Results of a CIE Detector Response Intercomparison

A total of fifteen laboratories participated in the CIE detector response intercomparison which was designed to assess the level of agreement among participating laboratories in the absolute measurement (with respect to SI) of photodetector response in the visible spectral region. Most participants were either commercial laboratories or university laboratories with the National Institute of Standards and Technology (NIST) serving as the host laboratory. Each laboratory determined the absolute response of each of two silicon photodiode radiometers which were designed for the intercomparison by NIST. Approximately two-thirds of the laboratories reported response values which agreed with the NIST values to within ±1.0% at the two wavelengths of 488 and 633 nm.3


Introduction
This report provides the final results of a detector response intercomparison under the aegis of CIE Technical Committee TC 2-06 on Absolute Spectral Responsivity of Detectors. Members of the Technical Committee are listed in Appendix B of this report.
The primary purpose of the intercomparison was to assess the level of agreement among participating laboratories in the absolute measurement (with respect to SI) of photodetector response (A/W) in the visible spectral region. The method chosen to accomplish this is to have these laboratories measure the absolute response of selected radiometers '  at two specific wavelengths near each end of the visible spectrum. The wavelengths selected are those of the helium-neon laser (632.8 nm) and the argon ion laser (488.0 nm).
The intercomparison was implemented on the basis of the National Institute of Standards and Technology (NIST) serving as the host laboratory and providing (a) the radiometers to be used in the intercomparison, (b) instructions to participating laboratories in the use of the radiometers in absolute response measurements, and (c) data analysis and a report of the results of the intercomparison.
The intercomparison was conducted in two stages: (1) intercomparison of U.S. laboratories and (2) intercomparison of laboratories outside of the United States. All participating laboratories except two are either commercial laboratories or university laboratories. The Electrotechnical Laboratory in Ibaraki, Japan and the Van Swinden Laboratory, The Netherlands, are national standards laboratories. The intercomparison was planned on the basis that NIST would measure the absolute response of all radiometers before shipment to the participating laboratories and then again after the radiometers were returned. The NIST absolute response value for each radiometer would be the average of the two NIST measurements.

The Radiometers
The radiometers used in this intercomparison were designed for ruggedness and ease of use and included commercially available silicon photodiodes. Each radiometer consists of a silicon photodiode and amplifier circuit mounted in a cylindrical aluminum housing and an external power supply.
Since it was expected that the majority of the laboratories would make their measurements using lasers, the photodiodes were not protected by a window. However, to protect each photodiode during non-use, the diodes were maintained in sealed compartments.
Two types of detectors were used: EG&G^ model UV-444B PN photodiodes and UDT model UV-100 inversion layer photodiodes. Radiometers PI-17, PI-19, PI-20, and PI-21 have the EG&G photodiodes while radiometers PI-25 through PI-32 have the UDT photodiodes. For the UDT photodiodes, a constant reverse bias voltage (4.5 V) was supplied by lithium batteries within each radiometer. The amplifier in each radiometer has gain settings from lO'* to 10' V/A with accuracies of ±0.03% except for the 10' range where it is ±0.5%.
Each participating U.S. laboratory received two radiometers: one with an EG&G photodiode and one with a UDT photodiode. The reason for requesting each laboratory to measure two radiometers is two-fold. (1) A second radiometer provides a backup for possible shipping damage and (2) there is a check on measurement repeatability.
After the first stage of the intercomparison was completed involving U.S. laboratories, a decision was made to use only the radiometers with the UDT photodiodes for the intercomparison involv-* Certain commerical equipment, instruments, or materials are identified in this paper to specify adequately the experimental procedure. Such identification does not imply recommendation or endorsement by the National Institute of Standards and Technology or by the CIE, nor does it imply that the materials or equipment identified are necessarily the best available for the purpose.
ing laboratories in other countries. This decision was made when it was determined that the EG&G photodiodes exhibited a small but significant response drift at 488 nm over long periods of time (see sec. 4).

