Hyphenation of atmospheric pressure chemical ionisation mass spectrometry to supercritical fluid chromatography for polar car lubricant additives analysis

Journal of Chromatography A, 1216 (2009) 837–844

Contents lists available at ScienceDirect

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

Hyphenation of atmospheric pressure chemical ionisation mass spectrometry to supercritical fluid chromatography for polar car lubricant additives analysis Gwenaelle Lavison-Bompard a,1 , Didier Thiébaut a,∗ , Jean-Franc¸ois Beziau b , Bernadette Carrazé b , Pascale Valette c , Xavier Duteurtre c , Jean-Claude Tabet d a

Laboratoire Environnement et Chimie Analytique (UMR CNRS 7121), Ecole Supérieure de Physique et de Chimie Industrielles de Paris, 10 rue Vauquelin, 75231 Paris cedex 05, France PSA Peugeot-Citroën SA, route de Gisy, 78140 Vélizy-Villacoublay, France Renault S.A, Technocentre, 78288 Guyancourt Cedex, France d Laboratoire de Chimie Organique Structurale et Biologique, Université de Paris VI, 4 place Jussieu, 75252 Paris cedex 05, France b c

a r t i c l e

i n f o

Article history: Received 4 August 2008 Received in revised form 17 November 2008 Accepted 26 November 2008 Available online 13 December 2008 Keywords: Supercritical fluid chromatography APCI ion trap MS Car lubricant Car lubricant additives

a b s t r a c t Car lubricant additives are added to mineral or synthetic base stocks to improve viscosity and resistance to oxidation of the lubricant and to limit wear of engines. As they belong to various chemical classes and are added to a very complex medium, the base stock, their detailed chromatographic analysis is very difficult and time consuming. In a previous paper, it was demonstrated that supercritical fluid chromatography (SFC) allows the elution of common low-molecular-weight additives. Since their total resolution could not be achieved owing to the limited peak capacity of packed columns, the hyphenation of selective and informative detection methods such as atomic emission detection (AED) was required. Further to results obtained in SFC-AED, this work describes the hyphenation of SFC to atmospheric pressure chemical ionisation ion trap mass spectrometry (MS). SFC–MS hyphenation is detailed: temperature, flow rates of gas and mobile phase introduced in the source, position of the restrictor, ionisation additives and conditions of autotune are studied. Car lubricant monitoring requires negative and positive ionisation modes with or without the addition of ionisation auxiliary solvent according to the nature of additives. Moreover, when sensitivity is of major concern for a selected additive, the autotuning routine of the MS has to be performed in conditions as close as possible to analytical conditions, i.e. under subcritical conditions. Unambiguous identification and structure elucidation of several additives in formulated car lubricants are also presented. © 2008 Elsevier B.V. All rights reserved.

1. Introduction The comprehensive analysis of the components of lubricants is a very difficult task using current analytical tools: they should enable separation of base stocks components and additives. Base stocks can be mineral, synthetic and hemi-synthetic. Thus, they are composed of both very numerous, non-polar and similar structure molecules. Additives belong to different chemical families of variable polarity and molecular weight. The identification of additives in a lubricant of “unknown” composition is of major importance. It could lead (i) to a better understanding of their behaviour, (ii) to the relationship between the structure of a molecule added to the base stock and lubricant properties, and (iii) to the prediction of the major changes occurring during ageing of the lubricant.

∗ Corresponding author. Tel.: +33 1 40 79 46 48; fax: +33 1 40 79 47 76. E-mail address: [email protected] (D. Thiébaut). 1 Present address: CRECEP, 144-156 avenue Paul Valliant Couturier, 75014 Paris, France. 0021-9673/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.chroma.2008.11.103

