Method To Reduce Chemical Background Interference in Atmospheric

May 8, 2007 - Richard A. Scheltema , Anas Kamleh , David Wildridge , Charles Ebikeme , David G. Watson , Michael P. Barrett , Ritsert C. Jansen , Rain...
0 downloads 0 Views 169KB Size
Download PDF
Anal. Chem. 2007, 79, 4013-4021

Method To Reduce Chemical Background Interference in Atmospheric Pressure Ionization Liquid Chromatography-Mass Spectrometry Using Exclusive Reactions with the Chemical Reagent Dimethyl Disulfide Xinghua Guo,†,§ Andries P. Bruins,*,† and Thomas R. Covey‡

Mass Spectrometry Core Facility, University of Groningen, Antonius Deusinglaan 1, 9713 AV Groningen, The Netherlands, and MDS Sciex, 74 Four Valley Drive, Concord, ON, L4K 4V8, Canada

The interference of chemical background ions (chemical noise) has been a problem since the inception of mass spectrometry. We present here a novel method to reduce the chemical noise in LC-MS based on exclusive gasphase reactions with a reactive collision gas in a triplequadrupole mass spectrometer. Combined with the zero neutral loss (ZNL) scan of a triple-quadrupole mass spectrometer, the reactive chemical noise ions can be removed because of shifts of mass-to-charge ratios from the original background ions. The test on various classes of compounds with different functional groups indicates a generic application of this technique in LC-MS. The preliminary results show that a reduction of the level of LC-MS base-peak chromatographic baseline by a factor up to 40 and an improvement of the signal-to-noise ratio by a factor up to 5-10 are achieved on both commercial and custom-modified triple-quadrupole LC-MS systems. Application is foreseen in both quantitative and qualitative trace analysis. It is expected that this chemical noise reduction technique can be optimized on a dedicated mass spectrometric instrumentation which incorporates both a chemical reaction cell for noise reduction and a collision stage for fragmentation. During the development of the ionization and interface techniques in mass spectrometry there have been recurrent attempts to reduce the accompanying chemical interference. This is of particular importance for atmospheric pressure ionization (API) which is the method of choice for coupling liquid chromatography and mass spectrometry. Chemical background noise ions are unavoidable byproducts of electrospray ionization (ESI)1 and atmospheric pressure chemical ionization (APCI),2 partially, due to their soft, efficient, and generic nature. Any traces of ionized contaminants or “stable” cluster ions (because the ions survive * To whom correspondence should be addressed. Phone: +31-50-363-3262. Fax: +31-50-363-8347. E-mail: [email protected]. † University of Groningen. ‡ MDS Sciex. § Current address: Graz University of Technology, Institute of Analytical Chemistry and Radiochemistry, Technikerstrasse 4, 8010 Graz, Austria. (1) Yamashita, M.; Fenn, J. B. J. Phys. Chem. 1984, 88, 4451-4459. (2) Covey, T. R.; Lee, E. D.; Bruins, A. P.; Henion, H. D. Anal. Chem. 1986, 58, 1451A. 10.1021/ac062365t CCC: $37.00 Published on Web 05/08/2007

© 2007 American Chemical Society

under mild declustering conditions) can be detected in mass spectrometry as sources of chemical background noise ions. Nowadays the interference is still commonly observed in advanced LC-MS systems even when an improved interface for declustering and the proper high-purity HPLC solvents/additives are applied. This problem is worsened in the analysis of extracts or biological samples with very low concentrations of unknown components or with molecules having a poor ionization response because then these chemical background ions may completely overshadow the appearance of analytes in chromatograms and even in mass spectra. Further application of API LC-MS in bioanalysis (such as drug discovery and development, metabolite profiling, biomarker discovery, high-throughput screening, etc.) and chemical/food analysis (structure identification, impurity profiling, etc.) demands more selective and specific MS detection. For the automated MS/MS identification of unknowns in API LCMS, the capability of the LC-MS system to pick up the right molecular ions and/or trigger a proper MS/MS experiment for further structure studies of eluted trace components is highly dependent on the signal-to-noise (S/N) ratio of both chromatography and mass spectrometry. To reach the milestones of the current achievement in the sensitivity and the improved S/N ratio in API LC-MS, there have been dramatic efforts to improve the hardware for both ionization and ion transportation techniques. In addition, an add-on orthogonal laminar flow interface has been reported to improve the declustering condition in ESI and APCI.3 These techniques have reduced the interference of chemical background noise by focusing on ionization selectivity, declustering (desolvation) conditions, and preventing contaminants from ion sources and tubing, etc. The comprehensive tandem MS scan modes, such as selected ion monitoring (SIM), multiple reaction monitoring (MRM), neutral loss scan, product ion scan, and precursor ion scan, can further improve the MS detection specificity. However, these MS/ MS techniques are generally not suitable for trace analysis of unknown compounds. The recent application of high-field asym(3) Alavi, A.; Cousins, L. M.; Javahery, G.; Jolliffe, C.; Vuckovic, D. In Proceedings of the 52nd ASMS Conference on Mass Spectrometry and Applied Topics, Nashville, TN, May 23-27, 2004; MPH-27 (American Society for Mass Spectrometry, published on CD-ROM).

