Opto-Mechanical and Electronic Design of a Tunnel-Trap Si Radiometer - Statistical Data Included

Journal of Research of the National Institute of Standards and Technology, Nov, 2000 by George P. Eppeldauer, Donald C. Lynch

A transmission-type light-trap silicon radiometer has been developed to hold the NIST spectral power and irradiance responsivity scales between 406 nm and 920 nm. The device is built from replaceable input apertures and tightly packed different-size silicon photodiodes. The photodiodes are positioned in a triangular shape tunnel such that beam clipping is entirely eliminated within an 8[degrees] field-of-view (FOV). A light trap is attached to the output of the radiometer to collect the transmitted radiation and to minimize the effect of ambient light. The photodiodes, selected for equal shunt resistance, are connected in parallel. The capacitance and the resultant shunt resistance of the device were measured and frequency compensations were applied in the feedback network of the photocurrent-to-voltage converter to optimize signal-, voltage-, and loop-gain characteristics. The trap radiometer can measure either de or ac optical radiation with high sensitivity. The noise-equivalent-power of the optimized devi ce is 47 fW in dc mode and 5.2 fW at 10 Hz chopping. The relative deviation from the cosine responsivity in irradiance mode was measured to be equal to or less than 0.02 % within 5[degrees] FOV and 0.05 % at 8[degrees] FOV. The trap-radiometer can transfer irradiance responsivities with uncertainties comparable to those of primary standard radiometers. Illuminance and irradiance meters, holding the SI units (candela, color- and radiance-temperature), will be calibrated directly against the transfer standard trap-radiometer to obtain improved accuracy in the base-units.

Key words: detector; irradiance; photocurrent; photodiode; radiant power; reference detector; responsivity; spectral response; transfer standard.

Accepted: August 29, 2000

Available online: http://www.nist.gov/jres

1. Introduction

1.1 Light-Trap Standards

Light-trap detectors have been used as radiometric standards since 1983 [1]. At that time UDT UV100 [1] n-on-p inversion layer silicon photodiodes were used in either four element (Model QED-100) or three element (Model QED-200) reflectance-type light-trap configurations [2, 3]. These primary standard devices measured the total power of the incident, well collimated, radiation. Their power response relative standard uncertainty was 0.03 % between 440 nm and 460 nm where bias voltage was not applied to the photodiodes [4]. (Note that throughout this paper, all uncertainties are either relative standard uncertainties or standard uncertainties, i.e., one standard deviation estimates, and hence the coverage factor used is k = 1 [5].) These non-linear devices had a limited dynamic range of operation [6].

Later, Hamamatsu S-1337 p-on-n silicon photodiodes were used in the Model QED-150 trap-detectors in an arrangement similar to the QED-200. These detectors were called "quantum-flat" because they have external quantum efficiencies (EQE) that are constant to within 0.1 % from 550 nm to 860 nm. The spectral responsivity of quantum detectors (in A/W) is proportional to their EQE and wavelength. The proportionality factor is e/hc, where e is the elementary electron charge, h is the Planck constant, and c is the speed of light in vacuum. Using the silicon photodiode self-calibration technique [71 for a single element S-1337 photodiode, the quantum flatness of these trap-detectors could be extended to 400 nm [8]. The S-1337 trap-detectors with the constant relative spectral responsivity were calibrated against either a QED-200 between 440 nm and 460 nm or an electrical substitution cryogenic radiometer. The responsivity of S 1337 trap-detectors could be extrapolated from the 440 nm to 460 nm range to longer wavelen gths with very little loss of accuracy because the shape of the internal quantum efficiency (IQE) does not depend on typical diode-to-diode variations in the doping profile. IQE is the ratio of the number of collected electrons to the number of photons absorbed by the detector after the front surface reflection loss. EQE = (l-p)IQE where p is the reflectance. Since the spectral shape of the IQE of S1337 type photodiodes can be modeled with very small uncertainty [9, 10], the spectral responsivity of S1337 reflectance-type trap-detectors could be interpolated between 406 nm and 920 nm with a relative standard uncertainty of 0.03 % if two or more absolute tie points were measured.

1.2 Reflectance-Type Trap Detectors

The reflectance-type (three-element) trap detectors have polarization-dependent fractional response variations of about 1 X [l0.sup.4] [11]. The reflectance loss of the S-1337 reflectance-type trap detectors increases from 0.21 % at 920 nm to 1 % at 406 nm [10]. We measured the responsivity ratios of a QED-150 to a S1337 reflectance-trap versus wavelength. The ratio changed from 1.0005 at 920 nm to 1.0085 at 406 nm, indicating that the reflectances in the blue for the same S1337 photodiode model can be different by several tenths of a percent. The reflectance depends on the oxide thickness of the selected S1337 photodiode. A spatial response non-uniformity of 0.03 % was measured on the same S1337 reflectance-trap where the photodiodes were individually aligned to optimize the device field-of-view (FOV) [12]. Later, 0.2 % spatial response non-uniformities were measured on three different QED- 150 devices using the same characterization facility [13]. The measured active areas of the QED- 150 trap detectors we re neither symmetrical nor similar. The higher response non-uniformities and the different shapes of the measured areas indicated that either the FOV was smaller and non-symmetrical or the reproducibility of the photodiode device to device positioning was poor.