A new design for thermal-infrared radiation thermometer and sensors is described. Critical optical elements, such as the field stop, Lyot stop, collimating lens, and detector, are placed inside a thermally stabilized assembly that is controlled using thermo-electric coolers and thermistors. The assembled radiation thermometer is calibrated using both variable-temperature fluid-bath and heat-pipe blackbodies from −45 °C to 75 °C and the use of a modified-Planck function and these blackbodies. The size-of-source effect both with and without the Lyot stop has been measured. This new design, during operations without the need for cryogenic cooling, demonstrates sub millikelvin temperature measurement resolution with few millikelvin, week-long stable operations while measuring room-temperature objects.
Most commonly used radiation thermometers operate in the 8 µm to 14 µm wavelength range to measure objects which are near or at room temperatures including human body temperatures . These thermal-infrared radiation thermometers (TIRT) can range from inexpensive, hand-held designs to much more expensive transfer standard quality types . The uses of TIRTs vary greatly, for example, ranging from routine non-contact diagnostics of equipment to assessments of food temperatures during transport and storage. Some of the most demanding uses of TIRTs are in measurements of sea-surface temperatures to validate remote sensing weather and climate satellite sensors .
Although TIRTs have been extensively utilized since the early 1960s, their optical designs have not evolved during this period. Commercial TIRTs have simple optical designs with sometimes just a lens and a detector, and these TIRTs can suffer from poor long-term stabilities and large size-of-source effects (SSE) . Some of these TIRTs have been characterized for SSE but since their designs cannot be modified, it is difficult to determine the exact cause of the SSE. Furthermore, optical designs developed for improving visible and near-infrared radiation thermometers have not been implemented in the thermal infrared devices or have only been partially implemented in standards-quality radiation thermometers which use off-axis reflective optics .
In this work, we describe the construction and characterization of an Ambient-Radiation Thermometer (ART) which implements the SSE reduction designs of visible and near-infrared radiation thermometers . To increase the long-term stability of the responsivity, optical components within the field-of-view of the detector and the detector are thermally stabilized using three separate thermo-electric control setups. The ART is constructed with zinc-selenide (ZnSe) lenses for collecting and focusing the thermal-infrared radiation onto the pyroelectric detector. The pyroelectric detector has a window with a 8 µm to 14 µm filter for spectral selection, and the detector and preamplifier are packaged into a hermetically sealed container. Due to the use of a pyroelectric detector, the radiation is modulated using a rotating chopper wheel, and the modulated signals are measured using a commercial lock-in amplifier. The SSE of the ART was measured, and the ART was calibrated using variable-temperature water-bath and a lower-temperature ammonia heat-pipe blackbodies, and both short- and long-term stabilities were assessed using those blackbodies .
2. Experimental setup
2.1. Radiation thermometer design
The optical design of the ART is shown in Fig. 1, and the detailed specifications of the ART are listed in Table 1 Optical elements are selected for optimal operation in the 8 µm to 14 µm spectral range. The optical elements are placed on a small breadboard for quick changes to the design, and the setup is enclosed on all sides using anodized-aluminum sheets as light-tight covers. Plano-convex, anti-reflection coated ZnSe lenses are used for this on-axis optical setup. The 50 mm diameter ZnSe objective lens is placed at a distance of 40 cm from the blackbody opening to form a focused image at a distance of 25 cm from the center of the objective lens to the field stop. Both sides of the chopper wheel are covered with reflective, aluminum tape to reduce the emittance of the wheel, and the wheel is placed so that the blade completely covers the opening of the detector assembly when the blade is in the closed position. Due to the slow response time of the pyroelectric detector, the chopper wheel is operated at 4 Hz.