Analytical Chemistry, Vol. 79, No. 11, June 1, 2007 4013

metry ion mobility spectrometry (FAIMS)4 as an interfacing technique in MS5,6 is also aimed at the reduction or separation of chemical noise ions to enhance sensitivity. In addition to these hardware improvements, (off-line) software approaches have also been developed and implemented into some commercial systems. The component detection algorithm (CODA),7 the matched filtration with experimental noise determination (MEND),8 the sequential paired covariance,9 and other noise filtration methods10 have been reported and applied in API LC-MS, along with some standard noise subtraction/smoothing algorithms implemented in the manufacturers’ application programs for postacquisition data processing. In the mean time, software-assisted approaches such as dynamic background subtraction,11 active chemical background noise reduction,12 and windowed mass selection13 have also been applied as on-line chemical noise reduction techniques. Software approaches are also reported which remove mainly periodic and/ or regular background noise in time-of-flight (TOF) mass spectrometry.9,14 However, the above-mentioned techniques have not really solved the problem of interference from chemical background noise.15 This is due to the fact that some trace contaminants16 in HPLC reagents,17 tubing materials, MS ion sources, and even the laboratory air can result in significant contribution to background ions,18 especially in trace analysis. It is also well-known that the chemical interference varies (sometimes significantly) with HPLC mobile phases, matrices, and MS ionization conditions. The best solution to this problem is a selective removal or reduction of chemical background ions before they can reach the MS detector, rather than postacquisition data processing. Interestingly in inductively coupled plasma mass spectrometry (ICPMS) for inorganic elemental analysis, the reduction of isobaric polyatomic chemical interferences has been realized using reactive collisions between the polyatomic interference ions and a gas (H2, He, O2, or NH3, etc.).19 Different manufacturers have implemented several configurations, such as dynamic reaction cell (DRC)20 and octopole reaction system (ORS).21 Fundamentally it consists of (4) Purves, R. W.; Guevremont, R. Rev. Sci. Instrum. 1998, 69, 4094-4105. (5) Guevremont, R.; Purves, R. W. U.S. Patent 6,504,149, 2003. (6) Venne, K.; Bonneil, E.; Eng, K.; Thibault, P. Pharmagenomics 2004, 4 (4), 30-40. (7) Windig, W.; Phalp, J. M.; Payne, A. W. Anal. Chem. 1996, 68, 3602-3606. (8) Andreev, V. P.; Rejtar, T.; Chen, H.-S.; Moskovets, E. V.; Ivanov, A. R.; Karger, B. L. Anal. Chem. 2003, 75, 6314-6326. (9) Muddiman, D. C.; Rockwood, A. L.; Gao, Q.; Severs, J. C.; Udseth, H. R.; Smith, R. D. Anal. Chem. 1995, 67, 4371-4375. (10) Visentini, J.; Kwong, E. C.; Carrier, A.; Zidarov, D.; Bertrand, M. J. J. Chromatogr., A 1995, 712, 31-43. (11) Le Blanc, Y.; Bloomfield, N. Real-Time Dynamic Background Subtraction: Improving the Automated Ion Selection Process. Technical Note 114TN0201; Applied Biosystems/MDS Sciex. (12) Ramsey, R. S.; Goeringer, D. E.; McLuckey, S. A. Anal. Chem. 1993, 65, 3521-3524. (13) Fleming, C. M.; Kowalswi, B. R.; Apffel, A.; Hancock, W. S. J. Chromatogr., A 1999, 849, 71-85. (14) Rejtar, T.; Chen, H.-S.; Andreev, V.; Moskovets, E.; Karger, B. L. Anal. Chem. 2004, 76, 6017-6028. (15) Marquet, P. Ther. Drug Monit. 2002, 24, 125-133. (16) Ende, M.; Spiteller, G. Mass Spectrom. Rev. 1982, 1, 29-62. (17) Mabic, S.; Regnault, C.; Krol, J. LC‚GC Eur. 2005, 18, 410-414. (18) Guo, X.; Bruins, A. P.; Covey, T. R. Rapid Commun. Mass Spectrom. 2006, 20, 3145-3150. (19) Eiden, G. C.; Barinaga, C. J.; Koppenaal, D. W. Rapid Commun. Mass Spectrom. 1997, 11, 37-42. (20) Tanner, S. D.; Baranov, V. I. J. Am. Soc. Mass Spectrom. 1999, 10, 10831094.

