Intramural comparison of NIST laser and optical fiber power calibrations

Journal of Research of the National Institute of Standards and Technology, March-April, 2004 by John H. Lehman, Igor Vayshenker, David J. Livigni, Joshua Hadler

The responsivity of two optical detectors was determined by the method of direct substitution in four different NIST measurement facilities. The measurements were intended to demonstrate the determination of absolute responsivity as provided by NIST calibration services at laser and optical-communication wavelengths; nominally 633 nm, 850 nm, 1060 nm, 1310 nm, and 1550 nm. The optical detectors have been designated as checks standards for the purpose of routine intramural comparison of our calibration services and to meet requirements of the NIST quality system, based on ISO 17025. The check standards are two optical-trap detectors, one based on silicon and the other on indium gallium arsenide photodiodes. The four measurement services are based on: (1) the laser optimized cryogenic radiometer (LOCR) and free field collimated laser light; (2) the C-series isoperibol calorimeter and free-field collimated laser light; (3) the electrically calibrated pyroelectric radiometer and fiber-coupled laser light; (4) the pyroelectric wedge trap detector, which measures light from a lamp source and monochromator. The results indicate that the responsivity of the check standards, as determined independently using the four services, agree to within the published expanded uncertainty ranging from approximately 0.02% to 1.24%.

Key words: absolute responsivity; calorimeter; cryogenic radiometer; intercomparison; laser; optical power; optical fiber; pyroelectric detector; spectral responsivity.

**********

1. Introduction

The responsivity of a single optical detector determined from independent comparisons is a means of assuring that our stated uncertainties for a given measurement service are reasonable. Furthermore, such comparisons are a means of complying with the ISO 1725 quality system, the acceptance of which has been agreed upon by the world's National Measurement Institutes [1]. Through these intercomparisons, we have become a customer of our own services and are able to rigorously evaluate our performance.

We presently have four measurement systems for measuring continuous-wave (CW) laser power at relatively low power levels (milliwatts and less). The oldest among the services was established in the late 1960s and is based on an isoperibol calorimeter, which we call the C-series calorimeter. This measurement device is electrically calibrated and is traceable to electrical standards (the volt and ohm). Since the C-series calorimeter was developed and as the demand for higher accuracy continues, we have more recently developed a measurement service based on a cryogenic radiometer as a primary standard; the Laser Optimized Cryogenic Radiometer (LOCR). To meet growing customer demand for routine calibration of laser and optical-fiber power meters (OFPMs), we have developed two additional calibration services based on comparisons with pyroelectric detectors for absolute responsivity of fibercoupled power meters and relative spectral responsivity. Among these four calibration services, absolute responsivity of fiber-coupled power meters, or OFPMs, at common telecommunications wavelengths (for example, 850 nm, 1310 nm, and 1550 nm) is the most frequently requested. The demand for OFPM calibrations is approximately 75 calibrations per year and continues to increase.

For the intramural comparison we used transfer standards capable of low measurement uncertainty [2,3]. These transfer standards are intended to have very high coupling efficiency, so that they may be used with the four measurement systems having various input-beam geometries, as shown in Fig. 1. These input geometries are: (1) free-field, nearly collimated laser light input; (2) laser light transmitted by single-mode fibers coupled with FC-type fiber connectors; (3) moderately diverging light from a lamp and monochromator. Where possible, we sought to repeat the responsivity measurements with the three laser-based measurement systems, using laser sources with nearly the same wavelengths. Nominally these wavelengths are: 514 nm, 633 nm, 850 nm, 1064 nm, 1310 nm, and 1550 nm. The spectral responsivity measurement system is capable of wavelength adjustment precision of about [ or -]0.1 nm to approximate the wavelength of any of the laser sources, but the bandwidth is approximately 6 nm [4].

The results of this intramural comparison are given in several subject areas: a description of the two transfer standards, description of the four measurement systems with a statement of the measurement uncertainty, and a summary of results. The uncertainty is given with coverage factor, k = 2, in every case. The coverage factor corresponds to a level of confidence for the relative expanded uncertainty that is approximately 95% [5].

2. Transfer Standards

The transfer standards, or check standards, for this comparison are photodiode-based optical detectors designed and built at NIST [2,3]. For convenience, we use the colloquial term "device under test," or DUT, to identify these detectors. The optical configuration of each DUT is based on an optical trap having two photodiodes and a spherical mirror. This basic design has been employed in the past using three diodes (and no mirror) [6]. The presence of the spherical mirror reduces the external quantum efficiency of the trap (compared with the three-diode design), but increases the coupling efficiency for larger values of numerical aperture (NA) [3]. The trap based on silicon (Si) photodiodes is suitable for measurements requiring an NA as large as 0.26. The trap based on indium gallium arsenide (InGaAs) photodiodes is suitable for a slightly lower NA because of the diode packaging constraints (the size of the chip carrier), and the choice of the spherical mirror having a larger radius of curvature. DUT1 designates the detector based on Si photodiodes and DUT2 designates the detector based on InGaAs photodiodes.