Rapid Screening of PET Bottles for Residual Cleaning Solvents Using a LONESTAR TM Portable Analyzer

Transcription

1 Rapid Screening of PET Bottles for Residual Cleaning Solvents Using a LONESTAR TM Portable Analyzer Detecting dimethyl sebacate and DBE-3 (dimethyl adipate and dimethyl glutarate mixture) Contact us

2 Contents Executive Summary... 3 Introduction... 4 The Lonestar Platform... 5 Experimental Setup... 6 Method and Results... 8 Standard solution preparation... 8 Lonestar parameters... 8 Peak identification with Mass spectrometry... 8 Initial DF spectra DBE-3 low concentration calibration Method optimisation Background testing of PET bottles Bottle Interface with Lonestar Conclusions Appendix A: FAIMS Technology at a Glance Sample preparation and introduction Carrier Gas Ionisation Source Mobility Detection and Identification P a g e

3 Executive Summary This report summarises the work undertaken to develop a rapid test method for the detection of bottle rinse residual chemicals dimethyl sebacate and DBE-3 (a mixture of dimethyl adipate and dimethyl glutarate) in polyethylene terephthalate (PET) bottles, using a Lonestar portable analyser. Also included are details of a plastic drinks bottle attachment system for the Lonestar analyzer. Overall, this work represents an exceptional result (in terms of sensitivity, selectivity and speed) and reaffirms the unique ability of Owlstone s portable, chemical sensor (Lonestar) to provide a real-time detection solution surrounding the chemicals of interest at the point of need. The key findings are presented below: Using Lonestar in a standard configuration dimethyl sebacate and DBE-3 can both be detected at concentrations below 50 ppb in a sealed vessel in less than 2 minutes. Significantly faster sampling times could be achieved depending upon the exact requirements of the system. Quantification of the analytes can be achieved across a range of concentrations from 50 ppb to approximately 20 ppm. This range could be extended upwards by using a split flow to dilute the sample flow drawn into the Lonestar. No other significant volatile compounds were generated from the sampled PET bottles and consequently there are no additional issues in detecting the dimethyl esters in the background of PET bottles. The only other analyte expected is peracetic acid and this has been shown to generate negative ions when ionised and therefore will not interfere with the positive ions of the dibasic esters. A simultaneous calibration of peracetic acid would be available but was not investigated in this study. The residual solvent responses can be differentiated from one another with adjustment of experimental parameters, particularly humidity. However, at higher concentrations of dimethyl sebacate, the response from DBE-3 can be masked, limiting the dynamic range over which the DBE-3 can be quantified as only the monomer peak of dimethyl glutarate can be used for quantification. 3 P a g e

4 Introduction Dimethyl sebacate (DMS) and DBE-3 are dibasic esters commonly used as solvents in the cleaning of PET bottles prior to filling. DBE-3 is a combination of two dibasic esters: 90% dimethyl adipate (DMA) and 10% dimethyl glutarate (DMG). The properties of these compounds are given intable 1. All three are made up of a carbon chain with a methyl ester functional group at each end. The US Food and Drug Administration (FDA) operates a food contact substance notification program under which potentially harmful substances that come into contact with foodstuffs are regulated via Food Contact Notices (FCN). DMS and DBE-3 are be subject to FCNs which stipulate the maximum permissible concentrations of the compounds in soft drinks contained in PET bottles. Compound Formula Length of carbon Molecular weight / g mol -1 chain DMS C 12 H 22 O DMA C 8 H 14 O DMG C 7 H 12 O Table 1 Properties of dibasic esters. All three compounds gain a positive charge during ionisation and therefore appear in the positive mode of the Lonestar spectra. Peroxyacetic acid may also be present in PET bottles; however it gains a negative charge upon ionisation and hence appears in the negative ion mode spectra produced by the Lonestar device, and thus does not interfere with the dibasic ester peaks. 4 P a g e