4014

Analytical Chemistry, Vol. 79, No. 11, June 1, 2007

an ion guide device before the (quadrupole) mass analyzer, which is enclosed in a chamber that can be pressurized with a reagent gas. The reagent reacts with the ions of chemical interferences to convert them into a different species. Alternatively, a nonreactive gas just collides with ions and takes away the kinetic energy of the polyatomic interferences that finally are removed by kinetic energy discrimination.22 It has been concluded20 that the use of ion-molecule chemistry is more efficient than that of collisional fragmentation/cooling. This is mainly due to the chemical resolution resulting from the selective reactions between the chemical interferences and the reagent gas. Apparently, the overwhelming structure difference between polyatomic ions and atomic (metal) ions in ICPMS contributes to the significantly different reactivity with a chosen reagent gas. As for the commonly observed chemical interference ions in API LC-MS for organic analysis, our recent study18 has shown that the majority of the persistent interfering ions still have significant characteristics in their structures compared to that of conventional protonated species. These chemical interferences are mainly classified either as a cluster-type of ions involving HPLC solvent/additive molecules or as fragment intermediates of contaminants. Furthermore, there are also some sodiated (cluster) ions instead of protonated ones. The unique structure differences are an indication of possible differences in reactivity between background ions and conventional protonated molecules. In this paper we present our preliminary investigation on the development of a method to reduce chemical interferences in API LC-MS using exclusive ion/molecule reactions with dimethyl disulfide (DMDS). EXPERIMENTAL SECTION API MS and API LC-MS experiments (positive mode) were carried out on either a commercial API3000 or a custom-modified API365 triple-quadrupole mass spectrometer (MDS Sciex, Concord, ON, Canada) coupled with a Perkin-Elmer 200 series HPLC. For comparison, some experiments were performed on a MDS Sciex QSTAR coupled with an Agilent 1100 nanoLC system (Agilent) and MDS Sciex API4000. The collision gas inlet of the mass spectrometers used in this work was modified in order to introduce a volatile chemical reagent into the rf-only quadrupole (Q2) collision cell. This modification allows freeze, pump, and thaw cycles needed to degas the chemical reagents and a quick switch between the chemical reagent and the normal nitrogen collision gas. Various LC eluents with different organic compositions, additives, elution conditions (gradient and isobaric), and MS ionization and acquisition modes were investigated with regard to chemical background ionic noise (LC column: Vydac C18 200 mm long × 2 mm i.d.). Both positive ESI, nanoESI, and APCI were used. To study only the gas-phase reactions, ions of the background interference or the protonated analytes were first mass-selected by the first mass-resolving quadrupole (Q1). Then they were allowed to undergo reactive collisions with the reagent DMDS present in the second quadrupole (rf-only collision cell). The amount of DMDS introduced into the collision cell was adjusted (21) 7500cs Octopole Reaction System ICP-MS. Technical Note 5988-9881EN; Agilent Technologies, 2003. (22) Yamada, N.; Takahashi, J.; Sakata, K. J. Anal. At. Spectrom. 2002, 17, 12131222.

Figure 1. Mass spectra of the reactions of the chemical background ions at m/z 99 with the reactive collision gas dimethyl disulfide (DMDS) at the partial pressure of (A) 0.4 × 10-5, (B) 0.7 × 10-5, and (C) 1.0 × 10-5 torr. The background pressure before introducing DMDS was 0.6 × 10-5 torr. Partial pressure is defined as the increment of the ion gauge readout, see the Experimental Section.