The Intercomparison
A total of six U.S. laboratories and nine laboratories in other countries participated in the intercomparison. Each laboratory was asked to complete a questionnaire concerning detailed information about their detector response measurement system and to use standard data forms for reporting their results. Tables 1 and 2 list the U.S. laboratories and the information each submitted about their measurements at 488 and 633 nm. Tables 3 and 4 list the corresponding information for laboratories in other countries. The information submitted covered eight measurement parameters: (1) absolute base (absolute standard(s) used), (2) standard deviation of the measurements, (3) number of measurements per radiometer, (4) type of radiation source used, (5) beam diameter of the source, (6) radiant power level at the radiometer, (7) ambient temperature during measurements, and (8) estimated uncertainty (with respect to SI) of the absolute standards used. Some laboratories used a single silicon photodiode as an absolute (standard) base for their measurements. The absolute response of these photodiodes was determined using the self-calibration method [1,2]. Two laboratories made measurements only at 633 nm. Of the fifteen laboratories participating in the intercomparison, five laboratories used lasers as radiation sources at both wavelengths and five used a tungsten lamp/filter/ monochromator system at both wavelengths. The remaining laboratories used various combinations of these sources. Radiant power levels ranged from 0.16 jaW to 0.7 mW.
The measurement system used at NIST for this intercomparison consists of He-Ne and Argon ion lasers, laser stabilizer, spatial filter, beam splitter, and a silicon photodiode monitor detector. Three UDT QED-200 absolute radiometers [3] were used as base standards. Figure 1 is an illustration of the system components. The NIST procedures for determining the absolute response of the intercomparison radiometers consisted basically of two steps: (1) measuring the ratio of the photocurrent of each UDT QED-200 radiometer to the photocurrent of the monitor detector at a particular laser power setting and (2) measuring the ratio of     the photocurrent of the intercomparison radiometers to the monitor detector at the same power level in (1). Since the UDT QED-200 radiometers are 100% quantum efficient (with voltage bias) at the wavelengths and power levels stated, the power (watts) can be accurately measured and the absolute response (amperes/watt) of each of the intercomparison radiometers can be determined. Details concerning the system and the measurement procedure are further described in [4]. The NIST absolute base was compared to other international standards laboratories in a recent detector response intercomparison sponsored by the Consultative Committee on Photometry and Radiometry (CCPR) [5]. In the CCPR intercomparison, the absolute response of a select group of silicon photodiode radiometers were measured by 10 international standards laboratories and also by NIST which served as the host laboratory. The ratios of the NIST response values to the mean of the response values of the other participating laboratories were 1.0011+0.0035 and 1.0014+0.0037 at the two wavelengths of 488 and 633 nm, respectively. Since the absolute response values reported by each of the participating laboratories were com-pared to the response values determined at NIST, it was essential for NIST to measure the response of each set of radiometers before it was shipped to the participating laboratory and then measured again after the radiometers were returned. The before and after measurements by NIST were made to determine if any significant changes occurred in the radiometers during shipment.

Data Analysis
Tables 5 and 6 list the laboratory designations, date of measurement, radiometer descriptions, and absolute responsivities reported by the U.S. laboratories and laboratories in other countries, respectively. Each set of response values for a participating laboratory includes the corresponding before and after values determined by NIST. The NIST value for each radiometer was taken as the average of the before and after respective values. The before and after NIST values indicate that some of the radiometers had undergone a small but significant change in response between shipments to and from the laboratories. For example, at 488 nm, the response value for radiometer PI-20 (laboratory C, table 5) decreased from 0.2814 to 0.2787 over the period 7/87 to 2/88 as measured by NIST. This is a decrease of 0.96%. All ratios reported represent an average of the before and after values.
Since three of the four radiometers with the EG&G type photodiode showed small but significant decreases in response at 488 nm over a 7month period, it was decided to use only the radiometers with the UV-100 type photodiodes for the second phase of the intercomparison (foreign laboratories). Table 7 is a listing of the participating laboratories by code letter, the absolute response values reported by each laboratory, the absolute response values as determined by NIST, and the ratios of the response values. Figures 2 and 3 are plots of the ratios of the response values (A/W) determined by each of the participating laboratories to the respective response values (A/W) determined by NIST at 488 and 633 nm. The solid line on each plot is the mean of all the ratios at the respective wavelength and the dashed lines are the standard deviation of the mean. Table 8 is a summary of the standard deviations of the measurements and the estimated uncertainty (with respect to SI) of the absolute standards used by each of the participating laboratories. Also listed are the before/after change in absolute response for each detector as measured by NIST and     Figure 3. Ratio of the participant laboratory spectral response to that determined by NIST at 633 nm. The error bars indicate the quadrature summation of the measurement and absolute uncertainties of each participant laboratory, the before/after response change for each radiometer, and the NIST measurement and absolute uncertainties. The dashed lines indicate the standard deviation of the ratio values.  the absolute response ratio uncertainty. The absolute response ratio uncertainty is the quadrature summation of the measurement and absolute uncertainties of each participant laboratory, the before/ after response change for each radiometer, and the NIST measurement and absolute uncertainties. The error bars in figures 2 and 3 indicate the absolute response ratio uncertainty for each laboratory.

Conclusion
In general, it can be concluded that most of the response values reported by the laboratories were in good agreement with NIST. At 488 nm, the mean of all participating laboratories was 0.71% higher than the corresponding NIST values with a standard deviation of 1.39%. Similarly, at 633 nm, the mean of all laboratory values was higher than the NIST values by 0.36% with a standard deviation of 1.07%. All laboratories participating in this intercomparison (except laboratory N) reported values at both wavelengths within ±2.0% of the NIST values and nine of the 14 laboratories reported values at both wavelengths within ±1.0% of the NIST values. This can be considered good agreement among the laboratories when one considers the variety of sources, procedures, and testing environments involved in this intercomparison.

Resume
Quinze laboratoires situes tant aux Etats-Unis que dans d'autres pays du monde entier ont pris part, dans le cadre de la CIE, a une comparaison de mesures de sensibilite de detecteurs qui avait pour but de determiner le niveau d'accord existant entre les laboratoires participants, pour la mesure de la sensibilite absolue (par rapport au SI) des photodetecteurs dans le domaine visible. La plupart des particpants etaient des laboratoires industriels ou des laboratoires universitaires. Le National Institute of Standards and Technology (NIST) jouait le role de laboratoire pilote. Chaque laboratoire a determine la sensibilite absolute de deux radiometres equipes de photodiodes au silicium, et specialement realises pour cette comparaison par le NIST. Les resultats fournis par environ les 2/3 des laboratoires sont en accord avec ceux du NIST dans la limite d'incertitude de ±1% pour les longueurs d'onde de 488 et 633 nm.