The Case for Fast micro GC and Chemometrics

1. The Case for Fast micro GC and Chemometrics Retention Time Alignment 6/9/2009

2. FalconFast microGC, the Clear Solution for “the Few of US” UBS newsletter, Investment Intelligence: As baby boom US workers retire, we create a shortage that follow-on generations have neither the numbers nor the skills to fill. Randy Shearer, GE: We are graduating fewer scientists and engineers in the US. We need to apply the continuing improvements in computer technology. Chromatography is a key area In GC, that means faster analysis Bill Winniford, DOW: We have to make do with a lot fewer people and they will have more to do. We are entering a time we have never seen before. Chief frustration is in data processing Not good at capturing knowledge from the experienced workforce 2

3. 3 Future Trends Miniaturized instrumentation Fast instrumentation Remote access to sample points Abundant data rich analyzers More precise control needed Changing regulations Changing products Convergence of labs and process analytics Resource limitations manpower, skill sets, materials, …

4. 4 Future Solutions To meet these challenges, we need: to implement analyzers capable of quickly measuring more critical process parameters; to extract more information from those analyzers; to better utilize computational advances; and to put more smarts into our analyzers. to recognize common instrument faults and communicate corrective actions to the appropriate party This will make our jobs manageable!

5. 5 The Clear Solution but What Does Use of Fast GC Mean? Lots of data: for example, 20 results per hour (480 per day) Lots of interpretation (are the results correct?) Lots of opportunity for errors without human intervention All this drives the need for automation analysis data collection preprocessing interpretation system suitability assessment results result validity fault identification pass/fail This means… Chemometrics!

6. Real World Example Set up at trade show AC Power Available Portable air pump H2 cylinder gas Resembles field environment Not a lab, less than ideal for stability Lots of ambient variables that can mask real sample variability.

7. Gasoline, Single Run from day 1, Channel 1-DB5, Channel 2-VB1701 This is an example of the gasoline sample runs.This is an example of the gasoline sample runs.

8. Gasoline, Single Run from Day 1 Overlaid on Single Run from Day 2, Channel 1-DB5, Channel 2-VB1701 The samples were properly sealed and stored. The unit was turned off each evening and restarted the next morning. This is an overlay of one of the gasoline runs from day one on top of a run form day two. It demonstrates that on the same sample, one can get good day to day reproducibility. Nevertheless, there is a surprise for us in these runs. The next slide shows the surprise but Brian’s talk is about finding the surprise.The samples were properly sealed and stored. The unit was turned off each evening and restarted the next morning. This is an overlay of one of the gasoline runs from day one on top of a run form day two. It demonstrates that on the same sample, one can get good day to day reproducibility. Nevertheless, there is a surprise for us in these runs. The next slide shows the surprise but Brian’s talk is about finding the surprise.

9. Fast GC – all 94 gasoline injections (all raw data from day 1 & day 2 using a single sample) Run-to-run retention time instability affects all forms of chromatography. As the number of peaks in a chromatographic run increases, the possibility of misassigning those peaks due to a time-axis shift increases as well. Run-to-run retention time instability affects all forms of chromatography. As the number of peaks in a chromatographic run increases, the possibility of misassigning those peaks due to a time-axis shift increases as well.

10. Fast GC – all 94 gasoline injections aligned A software algorithm can be called as part of the chromatographic method and aligns newly collected data to a target chromatogram. In this way, retention time shifts are corrected immediately without operator involvement and prior knowledge of the retention shift is not required. The software used is based on maximizing the correlation of one chromatogram to a “gold standard” chromatogram of a similar material. This is practical for shifts of up to 2 peak widths. A software algorithm can be called as part of the chromatographic method and aligns newly collected data to a target chromatogram. In this way, retention time shifts are corrected immediately without operator involvement and prior knowledge of the retention shift is not required. The software used is based on maximizing the correlation of one chromatogram to a “gold standard” chromatogram of a similar material. This is practical for shifts of up to 2 peak widths.

11. PCA scores before and after alignment To visualize the improvement in consistency, I have taken the chromatograms and did a PCA. Here, every chromatogram is represented by a single point and the closer two points are to one another, the more similar the chromatograms. The lion’s share of the variability in the original chromatograms is clearly due to retention shifts. To visualize the improvement in consistency, I have taken the chromatograms and did a PCA. Here, every chromatogram is represented by a single point and the closer two points are to one another, the more similar the chromatograms. The lion’s share of the variability in the original chromatograms is clearly due to retention shifts.

12. PCA scores, aligned data only When this shift is removed and the PCA re-applied, the tight knot of data does separate, one cluster for each day. When this shift is removed and the PCA re-applied, the tight knot of data does separate, one cluster for each day.

13. Alignment simplifies identification of sample changes (between retention times 30 & 50 seconds, some components “grew” overnight) Examining the tail end of the chromatogram gives the reason for the compositional shift. Essentially, we are looking at the freshness of the gasoline. We are able to see compositional changes likely due to oxidation impacting peaks whose intensity are less than 1% of the typical peaks.Examining the tail end of the chromatogram gives the reason for the compositional shift. Essentially, we are looking at the freshness of the gasoline. We are able to see compositional changes likely due to oxidation impacting peaks whose intensity are less than 1% of the typical peaks.

14. What about Variances Instrument to Instrument? 3 instruments 3 different columns 3 different analysts 3 different days All measuring C8 to C19 hydrocarbons Alignment Eliminates retention time variances Readily enables method transferability to multiple instruments

15. So… after you’re finished… after… Methods development Rollout Training… After it all becomes routine… Who is watching? 480 per day 365 days per year 175,200 chromatograms per instrument This is a big problem.

16. And the Solution Is… LineUp to eliminate retention time variances Simple outlier detection using Pirouette chemometric models Automating analysis data collection preprocessing interpretation system suitability assessment results result validity fault identification pass/fail Enabling Notification of the human when intervention is suggested Presentation of results known to be statistically valid Elimination of unnecessary “sample checking” in short… Making our jobs manageable!

17. Falcon/Infometrix Solutions LineUP