The potential of supercritical fluid chromatography (SFC) has been demonstrated for the analysis of additives and lubricants [1–5]. Recently [1] various stationary phases were tested in order to provide elution and improve resolution of main lowmolecular-weight additives. However, comprehensive separation of the additives in the base stock could not be achieved using SFC hyphenated to flame ionisation detection (FID) or UV detection. Two-dimensional separations using bare silica and reversed phase supports improved the separation but could not avoid peak overlapping. At this moment, SFC separation of lubricant additives still lacks resolution and cannot provide comprehensive separation. Thus, identification of compounds cannot be easily performed. Extra selectivity, required for a deeper insight into the lubricant composition, was partly obtained with atomic emission detection (AED) [2]. Nevertheless, structural information and elucidation can only be achieved by the implementation of selective and informative detection such as mass spectrometry (MS). MS, including atmospheric pressure chemical ionisation (APCI) MS, has been hyphenated to SFC with different sources and configurations [6–9] mainly for pharmaceutical applications [10–14].

838

G. Lavison-Bompard et al. / J. Chromatogr. A 1216 (2009) 837–844

Fig. 1. Scheme of the SFC–MS interface.

This paper describes the hyphenation of APCI–MS to SFC–UV–FID system already used for lubricant additives analysis [1]. Owing to the structure and the low-molecular weight of lubricant additives to be analysed, APCI is expected to provide a wider scope than electrospray ionisation (ESI) or APPI: ESI works in “liquid” like conditions; this means a higher amount of modifiers or ionisation fluid is required; moreover, better ionisation of investigated additives was expected with APCI than with ESI. APPI could also be involved and should lead to similar hyphenation conditions and results as APCI (this source was not available with our MS system). As the mass detector involved in this project is not configured for SFC hyphenation, the major parameters affecting the response are studied. A special attention is paid to infusion conditions for the mass spectrometer tuning in order to mimic the supercritical chromatographic conditions [15]. Finally, it is shown that SFC–MS enables the characterisation of main classes of non-polymeric additives in the range of concentration they are added in formulated lubricants.

2. Experimental

50 ␮m I.D. deactivated fused silica capillary tubing from S.G.E. (Villeneuve St. Georges, France) for transferring ca. 1% of column effluent to the FID system, (ii) a linear restrictor used as the transfer line to the mass spectrometer (ca. 50% of column effluent) and (iii) the diode array UV detector (ca. 50% of column effluent). The backpressure regulator of the SFC–FCM 1200 was placed downstream the UV detector and controlled the overall pressure in the system. The fixed restrictors controlled the mobile phase flow rates entering FID and MS systems; thus, the restrictors controlled the flow rate ratio between the three detectors. As neat CO2 was used for the separation, prior to entering the APCI source of the mass spectrometer auxiliary ionisation fluid could be added to the mobile phase via a SFC-300 Carlo Erba syringe pump connected to a T piece (Fig. 1). Global CO2 and ionisation fluids flow rate strongly depended (i) on the length of the linear restrictor and (ii) on the ionisation fluid flow rate. If one wanted to vary ionisation fluid flow rate at CO2 constant flow rate, one had to adapt the length of the restrictor, keeping in mind that the minimum length to be introduced in the APCI was 12 cm in the standard position. 2.2. MS

2.1. Supercritical fluid chromatograph A Berger Instruments supercritical fluid chromatograph SFC FCM-1200 supplied by Mettler Toledo AutoChem (Viroflay, France) was used. A dual-pump fluid control module delivered carbon dioxide and modifier. It included a FID system and a HP 1050 multiwavelength photodiode array detection (DAD) system. Column outlet pressure was controlled with the backpressure regulator and injections were performed using a Rheodyne Model 7413 valve with a 5-␮L internal loop. The injector was air actuated to enable a 4-s injection. Prior entering the UV detector, column effluent was split via a cross connected to (i) a laboratory made integral restrictor using

A LCQ ion trap mass spectrometer (Thermo Scientific, Les Ulis, France) was hyphenated to the SFC system. APCI source was used for positive and negative ionisation modes. Ionisation fluids were mainly methanol, dichloromethane or acetonitrile; they could be modified by acetic acid, trifluoroacetic acid and sodium or ammonium acetate. 2.3. APCI source Nebulisation of the column effluent was achieved using nitrogen introduced in two places as shown in Fig. 2: they are defined by the manufacturer as auxiliary and drying gases. In the commercial

Fig. 2. Scheme of APCI PROBE – profile (a) and face (in the source) (b) – and the three studied restrictor positions.