The design of the detector assembly, consisting of the field stop, two lenses, the Lyot stop and the detector itself, is critical to the low-noise, stable operation of the ART. The approximately 30 cm-long anodized aluminum tube that contains the detector assembly is temperature stabilized to 23 °C using three separate thermo-electric (TE) coolers. The three coolers are attached to the sections with the field stop, the Lyot stop, and detector and denoted as the front, back, and detector assemblies, respectively. Separate coolers are used so that each section can be optically aligned and to avoid temperature gradients. The temperature sensing of the two sections and the detector are performed using three separate glass-encapsulated thermistors , and commercial TE controllers are used to stabilize the assembly. As shown by the display on the TE controllers, the temperatures of the assemblies are stable to < 1 ohm which corresponds to < 2 mK at the set temperatures. The temperatures are set by letting all the sections equilibrate with the lab temperature and then decreasing the set point to 500 ohm below the equilibrium thermistor resistance. This corresponds to a setpoint of about 1 °C higher temperature than the laboratory temperature. The stabilization temperature is set to be just slightly above the lab temperatures so that convection effects can be minimized. The low-power temperature stabilization circuit can be operated for long periods of time without any operator intervention. Since the front and the back sections with the two lenses need to be separatly aligned, the two sections are separated by a gap which is later covered with a thin anodized aluminum tubing and then insulated using commercial foam-pipe insulation.
2.2. Choice of detector
Detectors which are sensitive to radiation in the 8 µm to 14 µm wavelength range are either classified as thermal devices such as thermopiles, pyroelectrics, and bolometers, or quantum devices such as HgCdTe detectors. The detectors which required cryogenic temperatures were not used for this application since we wanted to develop a radiation thermometer which can be stable and operational over extended periods of time without the need for cryocooling. A custom pyroelectric detector was developed in conjunction with a vendor to obtain high gains with low-noise performance. The current-to-voltage preamplifier with a 100 GΩ feedback resistor was integrated onto the detector into a single, electronic sealed package to obtain low noise levels. The spectral responsivity of the detector as measured in the NIST Infrared Spectral Comparator Facility (IR-SCF)  is shown in Fig. 2. The spectral responsivity is flat with a standard deviation of about 4% over the spectral region of interest from 8.5 µm to 13.5 µm.
From the calibrated spectral power responsivity, the noise-equivalent power (NEP) was determined so that comparisons to other detectors could be performed. As shown in Table 2, the NEP is dependent on the filter time constant setting on the lock-in amplifier. The noise-equivalent power of the chopped radiation is given by the NEP divided by the square root of the chopping frequency, and detectivities are obtained from multiplying by the square root of the area of the detector. These performance metrics are commonly used to compare detectors that can vary in detection wavelengths and active areas. The detectivity of our detector is roughly a factor of two higher than other commonly available pyroelectric detectors .
2.3. Optical design
The optical performance of the ART was assessed using a commercial, optical-modeling program. The specifications of the plano-convex ZnSe objective lens were input into the program, and the optimal focus distance from the back edge of the lens was determined using the program to find the minimum root-mean-squared-spot size at the wavelength of interest. Optimized distances were calculated at both visible wavelength of 550 nm and at 10 µm, and the distance offsets were used to guide the physical alignment of the ART. The field stop was aligned using visible radiation and then moved by the offset distance calculated using the optical modeling software. Strehl ratios shown in Table 3 are measures of the image quality and indicate that the ZnSe lens material is not optimally suited for collection of broad-band thermal radiation from 8 µm to 14 µm . It was calculated at the optimized wavelength of 11 µm and at the short and long wavelengths of the filter transmittances. The image quality rapidly degrades due to chromatic aberrations of the ZnSe lens, and the image quality is nearly diffraction limited only at the optimization wavelength. In retrospect, the use of Germanium lenses would have been more suitable.
3. Radiation thermometer calibration
3.1. Radiation thermometer setup
The schematic of the ART and the water-bath and heat-pipe blackbodies used for these studies is shown in Fig. 3. These blackbody measurements were performed in the NIST Advanced infrared radiometry and imaging facility(AIRI), and the ART was calibrated using the standard platinum resistance thermometers in both water-bath (WBBB) and ammonia heat-pipe blackbodies (AHPBB) . The water-bath blackbody can be operated from 15 °C to 75 °C. The NIST design has been extensively studied , and similar blackbodies have been constructed for laboratory and portable uses. At lower temperatures from −46 °C to 20 °C, an ammonia heat-pipe blackbody, with temperature control provided by an external, ethanol-bath circulator was used to calibrate the ART. The distance from the opening of the blackbody to the front surface of the lens is set at 40 cm.
3.2. Radiation thermometer calibrations
The temperatures of the fluid in the water bath and the ammonia heat pipe were measured using calibrated standard platinum resistance thermometers (SPRT), and these known temperatures are used to convert the optical signals from the ART to temperatures. For these initial measurements, no corrections were made for the spectral emissivity of the respective cavities, and the emissivities were assigned to be unity. For these initial measurements, we estimated the uncertainties (see Sec. 5). The interpolation function to convert signals shown in Fig. 4 to temperatures is obtained using two steps.
