light beam and the higher total amount of photons emitted by the rf lamp (Figure 2b).

Figure 3. Structures of the analytes.

Figure 4. Dependence of benzo[a]pyrene M+• signal intensity on the distance of the dc and rf lamp windows from the sample plume in μAPPI with toluene as the solvent. The shown values are averages for approximately 0.5 min of direct infusion of the 1 μM mixture at 3 μL/ min.

were similar with both lamps. A major difference between the lamps was found in toluene reactant ion production in μAPPI. Toluene M+• ion at m/z 92 was detected with both lamps. In addition to this, ions at m/z 108 and 199 were seen in high abundance with the rf lamp, whereas they were at the level of 1411

dx.doi.org/10.1021/ac2024574 | Anal. Chem. 2012, 84, 1408−1415

Analytical Chemistry

Article

Scheme 1. Photoionization Reactionsa

a

M = analyte, S = solvent, IE = ionization energy, PA = proton affinity, EA = electron affinity, ΔGacid = gas-phase acidity.

In DAPPI the intensities of the reactant and background ions were generally lower than in μAPPI. Methanol and methanol/ water (1:1) were ionized with the rf lamp, but not with the dc lamp. Acetone, methanol, and methanol/water (1:1) produced the same protonated reactant ions as in μAPPI. However, in DAPPI the background ion spectra recorded for toluene and hexane were similar with both lamps, but different from the μAPPI spectra. These results suggest that the sample plate material itself and impurities adsorbed on the plate may cause changes to the reactant and background ion composition. It can also be concluded that the impurities are nonpolar low proton affinity (PA) compounds, since the background was similar in DAPPI and μAPPI when using solvents that produced protonated solvent species. The analyte ions observed with the dc and rf lamps in positive ion μAPPI and DAPPI are presented in Table 2 and S2 in Supporting Information, respectively. Each studied solvent composition gave the same analyte ion species with the dc and rf lamps in μAPPI and in DAPPI, except DAPPI with methanol and methanol/water as solvents and solvent-free DAPPI, which showed no ions with the dc lamp. All the solvents that produced MH+ (acetone, methanol, methanol/water (1:1), and methanol/toluene (9:1)) or [M − H]+ (hexane) reactant ions produced mainly abundant protonated molecules of the test compounds (Table 2). The protonated analyte molecules can be formed via proton transfer reaction between the protonated solvent and the analyte (Scheme 1, reaction 2) or by direct photoionization of the analyte followed by fast hydrogen transfer from a protic solvent, such as methanol, to the M+• ion

Figure 5. (a) Relative signal intensity of all reactant ions in μAPPI at m/z 20−148. Note that the values between solvents should be compared with caution, since the mass spectrometer discriminates low m/z ions. The relative signal intensity of 1 μM (b) anthracene M+• and MH+ and (c) testosterone MH+ measured in μAPPI, and the average relative area of the spot signal of 10 pmol (d) anthracene M+• and MH+ and (e) testosterone MH+ in DAPPI using the dc and rf lamps. Abbreviations follow: tol = toluene, ace = acetone, hex = hexane, meoh = methanol, m/w = methanol/water (1:1), m/tol = methanol/toluene (9:1), s-f = solvent-free DAPPI.

(Scheme 1, reaction 3) as proposed earlier using an rf lamp.20 Thermodynamic calculations of reaction energy show that reaction 3 is favorable for testosterone, and unfavorable for the rest of the analytes (see Scheme S1 of Supporting Information). However, MH+ ions were detected for all studied analytes except anthracene in μAPPI or DAPPI when methanol was used as the solvent (Table 2 and S2). As the proton affinities for all analytes showing MH+ ions are above the PA of methanol, and the MH+ ion of methanol was observed in the reactant ion spectra (Table 1 and S1), proton transfer (reaction 2 in Scheme 1) is possible for all detected analytes with methanol and methanol/water. When toluene was used as the solvent, M+• ions of the compounds having low IEs (anthracene, benzo[a]pyrene, midazolam) were produced via charge exchange reaction 1412

dx.doi.org/10.1021/ac2024574 | Anal. Chem. 2012, 84, 1408−1415

Analytical Chemistry

Article

Table 2. Studied Compounds, Their Selected Gas-Phase Properties, and Their Ions Observed in Different Solvent Compositions in μAPPI with the Direct Current and Radio Frequency VUV Lampsa observed ions analyte Positive Ion μAPPI acetaminophen anthracene benzo[a]pyrene midazolam nicotine testosterone verapamil