by means of a needle valve and was monitored by reading the output of the Bayard Alpert gauge tube mounted on the vacuum chamber of the mass spectrometer, close to the turbomolecular pump. Under normal Q1 scan operating conditions the readout is 6.0 × 10-6 torr. When DMDS was introduced the pressure readout increased to 1.3 × 10-5 torr. The pressure increment (0.7 × 10-5 torr in this example) will be called “partial pressure” of DMDS. The experimental setup did not allow a direct pressure measurement inside the collision cell. The Sciex collision cell was designed for operation in the low-millitorr range, in order to take advantage of collisional focusing and slowdown of product ions prior to exiting the cell so that a narrow beam of slowly moving ions is accepted by Q3. In this way sensitivity of MS/MS and resolution of product ions in Q3 is improved. In view of the normal operating pressure range with nitrogen in the collision cell, the pressure of DMDS inside the collision cell is estimated to be a few millitorr. No corrections were made for the relative response factors of nitrogen and DMDS in the ion gauge readout. It is appreciated that the ion gauge readout for organic molecules can be several times as high as the real pressure. The resulting ion/ molecule reaction mixture was monitored by means of a product ion scan of the mass-resolving quadrupole Q3. To study the reduction of chemical interferences through full scan mass spectra, the mass spectrometer was operated in the zero neutral loss (ZNL) scan mode with the chemical reagent present in the Q2 collision cell, while the HPLC was run at the normal mode and conditions. Zero neutral loss scan is a special case of the well-known constant neutral loss scan with neutral loss set to zero (collision energies, 3-5 eV; dwell time, 5-10 ms). For comparison, the conventional collision gas N2 was also used

in some experiments to acquire normal Q1, Q3, and ZNL scan data for both infusion and LC-MS studies. The chemical reagent dimethyl disulfide (CH3S-SCH3, DMDS, 99%) was purchased from ACROS Organics. The tested compounds and the HPLC grade solvents and additives were commercially available and used as received. For infusion, a solution at a final concentration about 5-50 µM was used. About 10 ng of each compound was injected for LC-MS analysis. RESULTS AND DISCUSSION Reactions of Commonly Occurring Chemical Background Ions with the Chemical Reagent Dimethyl Disulfide. During the past decades of ion/molecule reaction studies in mass spectrometry, enormous efforts have been focused on investigating specific ion/molecule23 or ion/ion reactions24 for a class of compounds25 or functional groups26,27 in order to facilitate its analytical chemical application for either structure characterization or compound screening. The specificity of a successful application of ion/molecule or ion/ion reactions has limited its generic analytical application. Our initial attempts to enhance the specificity of detection of analyte ions in triple-quadrupole MS started with searching for neutral reagents that react specifically with a certain type of analyte ions. However, shortly after a few unsuccessful tryouts, we were inspired by the surprising observa(23) Brodbelt, J. S. Mass Spectrom. Rev. 1997, 16, 91-110. (24) McLuckey, S. A.; Stephenson, J. L. Mass Spectrom. Rev. 1998, 17, 369407. (25) Green, M. K.; Lebrilla, C. B. Mass Spectrom. Rev. 1997, 16, 53-71. (26) Eberlin, M. N. Mass Spectrom. Rev. 1997, 16, 113-144. (27) Stirk, K. M.; Orlowski, J. C.; Leeck, D. T.; Kentta¨maa, H. I. J. Am. Chem. Soc. 1992, 114, 8604-8606.

Analytical Chemistry, Vol. 79, No. 11, June 1, 2007

4015

Figure 2. Mass spectra of the reactions of the chemical background ions at m/z (A) 83, (B) 115, (C) 143, and (D) 159 with the reactive collision gas dimethyl disulfide (DMDS) at the partial pressure of 0.7 × 10-5 torr. The background pressure before introducing DMDS was 0.6 × 10-5 torr. Partial pressure is defined as the increment of the ion gauge readout, see the Experimental Section. Note that these chemical background ions were generated from MeOH/water/acetic acid (50:50:0.1 vol %).

tion of exclusive reactions of some neutral reagents with chemical background ions rather than with protonated analytes. The vast majority of the cluster-type (or related)18 chemical background ions react very efficiently with the neutral chemical reagent DMDS by forming clusters with up to three neutral molecules of DMDS. The typical mass spectra of the reactions of chemical background ions ([H3PO4 + H]+ at m/z 99, observed when the LC-MS eluent MeOH/water/acetic acid (50:50:0.1 vol %) was used, are given in Figure 1. With the increase of the partial pressure of DMDS (which has a vapor pressure of 29 torr at 25 °C) in the collision cell, ions at m/z 99 are converted gradually to the reaction products [99 + nDMDS]+. On top of the reactions with DMDS there are also accompanying clustering reactions of neutral water molecules with these background ions and their products with DMDS. Water was present either as an impurity of DMDS introduced into the collision cell or simply as a background vapor in the reagent gas handling system (water has a vapor pressure of 24 torr at 25 °C). The combination of these two reactions makes the reaction products a very complex mixture. All reactions with either DMDS or water result in the significant shift of the mass-to-charge ratios of the reactive chemical background ions. The involvement of the neutral water molecules in the reactions has also been observed for some other chemical 4016

