Technical Papers

» Fiber Optic Calorimeter

P.G. Davis and I.J. Bush
Optiphase Inc.
7652 Haskell Ave.
Van Nuys, CA 91406

S. Bayliss and C. Rudy
NIS-5 Safeguards Science and Technology
Los Alamos National Laboratory
Los Alamos, NM 87545

Abstract

A twin-bridge fiber optic calorimeter has been built and is currently being tested at Los Alamos National Laboratory (LANL). The intrinsic optical phase shift induced by changes in temperature is measured in both a reference canister and a sample canister. This system incorporates two Michelson interferometers each with an optical path mismatch of 1.6 km. A digital demodulation scheme is used which produces a 32bit phase word which tracks up to 500,000 fringes with a resolution of 10^-4 fringe, giving the system a dynamic range > 10⁹. Both interferometers are demodulated simultaneously at a rate of 83 kHz. The phase difference between the reference and sample interferometers is proportional to the temperature difference between the canisters and therefore correlates to the power produced by the sample in question. The optical system performance will be described along with preliminary calorimetric measurement results.

I. INTRODUCTION

Calorimeters are used to accurately measure amounts of nuclear material nondestructively. Using a calorimeter, the total thermal power produced by a sample containing nuclear material is measured. The precise amount of the radioactive material in the sample can then be determined. Other nondestructive assay (NDA) techniques often cannot provide an absolute measure of the amount of any particular element due to a dependence on the sample geometry and the material matrix. Therefore, by using a calorimeter, absolute amounts of radioactive isotopes can be accurately measured, regardless of the sample geometry and matrix. Calorimetry can provide standards that are then used to calibrate other radiation detectors for specialized geometries and matricies.

A typical calorimeter system is the twinbridge calorimeter depicted schematically in Figure 1. A conventional system consists of two identical canisters or "thermels" with lengths of temperature sensors wound around the thermel exteriors for measuring the heat flux from each thermel. These thermels are separated by a thermal resistance, usually a narrow air gap, from thermally stabilized environment, typically a water bath. The radioactive sample is placed within one thermel while the other is used as a reference in order to correct for small variations of the water bath temperature. Since the sample is radiating heat due to the radioactive decay of the nuclear materials within the sample, the sample chamber is heated relative to the reference chamber (which is far enough away as to be thermally isolated from the sample thermel). The sample chamber heats up relative to the reference chamber until thermal equilibrium is achieved. The net average temperature change and thus sensor response is then directly related to the thermal power produced by the sample. For large volume calorimeters (>3L) a single measurement may take 2-24 hrs depending one the sample initial temperature, specific heat, mass, and thermel conductivity.

Large volume (>3L) calorimeters using electrical resistance sensors have been used to measure Plutonium and tritium materials with thermal powers as low as 100 mW with an accuracy of 1%¹. Lower power nuclear materials, such as high-enriched Uranium, generate lower thermal powers that can not be measured accurately with electrical resistance sensors. In order to improve the measurement accuracy of nuclear materials and also to measure lower power samples a more sensitive calorimeter is required. The fiber optic calorimeter should result in more than an order of magnitude improvement over conventional technology.

II. CALORIMETER CONSTRUCTION

The fiber optic calorimeter utilizes the same twin-bridge design described above. However, instead of using conventional electrical resistance thermometers, the intrinsic thermal sensitivity of optical fiber is used to measure heat flow. The 400 m of Nickel resistance thermometer wire was replaced with 846 m of telecommunications grade single mode fiber. The fiber was precision wound along the entire length of the 0.6 m long thermel. The sensitivity of the fiber wrapped thermel was measured and is shown in Figure 2. Given a phase resolution of 5 mRad, with a sensitivity of 471,000 Radians/degree C we can expect a temperature resolution of 10^-8 C that is about 4 orders of magnitude better than conventional electrical resistance thermometers. In addition, the cost of a fiber optic calorimeter is much less than an electrical one. The fiber optic calorimeter uses 846 m of fiber for each thermel at a cost of only 5 ¢/m whereas the electrical calorimeter uses about 400 m of Nickel wire at a cost of $10/m. Furthermore, the fiber optic calorimeter is insensitive to EMI, whereas with the long resistance thermometers special care must be taken in order to attain the 10^-4C temperature resolution. The long wire windings must be symmetrically wound to minimize inductive effects and the sensor leads must be well shielded to the voltmeter. Unfortunately, the fiber optic calorimeter does suffer from acoustic sensitivity. Due to the large amounts of fiber used and the inherent sensitivity of the optical phase measurement to stretching and compression of the fiber, the thermels are very sensitive to acoustics and care must be taken to acoustically isolate the thermels. Nevertheless, the fiber optic calorimeter promises to offer higher performance at a lower cost than a conventional calorimeter.

III. SYSTEM DESCRIPTION

The operation of the fiber optic calorimeter is based on the linear relationship between temperature and optical path in a fiber as shown above. This optical path length change with temperature is due to both lengthening of the fiber and changes in the refractive index of the fiber:

where q is the optical path length in radians,n0 is the laser light frequency, n is the index of refraction and L is the fiber length. To measure the optical phase change, the fiber wrapped thermel is assembled in one arm of a Michelson interferometer while the other arm is kept short and packaged at the bottom of the thermel. Both are terminated with Faraday mirrors in order to negate the birefringence of the fiber and preserve 100% fringe visibility. The optical schematic is shown in Figure 3.

