Journal of Chromatography A, 1356 (2014) 211–220

Contents lists available at ScienceDirect

Journal of Chromatography A journal homepage: www.elsevier.com/locate/chroma

Hydrophilic interaction chromatography versus reversed phase liquid chromatography coupled to mass spectrometry: Effect of electrospray ionization source geometry on sensitivity夽 Aurélie Periat a , Isabelle Kohler a , Aurélie Bugey b , Stefan Bieri b , Franc¸ois Versace c , Christian Staub a,c , Davy Guillarme a,∗ a

School of Pharmaceutical Sciences, University of Geneva, University of Lausanne, 20, Boulevard d’Yvoy, 1211 Geneva 4, Switzerland Official Food Control Authority and Veterinary Affairs of Geneva, 22, Quai Ernest-Ansermet, 1211 Geneva 4, Switzerland c University Center of Legal Medicine, Forensic Toxicology and Chemistry Unit, 1, Rue Michel Servet, 1211 Geneva 4, Switzerland b

a r t i c l e

i n f o

Article history: Received 5 March 2014 Received in revised form 14 May 2014 Accepted 16 June 2014 Available online 26 June 2014 Keywords: Hydrophilic interaction chromatography HILIC–MS Sensitivity Electrospray ionization Source geometry ESI

a b s t r a c t In this study, the influence of electrospray ionization (ESI) source design on the overall sensitivity achieved in hydrophilic interaction chromatography (HILIC) and reversed phase liquid chromatography (RPLC), was investigated. State-of-the-art triple quadrupole mass analyzers from AB Sciex, Agilent Technologies and Waters equipped with brand specific source geometries were tested with various mobile phase pH on 53 pharmaceutical compounds. The design of the ESI source showed to strongly influence the gain in sensitivity that can be achieved in HILIC compared to RPLC mode. The 6460 Triple Quadrupole LC/MS system from Agilent Technologies was particularly affected by mobile phase settings. Indeed, compared to RPLC conditions, 92% of the compounds had an increased signal-to-noise ratio at a flow rate of 300 ␮L/min in HILIC mode at pH 6, while this percentage dropped to only 7% at 1000 ␮L/min and pH 3. In contrast, the influence of flow rate and mobile phase pH on the gain in sensitivity between RPLC and HILIC was found very limited with the API 5000TM LC/MS/MS system from AB Sciex, as only 15 to 36% of the tested compounds showed an enhanced sensitivity in HILIC mode. With the Xevo TQ-S instrument from Waters, superior sensitivity in HILIC was noticed for 85% of the compounds with optimal conditions (i.e., pH 3 and 1000 ␮L/min), whereas at sub-optimal conditions (i.e. pH 6 and 300 ␮L/min), it represented less than 50%. The gain in sensitivity observed in HILIC was found less significant with the recent LC–MS platforms used in this study than for old-generation instruments. Indeed, the improved ESI sources equipping the recent mass analyzers allow for enhanced evaporation efficiency, mainly for RPLC mobile phases containing high proportion of water and this even at high flow rates © 2014 Elsevier B.V. All rights reserved.

1. Introduction Hydrophilic interaction chromatography (HILIC) appears today as an attractive alternative to reversed phase liquid chromatography (RPLC) for analyzing polar and/or ionizable compounds [1–4]. This chromatographic mode is characterized by a polar stationary phase and aqueous-polar organic mobile phase containing an important proportion (>60%) of organic modifier, usually acetonitrile. This highly volatile organic mobile phase

夽 Presented at the 13th International Symposium on Hyphenated Techniques in Chromatography and Separation Technology, Bruges, Belgium, 29–31 January 2014. ∗ Corresponding author. Tel.: +41 22 379 34 63; fax: +41 22 379 68 08. E-mail address: [email protected] (D. Guillarme). http://dx.doi.org/10.1016/j.chroma.2014.06.066 0021-9673/© 2014 Elsevier B.V. All rights reserved.

