For over 10 years, several bolometer sensors with different properties have been tested in the IBOVAC facility. The aim has been to develop a bolometer sensor that can be operated in ITER and can withstand harsh operating conditions. For this purpose, important physical properties of the sensors, i.e., cooling time constant τ, normalized heat capacity κ, and normalized sensitivity sn, have been characterized in a vacuum condition and at various temperatures up to 300 °C. The calibration is achieved by ohmic heating of the sensor absorbers by applying a DC voltage and recording exponential current fall during heating. Recently, a Python program was developed to analyze the data and extract the above mentioned parameters including the uncertainties from recorded currents. In the present series of experiments, the latest prototype sensors developed for ITER are tested and evaluated. These include three different sensor types: two with Au absorbers on ZrO2 membranes (self-supporting substrate sensors) and one with Au absorbers on Si3N4 membranes supported by a Si frame (supported membrane sensors). Tests revealed that the sensor with ZrO2 substrate can only be operated up to 150 °C, while the supported membrane sensors passed the tests up to 300 °C successfully. These results will be used, together with other upcoming tests, such as irradiation testing, to select the most suitable sensors to be employed in ITER.

Bolometers are used to measure the incident radiation power over a wide range of the electromagnetic spectrum. In these sensors, a thin layer absorbs electromagnetic radiation, resulting in increased temperature. The radiation power can be measured by the change in the electrical or physical properties, such as resistance or bending, of the absorber itself, or by measurement of the absorber temperature from the opposite side, e.g., by a temperature dependant resistor.1–3 

Metal resistive bolometers were first developed by Langley in the 19th century.4 The main parts consist of a thin metal layer deposited on a membrane and resistors that are in thermal contact with the membrane: two resistors are deposited under an absorber layer exposed to radiation, and these sense temperature changes (called measurement absorber and meanders). Another two resistors are deposited under an absorber shielded from incident radiation (called reference absorbers and meanders). These four resistances are connected in a Wheatstone bridge. When incident radiation power heats the measurement absorber, heat is transferred through the membrane to the measurement meanders. Then, in these measurement meanders, the temperature increases, while in the reference meanders, it stays constant. This causes an imbalance in the Wheatstone bridge voltage, which can be used to sensitively measure the incident radiation power.

Bolometer sensors have been widely used in fusion reactors to measure the total radiation power loss from the plasma and its spatial distribution. This is an important parameter since it can be used to control reactor operation and also study various scenarios in fusion science. Measuring changes in radiation from the plasma can give deep insight into how different parameters and inputs, such as applied power and neutral gas injection, affect the plasma.5–7 

Bolometers are considered an important diagnostic tool for the ITER fusion reactor as well.8,9 However, operational conditions in ITER, the largest and most powerful fusion reactor ever to be built, are much harsher compared to the previous fusion reactors. In ITER, the first wall temperatures are expected to reach up to 400 °C and up to 250 °C at the bolometer sensors, which are shielded by first wall components and camera housings. Also, more intense electromagnetic and neutron radiations are to be dealt with. These are huge challenges for bolometer cameras and sensors.10 To develop the most suitable bolometer sensors for ITER, different concepts of bolometers sensors with different materials for absorbers, meanders, and membranes have been designed, prototyped, and tested in the IBOVAC facility at Max Planck Institute for Plasma Physics11–14 and other facilities.15–17 Here, we present the calibration and thermal test results of the bolometer sensors developed for ITER in the final prototyping phase. These results together with upcoming studies, such as neutron irradiation tests, will determine the most suitable bolometer sensor to be employed in ITER.

This work is organized as follows: Sec. II presents the various prototype sensor types and their properties. Section III B describes the calibration facility and preparation of sensors for testing. Section IV presents the theoretical background for the calculation of sensor parameters and explains the computer code used for data analysis. Section V presents and discusses the experimental results. Section VI outlines the conclusions.

The main requirements for the ITER bolometer sensors were defined based on the considered operational conditions and the temperatures and loads obtained from simulations. These are briefly explained below:

  1. Spectral response: The spectral range and power of radiation from the plasma during the ITER operation were the main factors for the design of the sensor absorbers. The main design parameters were absorber material and thickness. The thickness needed a compromise between a thicker design to cover a higher range of the radiation spectrum and a thinner design, given the manufacturing challenges and sensor durability.