To fit the signals with an interpolation function, negative signals must be converted to positive signals in order to have a singular, positive dependence on absolute temperature, K. An additive-constant voltage, D, is determined by minimizing the differences between the measured signals and scaled band-integrated Planck integrals at the measured temperatures where T1 is chosen to be 298 K or 25°C, S(λ) is the spectral responsivity plotted in Fig. 2 and L(λ,T) is the Planck function,
This step is performed to decrease the residuals between the interpolation function and their measured data points, and because the commercial fitting software could not converge on the optimized values with sufficiently small residuals The measured lock-in amplifier voltages, v, at various temperatures are fitted using a modified-Planck function ,Eq. (1).
The differences in temperatures from the fit or residual temperature from Fig. 5 are shown in Fig. 6. The residual temperatures are larger in the region where the AHPBB was used indicating both worse signal-to-noise at the lowest temperatures and difficulties in controlling the low-temperature blackbody due to condensation and increased convection effects. In the range where the WBBB was used for the calibrations, the residuals are below 10 mK from 20 °C to 75 °C.
4. Radiation thermometer characterizations
4.1. Temporal stability
One of the critical issues with thermal infrared devices and radiation thermometers is the poor long-term stability of these devices . This is mostly due to the sensitivity of the detector to the thermal radiation from the constantly changing ambient background and changing device temperatures. Unlike detectors which operate in the visible radiation range, detectors that are sensitive to thermal radiation cannot be shielded from ambient sources unless they are placed in cryogenically cooled, vacuum environments. Thus, the challenge in working with thermal detectors is to stabilize the thermal background radiation and yet have the detector be sensitive to the relevant sources of thermal radiation. The effectiveness of the temperature-stabilized assembly shown in Fig. 1 can be seen in Fig. 7. For these measurements, the ART was used to measure the radiation from the water-bath blackbody, which was set at a nominal temperature of 45 °C and controlled to an uncertainty of < 2 mK over the entire time interval. For these measurements, the initial temperature using the ART was found to be 45.342 °C. The temperature-stabilization circuit, which was on prior to these measurements, was turned off at the initial start of the measurements shown in Fig. 7, and the optical assembly was allowed to equilibrate to the varying ambient temperature of the laboratory, which was about 1.0 °C to 0.5 °C lower than the assembly temperature. Due to the poor thermal coupling between the optical assembly and the surroundings, the temperatures measured by ART increases slowly in about an hour to 0.5 °C higher than the initial temperatures. Without the feed-back temperature implemented, the ART measured temperatures change by about 100 mK during this 15 hour interval. The rapid response of the ART to the stabilization circuit being turned on is shown at the 16 h mark. The measured temperatures return to the initial values to within 2 mK.
Measurements were performed to determine the time constant for the thermal stabilization circuit to bring the ART to equilibrium conditions. In Fig. 8, the thermal stabilization was turned off at the 3 h mark and turned back on at the 4 h mark. The measured temperatures came back to < 5 mK within 30 min of the resumption of the stabilization loop.
Long-term measurements were also performed to test the limits of the ART stabilization circuit as shown in Fig. 9. The WBBB was set at 20 °C and the ART was used to perform measurements over a 48 h interval. Using a contact thermometer, the WBBB has been determined to be stable to < 1 mK over this interval due to the large volume of fluid (42 l) in the reservoir. Non-contact, ART temperatures were found to be stable to < 3 mK for 48 h. The laboratory temperature was not stabilized and could have been changing by a few degrees Celsius over the 2 days of measurements. The outside chassis of the ART is not temperature stabilized and was allowed to drift with the environmental temperatures. The stability of the ART directly results from the thermal stability of the detector compartment shown in Fig. 1.
The ART was also used to measure sources whose temperatures are lower than 20 °C. Figure 10 shows the measurements of the AHPBB set at a temperature of −30 °C. The AHPBB is cooled using a recirculating ethanol bath whose temperature is set from a contact sensor on the AHPBB. Chilled ethanol is circulated into the AHPBB using insulating hoses which are about 2 m in length. The oscillations are from the bath and are results of the control algorithm which could not be optimized further. The ART temperatures are digitized to about 10 mK due to the resolution limit of the lock-in amplifier.