5 The Lonestar Platform Lonestar is a powerful and adaptable chemical monitor in a portable self contained unit. Incorporating Owlstone s proprietary FAIMS technology, the instrument offers the flexibility to provide rapid alerts and detailed sample analysis. It can be trained to respond to a broad range of chemical scenarios and can be easily integrated with other sensors and third party systems to provide a complete monitoring solution. As a result, Lonestar is suitable for a broad variety of applications ranging from process monitoring to lab based R&D. Figure 1 Lonestar connections and dimensions 5 P a g e

6 Experimental Setup Due to the form of the sample (PET bottle) and in order to keep the sampling mechanism simple, a positive pressure flush of the bottles into the Lonestar unit using clean, dry air was chosen as the sampling method for this application. To investigate suitable parameters for the detection method clean dry air was passed through the headspace of a 500 ml Duran bottle containing a dilution of DMS, DBE-3 or both in water and into the Lonestar (Figure 2). A hotplate was used to maintain the temperature of the sample at 30 C, removing sample temperature as a variable. Figure 2 Setup for sampling the headspace of a solution of analyte. For some of the work, Owlstone Vapour Generators (OVGs) with DBE-3 and DMS permeation sources were used to provide a constant, known concentration of the analytes to the Lonestar. 6 P a g e

7 This setup is shown in Figure 3. Either or both of the two OVGs can be connected to the inlet of the Lonestar, with a makeup flow from the lab air line to make up a consistent total flow. The flow can also be directed through the headspace of a Duran bottle containing filtered water, to increase the humidity if necessary. Figure 3 - Setup with OVGs. Pressure and total flow through Lonestar are as shown in previous figure. 7 P a g e

8 Method and Results Standard solution preparation The maximum permissible concentrations of DBE-3 and DMS in PET bottles, once they have been filled with soft drink, is 750 μg l -1 and 6000 μg l -1 respectively. The volume of the smallest size of soft drink bottle trialled was 237 ml. It was decided to make up standard solutions that would give 10% less than these concentrations (675 μg l -1 DBE-3 and 5400 μg l -1 DMS) when 1 ml of standard solution was added to 236 ml water. This would represent the lowest concentration that needed to be detected. Accordingly, the following standard solutions were made up: μg l -1, or 160 ppm, DBE μg l -1, or 640 ppm, DMS At this concentration, DMS crystallised out of the solution. It was necessary to warm the solution above 30 C in a water bath and sonicate it before it could be used to make up solutions at lower concentrations. Lonestar parameters The operational settings of the Lonestar are seen in Table 2 below. Gas flow rate / SLPM 2.05 ± 0.05 Pressure / bar g 1.3 Field intensity range / % No. of CV sweeps to average 1 No. of lines in DF matrix 31 Gas recirculation Off Filter heater On Sensor heater On Table 2 Lonestar parameters during data collection. SLPM = standard litres per minute, CV = compensation voltage, DF = dispersion field. Peak identification with Mass spectrometry It is useful to identify the ions produced by DME-3 and DMS to facilitate parameter selection for selectivity. Monomer, dimer and hydration level of ion affect the trajectory through the ion selection field and the ion formed is analyte dependent. Due to the nature of the measurement dimer formation will always be suppressed when using the mass spectrometer, so the results can only be used as a guide. Gas phase ion analysis was carried out using permeation sources 8 P a g e