G. Lavison-Bompard et al. / J. Chromatogr. A 1216 (2009) 837–844

839

Fig. 3. Experimental instrument for sub- and supercritical phase infusion.

configuration of the detector, gases flow rates were pressure regulated. In this work, ball flow-meters/regulators were added after the regulation valves of the instrument. Thus, gas flow rates were measured and controlled continuously. Pressure variation in the source could be compensated in order to keep constant the flow rates. The temperature of the probe was varied from ambient to 400 ◦ C and the intensity of the Corona discharge from 0 to 8 ␮A. Automatic gain control was turned on. The potentials of capillary transfer lens (10–100 V) and intermultipole lens (10–200 V) have been optimised for best transfer conditions. 2.4. Direct infusion The LCQ was equipped with a direct infusion line: liquids could be directly introduced into the source to optimise the detection conditions. This operation required infusion of a suitable sample in constant conditions for a few minutes. Thus, the infusion line was modified to enable sub- and supercritical fluids to be infused in four steps as shown in Fig. 3: (1) The sample dissolved in methanol was loaded in a 150-␮L loop via a six-port valve. (2) The valve was switched: the loop was flushed by methanol at a low flow rate (10 ␮L/min) using a Gilson pump, model 308 (Villiers-le-Bel, France). (3) This liquid was mixed via a dynamic mixer (Gilson, model 811) to the CO2 dispensed at 150 bar by the Berger SFC pump of the chromatograph. (4) This subcritical mixture was introduced into the source using the unmodified transfer line. The ionisation fluids were also added before entering the source as for a regular chromatographic operation. Loop volume and flushing flow rate were adjusted in order to enable at least 10 min of sample infusion in constant operating conditions.

2.5. Data acquisition and apparatus control The Berger SFC system was personal computer controlled by Berger 3D SFC ChemStation software enabling collection of FID and UV signals. The LCQ mass spectrometer was piloted by Xcalibur software except for nitrogen flow rates as mentioned above. Total ion current (TIC) and extracted ions chromatograms have been plotted and integrated. 2.6. Columns and chemicals A Capcell Pak C18 column (250 mm × 4.6 mm, 5 ␮m) was used in all the experiments (AIT, Houilles, France). The mobile phase was 5.5 grade carbon dioxide from Messer France (Asnières, France). Nitrogen used for nebulisation was 4.5 grade from Messer France. Helium used in the ion trap was 5.0 grade (Messer France). Solvents were HPLC grade and purchased from Baker (Chromoptic, France). Acids and salts were analytical grade (Sigma, Saint Quentin Fallavier, France). The phenol derivatives (Lowinox TBP-6), triphenyl phosphate (TPP) and phenothiazine were purchased from Aldrich (Lyon, France); other additives were generously provided by private donators. The samples were dissolved in carbon disulfide (VWR, Fontenay-sous-bois, France) except for direct infusion. 3. Results and discussion As they belong to different major classes of lubricant additives, triphenyl phosphate, phenothiazine and Lowinox TBP-6 (Table 1) were selected as test compounds for optimisation of MS detection. For studying the effect of MS parameters on the response, no pressure programming of the mobile phase was used in order to avoid undesired variation of the CO2 flow rate in the detectors. Thus, the separation was performed at 80 ◦ C and 150 bar; the supercritical CO2 flow rate in the column was 2 mL/min.