4.2. Noise-equivalent temperature differences
A key metric of radiation thermometers is the noise-equivalent temperature difference or NETD. The NETD of the ART is difficult to assess due to many factors. Since the ART measures in the thermal infrared, a long-term stable source that is capable of low-noise operations and insensitive to the temperature fluctuations from the environment is needed. For the measurements shown in Fig. 11, the AHPBB was allowed to drift at room temperature without the recirculating ethanol bath, and its temperature measured using the imbedded SPRT. The large heat capacity of the heat pipe andthe deep blackbody cavity result in a stable source that slowly drifts over time. The AHPBB temperatures are also set close to the internal reference temperature of the ART so that the signal differences of the reference and measurement sources are small. In Fig. 11, the SPRT temperatures are digitized at 1 mK due to the resolution of the readout device. The ART NETD is estimated to be < 1 mK at 22 °C. The ART tracks the SPRT temperatures over the 5 h of measurements shown in Fig. 11. The NETD at lower than ambient temperatures are more difficult to assess due to the oscillations of the AHPBB. The NETD was assessed by differences of the SPRT and ART temperatures over a time interval where both measurements were smoothly, monotonically varying. The NETD based upon these measurements is found to be about 3 mK at −30 °C as determined from the width or the peak-to-valley of the differences shown in Fig. 12. The measurement of the NETD was performed in region where the effect of the quantization of the ART temperatures is reduced.
4.3. Size of source effect
The size-of-source effect (SSE) was determined for the ART using a commercial, flat-plate variable-temperature blackbody (FPBB) with a source diameter of 152.4 mm. A circular, metallic, variable-aperture iris was used to vary the source diameter, and the SSE was measured using the direct method . The FPBB was set at a constant temperature of 55 °C for these measurements, and to avoid emissivity changes with varying iris diameters, the FPBB was set at 20 degrees from perpendicular to avoid inter-reflections that could lead to a change in the FPBB emissivity. It was found that the FPBB aligned perpendicular to the optical axis of the ART resulted in an increase in the measured signals as the iris diameters were decreased. This increase in signal can be explained due to the increase in the FPBB emissivity from 0.95 to a higher value from the inter-reflections of the infrared radiation within the iris. The effectiveness of the Lyot stop in suppressing SSE can be observed in Fig. 13. With the detector placed directly at the field-stop location, the SSE is not suppressed and shows additional structure due to scatter from optical elements and the chassis. The SSE is measured using the direct method
5. Estimated uncertainties
A partial estimate of the standard and expanded uncertainties of the ART non-contact temperature measurement is given in Table 4. The traceability of the temperature measurements comes from the standard, platinum-resistance thermometer. The uncertainty value for the SPRT temperatures was obtained from manufacturer specifications. The WBBB emissivity is calculated to be 0.99997 when the opening is reduced to 50 mm diameter . Due to the depth of the respective WBBB and AHPBB cavities, the temperature differences between the SPRT and the cavity bottom and the non-uniformity of the cavity temperature are combined into the first term in the uncertainty budget. The uncertainty of the residuals of the fitted interpolation function is obtained from Fig. 6. The reproducibility of the ART temperatures is obtained from Fig. 9. The total uncertainties of the ART are only about a factor of 2 greater than that of industrial-grade standard platinum resistance thermometers. The uncertainty due to the chopper-surface temperature instability is reduced due to the coating of the surface of the chopper with an low-emittance aluminum foil tape. The long-term stable measurements shown in Fig. 9 or Fig. 11 would be not possible if the chopper surface temperature was varying by more than 1 mK. This is an incomplete estimate since we do not have long-term, month-long reproducibility measurements for this design, and furthermore uncertainties have not been assessed at the entire temperature range of the ART operations from −46 °C to 75 °C.