provides low backpressure due to its weak viscosity. In addition, it enhances droplet formation and desolvation efficiency, leading to a significant sensitivity improvement in electrospray ionization (ESI) [5,6]. This is also true for other detectors based on nebulization/evaporation processes (e.g. corona charged aerosol and evaporative light scattering detectors) [7]. Indeed, a gain in sensitivity for a large range of compounds has been reported several times in HILIC-ESI/MS compared to RPLC–ESI/MS [7–14]. A recent systematic study based on signal-to-noise ratios (S/N) comparison highlighted a ca. 4-fold median gain in sensitivity in HILIC vs. RPLC using a dataset of 56 basic drugs covering a wide range of physico-chemical properties [13]. Nevertheless, large differences were observed within the compounds set, with some of them even reaching a >100-fold sensitivity improvement in HILIC. Moreover, for ca. 95% of the compounds, highest S/N ratios were observed in HILIC mode.

212

A. Periat et al. / J. Chromatogr. A 1356 (2014) 211–220

Not only are the chromatographic conditions, e.g., mobile phase composition and pH, influencing the ionization efficiency, but also the ESI source geometry. Indeed, the multiple steps involved in ESI process, including droplet formation, desolvation, creation of gasphase ions and ion sampling, may be significantly affected by the design of the source [15]. Since the first ESI-MS experiments performed by Dole et al. in 1968, numerous developments have been carried out to improve the ionization efficiency, including modifications of the source geometry, the sprayer position relative to MS entrance, the sampling cone diameter, the number of transfer capillaries, as well as addition of heated gas [15–17]. Today, several source geometries are commercially available, i.e., (i) off-axis geometry, where the spray is at 30–45◦ to the x-axis between sampling capillary and the first quadrupole (e.g. Ion Spray interface), (ii) orthogonal geometry, where the spray is positioned at 90◦ relative to the x-axis (e.g., Jet Stream ESI source from Agilent Technologies or Turbo VTM source from AB Sciex), and (iii) Z-Spray geometry, which presents a double orthogonal sampling (e.g., Z-Spray from Waters). Orthogonal and Z-Spray configurations are usually providing better sensitivities than other source designs due to the prevention of clogging the MS orifice by non-volatile materials [15,17,19]. Until now, only a limited number of comparative studies have highlighted the effect of ESI source geometry on matrix effect or on signal suppression caused by ion-pairing reagent [16,18–21]. In the present work, the influence of ESI source design on the overall sensitivity observed between HILIC vs. RPLC was assessed using three state-of-the-art triple quadrupole (QqQ) mass spectrometers, i.e., Agilent 6460 Triple Quadrupole instrument from Agilent Technologies, API 5000TM System from AB Sciex, and Xevo TQ-S from Waters. Signal intensities, background noises and S/N ratios measured in HILIC and RPLC conditions were systematically compared using a training set of 53 basic drugs covering a broad range of lipophilicity and ionization properties. Besides, the influence of mobile phase pH and flow rate was also investigated. 2. Experimental 2.1. Chemicals and reagents Ultrapure water was supplied by a Milli-Q Advantage A10 purification system from Millipore (Bedford, MA, USA). Acetonitrile (ACN), methanol (MeOH), formic acid and acetic acid were of ULC–MS grade and purchased from Biosolve (Valkenswaald, Netherlands). The ammonium hydroxide (28%, m/v) solution was obtained from Sigma–Aldrich (Buchs, Switzerland). The 10 mM ammonium formate buffer at pH 3 was prepared with an adequate volume of formic acid and further adjusted to pH 3.0 with an ammonium hydroxide solution (28%, m/v). The pH values were measured with a SevenMulti pH meter (Mettler-Toledo, Schwerzenbach, Switzerland). 2.2. Compound dataset The training set of 53 basic drugs, covering a broad spectrum of pKa (i.e., between 6 and 11, except for a few compounds in the 2–6 pKa range) and log P values (i.e., between −1.2 and 5.6), was composed of the following compounds: 6-monoacetylmorphine, acebutolol, adenosine, alprazolam, alprenolol, amphetamine, antipyrine, bisoprolol, buprenorphine, bupropion, buspirone, clonidine, cocaethylene, cocaine, codeine, dextromethorphan, dibucaine, diltiazem, doxepin, fentanyl, flurazepam, heroin, hydroxyzine, imipramine, ketamine, N-methyl-1-(1,3-benzodioxol-5-yl)-2-butanamine (MBDB), 3,4methylenedioxyethamphetamine (MDEA), methadone, methamphetamine, methylephedrine, naloxone, naltrexone, norcocaine,