9 of DBE-3 and DMS incubated in Owlstone Vapour Generators (OVGs) to generate controlled ppb levels of analyte. The parameters and output, determined gravimetrically, of the permeation sources are shown in table 6 below. The output of the OVGs was introduced into a mass spectrometer fitted with a 63 Ni radiation source the same as that in a Lonestar. A make-up flow of 1.4 l min -1 was used. The resultant concentrations of DBE-3 and DMS are also given in the table. Permeation rate / Analyte Air flow rate / ml Concentration of Concentration of min -1 ng min -1 OVG output / ppb MS input / ppb DBE ± ± ± 1 DMS ± ± ± 2 Table 3 Properties of the dibasic ester permeation sources The molecular weight of DMS is g/mol. DBE-3 is made up of 90% dimethyl adipate (DMA), with molecular weight g/mol, and 10% dimethyl glutarate (DMG), molecular weight g/mol. When 62 ppb DBE-3 in air was introduced into the mass spectrometer, peaks were seen at and 160.9, as seen in Figure 4. There was a slight shift in m/z of around +0.7, indicating that the mass spectrometer needed recalibrating. As expected, the 174 peak was larger than that at 160. With 46 ppb DMS in air, a peak at 230 was seen ( Figure 5). Figure 4 - Mass spectrogram of dry 62 ppb DBE-3. Figure 5 - Spectrogram of dry 46 ppb DMS. When the air flow containing the analyte was passed through the headspace of a bottle of water before entering the mass spectrometer, the spectrograms in Figure 6 and Figure 7 below resulted. In the spectrogram for DBE-3 (Figure 6), peaks are seen at 174 and 365. (The peak at 230 is due to lingering DMS in the system.) This corresponds to a monomer of DMA and a dimer of DMA plus a water molecule. No peaks are seen at 160 or 337, which would be expected as the corresponding monomer and dimer of DMG, suggesting the charge has been transferred to the DMA via competitive ionisation. 9 P a g e

10 Figure 6 Mass spectrogram of 62 ppb DBE-3 in wet air. In the spectrogram for DBE-3 (Figure 7, below), peaks are seen at 230 and 477. This corresponds to a monomer of DMS and a dimer of DMS plus a water molecule. Figure 7 Mass spectrogram of 46 ppb DMS in wet air. The spectrogram resulting from the combination of 62 ppb DBE-3 and 46 ppb DMS in wet air is shown in Figure 8. Figure 8 Mass spectrogram of 62 ppb DBE-3 and 46 ppb DMS in wet air. Peaks are seen at 175 and 231 corresponding to DMA and DMS. There is a small peak at 477 and a very small one at 365, corresponding to the dimer plus a water molecule for each analyte. There are no peaks visible at 160 or 337. Sampling from the headspace of a solution of DBE-3 and/or DMS will give a concentration of the analytes at the Lonestar chip comparable to these 10 P a g e

11 concentrations or lower which gives a quick indication that the detection problem is viable. The additional sensitivity FAIMS can achieve over mass spec is partially due to the clustering of ions with water at the higher pressures, so dimer formation would be stronger than the data suggests and competition for charge between analytes is less. No unusual ion formation mechanisms were observed. Initial DF spectra Both DBE-3 and DMS showed up very strongly on Lonestar DF spectra. The peaks due to both analytes are in the positive mode (shown with a blue background). A blank matrix from before sampling began is shown in Figure 9. The Reactive ion peak (RIP) is labelled. This peak is due to the protonated water- hydronium generated from the carrier gas stream a similar peak in the negative mode is due to hydrated O 2 - ions. RIP Figure 9 Blank matrix (positive mode) before collecting data In Figure 10 below, equivalent DF matrices are shown for sampling from the headspace of a Duran bottle containing water samples with 1.6 ppm DBE-3 and 10 ppm DBE-3concentrations. The Lonestar shows a strong response. At 10 ppm DBE-3 the RIP has nearly disappeared, meaning the sensor is almost saturated. The large peak (1) was thought to be due to the dimer of DMA and the smaller peak (2) to the monomer of DMA. 11 P a g e

12 Figure 10 Positive mode matrix on sampling headspace of 1.6 ppm DBE-3 in water (left) and 10 ppm DBE-3 in water (right). In Figure 11 below, similar DF matrices are shown for 1.6 ppm DMS and 16 ppm DMS. In both matrices there is a strong peak due to the DMS dimer and a very faint peak due to the monomer Figure 11 Positive mode matrix on sampling headspace of 1.6 ppm DMS in water (left) and 16 ppm DMS in water (right). Initial sampling of a combination of the dibasic esters is shown in Figure 12 (1.6 ppm DBE-3 and 1.28 ppm DMS). In this situation two to four peaks might be expected, from the mass spectrometry work above: monomer and possibly dimer peaks of DMS and DMA, but not DMG. 12 P a g e