6. Discussion and conclusions
A thermally stabilized optical design for an infrared radiation thermometer has been constructed and characterized using variable-temperature blackbodies. The temperature stabilization of the field stop, collimating lens, Lyot stop, focusing lens, and detector within a common enclosure at a temperature close to ambient enables long-term stable operation of the radiation thermometer. Pyroelectric detectors must be operated using modulated signals since they respond to changes in the signal rather than steady state signal, and any changes in the background signal will lead to changes in the modulated signals. In this design, the chopper wheel has been covered with reflective aluminum tape to reduce the infrared emittance to low levels so that the detector will just measure the thermal radiation from the temperature-stabilized assembly shown in Fig. 1. The long-term stability of the calibration depends on the reproducibility of the detector response as well as the stability of the thermistors used for feed-back stabilization using the thermo-electric coolers. The few millikelvin stability of the ART measurements over days is attributed to the repeatability of these components. Since the reproducibility of thermistors over time has been well demonstrated, it is now possible to develop radiation thermometers which can be stable to few millikelvins over extended periods of time, such as months, and are easily transportable and deployable.
This new design can be immediately used to improve standards-quality radiation thermometers used to validate non-contact temperature scales at calibration laboratories. Incorporation of these design principles would also improve field-level radiation thermometers which are critical for assessing sea- and land-surface temperatures. Other thermal-infrared devices such as non-dispersive infrared (NDIR) gas sensors which operate in the mid- and long-infrared regions could also be improved with this design. Lastly, development of thermal imagers using the design in Fig. 1 is being explored.
1. P. Coates and D. Lowe, The Fundamentals of Radiation Thermometers (CRC Press, 2017).
2. O. Struss, “Transfer radiation thermometer covering the temperature range from −50 deg C to 1000 deg C,” in Temperature: Its Measurement and Control in Science and Industry Vol 7, D. C. Ripple, ed. (AIP, 2003), pp. 565–570.
3. J. P. Rice, J. J. Butler, B. C. Johnson, P. J. Minnett, K. A. Maillet, T. J. Nightingale, S. J. Hook, A. Abtahi, C. J. Donlon, and I. J. Barton, “The Miami2001 infrared radiometer calibration and intercomparison. Part I: laboratory characterization of blackbody targets,” J. Atmos. Ocean. Tech. 21(2), 258–267 (2004). [CrossRef]
4. Y. S. Yoo, B. H. Kim, S. D. Lim, S. N. Park, and S. C. Park, “Realization of a radiation temperature scale from 0 °C to 232 °C by a thermal infrared thermometer based on a multiple-fixed-point technique,” Metrologia 50(4), 409–416 (2013). [CrossRef]
5. J. P. Rice and B. C. Johnson, “The NIST EOS thermal-infrared transfer radiometer,” Metrologia 35(4), 505–509 (1998). [CrossRef]
6. H. W. Yoon, D. W. Allen, and R. D. Saunders, “Methods to reduce the size-of-source effect in radiometers,” Metrologia 42(2), 89–96 (2005). [CrossRef]
7. S. N. Mekhontsev, V. B. Khromchenko, and L. M. Hanssen, “NIST radiance temperature and infrared spectral radiance scales at near-ambient temperatures,” Int. J. Thermophys. 29(3), 1026–1040 (2008). [CrossRef]
8. S. D. Wood, B. W. Mangum, J. J. Filliben, and S. B. Tillett, “An investigation of the stability of thermistors,” J. Res. Natl. Bur. Stand. 83(3), 247–263 (1978). [CrossRef]
10. A. Rogalski, “Next decade in infrared detectors,” Proc. SPIE 10433, 104330L (2017).
12. P. Saunders, “General interpolation equations for the calibration of radiation thermometers,” Metrologia 34(3), 201–210 (1997). [CrossRef]
13. M. Battuello, F. Girard, T. Ricolfi, M. Sadli, P. Ridoux, O. Enouf, J. Perez, V. Chimenti, T. Weckstrom, O. Struss, E. Filipe, N. Machado, E. Van der Ham, G. Machin, H. McEvoy, B. Gutschwager, J. Fischer, V. Schmidt, S. Clausen, J. Ivarsson, S. Ugur, and A. Diril, “The European project TRIRAT: arrangements for and results of the comparison of local temperature scales with transfer infrared thermometers between 150 °C and 962 °C,” in Temperature: Its Measurement and Control in Science and Industry Vol 7, D. C. Ripple, ed. (AIP, 2003), pp. 903–908.
14. D. Lowe, M. Battello, G. Machin, and F. Girard, “A comparison of size of source effect measurements of radiation thermometers between IMGC and NPL,” in Temperature: Its Measurement and Control in Science and Industry Vol 7, D. C. Ripple, ed. (AIP, 2003), pp. 625–630.