THE EXTRACTION OF MAXIMUM INFORMATION FROM INDIVIDUAL ION ARRIVALS AND ITS APPLICATION TO EXTENDING THE DYNAMIC RANGE OF IMS-OATOF-MS DATA. Authors: Martin Green*1, Darrell Williams1, Garry Scott1, Tony Gilbert1, Martin Palmer1, Nick Tomczyk1, Keith Richardson1, Mark Wrona2 Affiliations: 1. Waters Corporation, Wilmslow UK, 2. Waters Corporation, Milford, MA, USA PURPOSE: Interrogate properties of individual ion arrivals in time of flight MS using fast FPGA processing. Digitised ion arrival events within single time of flight transients were processed using the FPGA and the presence of saturated ADC samples recorded and stored with each point in the summed spectra. If only a small percentage of ion arrivals contain saturated samples minimal distortion will be caused in the final data. bovine res test old det 2550V 100 bovine res test old det 2550V 100 % 250 Combined high dynamic range data High transmission data Low transmission data INTRODUCTION The emergence of orthogonal time-of-flight mass spectrometry has been made possible, in great part, by the rapid development of high-speed digital electronics. Developments include; increased speed and dynamic range of analogue to digital conversion, high speed data transfer protocols, large rapidly accessible memory and high-performance field programmable gate arrays (FPGAs). FPGA technology has allowed sophisticated data processing to be applied to individual ion arrivals, enhancing signal-to-noise-ratio and arrival time precision and enabling information about each ion arrival to be extracted and stored. In this paper, fast digital processing is used to examine each ion arrival and record information in the final mass spectrum. A method of increasing dynamic range utilising this information is described. METHODS All data were acquired using a Waters SYNAPT G2-Si mass spectrometer (Figure 1) using an ACQUITY UPLC system. 100 Memory 50 0 957 0 Time 0 Arrival Time Intensity Meta data 957 m/z 958 958 m/z Metadata: data:Stored Stored with Meta with each intensity / time each intensity / time pointininspectrum spectrum point Figure 2 Simplified schematic of ADC acquisition architecture. Figure 2 shows a simplified schematic of the Synapt data recording architecture. Signals from individual ion strikes at the electron multiplier detector were first digitised and then processed in real time within the FPGA before being summed in onboard memory. In addition to arrival time and intensity information for each transient, “metadata” (Table 1), describing other properties of the transients may be extracted and associated with each data point in the final summed spectra. These data are available for subsequent post processing. Area LHS, RHS Width at base Quantiles Width LHS, RHS Standard Deviation Skew Centre of mass Kurtosis Maximum Width FWHM Saturated points Table 1 Examples of metadata which may be stored. Table 1 lists some of the meta data which may be extracted and stored from individual ion arrivals within each time-offlight transient using the processing power of the FPGA. In the following example the presence of (or number of) ADC digitisation samples which exceed the vertical dynamic range of the ADC (saturated samples) were recorded for each detected ion arrival event and this information stored with each time-point in the summed spectra. UPLC Conditions: Small molecule Mix: Verapamil, Suphadimethoxine, Leucine Enkephalin, Caffeine, Acetaminophen. Column: Acquity BEH C18 1.7μm 2.1x50mm Mobile phase: A. Water +0.1% formic acid B. Acetonitrile +0.1% formic acid Gradient: 0 to 5mins 10% to 98% B RECORDING % SATURATION 250 200 150 100 UPLC Conditions: Propanolol in Human Plasma Figure 3 shows a representation of a signal produced by multiple ion arrivals within a single ToF transient. In this case the signal exceeds the vertical dynamic range of the ADC of 8bits i.e. 256. 50 Saturation of signals within time of Time flight transient at high ion input flux can lead to distortion in intensity and time of flight measurement. Figure 3 Ion arrival Extreme distortion results in errors in event exhibiting ADC quantification and mass measurement. intensity saturation. 0 Column: Cortecs UPLC C18+ 1.6μm 2.1x50mm Mobile phase: A. Water +0.1% formic acid B. Acetonitrile +0.1% formic acid Gradient: 0 to 4mins 2% to 60% B 4 to 7mins 60 to 95% B Once summation is finished, each point is marked with a saturation flag if the percentage of saturated events exceeds a preset threshold. Region of linear intensity response 0 In addition, calculation and storage of the ratio of these values can be calculated. For example the ratio of the maxima to the area for a given transient may be stored. Figure1. Synapt G2 Si Rather than record a saturation flag every time saturation has been detected, the percentage of saturation at each location in the summed spectra is calculated during summation. The number of saturated samples in an individual ion arrival may be recorded and used to indicate the extent of saturation. % arrivals saturating ADC RESULTS: Alternating high and low transmission data combined point by point based on saturation flags. 