13 Figure ppm DBE-3 and 1.28 ppm DMS In fact, only one peak was seen, probably caused by the peaks overlapping. Further work focussed on changing the experimental conditions to separate the peaks due to the DBE-3 and DMS. Peracetic acid was also sampled as a possible trace chemical which could be present in the PET bottles. Peracetic acid was found to produce no positive mode ion response; however a significant number of negative ions could be resolved in a 50ppm sample (Figure 13). Peracetic acid Figure 13 - Peracetic acid response at 50ppm in the negative ion mode 13 P a g e

14 Ion current / AU DBE-3 low concentration calibration An example instrument calibration for DBE-3 was performed. Known volumes of each standard solution were pipetted into filtered water to give give samples with the following concentrations: DBE-3: 50 ppb, 400 ppb, 1.6 ppm, 4 ppm, 5 ppm, 10 ppm, 20 ppm The first DF matrix at each concentration was discarded and the next three used to create the calibration. Owlstone s proprietary EasySpec software was used to extract the height of the major peak at 60% DF in each matrix. For DBE-3, this was the DMA dimer peak. The average peak height (in arbitrary units, AU) is plotted against the concentration to give the calibration. The DBE-3 calibration is linear up to about 5 ppm and then flattens out as the sensor is saturated. The calibration between 50 ppb and 4 ppm, within this linear range, is shown in Figure 14 with a line of best fit. DBE-3 low-concentration calibration y = 1.18E-04x E-01 R 2 = 9.94E Concentration / ppb Figure 14 DBE-3 calibration between 50 ppb and 4 ppm 14 P a g e

15 Method optimisation The pressure (number of collisions) and flow rate (ion velocity) through the Lonestar sensor along with the humidity of the carrier gas affect the ion trajectory through the dispersion field. Consequently they can be used to optimise selectivity by adjusting peak separation for different analyte and background combinations. Figure 15 shows DBE3 at 90% of the FCN limit under optimised pressure/flow/humidity conditions. As can be seen, the monomer/dimer structure was clearly resolved. Figure 15 DBE-3 with initial parameter optimisation boosting peak separation Applying the adjustments of input conditions to give the enhanced peak resolution also allows the mixture of dibasic esters to be resolved, Figure 16 is a DF matrix from headspace sampling of a 4.8 ppm DBE-3 and 0.6 ppm DMS solution, with humidity of around 19%. At 72% DF, the DMS and DMA peaks are well-defined. Since DMS has a much greater ionisation potential than DMA and DMG, a larger concentration of DBE-3 than DMS is required to resolve the DBE-3 peaks. 15 P a g e

16 DMS Dimers DMA Figure 16 Headspace sampling of 4.8 ppm DBE ppm DMS solution, humidity 19%. DF matrix Background testing of PET bottles Two holes were drilled in a bottle lid and fitted with grommets to accommodate PTFE tubing. The PET bottles could then be connected to the Lonestar to sample the flow passing through them. The backgrounds of one large (474 ml) and one small (237 ml) bottle were sampled, first dry and then with approximately 20 ml added water. A flow of 2.2 l min -1 and pressure of 1.3 bar g were used. Results are shown in Figure 17 to Figure 19 below. Figure 17 Sampling larger bottle: dry (left) and wet (right). 16 P a g e

17 Figure 18 Ion current vs. CV at 80% DF for the larger bottle: dry (left) and wet (right). Figure 19 Sampling smaller bottle: dry (left) and wet (right). Both bottles showed a weak background peak in the same place as the DBE-3 and DMS peaks. When sampled wet, there was some peak separation into the DMA monomer and combined dimer peaks: see Figure 18. Despite the bottles having been cleaned, a small amount of the DBE-3 and DMS may have permeated into the walls of the bottle which was then detectable (more in the larger bottle due to the larger surface area). The ion current values correspond to lower concentrations than the earlier calibration suggesting low ppb concentration of the solvents being present. In these DF matrices, the DMA peak is not intense enough to be used for quantification of the DBE-3 concentration, but optimisation work could improve this. The DMS peaks here could be used for quantification and could also be optimised. 17 P a g e