Mmonoisotopic 151.06 178.08 252.09 325.08 162.12 288.21 454.28

Negative Ion μAPPI acetaminophen 151.06 1,4168.02 dinitrobenzene 2-naphthoic acid 172.05 a

IE (eV) 7.624 7.424 7.124

PA (kJ/mol)

toluene

acetone

n.d. M+• M+• M+•, MH+ n.d. MH+ MH+, m/z 303

MH+ n.d. n.d. MH+ MH+ MH+ MH+

7.624

[M − H]− M−•, m/z 138

8.324

[M − H]−

8.027 9.211

24

877 88325 100226 96324 92526 98026

hexane

methanol

methanol/water (1:1)

methanol/toluene (9:1)

MH+ n.d. MH+ M+•, MH+ MH+ MH+ MH+, m/z 303

n.d. n.d. MH+ MH+ MH+ MH+ MH+

[M − H]− [M − H]− [M − H]− −• −• M , m/z 138 M , m/z 138 M−•, m/z 138

[M − H]− M−•, m/z 138

[M − H]− M−•, m/z 138

[M − H]−

n.d.

n.d.

n.d. MH+ MH+ MH+ MH+ MH+ MH+

[M − H]−

n.d. n.d. M+•, MH+ MH+ MH+ MH+ MH+

n.d.

IE = ionization energy, PA = proton affinity, n.d. = not detected.

Figure 6. Radio frequency lamp spectra of a mixture of acetaminophen, nicotine, anthracene, benzo[a]pyrene, testosterone, midazolam, and verapamil (10 pmol each) recorded with (a) solvent-free DAPPI and (b) DAPPI with toluene as the spray solvent.

(Scheme 1, reaction 4), and MH+ ions of compounds having high PAs (midazolam, testosterone and verapamil) were produced most probably by proton transfer from M+• ions of toluene (Scheme 1, reaction 5). Solvent-free DAPPI with the rf lamp was thought to produce analyte M+• ions by direct photoionization (Scheme 1, reaction 1), accompanied by proton or hydrogen transfer with unidentified ambient air or sample plate components or contaminants. A typical spectrum of this novel DAPPI mode is shown in Figure 6 together with DAPPI spectrum obtained using toluene as the DAPPI spray solvent.

When methanol or methanol/water (1:1) was used as the solvent, sodiated analyte ions and minor amounts of analyte MH+ ions were occasionally observed in both μAPPI and DAPPI when the VUV lamps were turned off. This was thought to result from thermospray, a phenomenon that has previously been reported using the heated nebulizer microchips in different experimental conditions.21 However, the ionization efficiency in lamp off mode (thermospray mode) was poor, and the signal was unstable compared to the lamp on mode (APPI). The ionization efficiencies of the rf and dc lamps were also compared in DAPPI-MS/MS mode by determining the limits of detection (LOD) using S/N ≥ 3 for nicotine, midazolam, 1413

dx.doi.org/10.1021/ac2024574 | Anal. Chem. 2012, 84, 1408−1415

Analytical Chemistry

Article

testosterone, and verapamil (Table S3 in Supporting Information). Acetone was used as the DAPPI spray solvent, since it gave high ionization efficiency and only MH+ ions were formed for each studied analyte. The LODs determined with the dc lamp were 1.5−5 fold higher than the values obtained with the rf lamp. When the analyte ion intensities and LODs (Figure 5b−e and Table S3) are compared, it can be concluded that the rf lamp, which has larger efficient ionization area, higher total emission of photons, and greater reactant ion production (Figures 2b and 5a), also provides more efficient analyte ionization than the dc lamp. It could be that a higher amount of reactant ions leads to more efficient ionization of the analytes, since the ionization efficiencies of the analytes in μAPPI (Figure 5b,c) correlate well with the reactant ion intensities (Figure 5a). In the case of methanol and methanol/water the dc lamp produced a significantly lower number of protonated solvent species than the rf lamp, and therefore, the ionization efficiency of the analytes with these solvents was clearly better with the rf lamp than with the dc lamp in μAPPI. For dopantlike solvents with low IEs, such as toluene and acetone, the difference in reactant and analyte ion intensities was smaller. Since in DAPPI no solvent reactants were observed for methanol with the dc lamp, analyte ions were not produced with the dc lamp, while the rf lamp with the higher photon emission was radiant enough to produce both solvent and analyte ions. In solvent-free DAPPI only the rf lamp provided repeatable ionization of the studied analytes, although the ionization efficiency with the rf lamp was 2−3 orders of magnitude higher when a suitable solvent was used than without a solvent (Figures 5d,e and 6).