5-10x improvement in dynamic range. FPGA 150 Relative intensity METHOD: Record saturated samples within individual ion arrivals. Calculate and store percentage of saturated arrivals with individual spectral points. Create extended dynamic range LC-IMS-MS data. ADC % 200 Detector 10 20 Arrival rate (ions / push) Figure 4. intensity vs average ion arrival rate 100 35% of all ion arrivals exceed the dynamic range of the ADC 80 60 40 20 0 10 0 10 20 30 40 time Figure 6 Chromatogram illustrating transmission switching dynamic range enhancement (DRE). Using the architecture described, each point in the continuum spectra carries a record of the percentage of saturated events. Combining the high- and low-transmission data may be performed directly by replacing flagged points in the high transmission data with corresponding points from the lowtransmission data Figure 9. This may be performed in real time as data are read from the ADC memory allowing high dynamic range LC-IMS-MS data to be produced. RESULTS: LC-IMS-MS 20 Arrival rate (ions / push) Figure 5. % saturated events vs average ion arrival rate Figure 4 shows a plot of relative intensity vs average ion arrival rate for an 8 bit ADC assuming an average height for a single ion arrival equivalent to 8 LSB and a Poisson ion arrival distribution. a: No DRE b: With DRE It can be seen that no significant distortion occurs in the measurement of response below an ion arrival rate 20 ions / push. Where ‘push’ refers to a time-of-flight transient recorded from a single orthogonal sampling of the axial ion beam. Figure 5 shows a plot of the proportion of ion arrivals which contain at least one ADC sample which exceeds 8 bits vs the ion arrival rate. At 20 ions per push approximately 35% of all ion arrivals are saturated to some degree. A threshold of >35% saturated events was chosen to determine when to associate a saturation flag with a location in the summed spectra. CONTINUUM DYNAMIC RANGE ENHANCEMENT (DRE) The transmission switching method (DRE) has been shown to increase the dynamic range of time of flight data1. a: No DRE Verapamil Error 15.8ppm b: With DRE TO DOWNLOAD A COPY OF THIS POSTER, VISIT WWW.WATERS.COM/POSTERS 4000 RESULTS: PROPRANOLOL IN HUMAN PLASMA 3000 2000 E Non-DRE IMS-MS data Figure 10a was acquired at a spectral rate of 5 spe/sec. In DRE mode Figure 10b, 100ms of full transmission data and 100ms of 5% transmission data were recorded at each collision energy. Data points in the 100% transmission data containing a saturation flag were replaced by corresponding points from the 5% transmission data scaled accordingly. Figure 10a. 100,000pg/ml Propanolol in Human plasma. Non –DRE 0 0 E HDMS was performed. In HDMS the post-IMS, CID fragmentation energy is switched repeatedly between low and high values while recording full IMS-MS nested data. High peak capacity precursor ion and product ion data are produced. Precursors and products may be associated by both retention time and IMS drift time or collision cross section. 20000 40000 60000 80000 100000 Concentration pg/ml Figure 11 quantification curve a) with and b) without DRE DISCUSSION In the LC-IMS experiments shown, 100 ms of IMS-MS data was acquired at each transmission value. 200 individual mass spectra were recorded within each 100ms IMS-MS nested data set. Combining low- and high-transmission data occurs in real time at a rate of 2000 MS spectra per second to produce a seamless, continuum data set with up to 10x increase in dynamic range and improved isotope and fragment ion ratios. This dramatic post-processing rate is made possible by the ability of the FPGA extract information from each ion arrival signal and record this data with each data point in the summed spectrum. As FPGA processing power continues to increase, the potential for more complex calculations within individual time-of-flight transients increases, improving mass-spectral data quality . CONCLUSION FPGA technology allows information from individual ion arrivals to be extracted and recorded with final spectra. Recording the proportion of ADC saturation with each mass spectral point allows high and low transmission IMS-MS data to be ‘stitched’ in real time. 10x dynamic range improvements achieved. Continuing development of FPGAs will facilitate more complex processing of individual ion arrivals, leading to higher quality time-of-flight data. Error 3.5 ppm Verapamil Figure 10b. 100,000pg/ml Propanolol in Human plasma. With –DRE b) With—DRE 1000 To demonstrate applicability to quantification of small molecules Propanolol was spiked into protein precipitated human plasma at a concentration of 100pg/mL to 100,000pg/ mL. Figure 7 LC-IMS-MS, DT vs RT a) no DRE, b) with DRE Figures 7 a and b show the results of LC-IMS-MS separation of a mixture of six small molecules at 500pg on-column without DRE and with DRE. A scan time of 100ms at full transmission and 100ms at 5% transmission was used. This results in an output spectral rate of 200ms / spectrum and an increase of 10x in dynamic range. Figure 8 shows an extracted mass chromatogram Verapamil m/z 455.2910 a) no DRE 15.8ppm mass measurement error and b) with DRE 3.5ppm mass measurement error. Figure 11 shows a plot of chromatographic response vs concentration (pg/ml) for Propanolol in human plasma without DRE [a] and with DRE [b] from 100pg/mL to 100,000pg/mL. The absolute sensitivity has been reduced by 2x for the DRE data due to the reduced duty cycle of this experiment. However, the linear dynamic range has been increased by between 5—10x from approx. 2 orders of magnitude to approx. 3 orders of magnitude. a) Without—DRE This produces wide dynamic range IMS-MS spectra at a spectral rate of 5 spec per second and with an overall duty cycle of approximately 50%. High and low transmission spectra are acquired on alternate scans. The high and low transmission data are then combined into a single, wide dynamic range spectrum replacing saturated peaks with corresponding peaks from the low transmission data scaled appropriately Figure 6. Previously this method was restricted to peak detected or centroided data. For LC-IMS-MS data up to 2000 spectra / second are produced making this method impractical to implement in ‘real-time’. Figure 9 Figure 9 shows a threedimentional LC-IMS-MS contour plot of the full transmission data in Figure 7a. Only the points containing greater than 35% of saturated ion arrivals are displayed. These points are replaced by the corresponding points in the 5% transmission data and scaled to give the high dynamic range data shown in Figure 7b. Response ACQUISITION ARCHITECTURE intensity OVERVIEW References Figure 8. extracted (m/z 455.2910),RT,DT plot m/z 500pg Verapamil a) without DRE, b) with DRE 1. Proceedings of the 49th ASMS Conference on Mass Spectrometry and Allied Topics, Chicago, Illinois, May 27-31, 2001 ©2014 Waters Corporation