840

G. Lavison-Bompard et al. / J. Chromatogr. A 1216 (2009) 837–844

Table 1 Structure, area and signal-to-noise ratios of the elution peak of model additives in SRM(+) mode, measured using liquid (L) and subcritical (SC) autotune calibrations. Analyte

Autotune phase

Triphenylphosphate, MW: 326, m/z: 327 Th

L

SC

Phenothiazine MW: 199 m/z: 200 Th

L

SC

Lowinox TBP-6, MW: 358, m/z: 359 Th

L

SC

Area (×109 )

S/N

4.5

17

10.4

26

1.7

4

13.1

10

7.8

4

16.0

11

Experimental conditions: Capcell Pak C18 (250 mm × 4.6 mm, 5 ␮m) column, pressure programming from 100 to 300 bar at 20 bar/min, 80 ◦ C, DCO2 : 2 mL/min (pressurised, in the column), FCO2 : 380 mL/min (gas flow rate in the interface), Dmethanol : 200 ␮L/min.

Otherwise mentioned, MS was performed in the positive mode as [MH]+ ions were the main observed ions for each test compound. Thus, the effect of the investigated parameters was determined by observing the evolution of the area and the signal-to-noise ratio of the elution peaks corresponding to the TIC or the [MH]+ ions. Effect of ionisation additives was studied in both positive and negative modes. 3.1. Study of hyphenation parameters 3.1.1. Temperature As APCI is a gas phase ionisation process, temperature had a major effect on the quality of the APCI–MS signal because the size of clusters can change with temperature. Moreover, when hyphenated to SFC, heating in the interface zone was required to avoid freezing effect owing to the adiabatic decompression of the gas like CO2 in the restrictor. The temperatures of two zones were studied using conditions mentioned above: (i) the temperature of the vaporiser, in which the restrictor was introduced and (ii) the temperature of the transfer capillary between atmospheric and low pressure zones of the mass spectrometer. As expected, when the vaporiser temperature was lower than 100 ◦ C, an erratic response was obtained: plugging of the restrictor was likely to occur. Above 100 ◦ C, the signal was stable. Areas and the signalto-noise ratios for each compound (TIC and SMR) slightly increased until decomposition of the more thermolabile compounds occurred, above 300 ◦ C. Thus, 300 ◦ C was selected as the temperature of the restrictor. The temperature of the transfer capillary was varied from 100 to 300 ◦ C, 150 ◦ C being the default value. Again, the peak areas and signal-to-noise ratios increased until they reached a maximum around 250 ◦ C. Then, they dramatically decreased. At 300 ◦ C, the signal was ca. 10 times lower than at 250 ◦ C for the three compounds.

The increase of the signal with the temperature was related to the more efficient desolvatation of ions. The 100 ◦ C difference compared to the default temperature was probably related to the nature of our test compounds, which were supposed to form stable charged aggregates. It is worth noticing that the response of phenothiazine and lowinox TBP-6 were respectively ca. 10 and 100 times lower than the response of TPP. 3.1.2. Gas flow rate 3.1.2.1. CO2 . The CO2 gas flow rate in the interface was adjusted between 350 and 650 mL/min by varying the length of the restrictor. The flow rate of methanol used as ionisation fluid was kept constant at 150 ␮L/min. Within the studied range, the response, area and signal-to-noise ratio, slightly varied (<20%), the maximum CO2 gas flow rate being around 380 mL/min. 3.1.2.2. Nebulisation gas flow rate. Nebulisation in the LCQ MS uses nitrogen introduced in two different locations in the interface (Fig. 2). One, the closest to the restrictor, is referred as nebulising gas. The other one, referred as the auxiliary gas, is supposed to focus the spray formed by the chromatographic effluent. The nebulisation gas flow rate was monitored in the range 1–5 L/min. The lower the nitrogen flow rate, the higher the response, except for phenothiazine for which ionisation was almost constant between 1 and 2 L/min. It must be pointed out that the optimum nitrogen flow rate was much lower in SFC than in LC (5–10 L/min). This illustrates that the decompression of the pressurised mobile phase in SFC favoured the nebulisation process. For following applications, the nebulisation gas flow rate was kept at 1.5 L/min. The effect of the auxiliary gas was found to be negligible. 3.1.3. Nature of ionisation additive In the positive mode, methanol or acetonitrile were added to the pure CO2 as ionisation fluids They could yield small protonated solvent clusters. For the three compounds, the best response was obtained using methanol, especially for TPP. No clear reason could be advocated since the low proton affinity of methanol is similar to that of acetonitrile. The addition of acids, sodium or ammonium acetate in the methanol decreased the intensity of the signal. The main ions formed remained (MH)+ . The negative mode was also studied using methanol, acetonitrile or dichloromethane as additives. Methanol was also modified using 0.1% of trifluoroacetic acid (TFA) or ammonium acetate. TPB-6 was selected as model compound since this phenolic derivative leads to good quality response in the negative ionisation mode. The three ions observed are described in Fig. 4. Ions I and II were produced by the reduction of C–S bond occurring in the gas phase leading to two complementary fragment ions only differing by the presence of one S atom and the position of the negative charge. Ions I and II are complementary ions and are formed from neutrals. For ion I, the formation mechanism could not be precisely defined: it could be formed directly from the precursor molecule TBP-6 or from ion II. Ion III is a tetramer of ion II. When the dimers were formed, cyclic tetramers could be obtained according to radical reaction in the source when reduction occurs. If they were observed on the APCI mass spectra, the abundance of ions (M−H)− , (M+Cl)− and (M+CH3 COO)− was very low compared to the abundance of the three main ions. Fig. 4 also shows the variation of peak areas measured for the three main ions. Using pure methanol, acetonitrile or dichloromethane as additives, ion II gave the highest response. It seems that the solvent