18 Bottle Interface with Lonestar In order to connect the bottles to the Lonestar quickly and ensure an airtight seal when sampling the residual solvents a custom bottle seal is required. Figure 20 shows the Lonestar bottle fitting mechanism, that allows quick one handed attachment of PET bottles to the instrument. Figure 20 Lonestar bottle fitting mechanism The user inserts the bottle into the inlet and then applies a clamp with a rotating handle. The large handle on the clamp easily provides a large torque which pushes the bottle up towards the Lonestar inlet, compressing the O-ring to make the seal. This system combines an easy to use mechanism with a simple inlet. The force applied to the O-ring seal is linear, preventing substantial wear to the seal, increasing O-ring lifetime. Figure 21 outlines the bottle fitting process. 18 P a g e

19 Figure 21 Lonestar bottle fitting process 19 P a g e

20 Conclusions The following conclusions can be drawn from the testing Dimethyl sebacate and DBE-3 can be detected at ppb concentration levels, significantly below the FCN contact limits. DMS and DBE-3 can be separated from one another, with adjustment of the sampling parameters, the monomer peaks of DMS and DMA are suitable peaks for selectivity, though the dimer gives more dynamic range. DMS gives a greater response than DMA, which gives a greater response than DMG. This is thought to be due to the increased stability of the molecular ions as the length of the carbon chain increases. Quantitation of the solvents has been demonstrated over a range from 50 ppb to 20 ppm for DBE-3. The lower limit of detection for DMS has not been determined but as the instrument is more sensitive to DMS will be significantly lower than 50 ppb, quantitation between 1 ppm and 16 ppm has been demonstrated. Lonestar is capable of detecting dimethyl esters in the background of PET bottles. The PET bottles provided showed trace solvent levels despite being cleaned. Three concepts for a custom sampling interface for attaching the bottles have been drafted and presented. 20 P a g e

21 Appendix A: FAIMS Technology at a Glance Field asymmetric ion mobility spectrometry (FAIMS), also known as differential mobility spectrometry (DMS), is a gas detection technology that separates and identifies chemical ions based on their mobility under a varying electric field at atmospheric pressure. Figure 22 is a schematic illustrating the operating principles of FAIMS. RF waveform 0V Pk to Pk V Ionisation source Electrode channel CV Ion count Air /carrier gas flow direction Figure 22 FAIMS schematic. The sample in the vapour phase is introduced via a carrier gas to the ionisation region, where the components are ionised via a charge transfer process or by direct ionisation, dependent on the ionisation source used. It is important to note that both positive and negative ions are formed. The ion cloud enters the electrode channel, where an RF waveform is applied to create a varying electric field under which the ions follow different trajectories dependent on the ions intrinsic mobility parameter. A DC voltage (compensation voltage, CV) is swept across the electrode channel shifting the trajectories so different ions reach the detector, which simultaneously detects both positive and negative ions. The number of ions detected is proportional to the concentration of the chemical in the sample Sample preparation and introduction FAIMS can be used to detect volatiles in aqueous, solid and gaseous matrices and can consequently be used for a wide variety of applications. The user requirements and sample matrix for each application define the sample preparation and introduction steps required. There are a wide variety of sample preparation, extraction and processing techniques each with their own advantages and disadvantages. It is not the scope of this overview to list them all, only to highlight that the success of the chosen application will depend heavily on this critical step, which can only be defined by the user requirements. There are two mechanisms of introducing the sample into the FAIMS unit: discrete sampling and continuous sampling. With discrete sampling, a defined volume of the sample is collected by weighing, by volumetric measurement via a syringe, or by passing vapor through an adsorbent for pre-concentration, before it is introduced into the FAIMS unit. An example of this would be attaching a container to the instrument containing a fixed volume of the sample. A carrier gas (usually clean dry air) is used to transfer the sample to the ionization region. Continuous sampling is where the resultant gaseous sample is continuously purged into the Detector Sample Ionization Ion separation based on mobility Detection Exhaust Preparation and introduction 21 P a g e