norephedrine, nortriptyline, noscapine, papaverine, perphenazine, pethidine, pindolol, prilocaine, propranolol, pseudoephedrine, pyrilamine, reserpine, salbutamol, sulpiride, terfenadine, tetracaine, thebaine, tolazoline, tramadol and triprolidine, which were either purchased in powder form from Sigma–Aldrich (Steinheim, Germany) or in methanolic solution (1 mg/mL) from Lipomed (Arlesheim, Switzerland). Stock solutions of compounds in powder form were prepared at 1 mg/mL in pure MeOH. Two mixtures of compounds were prepared. The first mixture was composed of 26 compounds including adenosine, alprenolol, bupropion, heroin, norephedrine, perphenazine, propranolol and terfenadine each at 1000 ng/mL; alprazolam, amphetamine and thebaine each at 500 ng/mL; naloxone at 400 ng/mL; dextromethorphan, methylephedrine and pseudoephedrine each at 200 ng/mL; dibucaine, doxepin and fentanyl each at 100 ng/mL; buspirone, MBDB, norcocaine, pethidine, prilocaine, sulpiride and tramadol each at 50 ng/mL and diltiazem at 30 ng/mL. The second mixture consisted of 27 compounds including nortriptyline at 2000 ng/mL, buprenorphine at 1500 ng/mL, codeine and reserpine each at 1000 ng/mL, clonidine, naltrexone and triprolidine each at 500 ng/mL; acetylmorphine, antipyrine and tolazoline each at 400 ng/mL; MDEA, methamphetamine, pindolol and pyrilamine each at 200 ng/mL; cocaethylene, flurazepam, hydroxyzine, ketamine, methadone, noscapine and salbutamol each at 100 ng/mL; cocaine, papaverine and tetracaine each at 50 ng/mL; imipramine and acebutolol each at 25 ng/mL; and bisoprolol at 5 ng/mL. Both samples were then diluted 10-fold for the analyses with Agilent and Sciex instruments, and 100-fold for the Waters detector. At least 95% ACN or water was used for dilution in HILIC and RPLC conditions, respectively, and 2 ␮L of sample volume were injected. 2.3. Instrumentation 2.3.1. Mass spectrometry Various QqQ mass analyzers were used for this study, namely (i) Agilent 6460 Triple Quadrupole System (Agilent Technologies, Santa Clara, CA, USA), (ii) AB Sciex API 5000TM System (AB Sciex, Concord, Canada) and (iii) Waters Xevo TQ-S (Waters Corporation, Milford, MA, USA). All experiments were carried out in positive ionization mode and Selected Reaction Monitoring (SRM) acquisition. Source parameters were systematically investigated in HILIC and RPLC mode using a mixture of 4 model compounds representative of the training set, i.e., adenosine, clonidine, doxepin and norcocaine at 600 ␮L/min and pH 3; optimal conditions are summarized in Table 1. Nitrogen was used as the nebulizing gas. Dwell times and inter-channel delays were both set at 5 ms, allowing for a sufficient number of data points across the narrow chromatographic peaks obtained with columns packed with sub-2 ␮m particles [17]. Cone voltages and collision energies were determined for each compound and are presented in Table S1 (Supplementary material). 2.3.2. Liquid chromatography An Agilent 1290 Infinity LC System (Agilent Technologies, Waldbronn, Germany) was used for the experiments with the Agilent 6460 Triple Quadrupole. Data acquisition, instrument control and data handling were performed with MassHunter software. A Thermo Scientific Dionex UltiMate 3000 RSLC System (Dionex, CA, USA) was used for experiments with AB Sciex API 5000TM . Data were acquired and processed with Analyst software. A Waters Acquity UPLCTM I-Class System from Waters (Milford, MA, USA) was used in combination with Waters Xevo TQ-S. Data acquisition, instrument control and data handling were performed with MassLynx Software. A Zorbax HILIC Plus Rapid Resolution HD column (50 mm × 2.1 mm i.d., 1.8 ␮m) from Agilent Technologies was