G. Lavison-Bompard et al. / J. Chromatogr. A 1216 (2009) 837–844

841

Fig. 4. Main negative ions of Lowinox TBP 6 and evolution of the corresponding peak areas (TIC and SMR) as a function of the ionisation fluid composition. Experimental conditions: Capcell Pak C18 (250 mm × 4.6 mm: 5 ␮m) column, 150 bar, 80 ◦ C, DCO2 : 2 mL/min (pressurised, in the column) FCO2 : 450 mL/min (gas flow rate in the interface), Dionisation fluid : 150 ␮L/min.

in the gas phase yielded a significant orientation for the competitive deprotonation process, leading respectively to ions I and III (Fig. 4). When adding TFA and ammonium acetate to methanol, ionisation suppression for ions I and II and enhancement of ion III are clearly observed. The best compromise for the detection of the additives was observed with pure methanol as ionisation fluid. Thus, methanol has been used in both positive and negative modes. The effect of methanol flow rate was investigated in positive mode at constant CO2 flow rate by modifying the length of the restrictor used as the transfer line. Thus, only four different methanol flow rates were studied. Fig. 5 shows the variation of peak area versus methanol flow rate. Two cases could be distinguished as follows:

When no methanol was added, ionisation still occurred. As the main ion was still (MH)+ , one can conclude that traces of water in the CO2 were able to form this ion as already described elsewhere [7,16,17]. (M)•+ ions, supposed to be formed by charge transfer mechanism with the CO2 •+ were not observed. One could notice that the TBP-6 response was much than four times higher without methanol.

When adding up to 200 ␮L/min of methanol “as a proton source”, the detection response increased. Above this value, the source was saturated and the response decreased. Thus, to ensure maximum information on the composition of unknown car lubricants, experiments have been carried out with and without methanol, the optimum methanol flow rate being around 200 ␮L/min. This flow rate was the best compromise between the ionisation efficiency and the source saturation. 3.1.4. Position of the restrictor Owing to the nebulising effect of CO2 decompression, it was supposed that better and faster vaporisation could be obtained in SFC–MS than in LC–MS. So, the duration of evaporation process in the source could be reduced compare to LC coupling. A longer part of the restrictor could be introduced in the vaporiser. Three positions of the restrictor were investigated (Fig. 2): the standard position (default for LC–MS coupling) at 12 cm from the inlet of the probe, a medium position at 14.8 cm from the probe and an extreme position located at 17.6 cm, almost at the outlet of the vaporiser. The best response was observed in the LC default position that was kept for further experiments. The nebulising properties of the CO2 mainly affected the nebulising gas flow rate rather than the position of the restrictor.