Figure 7. (a) The relative intensity of 10 μM 1,4-dinitrobenzene M−• in μAPPI, and (b) the average relative area of the spot signal of 1,4dinitrobenzene M−• (10 pmol) in DAPPI measured with the dc and the rf lamps. Abbreviations follow: tol = toluene, ace = acetone, hex = hexane, meoh = methanol, m/w = methanol/water (1:1), m/tol = methanol/toluene (9:1).

and thus improved ionization efficiency. The broader light beam of the rf lamp could also lead to more efficient release of electrons from the metal surfaces of the ion source. The dc and rf lamps were found to produce negative analyte ions most efficiently with acetone and toluene as solvents (Figure 7), and the ionization efficiency was over an order of magnitude higher than with, e.g., methanol (IE for methanol is 10.8 eV and for methanol dimer 9.8 eV).18 This can be explained similarly as in positive ion mode: the photoionization of a solvent and production of the thermal electrons are more efficient with the solvents having lower IE (toluene, acetone) than the solvents having higher IE (hexane, methanol). Since the release of thermal electrons depends on the used solvent, the production of thermal electrons must be largely due to photoionization of the solvent (Scheme 1, reaction 1), at least with the lower IE solvents. However, the release of thermal electrons from metal surfaces of the ion source was thought to be possible, since background ions were produced in negative ion DAPPI also without a spray solvent.

■

NEGATIVE ION MODE All studied solvents and both lamps gave the same ions for each analyte in negative ion μAPPI and DAPPI (Tables 2 and S2, respectively). 1,4-Dinitrobenzene produced negative molecular ions (M−•) and a fragment ion [M −NO]− (m/z 138), and acetaminophen and 2-naphthoic acid, both including an acidic group in their structure, produced deprotonated molecules ([M − H]−). Thermal electrons, which initiate the ionization process in negative ion APPI, are formed in photoionization of the solvent22 or released from the metal surfaces of the ion source.23 Since ambient oxygen is always present in the employed ion sources and oxygen has high electron affinity, the ionization reactions are likely to proceed via superoxide ion (O2−•) mediated reactions. Because 1,4-dinitrobenzene has high electron affinity, it is ionized most likely by charge exchange reaction with O2−• (Scheme 1, reaction 8), although electron capture reaction is also possible (Scheme 1, reaction 6). Acetaminophen and 2-naphthoic acid having acidic groups in their structures can be deprotonated by O2−•, which can act as a gas-phase base (Scheme 1, reaction 9). The ionization mechanism in APPI as well as the energetics of the reactants have been presented in more detail earlier.22 However, the m/z ratios of the reactant ions are low, and in this study, they could not be detected reliably due to discrimination of low masses by the ion trap MS. As in positive ion mode, the rf lamp provided higher ionization efficiency in negative ion mode than the dc lamp in both μAPPI and DAPPI (Figure 7). This is probably because the larger ionization area and higher amount of photons of the rf lamp provide more efficient photoionization of the solvent, which leads to formation of higher number of thermal electrons

■

CONCLUSIONS

We have shown that the rf VUV lamp gives higher ionization efficiency than the dc lamp in both μAPPI and DAPPI when solvents with high, above 10 eV, IE are used. This makes possible the use of solvents such as methanol or water in APPI without the addition of a low IE dopant. In DAPPI the rf lamp enabled photoionization even completely without a solvent. When using dopant-like solvents that have low IEs, such as toluene or acetone, there was no significant difference in the performance of the lamps. The difference between the ionization efficiencies of the dc and rf lamps was thought to result from the broader light beam of the rf lamp, which leads to larger efficient ionization area, higher total amount of photons emitted, and more efficient formation of solvent reactant ions. The comparison shows that photoionization techniques could benefit from the development of an even more intense photoionization lamp with optimized light beam shape that would provide enough photons for efficient operation with reversed-phase LC solvents. 1414

dx.doi.org/10.1021/ac2024574 | Anal. Chem. 2012, 84, 1408−1415

Analytical Chemistry

■

Article

(26) Ö stman, P.; Pakarinen, J. M. H.; Vainiotalo, P.; Franssila, S.; Kostiainen, R.; Kotiaho, T. Rapid Commun. Mass Spectrom. 2006, 20, 3669−3673. (27) Kuhn, W. F.; Lilly-Leister, D.; Kao, J.; Lilly, A. C. J. Mol. Struct. 1989, 212, 37−44.

ASSOCIATED CONTENT

S Supporting Information *

Additional information as noted in the text. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*Phone: +358 9 191 59169. Fax: +358 9 191 59556. E-mail: [email protected].