Analytical Chemistry, Vol. 79, No. 11, June 1, 2007

background ions. No efforts have been made in this study to remove water simply because of the intention to take advantage of reactions with water to further increase reaction efficiencies. On the other hand, one may examine the possibility of applying water as a chemical reagent to react with the chemical background ions. Similar clustering reactions of DMDS and water with other background ions at m/z 83 ([AcOH + Na]+), 115 (structure not assigned yet), 143 ([2AcOH + Na]+), and 159 ([H3PO4 + AcOH + H]+ are shown in Figure 2. It should be noted that not only the cluster-type background ions but also those ions related to contaminants react efficiently with DMDS. Besides clustering with DMDS, other reactions of chemical background ions with DMDS include the formation of the product ions at m/z 141 ([DMDS + SCH3]+) and a small amount of charge/proton-transfer reactions to form [DMDS]+ or [DMDS + H]+. In agreement with the literature about reactions of DMDS with small organic ions,28 it is suggested that ions at m/z 141 are formed through first charge/proton transfer from background ions to DMDS, followed by reactions with another DMDS neutral molecule to yield products [DMDS + SCH3]+. A detailed discussion about the formation and structure of the ion (28) Watkins, M. A.; WeWora, D. V.; Li, S.; Winger, B. E.; Kentta¨maa, H. I. Anal. Chem. 2005, 77, 5311-5316.

Figure 4. Typical mass spectra of reactions of the isolated ions of (A) protonated prazepam [M + H]+, m/z 325 and (B) doubly protonated peptide angiotensin II [M + 2H]2+, m/z 524, with dimethyl disulfide (DMDS) at the partial pressure of 0.7 × 10-5 torr. Partial pressure is defined as the increment of the ion gauge readout, see the Experimental Section.

Figure 3. Mass spectra of the reactions of the chemical background ions at m/z (A) 60, (B) 78, (C) 83, (D) 149, and (E) 205 with the reactive collision gas dimethyl disulfide (DMDS) at the partial pressure of 0.7 × 10-5 torr. Partial pressure is defined as the increment of the ion gauge readout, see the Experimental Section. The background pressure before introducing DMDS was 0.6 × 10-5 torr. Note that these chemical background ions were generated from ACN/HCOOH/ H2O.

at m/z 141 is presented by Jarvis et al.29 Some typical examples of this type of reactions of the ions at m/z 60 [ACN + H2O + H]+, 78 [ACN + 2H2O + H]+, 83 [2ACN + H]+, 149 (protonated phthalic anhydride), and 205 (protonated dibutyl phthalate minus butanol, [149 + C4H8]+) generated from ACN/water/HCOOH (50: 50:0.1 vol %) are given in Figure 3. The (small) cluster background ions involving acetonitrile molecules and the fragment background ions from contaminants (such as m/z 205, etc.) often react with DMDS to form products at m/z 141. The proton affinity of DMDS is about 195 kcal/mol,30 which may play a role in formation of (29) Jarvis, M. J. Y.; Koyanagi, G. K.; Zhao, X.; Bohme, D. K. Anal. Chem. 2007, 79, 4006-4012.

m/z 141 and in governing reactions of clustering with DMDS or charge/proton transfer. Some isobaric background ions such as that at m/z 83 in Figure 2A ([AcOH + Na]+) and Figure 3C ([2ACN + H]+) undergo significantly different reactions with DMDS due to their different structures. The data presented here are the results of reactions in Q2 of a triple-quadrupole mass spectrometer. The translational energy (Elab) of ions entering Q2 is a few electron volts. A study on some of the reactions mentioned above is presented by Jarvis et al. who have used a selected ion flow tube (SIFT) system that permits kinetic studies under carefully controlled, thermal equilibrium conditions.29 For example, the ion [2ACN + H]+ does react with DMDS in Q2 of our triple quadrupole, but a very low reactivity is observed in the SIFT instrument. Reactions of Analyte Ions with Dimethyl Disulfide and a Test on a Variety of Compounds Encountered in LC-MS. In contrast with extensive reactions of DMDS with the majority of background ions as shown in the previous section, there is a generally low reactivity of DMDS toward protonated analytes. Typical mass spectra of the reactions of the isolated ions of protonated prazepam and doubly protonated peptide angiotensin II are given in Figure 4. Under the experimental condition, only 1% of protonated prazepam reacts with DMDS to form products at [M + H + DMDS]+ together with some minor dissociation products [M - C4H8]+ due to collisions with DMDS. [M + 2H]2+ from angiotensin II does not react with DMDS at all. A brief summary of a test of protonated molecules of a variety of compounds that are encountered in LC-MS analysis is given in Table 1, which includes compounds with different functional groups, proton affinities (basic to acidic compounds), and molecular weights. Data on a wide range of compounds, including some peptides, are available as Supporting Information. As a general (30) Hunter, E. P.; Lias, S. G. Proton Affinity Evaluation. In NIST Chemistry WebBook, NIST Standard Reference Database; Linstrom, P. J., Mallard, W. G., Eds.; National Institute of Standards and Technology: Gaithersburg, MD, March 2003; No. 69 (http://webbook.nist.gov).

