Abstract
In case the primary realization of the spectral responsivity scale is not conducted at all target wavelengths but at only a small part of them, one needs to extrapolate values at the specific wavelengths to an extended range. In this work, we present a fully experimental procedure to extrapolate a single value of spectral responsivity at 633 nm into the whole working wavelength range (250 – 1100) nm of Si photodiodes. It is based on spectral responsivity comparison between a Si trap detector and a low-NEP pyroelectric detector of nearly flat spectral response. For this purpose, we developed a setup specialized to compare a Si-trap detector of dc-current output with a pyroelectric detector of ac-voltage output by using a modulated probing light source and a monitoring technique. To keep the probing light chopped even for the dc-photocurrent readout, we adopted a low chopping frequency of 4 Hz and a triggered readout for the Si-trap detector, which leads to a speedy comparison between the Si-trap detector and the pyroelectric detector. For the reference pyroelectric detector, we characterized the spectral absorptivity of the black-coating and the nonlinearity of the lock-in amplifier readout. Compiling all the required information, the spectral responsivity of the Si trap detector could be measured with the minimum uncertainty of 0.3 % (k = 2), which was validated by comparing with that of our previous method based on a numerical extrapolation.
NIST reference spectral responsivity scales for improving the SI candela and kelvin