Liquid chromatography^tandem mass spectrometric analysis of surface and waste. (LC)^atmospheric pressure chemical ionisation

Transcription

1 trends in analytical chemistry, vol. 19, no. 4, Liquid chromatography^tandem mass spectrometric analysis of surface and waste water with atmospheric pressure chemical ionisation I: instrumentation Paul G.M. Kienhuis*, Rene B. Geerdink RIZA, P.O. Box 17, 8200 AA Lelystad, The Netherlands Liquid chromatography (LC) coupled to tandem mass spectrometry (MS /MS), i.e. LC^MS /MS, is a technique for the analysis of medium polarity to ionic compounds. This and the following paper highlight the potential and limitations of atmospheric pressure chemical ionisation (APCI) combined with MS /MS in the analysis of water. Part I describes the APCI interface and the scan modes of a triple quadrupole mass spectrometer. It gives a general introduction to the instrumentation, from a practical point of view. Part II gives more detailed information, in which the instrumentation is optimised for and used in the analysis of surface water and waste water. Attention is paid to the use of methanol versus acetonitrile as organic modi er in combination with APCI, the application of the RF-only daughter scan mode, and the quantitation of known compounds and the identi cation of unknown compounds found in surface and waste water. z2000 Elsevier Science B.V. All rights reserved. Keywords: Liquid chromatography^tandem mass spectrometry; Thermospray; Tandem mass spectrometry; RFD scan mode; Instrumentation; Atmospheric pressure chemical ionization; Daughter scan mode; Collisionally induced dissociation 1. Introduction *Corresponding author In order to analyse surface and waste water at the Institute for Inland Water Management and Waste Water Treatment (RIZA) in The Netherlands, several methods based on mass spectrometric detection are used to cover the range from volatile to medium polar compounds. The volatile ones (with a target list of about 60 compounds) are analysed by on-line purge and trap concentration followed by gas chromatography (GC)^mass spectrometry (MS) detection. The less volatile compounds (about 400) are analysed by off-line extraction on XAD-4, or liquid^liquid extraction with dichloromethane, followed by GC^MS detection. The thermolabile or medium polar compounds (about 70) are analysed by on-line SPE extraction or large loop injections, followed by liquid chromatography (LC)^atmospheric pressure chemical ionisation (APCI)-tandem MS (MS /MS) detection. The aim of each method is to concentrate or extract as many compounds as possible, to quantify the target compounds and to identify the unknown peaks. This article highlights the LC^APCI-MS /MS analysis. We started in 1990 with a combination of LC^ UV^Thermospray-MS /MS (on a triple-stage quadrupole (TSQ)-70 instrument, Finnigan MAT) and subsequently purchased an atmospheric pressure ionisation (API) interface in 1996 in order to improve the ionisation of the more polar and ionic compounds. API has largely replaced the thermospray interface, because of its ease of operation, the possibility of ionising large molecules, the improved ability to handle higher ow-rates, the stability of the ionisation process, the possibility of generating fragment ions on a single quadrupole mass spectrometer, and the introduction of relative inexpensive `desktop' instruments. The API interface can be used in two different ionisation modes: APCI which, just like thermospray, is suitable for the neutral compounds, and electrospray ionisation /00/$ ^ see front matter ß 2000 Elsevier Science B.V. All rights reserved. PII: S (99)

2 250 trends in analytical chemistry, vol. 19, no. 4, 2000 (ESI), which is suitable for the ionic compounds [ 1^4 ]. Both thermospray and API are soft ionisation techniques. In the mass spectra, ions related to the intact molecule of a compound are observed mainly, ionised by detachment [M3H] 3 or attachment of a proton [M+H ], ammonia ([M+NH 4 ], ammoniated molecule) or other cations or anions. However, in order to identify a compound and to improve the reliability of an analysis, more speci c ions are necessary [ 5 ]. Fragmentation by means of collisionally induced dissociation (CID) is, at the moment, the most common way to get additional ions. CID can be performed either internally in an API source ( by a process called source CID [ 6 ] or cone voltage fragmentation [ 7 ]) or in a tandem mass spectrometer. This article is divided into two parts, and describes the design, optimisation and application of the LC^APCI-MS /MS instrumentation. Part I gives a general introduction and rules-of-thumb to optimise the APCI interface and the scan modes and settings of a triple quadrupole mass spectrometer. Detailed discussions on the instrumentation, the theoretical principles, and mechanisms can be found elsewhere [ 7^10 ]. Part II gives more speci c information concerning the application of the instrumentation to the analysis of surface and waste water. As a reference, the method used in combination with thermospray is discussed [ 11], followed by the problems (such as the choice of the organic modi er) encountered while changing from thermospray to APCI. Attention is paid to the use of several scan modes, the analysis and identi- cation of target compounds, of compounds often found in surface water, and the so-called CI^CID spectrum library. 2. Ionisation In GC^MS, a low ow of an inert gas (helium) carries the compounds into the ion source. With LC^MS, a relatively large amount of solvent enters the ionisation interface and strongly affects the ion- Fig. 1. In the Finnigan MAT APCI interface, the mobile phase (L) is passed into the atmospheric pressure region (AP) of the interface through a fused silica capillary (150 Wm ID) as sample inlet tube. A sheath-gas (Sh) ow (N 2 ), and eventually an auxiliary (A) ow (N 2 ), is used to nebulise the solvent and to keep the interface dry. The resultant spray is heated by a heater block (V, vaporiser) and sprayed against a spray shield. The nebulised solvent and analyte molecules are partly ionised by a corona discharge (D). The extent of solvent-vapour mixture in the interface is directed to waste, through two tubes (W) into a hood. The droplets formed in the tubes are combined in a ask. To prevent back-pressure, a large-bore tube and ask (which has been used previously in our laboratory as a cold nger in the thermospray exit line) is applied. Some other users apply a small suction pump to the drain connection. The heated capillary (C) samples the atmospheric-pressure region of the interface continuously. Ions and neutrals in the gas or liquid phase are drawn into the capillary and transported to the reduced pressure (RP) region by a reducing pressure gradient. The capillary is heated to reduce the amount of solvent. A tube-lens (T) directs the ions towards the skimmer (Sk) ori ce, which is off-axis of the capillary in order to reduce the amount of aerosol particles entering the high-vacuum (HV) region of the mass spectrometer. Between the skimmer ori ce and the rst mass lter, an octapole (O) is used to focus and transport the ions.

3 trends in analytical chemistry, vol. 19, no. 4, Table 1 Overview and general instrumental settings of the more important parameters that will in uence the sensitivity of the Finnigan MAT APCI interface Parameters Instrumental settings Mobile phase parameters % Organic modi er Type of modi er Other eluent constituents such as salt ( molarity), acidity Instrumental parameters Flow-rate Up to 2 ml / min; generally 0.1^1 ml / min Corona current 4^6 WA Sheath-gas ow-rate 10^100 psi Auxiliary gas ow-rate Normally not used Vaporiser temperature 350^425³C Heated capillary temperature 170^200³C Capillary voltage Dependent on most of the other parameters Tube-lens voltage Strongly dependent on most of the other parameters isation. The way in which the eluent composition affects the ionisation process is different for thermospray and APCI. Thermospray ionisation is performed in the highvacuum envelope of the mass spectrometer. To prevent freezing at the ori ce, the eluent is heated in a vaporiser before entering the source. To obtain the best response, the extent of heat must be such that only part of the solvent has been evaporated before leaving the vaporiser as a spray of droplets. Evaporation of the droplets continues in the high vacuum of the source whereas ion^molecule reactions are taking place during evaporation and in the gas-phase. The ion residence time in the ion source is short, i.e. of the order of Ws [9,10]. APCI is performed at atmospheric pressure. The heat, necessary to evaporate the eluent, is now admitted after nebulisation of the eluent. Ions are formed by means of a corona discharge. The ion residence time in the source is much longer, of the order of ms. Ion^molecule reactions in the gas-phase dominate the ionisation process [ 9,10,12,13 ]. These differences in ionisation and reaction time result in different in uences of the analyte and solvent properties on the ions formed during ionisation. The Finnigan APCI interface has a heated