842

G. Lavison-Bompard et al. / J. Chromatogr. A 1216 (2009) 837–844

Fig. 5. Evolution of the elution peak area of each of the three test compounds in the positive mode as a function of the methanol flow rate introduced in the linear restrictor. Experimental conditions as in Fig. 4 except FCO2 : 380 mL/min (gas flow rate in the interface).

3.1.5. Other parameters Corona discharge intensity (2–8 ␮A), potentials of capillary transfer lens (10–100 V) and inter-multipole lens (10–200 V) were also studied. As the response decreased when the intensity of corona discharge increased, 2 ␮A was selected. Concerning the two lenses, the default values, 30 V each, were found to give the best results and were kept for further experiments. At low voltage, desolvation process of charged aggregates was partial while at higher potential, a total desolvation took place, collision-induced dissociation occurred. 3.2. Autotune in SFC conditions For a number of parameters, i.e. transfer capillary tension, capillary lens offset, multipole lens and inter-multipole lens offsets, the autotune routine of Xcalibur software was used to optimised MS Table 2 Detection limits (S/N = 3) and repeatability with and without addition of methanol for MS detection. Molecule

LOD (ng) and repeatability This work

Methyl benzoate Diphenylamine Anthracene Nitrobenzene

Lit. [Ref.]

With methanol (RSD %)

Without methanol (RSD %)

34 (1.5) 122 (2.4) 7 (1.8) 110 (2.3)

4 (3.8) 1.7 (3.2) 0.4 (3.0) 237 (2.9)

– 0.5 [12] 1 [11] 125 [12]

Experimental conditions as in Table 1.

conditions for a selected ion. This involved direct infusion of the selected molecule during the whole process in constant conditions as close as possible to chromatographic ones. Obviously, the use of liquid phase infusion may be not the best medium for autotune in SFC–MS. Thus, the system was modified to enable infusion in supercritical conditions (Fig. 3). Indeed, the Berger SFC system was used for infusion. It was modified to enable large volume injection (150 ␮L) as described in Section 2. So, the solute was solvated by both the CO2 and the methanol in conditions similar to those used for the chromatographic separation despite there was no column. However, direct infusion in neat CO2 is unachievable using this simple approach because the sample is a liquid. For each compound, two series of autotune calibrations were performed in the positive mode on the selected (MH)+ ions: one in liquid and the other one in the subcritical medium. Then, for each compound, these two sets of conditions were applied for SFC–MS analysis. The comparison of the results is reported in Table 1: in all the cases, peak area and signal-to-noise ratios were higher when the subcritical autotune was used. The enhancement factor varied from 1.5 to 7.7 depending on the compound investigated. This demonstrated the potential of this approach for enhancing the sensitivity of the detector. Keeping in mind the complexity of the formulated lubricants to be investigated, this approach could be used when the maximum sensitivity was required towards a selected target. 3.3. Performances 3.3.1. Detection limit Methyl benzoate, diphenyl amine and anthracene were used as test compounds to determine the system detection limit without the use of subcritical autotune. Table 2 shows the detection limits (S/N = 3) obtained with and without addition of methanol for MS detection. As mentioned above, addition of methanol did not improve sensitivity for all the test compounds. Indeed, the sensitivity only increased by a factor of 2 for nitrobenzene. The values reported in this work were in good agreement with the literature [18]. It must be pointed out that Morgan et al. used a different column with pressure programming. This favoured apparent efficiency and, thus, sensitivity. Furthermore, experiments were not carried out using the same mass spectrometer; thus, desolvation conditions were unknown and results were difficult to compare. 3.3.2. Repeatability The repeatability was calculated on a set of eight successive injections (Table 2). In all the cases, RSD were lower than 4% illustrating the good performances of the system. One should mention the better repeatability when methanol was added: this result was difficult to explain as it could be related to a better decompres-

Table 3 Peak area and signal-to-noise ratio of Lowinox TBP-6 and Lowinox TBM-6. Additive

Lowinox TBP-6

Lowinox TBM-6

1.7 24

10.6 540

Structure

Area (×109 ) S/N Experimental conditions: same as in Table 1.