The optical phase difference between the two arms of the Michelson is then measured utilizing a phase generated carrier (PGC) technique². The carrier signal is generated via a frequency modulation of the laser. Since the Michelson interferometer has a large optical path difference, about 1.6 km, a small frequency modulation will produce the required p radians for demodulation:

Using an Optiphase Inc. OPD-200 digital demodulator, a dynamic range of 10⁹ can be achieved. The demodulator output has been configured such that the LSB of the 32-bit word output corresponds to 0.8 mRad. This results in the ability to count over 500,000 fringes with a resolution of < 1 mRad. Two demodulators are used in the system, one for each thermel. The demodulators are synchronized so that the digital outputs are simultaneously read by a PC at 10 kHz. The sample and reference phases are subtracted in software and the resulting phase difference represents the temperature difference between the thermels.

IV. OPTICAL PATH MATCHING

The purpose of the reference thermel is not only to compensate for thermal fluctuations of the water bath but also to compensate for frequency drift of the laser. As was described above and in equation (2), a change in the laser frequency will cause a change in the optical phase as measured by the interferometer. This effect is used to create the PGC signal for the OPD-200 demodulator. However, frequency drift of the laser will also induce a phase drift which if not compensated correctly will appear to be a temperature drift. The Lightwave diode pumped YAG (DPY) laser used has a specified frequency drifted of less than 100 MHz/hr. Given the 1.6 km OPD of the Michelson interferometer, this could result in over 5000 radians of phase drift and would limit the resolution of the system to about 0.01 C. In order to compensate for this frequency drift the reference thermel interferometer must be path matched with the sample interferometer. It was desired to achieve an optical phase resolution of < 5 mRad, therefore, the 1.6 km OPDs of the two interferometers needed to be matched to within 1 mm. To achieve this accuracy, the calorimeter optical system shown in Figure 3 was used. A large laser frequency drift was simulated by modulating the laser temperature at 10 Hz. A pk-pk frequency modulation of 17 GHz was achieved in this manner and therefore produced a pk-pk phase modulation in both the reference and sample interferometers of 150,000 fringes pk-pk as shown in Figures 4 and 5. The difference between the two interferometers was then calculated in software and resulted in a signal whose pk-pk is proportional to the OPD difference between the interferometers. The variation in the difference signal of Figure 4 is 8.15 fringes pk-pk (ignoring the dc drift caused by temperature drift). From this measurement the OPD difference between the two interferometers was calculated to be 9.55 cm.

The interferometer length of the reference thermel was adjusted to minimize the OPD difference between the thermels until the difference measurement resulted in <0.01 fringes pk-pk (Figure 5) which corresponds to an optical path difference of < 1mm. Therefore, a phase resolution of < 5 mRad should be achievable which corresponds to a temperature resolution of 10^-8 C.

V. INSTALLATION AT LANL

The thermels were transported to LANL for installation. The installation procedure consisted of placing the thermels inside the stainless steel container shown in Figure 6. The bottom plate was then sealed and the entire assembly was placed within a large water bath. The coupler and faraday mirrors were stored at the bottom of the stainless steel enclosure and the insensitive leads from both the reference and sample chambers were fed up through the narrow stainless tubing in the center of Figure 6. An FC adapter was installed at the top of the calorimeter assembly above the water line for connection back to the calorimeter electronics chassis shown in Figure 7. The top of the calorimeter was also covered with foam insulation in order to prevent thermal coupling with the room environment. Initial tests without top insulation showed sensitivity to the room air temperature and tracked the air conditioning cycles.

VI. CALORIMETRIC MEASUREMENTS

Some preliminary thermal sensitivity measurements have been made. For these measurements, a calibrated heat source was used to introduce known amounts of heat in the reference thermel. A sample run is shown in Figure 8. In this run, 11 mW of power was applied for 3 hours, until equilibrium was reached and then the power was shutdown to allow the sample thermel to cool back down to the water temperature. The peak phase excursion is proportional to the thermal power applied. This measurement was repeated for 3 mW, and 44 mW and the results are plotted in Figure 9. A linear fit shows that the thermal sensitivity of the fiber optic calorimeter is 151 rad/mW. During the data runs, the phase drift was about 200 radians which limits the calorimeter resolution to about 1 mW. This resolution is comparable with current state of the art calorimeters of similar size. However, with improvements in the insulation and water bath stability it is expected that the resolution will improve dramatically. If the phase drift can be reduced to the laser frequency drift limit of 1 mRad, then thermal sensitivities of 7 nW could be achieved.

VII. CONCLUSIONS

A twin-bridge fiber optic calorimeter has been built and is currently being evaluated at LANL. Extremely high resolution calorimetry is possible due to the large dynamic range capabilities of the Optiphase Inc. OPD-200 digital demodulator, and due to the careful path matching of the reference and sample interferometers in order to reduce common mode errors. A thermal sensitivity of 1 mW has been demonstrated. However, if the long term phase drift can be reduced, resolutions as low as 7 nW are theoretically possible. Since the interferometer paths have been matched to within 1 mm, the observed phase drifts appear to be temperature induced from either water bath stability, thermal gradients in the water bath or some other environmental effect. The sources of the phase drift are still under investigation.

VIII. ACKNOWLEDGMENTS

This work is supported by the US Department of Energy, Office of Nonproliferation and National Security, Office of Security Affairs, Office of Safeguards and Security.

The authors wish to acknowledge Mirko Invancevic of Optiphase Inc. for his expert assembly of the fiber optic components and Cory Keitz of Optiphase Inc. for his work integrating the electronic components and fabricating the demodulators.

¹C. Rudy et. al. Los Alamos Report # LA-UR-97-4176
²J. Bush et. al. Third Pacific Northwest Fiber Optic Sensor Workshop, SPIE Proceedings Vol. 3180