  2. Robustness: The operation of ITER will cause different types of loads on components, the most relevant ones for the bolometer sensor being thermal and nuclear loads. The temperatures that bolometer cameras are expected to reach and the nuclear loads on the sensors were obtained by simulations. The prototype sensors will have to remain functional under these conditions for most of the lifetime of ITER.

  3. Electrical characteristics: A main design parameter is the resistance value of the meanders in sensors. It must be sufficiently large so that it is by far the most dominant (temperature-dependent) resistance value compared to the resistance of tracks in the sensors, cables in cameras, vacuum vessels, feedthroughs, ex-vessel cables, and all connectors. These are expected to add up to several tens of ohms. On the other hand, the resistances must not be too large to avoid, together with other capacitances in the system, RC effects from being noticeable. This should be taken into account since the sensors are normally operated with AC voltage. In addition, as will be discussed in Sec. IV, the sensitivity of sensors is inversely dependent on the resistance of meanders, and excessively high resistances would result in unacceptably low sensitivities. To determine resistance value, a compromise was needed between these considerations.

  4. Sensor parameters: The main requirements for the radiation power diagnostic in ITER are the time-resolution and accuracy of measurements. Thus, a range was determined for the main sensor parameters (cooling time constant and sensitivity), and the sensor was developed to fulfill this goal.

Based on these considerations and resulting requirements, two types of bolometer sensors were selected to be potentially employed in ITER. The most important parameters decided for the sensors were 20 μm gold absorbers and 1 kΩ meander resistances. The main challenge for the manufacture of these sensors was that they should withstand temperatures as high as 400 °C with much thicker absorbers than those manufactured previously. Both sensor types are described below.

It is considered that the sensor-holders of ITER bolometer cameras will have five channels. However, to avoid capacitive coupling and cross-talk between channels, the use of single channel bolometer sensors in ITER has been planned. For this reason, most prototype sensors were produced in single or five-channel layouts. Since, in the IBOVAC test facility, described in Sec. III A, only four channels can be connected and calibrated for each sensor holder, some sensors were produced in four-channel layout, which corresponds to that used on most active fusion devices.

As explained in Sec. I, the main part of resistive bolometers are a radiation absorbing metal layer, a membrane on which the absorber is deposited, and meanders that are deposited on the other side of the membrane under the absorber. A schematic of the self-supporting substrate sensor type is shown in Fig. 1. As can be seen, only a substrate supports the absorber and meanders without any additional supporting structure. To reduce the time of heat transfer from the absorber to the support structure (heat sink), a thin metal layer in the range of a few hundred nanometres can be deposited on the absorber side of the membrane, connecting the absorber to the heat sink. This layer is known as the heat conduction layer (HCL). The thickness of the HCL affects the sensor response considerably and is an important parameter in the sensor design.

FIG. 1.

Schematic of self-supporting substrate bolometer sensor type.

FIG. 1.

Schematic of self-supporting substrate bolometer sensor type.

Close modal

Self-supporting substrate bolometer sensors were developed for the ASDEX fusion reactor at the beginning of the 1980s18 and have since been widely used in various fusion reactors, such as JET, ASDEX Upgrade, Tore Supra, RFX, and FT-U.1,19–21 The original design consisted of 4.5 μm thick gold absorbers and 0.1 μm thick meanders with 5 kΩ resistance deposited on a 7.5 μm thick Kapton foil.18,19 Later, the design was changed to 8 μm thick gold absorbers deposited on a 20 μm thick mica substrate, with the meander resistance being 1.2 kΩ.22 Mica could withstand the higher temperatures in JET, which reached up to 320 °C.23 

During decision making for ITER bolometer sensors in 2016, the manufacturer of the mica sensors no longer produced them. An additional concern was that, since the mica is a natural product and each series can have different properties, the reproducibility of the sensors with mica substrate would be at stake. Furthermore, since the mica is hygroscopic, the vulnerability of the mica substrate layers to steam was concerning, particularly in the event of water ingress events.