G. Lavison-Bompard et al. / J. Chromatogr. A 1216 (2009) 837–844

843

Fig. 6. SFC/FID–UV–MS analysis of Irganox L57: (a) standard of Irganox L57 and (b) formulated lubricant: TIC and SRM chromatograms and possible corresponding molecular structure of m/z 226, 282, 338 and 394. Experimental conditions as in Table 1.

sion in the linear restrictor or to a better ionisation process when methanol is added. 3.4. Applications 3.4.1. Structure–retention–ionisation relationship In the major class of phenol derivatives, the two position isomers Lowinox TBP-6 and TBM-6 were found to exhibit very different behaviour [1]. The MS response (peak area) of TBP-6 was six times lower than the MS response of TBM-6, both being injected at the same concentration (Table 3). Molecular modelling suggested intramolecular bonding occurred between the two hydroxyl groups of TBP-6 whereas it was not the case for TBM-6. This result could be correlated to the chromatographic behaviour reported in Ref. [1] where it is shown that TBM-6 could not be eluted from a bare silica nor from a diol bonded silica column. Strongest interaction with the residual silanol groups of the stationary phase revealed the accessibility and reactivity of the hydroxyl groups of the additive. Interestingly, the MS responses showed some similarity with chromatographic behaviour. 3.4.2. Elucidation of structure of reference additives Irganox L57 is an antioxidant additive employed in many car lubricant formulations as radical inhibitor. As illustrated in Fig. 6a, the SFC–UV chromatogram of the reference additive exhibited at least eight peaks. Using the SFC–FID–UV–MS system, it was possible to determine the total number of carbons of the side chains using the corresponding MS spectra. This number is indicated on the chromatogram above each peak. Four groups of compounds were identified having side chains ranging from 4 to 16 atoms of carbon. In three groups, one major peak suggested that there was a

main isomer. SFC gave a fingerprint of the additive but, as expected for hydrocarbons isomers elucidation, MS failed to provide precise structural information. Nevertheless, the SFC hyphenated system could be used for monitoring the ageing of the car lubricants or to assess the origin of additive. 3.4.3. Identification and quantification of reference additives in formulated lubricants As an example, Irganox L57 was analysed in a formulated car lubricant; Fig. 6b shows the corresponding TIC and reconstructed chromatograms on majors ions (226, 282, 338, 394 Th). Once the elution zones of the different species corresponding to the ions mentioned above were defined, MS–MS experiments were performed by collision-induced dissociation. The comparison of the spectra to reference material was carried out. It confirmed the identification of Irganox L57 homologues, having side hydrocarbon chains ranging from C4 to C16. Moreover, quantification in UV and MS could be performed using external standard calibration and standard addition. As shown in Table 4, results were in Table 4 UV and MS quantification results of Irganox L57 in a formulated lubricant. Quantification method

UV MS MS2

External standard calibration

Standard addition

% (w/w)

RSD (%)

% (w/w)

RSD (%)

0.24 0.28 0.25

5 5 6

0.26 0.26 0.27

5 4 7

Experimental conditions: Capcell Pak C18 (250 mm × 4.6 mm, 5 ␮m) column, 150 bar, 30 ◦ C; DCO2 : 2 mL/min (pressurised, in the column); FCO2 : 380 mL/min (gas flow rate in the interface); Dmethanol : 200 ␮L/min.