Analytical Chemistry, Vol. 79, No. 11, June 1, 2007

4017

Table 1. Reactions of DMDS with Various Types of Analyte Compounds

a The abundance of reaction product ions relative to the precursor ion (percent). b The underlined ones are due to dissociation rather than adduct formation.

conclusion, reactions of analyte ions [M + H]+ tested so far result in no or very little products, which indicates that reactions of DMDS with background ions are very exclusive. Only a few analytes [M + H]+ demonstrate a reaction efficiency of more than 15%, including the dissociation of some fragile analyte ions (such as [M + H]+ of simazine and omeprazole). The only protonated analyte found so far to react with DMDS to form [M + H + 4018 Analytical Chemistry, Vol. 79, No. 11, June 1, 2007

DMDS]+ with 20% relative abundance is nicotinic acid ([M + H]+, m/z 124). Besides the protonated analytes, other ions from some analytes such as sodium adducts, protonated dimers, and analogues of analytes are also studied. In accordance with the efficient reactions of some sodiated chemical background ions, all sodium adducts of analytes studied so far react efficiently (65-75% relative

abundance) with DMDS to form [M + Na + DMDS]+. The interesting difference in reactivity between sodiated and protonated analytes may provide an effective method for selective removal of unwanted sodium adducts of analytes in LC-MS. In addition, the protonated dimers of a few analytes dissociate during interaction with DMDS. The question arises why background ions and analyte ions are so different in their reactivity toward DMDS. One explanation is that background ions are for the most part cluster ions and fragment ions formed by mild collision-induced dissociation (CID) in the atmospheric pressure to vacuum interface of the API source. The [M + H]+ ions of analytes tested so far are different and more stable chemical entities, compared with background ions. A clear understanding of the difference in reactivity is yet to be elucidated. The application of exclusive reactions of DMDS with background ions is discussed in the following sections. Instrumentation for the Reduction of Chemical Background Noise. As shown in Figures 1-3, reactions of DMDS with chemical noise ions result in products with mass-to-charge ratios shifted to either higher or lower ones than the original ions concerned. Without removing them, these product ions will still interfere MS or MS/MS analysis at other m/z values. It is very desirable to remove these products derived from chemical noise. The DRC20 used in ICPMS to remove reactive chemical noise is a pressurized rf-dc quadrupole band-pass filter with low resolution. The transmission bandwidth is several m/z units wide. It can be scanned in concert with the analyzing quadrupole and transmits ions that have not reacted but rejects all product ions that fall outside the transmission bandwidth. Although such a band-pass filter20,31 is not available on the triple-quadrupole mass spectrometer used in our study, it is in fact possible to perform ZNL scans to remove the ions which change their m/z during reactions/collisions with DMDS. Zero neutral loss scans in tandem mass spectrometry have not been used in real applications. However, it has been indicated in the literature that a combination of ZNL with CID on a triple-quadrupole mass spectrometer32 may reduce the intensities of dissociable chemical noise ions if analyte ions can survive the collision energy needed to fragment noise ions. In our study changes of m/z of background ions are the result of exclusive reactions with a reagent. Therefore, the ion energy in the collision cell can be kept relatively low in order to prevent dissociation of analyte ions. After a mixture of isobaric reactive and unreactive ions has passed through the first quadrupole mass analyzer (Q1), the reactive ones will react with the neutral reagent present in the rf-only collision cell to form some products at a different m/z value. The unreactive ions remain unaffected. Then the second quadrupole mass analyzer (Q3) allows only the ions, which retain the original m/z, to go through it. The reactive ions will be filtered out because of the m/z shifts after reactions with the reagent gas. A comparison of typical ZNL mass spectra of pure chemical background noise before and after introducing DMDS into the collision cell is shown in Figure 5. In the normalized spectra, a significant reduction of the intensities of the background ions is observed in Figure 5B when DMDS is used to react compared to that in Figure 5A without using DMDS (instead, N2 was used to (31) Tanner, S. D.; Baranov, V. I. U.S. Patent 6,140,638, 2000. (32) Hager, J. U.S. Patent 6,700,120, 2004.