capillary to guide the ions into the high-vacuum envelope of the mass spectrometer (Fig. 1). The settings discussed below are speci c for this interface: other types of API interfaces may act in different ways. The rst choice to be made is in the selection of the eluent composition (Table 1). It is important to keep in mind that ionisation via APCI is chemistry in the gas-phase. The process starts with the generation of N 2 and O 2 3 ions by means of a continuous corona discharge. These ions react with the solvent and analyte molecules. Owing to the relatively long residence time in the APCI source, the ions measured are the result of an equilibrium in the gasphase. The eluent composition affects this equilibrium; for example, more compounds can be ionised in the positive-ion mode when methanol is used as organic modi er instead of acetonitrile. An explanation can be found in the proton af nity of the compounds involved (Table 2). The proton af nity of methanol is lower than for acetonitrile, so more analytes can be ionised by proton attachment in the presence of methanol. The proton af nity, as such, does not fully explain the ionisation process, but it is important to realise that solvents and compounds in the eluent can affect the ionisation process. Therefore, in the analysis of positive ions, it is important that the components of the applied solvent composition have a lower proton af nity than the analyte(s) to be measured. On the other hand, Table 2 Proton af nity and gas-phase acidity of LC^MS-related compounds [ 14 ] Compound Proton af nity in kj / mol Ammonia Acetic acid Acetonitrile Methanol Formic acid Water Gas-phase acidity in kj / mol

4 252 trends in analytical chemistry, vol. 19, no. 4, 2000 Table 3 In uence of several instrumental parameters on the ion current of m/z215 [M+H ] of metribuzin continuously infused at a concentration of 200 ng / ml Flow-rate (ml/min) Heated capillary (³C) Sheath-gas pressure (PSI) Capillary voltage (V) Tube-lens voltage (V) m/z215, countsu After each change of the ow-rate, temperature of the heated capillary, or the sheath-gas ow, the intensity of m/z215 is shown before and after the capillary and tube-lens voltages have been optimised. Mobile phase methanol /water (50% v /v, 1 ml acetic acid /l). Vaporiser 450³C, No auxiliary gas ow. when a target compound has a high proton af nity (in general, nitrogen-containing compounds [ 12,13 ]), it might be useful to consider acetonitrile in order to reduce the ionisation of interfering compounds. The choice of a solvent also affects the amount of internal energy an ion gains during ionisation, and thus the degree of CID fragmentation later on [ 15 ]. In the negative-ion mode, the gas-phase acidity affects the ionisation process. An increased tendency of a compound to ionise by detachment of a proton is related to a lower value of the gas-phase acidity. Therefore, when a lower ph of the high performance liquid chromatography (HPLC) eluent is necessary for separation of acidic compounds, the S /N ratio is better if acetic acid is used to acidify the HPLC mobile phase instead of the stronger formic acid, although the difference in gas-phase acidity is small (Table 2). Examples of these equilibrium phenomena will be given in Part II. The Finnigan APCI interface can handle a maximum ow-rate of about 2 ml / min. Flow-rates between 0.1 and 1 ml / min are generally used. No large differences in sensitivity are observed at different ow-rates after optimisation of all parameters at each ow-rate. It is important to know that almost all the instrumental parameters indicated in Table 1 are dependent on the solvent ow-rate, except for the corona discharge current. The auxiliary gas ow (N 2 ) is generally not necessary. For many compounds it has a large effect, with `no auxiliary gas' as the best option. It is also useful to optimise the auxiliary gas ow in the LC mode (where the compound is only in the ion source for a short time) and not in the FIA mode (where the compound is continuously available). The sheath-gas (N 2 ) pressure is more important, and should be increased at increasing ow-rates. The whole pressure-gauge range of 0^100 psi has to be used to optimise the signal intensity and stability. The optimum values for the capillary and tubelens voltage are affected by most of the parameters, but especially by the vaporiser and capillary temperature. The optimal range of the tube-lens voltage is rather small, and must be checked after each change of a setting. An example of the in uence of some parameters on the signal