22 FAIMS unit and either is diluted by the carrier gas or acts as the carrier gas itself. For example, continuously drawing air from the top of a process vat. The one key requirement for all the sample preparation and introduction techniques is the ability to reproducibly generate and introduce a headspace (vapour) concentration of the target analytes that exceeds the lower limits of detection of the FAIMS device. Carrier Gas The requirement for a flow of air through the system is twofold: Firstly to drive the ions through the electrode channel to the detector plate and secondly, to initiate the ionization process necessary for detection. As exhibited in Figure 23, the transmission factor (proportion of ions that make it to the detector) increases with increasing flow. The higher the transmission factor, the higher the sensitivity. Higher flow gives a larger full width half maximum (FWHM) of the peaks but also decreases the resolution of the FAIMS unit (see Figure 24). Figure 23 Flow rate vs. ion transmission factor The air/carrier gas determines the baseline reading of the instrument. Therefore, for optimal operation it is desirable for the carrier to be free of all impurities (< 0.1 ppm methane) and the humidity to be kept constant. It can be supplied either from a pump or compressor, allowing for negative and positive pressure operating modes. Ionisation Source There are three main vapor phase ion sources in use for Figure 24 FWHM of ion species at atmospheric pressure ionization; radioactive nickel-63 set CV (Ni-63), corona discharge (CD) and ultra-violet radiation (UV). A comparison of ionization sources is presented in Table 4. Ionisation Source Mechanism Chemical Selectivity Ni 63 (beta emitter) creates a positive / negative RIP Charge transfer Proton / electron affinity UV (Photons) Direct ionisation First ionisation potential Corona discharge (plasma) creates a positive / negative RIP Table 4 FAIMS ionization source comparison Charge transfer Proton / electron affinity 22 P a g e

23 Ni-63 undergoes beta decay, generating energetic electrons, whereas CD ionization strips electrons from the surface of a metallic structure under the influence of a strong electric field. The generated electrons from the metallic surface or Ni-63 interact with the carrier gas (air) to form stable +ve and -ve intermediate ions which give rise to reactive ion peaks (RIP) in the positive and negative FAIMS spectra (Figure 25). These RIP ions then transfer their charge to neutral molecules through collisions. For this reason, both Ni-63 and CD are referred to as indirect ionization methods. For the positive ion formation: N 2 + e- N e- (primary) + e- (secondary) N N 2 N N 2 N H 2 O 2N 2 + H 2 O + H 2 O+ + H 2 O H 3 O + + OH H 3 O+ + H 2 O + N 2 H + (H 2 O) 2 + N 2 H + (H 2 O) 2 + H 2 O + N 2 H + (H 2 O) 3 + N 2 For the negative ion formation: O 2 + e- O 2 - B + H 2 O + O 2 - O 2 - (H 2 O) + B B + H 2 O + O 2 - (H 2 O) O 2 - (H 2 O) 2 + B The water based clusters (hydronium ions) in the positive mode (blue) and hydrated oxygen ions in the negative mode (red), are stable ions which form the RIPs. When an analyte (M) enters the RIP ion cloud, it can replace one or dependent on the analyte, two water molecules to form a monomer ion or dimer ion respectively, reducing the number of ions present in the RIP. Monomer Dimer H + (H 2 O) 3 + M + N 2 MH + (H 2 O) 2 + N 2 + H 2 O M 2 H + (H 2 O) 1 + N 2 + H 2 O Dimer ion formation is dependent on the analyte s affinity to charge and its concentration. This is illustrated in Figure 25A using dimethyl methylphsphonate (DMMP). Plot A shows that the RIP decreases with an increase in DMMP concentration as more of the charge is transferred over to the DMMP. In addition the monomer ion decreases as dimer formation becomes more favourable at the higher concentrations. This is shown more clearly in Figure 25B, which plots the peak ion current of both the monomer and dimer at different concentration levels. 23 P a g e