Figure 5. Comparison of zero neutral loss (ZNL) electrospray mass spectra of pure chemical background derived from ACN/H2O/TFA ) 50:50:0.1% (A) before introduction of DMDS into the collision cell (pressurized with N2 instead); (B) after introduction of DMDS into the collision cell. The partial pressure of N2 and DMDS is 0.7 × 10-5 torr. Partial pressure is defined as the increment of the ion gauge readout, see the Experimental Section. The two spectra are normalized to the base peak in (A).

compensate the total pressure in the collision cell for a reasonable comparison). It is concluded that the change of mass spectra is due to the efficient reactions of DMDS with the noise ions. Furthermore, the removal of the majority of chemical background ions also results in a dramatic drop of the level of chromatographic baseline in LC-MS, which is advantageous for the detection and isolation of trace components in LC-MS. It is essential to investigate further how analyte ions behave during the noise removal process. A comparison of typical ZNL mass spectra of midazolam ([M + H]+, m/z 326) is shown in Figure 6 obtained without (Figure 6A) and with (Figure 6B) the use of the neutral reagent gas DMDS to reduce chemical noise. It is clear that the application of DMDS results in a selective reduction of background noise and leaves the analyte ions mostly unaffected because protonated midazolam does not react with DMDS (the mass spectrum of the reaction products is not shown). Similarly for the tested compounds summarized in Table 1 (a full list of the analytes is available as Supporting Information) a selective removal of chemical background noise can be observed using DMDS. Although the test does not provide a comprehensive list for all compounds encountered in LC-MS, the observation of exclusive reactions of DMDS with background ions so far indicates the generality of the application of the technique to reduce chemical noise in LC-MS analysis. Reduction of Chemical Background Noise in Mass Spectra and Chromatograms in LC-MS. A typical example of the application of the noise reduction technique is given in Figure 7 for turbo ionspray LC-MS analysis of a mixture of four pharmaceutical compounds: nicotinamide, etamivan, flunitrazepam, and testosterone. The mixture was diluted on purpose to a concentration where the detection of some components was seriously impaired by abundant chemical background noise ions. It is important to note that in Figure 7 the ZNL scan mode was used for the background reduction acquisition using DMDS (Figure Analytical Chemistry, Vol. 79, No. 11, June 1, 2007

4019

Figure 6. Comparison of zero neutral loss (ZNL) electrospray mass spectra of diluted midazolam ([M + H]+, m/z 326) (A) before introduction of DMDS to the collision cell (pressurized with N2 instead); (B) after introduction of DMDS to the collision cell. The partial pressure of N2 and DMDS is 0.7 × 10-5 torr. Partial pressure is defined as the increment of the ion gauge readout, see the Experimental Section. The two spectra are normalized to the base peak in (B).

7D-F) and that Q3 single MS scans were done for the standard acquisition (Figure 7A-C). Approximately a loss of 2-3 times in signal intensities for all ions is observed due to transmission losses in ZNL mode. The LC-MS base-peak chromatogram before using DMDS (N2 was used instead to compensate the pressure for comparison) is shown in Figure 7A, where only two out of the four compounds were detected. The low-abundant components nicotinamide (eluting at the retention time of 2.10 min) and testosterone (17.45 min) were not detectable because they were less abundant than the base-peak chemical background ions as indicated in Figure 7C for the mass spectrum of testosterone. Base-peak chromatograms are often used to reveal the trace components in LC-MS analysis to localize/identify unknown species. This is mainly to prevent the significant contribution of chemical background ions that would otherwise give a high baseline in total ion current (TIC) chromatograms and may totally overshadow the appearance of those low-abundant analytes. When DMDS was introduced to reduce chemical background noise, a significant improvement of the base-peak chromatogram is shown in Figure 7D. The fluctuating baseline before the noise reduction (Figure 7A) becomes a relatively flat line after the noise reduction. The level of the baseline drops to 1/24 of that in Figure 7A. As a result, the two minor components nicotinamide and testosterone were detected at 2.10 and 17.45 min, respectively, with sufficient S/N ratios, in contrast to the analysis without the chemical noise reduction. The overall improvement of the chromatographic S/N ratios for all components is at least 5-7 times in this case. This is due to the selective removal of reactive chemical background ions during reactions with DMDS, which also results the changes of the background mass spectra (from Figure 7 part B to part E) and that of analytes (from Figure 7 part C to part F). In our preliminary study, both the TIC and the level of the baseline in base-peak chromatograms are reduced by at least a factor of 10 and the S/N ratio is improved by a factor of 5-20. 4020 Analytical Chemistry, Vol. 79, No. 11, June 1, 2007