intensity of m/z215 of metribuzin is shown in Table 3. On the left side of the table, the parameters: ow-rate, capillary temperature and sheath-gas pressure, are shown. In the middle are the applied capillary and tube-lens voltages, and to the right the resulting signal intensity of m/z215. At the chosen parameter settings of the rst line, a signal intensity of 75U10 6 is reached after optimisation of the voltages. On the second line, the ow-rate has been changed to 0.5 ml / min. A signal intensity of 37U10 6 counts is reached at the same voltages as in the rst line, but the intensity increases to 40U10 6 in the third line after optimisation of the capillary voltage and the tube-lens voltage. In the same way, the sheath-gas ow is changed at lines 4^5, and the capillary temperature at lines 6^7. Every time, the signal intensity before and after the optimisation of the capillary and the tube-lens voltages are shown. The optimum values for the vaporiser and capillary temperature (after optimisation of the capillary- and tube-lens voltage) are rather at, and less ow-rate dependent than the sheath-gas ow. A vaporiser temperature of 350^425³C and a heated capillary temperature of 170^200³C are common

5 trends in analytical chemistry, vol. 19, no. 4, Fig. 2. Scan modes of a triple quadrupole mass spectrometer. settings, but for speci c compounds, a higher capillary temperature may be favourable [ 16 ]. Garcia et al. [ 17 ] reported the optimisation of 10 variables of the Finnigan APCI interface in the positive-ion mode for ibuprofen, 2-(4-isobutylphenyl)propionic acid, as the target compound, using fractional factorial- and response-surface modelling. Their results differ from the settings shown in Table 1. They found a ow-rate of 0.1 ml / min or lower, in combination with a high water content, a high auxiliary gas ow, a low tube-lens voltage of 55 V, and a heated capillary temperature of 225³C, to be optimum. As indicated by the authors, these optimised parameters might be speci c for ibuprofen. Schaefer and Dixon [ 18 ] investigated the quenching effect of several common buffers on the ionisation of ibuprofen and other model compounds, in the negative-ion mode with APCI on a Sciex triple quadrupole mass spectrometer. The

6 254 trends in analytical chemistry, vol. 19, no. 4, 2000 results show the quenching effect of the more strongly acidic buffer components on the weakly acidic compounds. They advise the use of the base, N-methylmorpholine, as buffer component in the analysis of weakly acidic compounds. Spliid and Koppen [ 19 ] described the optimisation of the Finnigan APCI interface for a group of pesticides at a ow-rate of 0.2 ml / min and with methanol as organic modi er. The optimisation of the inlet parameters such as capillary temperature, vaporiser temperature, corona current and sheathgas pressure was discussed and the results are comparable with our own ndings. In our laboratory, the mass spectrometer, including the analyser part, is serviced twice a year. After each service, the whole instrument has to be set properly. In order to tune and calibrate the mass spectrometer in combination with the APCI interface in the desired mass range (10^700 m/z), a mixture of analytes at a concentration of 200 Wg/l is used both for the positive and negative-ionisation mode (Table 4). The mixture is continuously introduced at a ow-rate of 0.5 ml / min at the optimal APCI settings speci c for the mixture. The resulting general tune- le is used as starting point to optimise the source parameters for the different applications for which the mass spectrometer is used. In this way, for each application, a tune- le is created which is speci c for each application at the same ow-rate, solvent composition, and temperature settings as applied in the analysis. After tuning the APCI with the analyte mixture, no contamination of the inlet system was seen after cleaning of the spray shield and ushing of the pump, tubing and heated capillary with methanol: water (50/50, v /v). Tuning and calibration of the whole instrument are normally only performed after cleaning of the whole analyser part of the TSQ, and proved to be very stable. The source (vaporiser, corona needle, heated capillary, skimmer and rst lenses) is cleaned more often. After cleaning, the source parameters are checked. Generally, no large differences in settings are found after cleaning. 