SED navigation bar go to SED home page go to SED publications page go to NIST home page SED Home Page SED Contacts SED Projects SED Products and Publications Search SED Pages


contents     previous     next


3.2.1 SRM 2035: Near Infrared Wavelength Standard

James J. Filliben
Statistical Engineering Division, ITL

Steven J.  Choquette
Analytical Chemistry Division, CSTL

SRM 2035 certifies peak locations in optical filters. NIST is certifying 99 such physical filters- all cut from the same block of glass. Steve's work in the near-infrared is a natural extension of Leonard Hannsen's SRM 1921 infrared wavelength standard and John Travis' UV visible wavelenth standard. The near-infrared standard of Steve's is particularly valuable for the calibration of FT instruments in the pharmaceutical industry and the FDA.

The raw data for this experiment is a series of spectra; and the wavelength locations of the 7 peaks of these spectra are the parameters of interest. A "run" in this experiment takes about 4 hours and involves evacuated purging (via liquid nitrogen) to remove water vapor-the primary contaminant in this experiment. The usual design of experiment priniciples of

1. randomization;
2. replication;
3. blocking; and
4. use of a control

were applied to formulate the experimental plan. Over the course of the 4 hours, multiple measurement passes were made with peak locations being measured at each pass. As the passes were being made, the water vapor in the system was being reduced, and thus the peak locations were (at a very low level) changing as they were being remeasured. The primary question of interest was how much the 7 peak locations changed/drifted as a function of pass, how the uncertainties for these location estimates changed with pass, and how to arrive at final certificate values in this low-level drifting environment.

The final design was a balanced, 2-week, morning-and-afternoon, with a control, appropriate randomization and replication factorial design.

The design and subsequent analysis yielded the appropriate certificate values, along with other information about outlying filters, and hyper-sensitive peaks. The multiplot on the opposite page illustrates the effect of pass (= time) on the 7 peak locations.


\begin{figure}
\epsfig{file=/proj/sedshare/panelbk/99/data/projects/stand/choquett.ps,width=6.0in}\end{figure}

Figure 7: The above plots show the drift in peak location as a function of pass (= time). Such drift occurs with both mean and median as location estimators. Based on these plots, appropriate certificate values for peak wavelength location were obtained for all 7 peaks.



contents     previous     next

Date created: 7/20/2001
Last updated: 7/20/2001
Please email comments on this WWW page to sedwww@nist.gov.