A. Periat et al. / J. Chromatogr. A 1356 (2014) 211–220

213

Table 1 Settings of source parameters. TurboVTM Source (AB Sciex API 5000TM )

RPLC

HILIC

Jet Stream source (Agilent 6460 Triple Quadrupole)

RPLC

HILIC

Z-spray Source (Waters Xevo TQ-S)

RPLC

HILIC

Spray voltage (V)

3000

5000

Capillary voltage (V)

3500

3000

Nebulizer gas pressure (psi)

40

30

Nebulizer gas pressure (psi)

50

40

3000 50 7

2000 50 7

Heater gas pressure (psi)

50

40

Sheath gas flow rate (L/min)

11

11

Capillary voltage (V) Cone voltage (V) Nebulizer gas pressure (bar) Desovation gas flow rate (L/h) Desovation gas temperature (◦ C) Cone gas flow rate (L/h)

1000

1000

600

600

150

150

◦

750

650

Sheath gas temperature ( C)

400

400

Curtain gas pressure (psi)

20

20

Drying gas flow rate (L/min) Drying gas temperature (◦ C)

10 300

10 300

Heater gas temperature ( C)

◦

used for HILIC experiments while an Acquity UPLC BEH C18 (50 mm × 2.1 mm i.d., 1.7 ␮m) from Waters was employed for all RPLC experiments.

2.4. Systematic evaluation of sensitivity The 53 model compounds were analyzed in HILIC and RPLC modes using generic stationary phases, i.e., C18 and unmodified bare silica, respectively. Two mobile phase pH, i.e., pH 3 and 6, were investigated in both modes. Experiments were conducted at three different mobile phase flow rates, namely 300, 600 and 1000 ␮L/min. For HILIC experiments at 600 ␮L/min, the gradient profile was the following: 95% ACN for 1 min, 95–65% ACN in 2.2 min, and a final isocratic step at 65% ACN for 0.5 min. The same gradient profile was used for experiments at 300 and 1000 ␮L/min but the times were geometrically scaled leading to the following gradient: 95% ACN for 2 min, 95–65% ACN in 4.4 min, and 65% ACN for 1 min for experiments at 300 ␮L/min; 95% ACN for 0.6 min, 95–65% ACN in 1.32 min, and 65% ACN for 0.3 min for experiments at for 1000 ␮L/min. For RPLC experiments at 600 ␮L/min, the gradient profile consisted of 2% ACN for 1 min, 2–70% ACN in 5 min, and a final isocratic step at 70% ACN for 0.5 min. The same gradient profile was employed for the two other flow rates, leading to the following gradients: 2% ACN for 2 min, 2–70% ACN in 10 min, and 70% ACN for 1 min for experiments at 300 ␮L/min; 2% ACN for 0.6 min, 2–70% ACN in 3 min, and 70% ACN for 0.3 min for experiments at 1000 ␮L/min. Both mixtures of compounds were injected in triplicate under each experimental condition. In order to ensure the validity of the methodology, a system suitability experiment (SSE) consisted in an injection of a mixture of 4 analytes (i.e., adenosine, clonidine, doxepin and norcocaine) was performed every 18 analysis with a systematic evaluation of chromatographic performances (e.g., retention times variability, peak width, etc.) and MS signal intensity. Prior to each SSE, a blank injection was performed to avoid any carry-over effect. The S/N values of individual compounds were reported for each condition and automatically calculated by the data acquisition software, considering the same methodology for noise determination, i.e., the selected background noise zones for flow rates of 300, 600 and 1000 ␮L/min were between 0.5 and 2.5 min; 0.5 and 1.5 min and 0.6 and 1.0 min, respectively. For the compounds which elute in this region, the noise area was between 3.0 and 5.0 min; 1.5 and 2.5 min; and 1.0 and 1.5 min for 300, 600 and 1000 ␮L/min respectively. This systematic procedure was strictly similar to the one implemented in a previous study described elsewhere which dealt with a previous-generation MS instrument [13]. The results obtained in the present study can thus also be compared to the previous one, increasing the number of tested QqQ instruments to four.