3. Fragmentation With electron-impact ionisation, both ionisation and fragmentation of the molecules occur at the same time. The electron beam `hits' the molecules. An electron is lost, but the total process is so energetic that most of the ionised molecules fragment. These fragment ions are essential to con rm the identity of a compound [ 5,20 ]. Soft ionisation processes are less energetic and produce mainly ions containing the intact molecule. Additional processes are required to produce fragment ions. CID is the most common way to do so. The ions are accelerated to increase their kinetic energy and pass through a chamber with a collision gas. In an API interface, CID can be performed in the API source [ 6,7 ]. In a triple-stage quadrupole mass spectrometer, a quadrupole or octapole acts as the collision chamber, with an inert gas as collision gas. Parent ions enter the collision chamber and the fragment ions generated are called daughter ions Source CID By the application of an additional voltage in the region between the heated capillary and the skimmer, or in the region between the skimmer and the octapole, the ions are accelerated (Fig. 1). Nitrogen from the sheath-gas and solvent molecules acts as collision gas. Source CID offers the possibility of fragmenting ions in combination with a single quadrupole mass spectrometer, but only one `mode' is available. All ions are involved in the CID; they cannot be selected as in a tandem mass spectrometer. Table 4 Tune masses of the mixture used to tune and calibrate the mass spectrometer in the positive- and negative-ion modes Compounds Positive ions m/z Negative ions m/z Ethylenethiourea [M+H ] (ETU) DNOC [M3H] 3 Metribuzin [M+H ] Warfarin [M3H] 3 Tris(2-butoxyethyl) [M+H ] phosphate Fluazinam [M3H] 3 Reserpine [M+H ] [M3H] 3 Ions of the solvent: Water [H 2 O] Acetic acid [M3H] 3 A mixture of seven compounds is used at a concentration of 200 ng /l each, dissolved in methanol /water (50:50, v /v) with 1 ml acetic acid /l.

7 trends in analytical chemistry, vol. 19, no. 4, Fig. 3. Quadrupole offset voltages in the MS /MS modes. The offset voltage of a scanning quadrupole is about 3^5 V. Therefore, after leaving the source, the ions will gain an additional kinetic energy of 3^5 ev before passing the quadrupole. In the MS /MS modes, especially, the offset voltage of Q3 is important. Graph A shows a situation for positive ions without collision gas. The ions are accelerated to an additional kinetic energy of 20 ev before passing the collision cell, and reduced back to 5 ev by the offset voltage of Q3. In Graph B, collision gas at a low pressure of about 0.5 mtorr argon is present in the collision cell. The kinetic energy of the ions will be reduced by one or some collisions resulting in a divergent and insuf cient remaining energy to pass Q3 properly. To compensate for this effect, the offset voltage of Q3 must be increased ( Graph C). For the different MS /MS scan modes, Eq. 1 DOFF ms=ms ˆ DOFF ms COFF 13 M d =M p MSMSC=100 Š 1 is used by the instrument to calculate the offset voltage of Q3, where DOFF ms=ms represents the offset voltage of Q3 in the MS /MS scan modes; DOFF ms represents the offset voltage of Q3 in the Q3MS scan mode; COFF the applied collision-offset voltage; M d the m/zof the daughter ion; M p the m/zof the parent ion, and MSMSC the MSMS correction factor. TheMSMSCrangesfrom0to100anddependsonthepressureandcompound[ 26 ]: acommonvalueatalowgaspressureisabout70. At higher collision gas pressures, the ions will have multiple collisions and therefore their remaining kinetic energy is much lower. In this situation (Graph D), the offset voltage of Q3 is added to the actual collision offset voltage using an MSMSC=0, resulting in Eq. 2: DOFF ms=ms ˆ DOFF ms COFF 2 In the RFD scan mode, an MSMSC = 0 must be used because the cut-off mass value is used by the TSQ as M p, resulting in unrealistic offset voltages. As an example, in the positive-ion mode: DOFF ms = 35V,COFF=320 V, M p = cut-off mass = 70, M d = 200 m/zand MSMSC = 70. The resulting DOFF ms=ms will be +15 V and the positively charged daughter ions cannot pass Q3. The voltage difference can be changed from scan to scan. This can be used to cover a voltage range to generate a whole range of fragment ions. Josephs [ 21] uses this option to acquire spectra of drugs at CID offset voltages of 0, 320, 330 and 340 V on alternate scans. The averaged spectrum of a peak is used to build a spectral library, which is then used for automated

8 256 trends in analytical chemistry, vol. 19, no. 4, 2000 Fig. 4. In uence of the collision-gas (argon) pressure on the peak performance of atrazine ( m/z216/218) at 0, 1, 2 and 5 mtorr. Collision offset 320 V, MSMSC = 0.