■

ACKNOWLEDGMENTS The Academy of Finland and CHEMSEM graduate school are acknowledged for funding the study.

■

REFERENCES

(1) Robb, D. B.; Covey, T. R.; Bruins, A. P. Anal. Chem. 2000, 72, 3653−3659. (2) Syage, J. A.; Evans, M. D.; Hanold, K. A. Am. Lab. 2000, 32, 24− 29. (3) Robb, D. B.; Blades, M. W. Anal. Chim. Acta 2008, 627, 34−49. (4) Marchi, I.; Rudaz, S.; Veuthey, J. Talanta 2009, 78, 1−18. (5) Kauppila, T. J.; Ö stman, P.; Marttila, S.; Ketola, R. A.; Kotiaho, T.; Franssila, S.; Kostiainen, R. Anal. Chem. 2004, 76, 6797−6801. (6) Haapala, M.; Luosujärvi, L.; Saarela, V.; Kotiaho, T.; Ketola, R. A.; Franssila, S.; Kostiainen, R. Anal. Chem. 2007, 79, 4994−4999. (7) Haapala, M.; Pól, J.; Saarela, V.; Arvola, V.; Kotiaho, T.; Ketola, R. A.; Franssila, S.; Kauppila, T. J.; Kostiainen, R. Anal. Chem. 2007, 79, 7867−7872. (8) Kauppila, T. J.; Bruins, A. P.; Kostiainen, R. J. Am. Soc. Mass Spectrom. 2005, 16, 1399−1407. (9) Short, L. C.; Cai, S.; Syage, J. A. J. Am. Soc. Mass Spectrom. 2007, 18, 589−599. (10) Itoh, N.; Aoyagi, Y.; Yarita, T. J. Chromatogr., A. 2006, 1131, 285−288. (11) Kauppila, T. J.; Kostiainen, R.; Bruins, A. P. Rapid Commun. Mass Spectrom. 2004, 18, 808−815. (12) Kauppila, T. J.; Kuuranne, T.; Meurer, E. C.; Eberlin, M. N.; Kotiaho, T.; Kostiainen, R. Anal. Chem. 2002, 74, 5470−5479. (13) Robb, D. B.; Smith, D. R.; Blades, M. W. J. Am. Soc. Mass Spectrom. 2008, 19, 955−963. (14) Kumar, V.; Datta, A. K. Appl. Opt. 1979, 18, 1414−1417. (15) Ahonen, L. L.; Haapala, M.; Saarela, V.; Franssila, S.; Kotiaho, T.; Kostiainen, R. Rapid Commun. Mass Spectrom. 2010, 24, 958−964. (16) Saarela, V.; Haapala, M.; Kostiainen, R.; Kotiaho, T.; Franssila, S. Lab Chip 2007, 7, 644−646. (17) Tubaro, M.; Marotta, E.; Seraglia, R.; Traldi, P. Rapid Commun. Mass Spectrom. 2003, 17, 2423−2429. (18) Kostko, O.; Belau, L.; Wilson, K. R.; Ahmed, M. J. Phys. Chem. A 2008, 112, 9555−9562. (19) Martrenchard, S.; Grégoire, G.; Dedonder-Lardeux, C.; Jouvet, C.; Solgadi, D. PhysChemComm 1999, 2, 15−19. (20) Syage, J. A. J. Am. Soc. Mass Spectrom. 2004, 15, 1521−1533. (21) Keski-Rahkonen, P.; Haapala, M.; Saarela, V.; Franssila, S.; Kotiaho, T.; Kostiainen, R.; Auriola, S. Rapid Commun. Mass Spectrom. 2009, 23, 3313−3322. (22) Kauppila, T. J.; Kotiaho, T.; Kostiainen, R.; Bruins, A. P. J. Am. Soc. Mass Spectrom. 2004, 15, 203−211. (23) Basso, E.; Marotta, E.; Seraglia, R.; Tubaro, M.; Traldi, P. J. Mass Spectrom. 2003, 38, 1113−1115. (24) NIST Chemistry WebBook, NIST Standard Reference Database Number 69; Linstrom, P.J., Mallard, W.G., Eds.; National Institute of Standards and Technology: Gaithersburg MD; http://webbook.nist. gov (retrieved August 15, 2011). (25) Strouk-Pointet, K.; Milliet, A.; Renou-Gonnord, M.-F. J. Mass Spectrom. 1995, 30, 1495−1504. 1415

dx.doi.org/10.1021/ac2024574 | Anal. Chem. 2012, 84, 1408−1415