844

G. Lavison-Bompard et al. / J. Chromatogr. A 1216 (2009) 837–844

Fig. 7. SFC/APCI(+)MS analysis of Irgalube TPP: (a) standard and (b) formulated lubricant. Experimental conditions as in Table 1.

very good agreement for both methods but MS allowed structure assignation and baseline resolution owing to mass range selection. Irgalube TPPT, an antiwear extreme pressure additive, has also been identified in the same lubricant. Fig. 7 compares chromatograms and related MS spectra of the reference additives and of the car lubricant. Without MS, the identification of this compound was very difficult because of the lack of resolution as it can be seen in the TIC chromatogram. Using the retention time of the standard and the very clear mass spectra, unambiguous identification of the additive in the real sample was easily achieved. 4. Conclusion Hyphenation of MS to SFC was found to be quite similar to LC–APCI–MS. However, this work pointed out three noticeable differences concerning: (i) the reduced nebulisation gas flow rate, (ii) the effect of the addition of ionisation fluid or not on sensitivity and repeatability and (iii) the need of subcritical conditions for autotuning of the MS for better sensitivity towards a target analyte. SFC–MS is a must for structure elucidation of car lubricant additives standards and their identification in a formulated lubricant. Using appropriate mass range, MS selectivity added essential discrimination dimension. Moreover, our system could be used for quantitative analysis of target additives. For more comprehensive analysis of car lubricant composition including the base stock, FID, UV and MS traces are complementary. This is the reason why this work is part of a unique multihyphenated system including AED [18] and FTIR [8]. In one single injection, one could obtain: (i) a universal trace using FID, mostly from the base stock, (ii) a UV trace of unsaturated compounds from the base stock

and additives, (iii) structural information from FTIR and MS and (iv) elementary confirmation with AED. Acknowledgements PSA-Peugeot-Citroën and Renault SA are kindly acknowledged for their technical and financial support. References [1] G. Lavison, F. Bertoncini, D. Thiébaut, J.-F. Beziau, B. Carraze, P. Valette, X. Duteurtre, J. Chromatogr. A 1161 (2007) 300. [2] F. Bertoncini, D. Thiébaut, M. Gagean, B. Carraze, P. Valette, X. Duteurtre, Chromatographia 53 (Suppl.) (2001) S427. [3] A.M. Barnes, K.D. Bartle, S. Christopher, C. Heathcote, J. High Resolut. Chromatogr. 23 (2000) 389. [4] B.N. Barman, V.L. Cebolla, L. Membrado, Crit. Rev. Anal. Chem. 30 (2000) 75. [5] S. Ashraf, K.D. Bartle, A.A. Clifford, R. Moulder, J. High Resolut. Chromatogr. 15 (1992) 535. [6] P.J. Sjörberg, K.E. Markides, J. Chromatogr. A 785 (1997) 101. [7] P.J. Sjörberg, K.E. Markides, J. Chromatogr. A 855 (1999) 317. [8] U. Just, D.J. Jones, R.H. Auerbach, G. Davidson, K. Käpler, J. Biochem. Biophys. Methods 43 (2000) 209. [9] P. Sandra, A. Medvedovici, Y. Zhao, F. David, J. Chromatogr. A 974 (2002) 231. [10] M. Tuomola, M. Hakala, P. Manninen, J. Chromatogr. B 719 (1998) 25. [11] M.C. Ventura, W.P. Farrell, C.M. Aurigemma, M.J. Greig, Anal. Chem. 71 (1999) 4223. [12] K. Dost, D.C. Jones, G. Davidson, Analyst 125 (2000) 1243. [13] S.H. Hoke, J.D. Pinkston, R.E. Bailey, S.L. Tanguay, T.H. Eichhold, Anal. Chem. 72 (2000) 4235. [14] Y. Zhao, P. Sandra, G. Woo, S. Thomas, K. Gahm, D. Semin, LC–GC Eur. 17 (2004) 224. [15] L.N. Tyrefors, R.X. Moulder, K.E. Markides, Anal. Chem. 65 (1990) 2835. [16] D.G. Morgan, K.L. Harbol, N.P. Kitrinos Jr., J. Chromatogr. A 800 (1998) 39. [17] F. Sadoun, H. Virelizier, P.J. Arpino, J. Chromatogr. 647 (1993) 351. [18] F. Bertoncini, D. Thiébaut, J.-F. Beziau, B. Carrazé, P. Valette, X. Duteurtre, J. Chromatogr. A 910 (2001) 127.