9 trends in analytical chemistry, vol. 19, no. 4, library searching of peaks detected in unknown formulations TSQ scan modes The TSQ is a tandem mass spectrometer with three quadrupoles (Fig. 2) in-line [ 22 ]. A quadrupole can act as a mass analyser by a combination of AC (RF range, about 1 MHz) and DC voltages [ 23 ]. If only an RF voltage is used, all ions above a certain m/zratio (the so-called `cut-off mass') are not only allowed to pass the quadrupole, but are also focused. This focusing property is used in the central quadrupole (Q2), which is pressurised with a collision gas and used as the collision chamber. The extent of fragmentation depends on the internal energy of the ions gained during ionisation, the voltage difference between the ion source and the collision chamber (collision offset), and the pressure and the properties of the collision gas. Typically, argon is used as collision gas. Recent versions of the TSQ use an octapole instead of a quadrupole to improve the focusing effect. If next to the RF an additional DC voltage is applied, the quadrupole is able to act as a mass analyser or mass lter. It actually works as a bandpass lter: only ions of one particular m/zratio are allowed to pass. All other ions will collide on one of the rods. At a resolution of one m/z(nominalmass resolution, peak-width at half height is 0.65^0.7 mass units), about 10% of the ions of the selected m/zvalue will pass the quadrupole The rst and third quadrupoles can arbitrarily be used as an RF-only quadrupole or as the mass lter, resulting in several single MS and MS /MS scan modes. In Fig. 2, most of the possible combinations are shown. In the single MS scan modes, either the rst (Q1MS) or the third quadrupole (Q3MS) acts as a mass lter, and no collision gas is admitted to the collision cell. The result resembles a single quadrupole system. The Q3MS gives a better sensitivity compared to Q1MS, because of the focusing effect of Q1 and Q2 [ 24 ]. In the RF-only daughter (RFD) scan mode, Q1 is operated as an RF-only device, the collision cell is pressurised with collision gas, and Q3 acts as a mass lter. All ions above a pre-set cutoff m/z are allowed to enter the collision cell and are allowed to fragment. Unfragmented parent ions are scanned together with the generated fragment (daughter) ions. This scan mode (SRM) can be compared to a single quadrupole mass spectrometer with a separated ionisation and fragmentation chamber. In the RF-only parent scan mode (not shown in Fig. 2), Q1 is used as mass lter, Q2 is pressurised and Q3 acts as an RF-only quadrupole. The ion current measured for each m/z selected by Q1 is acquired for that speci c mass. If no ions are lost in the collision cell, the resulting mass spectrum should be identical to a spectrum acquired in the Q1MS scan mode. No useful application for this scan mode is reported in the literature. In the MS /MS scan modes, both Q1 and Q3 are used as mass lters. The most common and speci c combination is the daughter-ion scan mode (DAU). Here, Q1 is set on a xed m/z value and the selected ions are passed into the collision cell. ThedaughterionsarescannedwithQ3atanappropriate scan range. In the parent-ion scan mode (PAR), Q1 is scanning and Q3 is set on a xed m/ z value. The result is a spectrum of all ions producing the speci c daughter ion. As an example, m/z 149 is used to search for phthalates. In the acquired spectrum, all parent ions producing m/z 149 are shown. In the neutral-loss scan mode (NEU), both Q1 and Q3 are scanning, but with a constant m/z difference. The result now is a spectrum of all parent ions losing a speci c neutral part in the collision cell. The selected reaction monitoring mode (SRM) can be compared with the selected ion monitoring mode of a mass spectrometer with a single mass analyser. By analysing only a selected number of ions, instead of a mass range, the analysis time for each ion is increased, improving the S /N ratio. This mode is very useful for analysing a limited number of target compounds. The Q1 selects the parent ions one by one, and the speci c daughter ions of each parent ion are acquired in Q3. It is also possible to adjust the collision offset for each single reaction to achieve the best response. The speci city of this mode is often used to reduce, or even skip, the chromatographic part of the analysis [25] Collision-gas pressure and collision-offset voltage In order to achieve an effective CID process, three parameters must be set: õ A collision gas must be present in Q2 at the right pressure.