24 Dimer RIP Monomer Figure 25 DMMP Monomer and dimer formation at different concentrations The likelihood of ionization is governed by the analyte s affinity towards protons and electrons (Table 5 and Table 6 respectively). In complex mixtures where more than one chemical is present, competition for the available charge occurs, resulting in preferential ionisation of the compounds within the sample. Thus the chemicals with high proton or electron affinities will ionize more readily than those with a low proton or electron affinity. Therefore the concentration of water within the ionization region will have a direct effect on certain analytes whose proton / electron affinities are lower. Chemical Family Example Proton affinity Aromatic amines Pyridine 930 kj/mole Amines Methyl amine 899 kj/mole Phosphorous Compounds TEP 891 kj/mole Sulfoxides DMS 884 kj/mole Ketones 2- pentanone 832 kj/mole Esters Methly Acetate 822 kj/mole Alkenes 1-Hexene 805 kj/mole Alcohols Butanol 789 kj/mole Aromatics Benzene 750 kj/mole Water 691 kj/mole Alkanes Methane 544 kj/mole Table 5 Overview of the proton affinity of different chemical families 24 P a g e

25 Chemical Family Electron affinity Nitrogen Dioxide Chlorine Organomercurials Pesticides Nitro compounds Halogenated compounds Oxygen Aliphatic alcohols Ketones 3.91eV 3.61eV 0.45eV Table 6 Relative electron affinities of several families of compounds The UV ionization source is a direct ionization method whereby photons are emitted at energies of 9.6, 10.2, 10.6, 11.2, and 11.8 ev and can only ionize chemical species with a first ionization potential of less than the emitted energy. Important points to note are that there is no positive mode RIP present when using a UV ionization source and also that UV ionization is very selective towards certain compounds. Mobility Ions in air under an electric field will move at a constant velocity proportional to the electric field. The proportionality constant is referred to as mobility. As shown in Figure 26, when the ions enter the electrode channel, the applied RF voltages create oscillating regions of high (+V HF ) and low (-V HF ) electric fields as the ions move through the channel. The difference in the ion s mobility at the high and low electric field regimes dictates the ion s trajectory through the channel. This phenomenon is known as differential mobility. +V HF +V HF -V LF Difference in mobility 0V -V LF d Duty Cycle = d/t t Pk to Pk V Figure 26 Schematic of a FAIMS channel showing the difference in ion trajectories caused by the different mobilities they experience at high and low electric fields Figure 27 Schematic of the ideal RF waveform, showing the duty cycle and peak to peak voltage (Pk to Pk V) The physical parameters of a chemical ion that affect its differential mobility are its collisional cross section and its ability to form clusters within the high/low regions. The environmental factors within the electrode channel affecting the ion s differential mobility are electric field, humidity, temperature and gas density (i.e. pressure). 25 P a g e