3. Results and discussion 3.1. Description of ESI sources This study compares the sensitivity achieved in RPLC and HILIC modes on three QqQ instruments representing the latest generation of MS instruments. These systems present obvious differences, mostly in their ionization source geometry, as illustrated in Fig. 1, which may significantly impact both RPLC–MS and HILIC–MS overall sensitivity. The ionization process of all these interfaces is pneumatically assisted by a nitrogen gas coaxially blowing around the spray capillary, to help droplet formation and desolvation [22]. Besides this nitrogen stream, additional features have been developed by the different manufacturers to support the droplet desolvation. The Turbo VTM source (Fig. 1A) has been developed by AB Sciex. It is compatible with flow rates of up to 3 mL/min and able to vaporize from 100% aqueous to 100% organic mobile phases. In this source geometry, a nebulizer gas (up to 90 psi) and a heater gas that originates from the turbo heaters (≤90 psi and temperature of up to 750 ◦ C) are added for a more efficient nebulization. The combination of the IonSpray effluent and the heated drying gas from the turbo sprayer are then projected at a 90◦ -angle to the MS sampling orifice [23]. Recent MS instruments from Agilent Technologies are equipped with the so-called Jet Stream thermal gradient focusing technology, illustrated in Fig. 1B, where a heated nitrogen sheath gas (up to 400 ◦ C and 12 L/min) is surrounding the nebulizer. In this configuration, according to the manufacturer, a 5–10-fold gain in sensitivity may be achieved for mobile phase flow rates between 0.25 and 2 mL/min which is attributed to a spray confinement and a more efficient desolvation process [24]. In the Z-spray source (Fig. 1C) proposed by Waters (Milford, MA, USA), the probe is positioned perpendicularly to the sampling cone, while the second extraction cone is also perpendicular to the ion beam, thus improving the mobile phase elimination. In addition, the source block is heated at 150 ◦ C and the heated nebulizing gas flow and temperature can be as high as 1200 L/h and 650 ◦ C, respectively. Under such extreme conditions, the negative impact of elevated mobile phase flow rates on sensitivity seems to be moderate.

3.2. Impact on sensitivity in RPLC and HILIC The results obtained for the evaluation of sensitivity in HILIC and RPLC modes are reported in Figs. 2–4 for Agilent, AB Sciex and Waters instruments, respectively, based on similar experiments. The whole series of experiments on a given device were systematically performed in a timeframe as narrow as possible to lower the time-dependant variability. The sensitivity was estimated at three different flow rates, i.e., 300, 600 and 1000 ␮L/min, adapted to the column dimensions, and with two mobile phase pHs, i.e., pH 3 and

214

A. Periat et al. / J. Chromatogr. A 1356 (2014) 211–220 analyte soluon

analyte soluon

nebulizing gas

heater

heater

B. curtain gas

A.

super-heated sheath gas heated drying gas

curtain gas

nozzle voltage heated nitrogen

spray

exhaust

C.

nebulizing gas

resisve sampling capillary

vent heated desolvaon gas

baffle plate

analyte soluon

nebulizing gas sampling cone

extracon cone

isolaon valve

ion block cone gas

Fig. 1. Graphical representation of the investigated source geometries. (A) TurboVTM source of AB Sciex API 5000TM LC/MS System; (B) Jet Stream source of Agilent 6460 Triple Quadrupole LC/MS system; and (C) Z-Spray source of Waters Xevo TQ-S. Adapted from [15] with permission.

6. In order to have comparable S/N ratios between the LC–MS systems, analytes were diluted 10-fold for use with Agilent and AB Sciex instruments, and 100-fold for Waters instrument which was more sensitive. The percentages reported in Figs. 2–4 correspond to the proportion of compounds among the set of 53 basic drugs presenting a specific behaviour. The group labeled in dark gray highlights the compounds giving a lower sensitivity in HILIC compared to RPLC (