10 258 trends in analytical chemistry, vol. 19, no. 4, 2000 Fig. 5. In uence of argon pressure on the signal intensity of atrazine at a low and a medium collision-offset voltage ( m/z 216, [M+H] at 37 V,andm/z174, fragment ion at 320 V). õ õ The ions must have enough kinetic and internal energy to get collisions resulting in fragment ions. The ions entering the third quadrupole must have such a kinetic energy that they can be analysed properly. For the collision-gas pressure, two options are available. Until about 1990^1994, it was quite common to apply a low gas pressure (about 0.5 mtorr), resulting in one or a few collisions per ion [ 26 ]. To produce fragment ions, a difference of 10^50 V is needed in the offset voltage between the source and the collision cell to accelerate the ions between the source and the collision cell. Owing to the low collision-gas pressure, only a few collisions can take place and therefore, most of the fragmented and unfragmented ions leave Q2 with a rather high velocity. In the Q3MS scan mode, the offset voltage of Q3 is set to 3^5 V. Using the same offset voltage in the MS /MS scan modes, the kinetic energy of the ions will be reduced before entering Q3. For example, with a collision offset of 320 V, positive ions will gain an additional kinetic energy of 20 ev between the source and the collision cell. With an offset of 35 V for Q3, the kinetic energy will be reduced by 15 ev, resulting in a remaining energy of 5 ev. Without collision gas in the collision cell, this works very well (Fig. 3A). However, when the cell is pressurised, the ions will lose some of their kinetic energy at each collision and therefore cannot enter Q3 properly (Fig. 3B). To compensate for this effect, a correction factor (MSMSC) is used (Fig. 3C). The actual value of the correction factor, however, depends strongly on the actual collision-gas pressure and the properties of the ions involved, resulting in a detection system which is dif cult to optimise, and not robust [ 27 ]. Nowadays, a higher collision-gas pressure (2^4 mtorr argon) is common. In this pressure range, the offset voltage of Q3 is calculated by adding the offset voltage normally applied in the Q3MS mode (3^5 V) to the actual collision-offset voltage (MSMSC = 0): thus, at a collision offset of 320 V, the offset voltage of Q3 is 323 to 325 V. Now, the collision gas must reduce the kinetic energy so that the ions will pass Q3 with the proper kinetic energy (Fig. 3D). The in uence of the collision-gas pressure on the resolution is shown in Fig. 4. The resolution of m/z216 and 218 of atrazine is shown at various collision-gas pressures, while a collisionoffset voltage of 320 V is applied. In the TSQ-700, a collision-gas pressure of about 1.5 mtorr is suf cient to establish the nominal mass resolution again. At higher pressures, up to 5 mtorr, the reso-

11 trends in analytical chemistry, vol. 19, no. 4, lution does not increase further. The collision gas is also needed to fragment ions. A low collision-offset voltage can be used to analyse the intact parent ions (RFD scan mode) or the higher mass daughters of very unstable molecules (RFD and MS /MS scan modes). Higher collision-offset voltages are needed to fragment the parent ions. The collisiongas pressure affects the signal intensity of the ions at the different collision-offset voltages. At a low collision-offset voltage, a high collisiongas pressure reduces the signal intensity of the ions. In Fig. 5, the signal intensity of m/z216 of atrazine is shown. At the applied collision-offset voltage of 37 V, the ions do not fragment. The signal intensity reduces at increasing collision-gas pressure, because not all ions are able to pass the collision cell at higher gas pressures. The optimum gas pressure is around 1.5 mtorr argon. The