In an automatic identification or screening process with LCMS it is important to trigger a tandem MS/MS scan to acquire further information on structures. These MS/MS experiments are performed on the base peak or the most abundant ions. However, if the intensities of the trace components are already lower than that of the major (base peak) chemical background ions in a mass spectrum, these minor ions will not be identified and picked up for a further MS/MS experiment. Other effects of chemical noise include isobaric interference and limitation on the S/N ratio. It has been shown in this study that in all cases using the noise reduction technique it is possible to detect more minor components than in a conventional LC-MS approach. Besides DMDS, several compounds were tested briefly with regard to their suitability as an alternative reagent gas for chemical noise reduction. This includes diethyl disulfide, methyl propyl disulfide, dimethyl trisulfide, 1,3-dioxolane, ethylene oxide, and butadiene monoxide. Most of them are eliminated for a further study because of low reactivity with chemical background ions, lower volatility, or dramatic loss of ions during the interaction with analyte ions. Ethylene oxide and butadiene monoxide show a similar reaction efficiency with background ions as DMDS does. However, when preliminarily tested on the reaction selectivity with some protonated analytes, both of them react to form products with a 20% relative intensity, which indicates more loss of analyte ions compared to DMDS. Although the lower reaction selectivity toward noise ions perhaps is a limitation for its application, it is worthwhile to carry out a further investigation. CONCLUSION A novel technique33 is reported for the reduction of chemical background noise in API LC-MS, based on exclusive gas-phase ion-molecule reactions of background ions with DMDS. Our study demonstrates that major chemical background ions react efficiently with the neutral reagent DMDS, and not with the vast majority of the protonated analytes tested so far, which indicates that a generic application is feasible. Apparently, significant structure characteristics of the persistent chemical background interferences compared to that of the vast majority of conventional protonated analytes, such as peptides, proteins, pharmaceuticals, and so on, result in the exclusive reactions. In addition, the novelty of the method lies in the discovery of a gas-phase reagent that could make this possible in combination with band-pass filters or mass spectrometric scanning. The technique may improve S/N ratios in all areas of quantitative and qualitative LC-MS trace analysis. Nowadays some software techniques (such as dynamic background subtraction11) are used, but because our new approach reduces chemical noise before detection occurs, it should provide improvements in addition to the software methods. In other words the use of this technique along with data processing methods should provide results better than either alone. Also this approach will work in situations where LC is not used as a means of sample introduction (nanoESI infusion-type methods) where background subtraction methods do not work because there are no analyte-free regions in the data from which to derive the background spectra. Another advantage over software methods is that, in principle, this technique can remove isobaric interfer(33) Guo, X.; Bruins, A. P.; Covey T. R. Provisional U.S. Patent Appl. No. 60/ 765809, 2006.

Figure 7. Comparison of turbo ionspray reversed-phase LC-MS analysis of a diluted mixture of four pharmaceutical compounds (A-C) before and (D-F) after introducing DMDS into the collision cell. Significant changes are observed between (A and D) base-peak chromatograms, (B and E) pure chemical background ions, and (C and F) mass spectra of one of the low-abundant components, testosterone. The mass spectrum E is normalized to the base peak in (B). The four compounds are nicotinamide (2.10 min), etamivan (10.15 min), flunitrazepam (15.40 min), and testosterone (17.45 min), respectively. About 10 ng of each compound was injected. The flow rate of 200 µL/min was split to 80 µL/min into the mass spectrometer. The partial pressures of N2 and DMDS were 0.7 × 10-5 torr. Partial pressure is defined as the increment of the ion gauge readout, see the Experimental Section.

ences when there are sufficient differences in ion chemistry. Although the technique has been implemented into both a commercial and a modified triple-quadrupole LC-MS to demonstrate its robustness and generality, an optimized performance is expected on a dedicated instrumentation where tandem MS can also be realized.

SUPPORTING INFORMATION AVAILABLE

ACKNOWLEDGMENT This study is the result of a research contract between Applied Biosystems/MDS Sciex (Concord, ON, Canada) and the University of Groningen.

Received for review December 14, 2006. Accepted March 26, 2007.

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

AC062365T

Analytical Chemistry, Vol. 79, No. 11, June 1, 2007

4021