Standard Reference Material 2881 - National Institute of Standards ...

More documents

Recommendations

Info

12. Discussion of Certified ValuesThe concentration β i for each oligomer (with associated uncertainties) for theOctylA polystyrene MMS converts it into SRM 2881 with a certified MMD. All Type Buncertainty is small compared to the two forms of Type A uncertainty, gravimetric andablation. Here we consider the issue of why the Type B uncertainty is small compared tothe Type A. In earlier SRM work using light scattering to determine the M w orosmometry to determine the M n , we generally were able to design the experiment so TypeA uncertainty was comparable or somewhat smaller then the Type B uncertainty. Herethis is certainly not the case. We may take it as lack of control over some parts of theexperiment—in particular the sample preparation. The sample preparation is done in away hoping to obtain a good sampling of the MMD of the original polymer by the use ofa spraying technique where it is expected that that the rapid droplet evaporation willallow us to obtain uniformity in the sample surface. But the electrospray may easily beaffected by a variety of things we did not control—humidity in room during samplepreparation and subtle changes in the electrospray cone character leading to uncontrolleddroplet size. Furthermore, the ablation process is a nonequilibrium event; making thechemistry of the plume is itself a variable process. Any expectations of high-precisionrepeatability are unlikely.There is a second source of Type A uncertainty. There is the source from therepeatability of the fundamental data, i.e., in β i , as discussed above. However, there isalso a source of uncertainty in the correction term for Q/k 0 . Table 3 gives the values ofthe mean value of the experimental β i the experimental fraction of the signal in peak ofmass m i , with its statistical standard deviation (the contribution to the Type Auncertainty). It is unclear that these are independent uncertainties but having no otherknowledge we must assume they are independent and add them in the usual way by usingthe sum of squares. (The reader is reminded that there are no contributions from the TypeB uncertainties because they are so small.) Table 1 offers a listing of the contributions tothe uncertainty from all sources. It is unclear that these are independent uncertainties buthaving no other knowledge we must assume they are independent and add them in theusual way by using the sum of squares. Table 3 offers the MMD for the polymer as35
certified by this method with the uncertainties of each β i for the cumulative distributionform 1 % to 99 %. Table 4 is a reduced version of this table found on the SRMcertificate. Figure 10 is a plot of the final certified values found in Table 4.13. ConclusionsWe conclude with a few final words about the Taylor’s expansion mathematicalmodel. The impetus for the model came from noticing in many earlier published reportsthat k i appeared to be a slowly-varying function of mass. A Taylor's expansion that islinear in mass creates a useful computational bridge from mass spectra to measurablemass moments. For this reason we chose to build on this foundation. However, if someother variable (e.g., time in time-of-flight mass separation) were shown to have a longerrange of linearity then an expansion in that variable would provide a more accurate fittingfunction.Our testing of the mathematical model is far from exact but appears to be robustprovided a few conditions are met. Most importantly, for a given set of instrumentparameters and for a sample preparation with some modicum of homogeneity andrepeatability, a fairly wide region of instrument space where the polymer-to-matrix massratio results in a proportional number of polymer ions arriving at the detector must exist.This proportionality constant must be a slowing varying function of mass so that it can belinearly approximated. Additionally, this region of concentration space must have a highenough polymer concentration for us to obtain a good signal-to-noise ratio. Furtherimprovements could be made in two areas: 1) improved reproducibility in the data likelythrough better sample preparation methods and a fuller understanding of the MALDIprocess, and 2) an estimate of the functional dependence of Q on the various machineparameters. This latter improvement can only come from a mathematical description ofthe total instrument behavior, from ablation through mass separation and detection tosignal digitization, which is currently beyond reach for MALDI-TOF mass spectrometry.36
Page 1 and 2: NISTIR 7512A Report on the Certific
Page 3 and 4: ABSTRACTThe certification of an abs
Page 5 and 6: 1. IntroductionThe certification of
Page 7 and 8: BHT), was used as the solvent. The
Page 9 and 10: anticipated to be a useful material
Page 11 and 12: sample using a separate method, or
Page 13 and 14: 6.0. Signal Axis Calibration Proced
Page 15 and 16: 6.2 Mathematical Model for Gravimet
Page 17 and 18: Here Q is a function of all the exp
Page 19 and 20: Let us consider this final equation
Page 21 and 22: GGexpTAexpTBGG0TA0TB=⎧⎨1+⎩0(
Page 23 and 24: were also used in this experiment:
Page 25 and 26: The matrix was all-trans retinoic a
Page 27 and 28: educes the data set to groups of th
Page 29 and 30: atios of the two sets of data is ag
Page 31 and 32: An analysis of variance (ANOVA) 5 w
Page 33 and 34: 11. Discussion of Type B Uncertaint
Page 35: Two other methods were used to chec
Page 39 and 40: References1. S. D. Hanton, I. Z. Hy
Page 41 and 42: Figures1. MALDI-TOF mass spectra of
Page 43 and 44: Figure 242
Page 45 and 46: Figures 5 (inset) and 6 (main)44
Page 47 and 48: Figure 946
Page 49 and 50: Tables1. Q/k 0 with its estimated u
Page 51 and 52: Table 2Instrument Parameter Optimal
Page 53 and 54: 103 10840.42 0.0063 8.93 % 14.72 %
Page 55 and 56: 99 10423.86 0.0139 0.0008 93.9 %100
Page 57 and 58: INSTRUCTIONS FOR USEStorage: The SR
Page 59: REFERENCES[1] See ASTM D5296 05 “

Standard Reference Material 2881 - National Institute of Standards ...

Create successful ePaper yourself

Delete template?

Save as template?