minimum gas pressure needed to give a normal resolution is also 1.5 mtorr, so a pressure between 1.5 and 2 mtorr is optimal. At a medium collision-offset voltage, it is better to use a higher collision-gas pressure. In Fig. 5, the signal intensity of m/z 174 of atrazine is shown. At the applied offset voltage of 320 V, m/z 216 fragments to m/z 174. The fragmentation process becomes more ef cient at higher collision-gas pressures, until an optimum is reached at 3^5 mtorr. When both low and higher collision-offset voltages have to be applied in the same run, a collision-gas pressure must be chosen to meet both requirements. A compromise is found between 2 and 3 mtorr. 4. Applications With the instrumentation and general rules described here, a variety of types of applications can be performed. APCI or ESI together with MS / MS are often used, and are very successful in the analysis of a limited number of target compounds in combination with short LC-columns and short analysis times. For these methods, the solvents and systemparameterscanbechosentobeoptimalforthe compounds involved. In the analysis of surface water and waste water, however, choices must be made in order to analyse as many compounds as possible in a limited numberofruns.thewaysinwhichwedealwiththis goal, together with the results of optimisation and analyses, are described in Part II. Acknowledgements The authors thank Wilfried Niessen of Hyphen MassSpec Consultancy, Leiden, The Netherlands, for critically reading the manuscript and for his helpful suggestions. References [ 1] I. Ferrer, D. Barceloè, Analysis 26 (6) (1998) M118. [ 2 ] R.D. Voyksner, Environ. Sci. Technol. 28 (1994) 118A. [ 3 ] A.P. Bruins, Trends Anal. Chem. 13 (1994) 37. [ 4 ] A.P. Bruins, Trends Anal. Chem. 13 (1994) 81. [ 5 ] L.Y.T. Li, D.A. Campbell, P.K. Bennet, J. Henion, Anal. Chem. 68 (1996) [ 6 ] A. Cary, Technical Report No. 620, 7000 Series Finnigan MAT, [ 7 ] B.L.M. van Baar, in: D. Barceloè (editor), Applications of LC^MS in Environmental Chemistry, J. Chromatogr. Lib., Vol. 59, Elsevier, Amsterdam, 1996, p. 71. [ 8 ] P.J. Arpino, Mass Spectrom. Rev. 9 (1990) 631. [ 9 ] W.M.A. Niessen and J. van der Greef, Liquid Chromatography-Mass Spectrometry, Chromatographic Science Series, Vol. 58, Marcel Dekker, New York, [ 10 ] W.M.A. Niessen, in: D. Barceloè (editor), Applications of LC^MS in Environmental Chemistry. J. Chromatogr. Lib., Vol. 59, Elsevier, Amsterdam, 1996, p. 3. [ 11] P.G.M. Kienhuis, R.B. Geerdink, A. Sijpersma, Anal. Methods Instrum. 2 (1995) 236. [ 12 ] J. Sunner, G. Nicol, P. Kerbarle, Anal. Chem. 60 (1988) [ 13 ] J. Sunner, G. Nicol, P. Kerbarle, Anal. Chem. 60 (1988) [ 14 ] S.G. Lias, J.E. Bartmess, J.F. Liebman, J.L. Holmes and R.D. Levin, J. Phys. Chem. Ref. Data 17, Suppl. 1 (1988). [ 15 ] R. Wolf, 13th LC /MS Symposium on Liquid Chromatography, Montreux, [ 16 ] A. Rosell-Meleè, J.F. Carter, J.R. Maxwell, J. Am. Soc. Mass Spectrom. 7 (1996) 965. [ 17 ] D.M. Garcia, S.K. Huang, W.F. Stansbury, J. Am. Soc. Mass Spectrom. 7 (1996) 59. [ 18 ] W.H. Schaefer, F. Dixon Jr., J. Am. Soc. Mass Spectrom. 7 (1996) [ 19 ] N.H.Spliid,B.Koppen,J.Chromatogr.A736 (1996)105. [ 20 ] P.G.M. Kienhuis, J. Chromatogr. 647 (1993) 39. [ 21] J.L. Josephs, Application Report No. 244, Finnigan MAT, [ 22 ] E. de Hoffmann, J. Mass Spectrom. 31 (1996) 129. [ 23 ] P.E. Miller, M. Bonner Denton, J. Chem. Educ. 63 (1986) 617. [ 24 ] P.H. Dawson, J.B. French, J.A. Buckley, D.J. Douglas, D. Simmons, Org. Mass Spectrom. 17 (1982) 205. [ 25 ] R.B. Geerdink, in: D. Barceloè (editor), Applications of LC^MS in Environmental Chemistry, J. Chromatogr. Lib., Vol. 59, Elsevier, Amsterdam, 1996, p [ 26 ] S.A.McLuckey,J.Am.Soc.MassSpectrom.3 (1992)599. [ 27 ] J. Abian, G. Durand, D. Barcelo, J. Agric. Food Chem. 41 (8) (1993) 1264.