26 The electric field in the high/low regions is supplied by the applied RF voltage waveform (Figure 27). The duty cycle is the proportion of time spent within each region per cycle. Increasing the peak-to-peak voltage increases/decreases the electric field experienced in the high/low field regions and therefore influences the velocity of the ion accordingly. It is this parameter that has the greatest influence on the differential mobility exhibited by the ion. It has been shown that humidity has a direct effect on the differential mobility of certain chemicals, by increasing/decreasing the collision cross section of the ion within the respective low/high field regions. The addition and subtraction of water molecules to analyte ions is referred to as clustering and de-clustering. Increased humidity also increases the number of water molecules involved in a cluster (MH + (H 2 O) 2 ) formed in the ionisation region. When this cluster experiences the high field in between the electrodes the water molecules are forced away from the cluster reducing the size (MH + ) (de-clustering). As the low field regime returns so do the water molecules to the cluster, thus increasing the ion s size (clustering) and giving the ion a larger differential mobility. Gas density and temperature can also affect the ion s mobility by changing the number of ion-molecule collisions and changing the stability of the clusters, influencing the amount of clustering and de-clustering. Changes in the electrode channel s environmental parameters will change the mobility exhibited by the ions. Therefore it is advantageous to keep the gas density, temperature and humidity constant when building detection algorithms based on an ion s mobility as these factors would need to be corrected for. However, it should be kept in mind that these parameters can also be optimized to gain greater resolution of the target analyte from the background matrix, during the method development process. Detection and Identification As ions with different mobilities travel CV = -5V CV -6V down the electrode channel, some will have trajectories that will result in ion annihilation against the electrodes, CV 6V CV = 0V Detector whereas others will pass through to hit CV -6V the detector. To filter the ions of different mobilities onto the detector plate a compensation voltage (CV) is CV 6V CV = 5V Detector scanned between the top and bottom CV-6V electrode (see Figure 28). This process realigns the trajectories of the ions to hit the detector and enables a CV spectrum CV 6V Detector to be produced. The ion s mobility is thus expressed as a Figure 28 Schematic of the ion trajectories at compensation voltage at a set electric different compensation voltages and the field. Figure 29 shows an example CV resultant FAIMS spectrum spectrum of a complex sample where a de-convolution technique has been employed to characterize each of the compounds. - 6V - 0V +6V - 6V - 0V +6V - 6V - 0V +6V 26 P a g e

27 represented by the color contours. P1 P2 P3 Figure 29 Example CV spectra. Six different chemical species with different mobilities are filtered through the electrode channel by scanning the CV value P4 P5 P6 Changing the applied RF peak-to-peak voltage (electric field) has a proportional effect on the ion s mobility. If this is increased after each CV spectrum, a dispersion field matrix is constructed. Figure 30 shows two examples of how this is represented; both are negative mode dispersion field (DF) sweeps of the same chemical. The term DF is sometimes used instead of electric field. It is expressed as a percentage of the maximum peak-topeak voltage used on the RF waveform. The plot on the left is a waterfall image where each individual CV scan is represented by compensation voltage (x-axis), ion current (yaxis) and electric field (z-axis). The plot on the right is the one that is more frequently used and is referred to as a 2D color plot. The compensation voltage and electric field are on the x, and y axes and the ion current is RIP PIP RIP PIP Electric Field Compensation voltage Figure 30 Two different examples of FAIMS dispersion field matrices with the same reactive ion peaks (RIP) and product ion peaks (PIP). In the waterfall plot on the left, the z axis is the ion current; this is replaced in the right, more frequently used, colorplot by color contours With these data rich DF matrices a chemical fingerprint is formed, in which identification parameters for different chemical species can be extracted, processed and stored. Figure 31 shows one example: here the CV value at the peak maximum at each of the different electric field settings has been extracted and plotted, to be later used as a reference to identify the same chemicals. In Figure 32 a new sample spectrum has been compared to the reference spectrum and clear differences in both spectra can be seen. 27 P a g e

28 DF / % DF / % PPIP 1 PPIP PPIP Pork at 7 Days PPIP CV / V NPIP NPIP 2 NPIP 1 NPIP NPIP 1 NPIP 3 Pork at 7 Days CV / V Figure 31 On the left are examples of positive (blue) and negative (red) mode DF matrices recorded at the same time while a sample was introduced into the FAIMS detector. The sample contained 5 chemical species, which showed as two positive product ion peaks (PPIP) and three negative product ion peaks (NPIP). On the right, the CV at the PIP s peak maximum is plotted against % dispersion field to be stored as a spectral reference for subsequent samples. Figure 32 Comparison of two new DF plots with the reference from Figure 10. It can be seen that in both positive and negative modes there are differences between the reference product ion peaks and the new samples 28 P a g e