Considering the above mentioned issues, it was decided to substitute mica with yttria-stabilized zirconia (YSZ). YSZ substrates are synthetic; hence, the properties can be kept close to each other in different production series. In addition, they are mechanically stable and resistant to neutron irradiation.

Two variants of the self-supporting substrate sensors were produced on the same specifications by two manufacturers, German Fraunhofer IMM and Swiss CSEM. The main differences between the manufactured prototypes (other than manufacturing tolerances) are the number of sensor channels, different track layouts, and platinum vs gold HCL for IMM and CSEM sensors, respectively. Both sensor variants are described in detail below.

An image of an IMM sensor without HCL is shown in Fig. 2. The membrane of these four-channel sensors is 20 × 23 mm2 large and 18–23 μm thick. The gold absorbers are 1.5 × 4 mm2 large, and their thickness is 20 μm. The platinum HCLs of the sensors have three thickness variants: 0, 200, and 350 nm. The platinum meander resistances are ∼1100 Ω at room temperature.

FIG. 2.

Image of an IMM self-supporting substrate sensor without HCL from the absorbers side.

FIG. 2.

Image of an IMM self-supporting substrate sensor without HCL from the absorbers side.

Close modal

An image of a CSEM sensor with 155 nm gold HCL is shown in Fig. 3. The five-channel sensors are 23 × 25 mm2 large and 19–23 μm thick. The gold absorber size is the same as the IMM one, and the substrate thickness varies between 17.5 and 20.5 μm. These sensor samples have either no HCL or feature a gold HCL with a thickness of 155 nm or 215 nm. The resistances of the platinum meanders are ∼1050 Ω at room temperature.

FIG. 3.

Image of a CSEM self-supporting substrate sensor with 155 nm gold HCL from the absorbers side.

FIG. 3.

Image of a CSEM self-supporting substrate sensor with 155 nm gold HCL from the absorbers side.

Close modal

Both manufacturers have also produced single-channel prototype sensors with the same characteristics as described above. They will be used in various sensor tests, such as neutron irradiation.

In supported membrane bolometer sensors, the membrane is a very thin layer (in the range of some micrometers) that is supported by a much thicker supporting structure (in the range of some hundred micrometers). A schematic of the sensor is shown in Fig. 4.

FIG. 4.

Schematic of supported membrane bolometer sensor type.

FIG. 4.

Schematic of supported membrane bolometer sensor type.

Close modal

These types of bolometer sensors were introduced by Giannone et al. in 2005.8 The first prototype consisted of a 1.5 μm thick SiN membrane and a 1.5 μm thick platinum absorber and platinum meanders with 1.2 kΩ resistance. The membrane was supported by a 600 μm silicon wafer.

Since 2008, the supported membrane bolometer sensors have been manufactured and further developed by Fraunhofer IMM in Mainz, Germany. The first version had a 4.5 μm thick platinum absorber deposited on a 1.5 μm thick SiN membrane and was tested in ASDEX Upgrade in 2009.11 In 2012, prototypes with 3 μm membrane thickness and the same absorber thickness were tested whereby mechanical failure of platinum absorbers over 200 °C was observed.12 Despite efforts to overcome the problem by using other manufacturing techniques, this problem could not be solved, and gold was chosen as the absorber material for the later prototyping.14 

Based on these insights, IMM has manufactured three combinations of the absorber and heat conduction layer thickness for the final prototyping phase for ITER: (1) 18.4 μm thick gold absorber without HCL, (2) 19.7 μm thick gold absorber and 200 nm platinum HCL, and (3) 16.4 μm thick gold absorber and 350 nm platinum HCL. Images of a sensor without HCL are shown in Fig. 5. The absorber size is the same as the supported-membrane sensors, i.e., 1.5 × 4 mm2. The thickness of the absorbers varies due to manufacturing procedures and tolerances. The sensors are single channel and are 5 × 23 mm2 large. The SiN membrane thickness is 3 μm, and the silicon support structure is 600 μm thick. The resistance of the platinum meanders is ∼920 Ω at room temperature.

FIG. 5.

Images of an IMM supported-membrane sensor without HCL from the meanders (left) and absorbers (right) sides.

FIG. 5.

Images of an IMM supported-membrane sensor without HCL from the meanders (left) and absorbers (right) sides.

Close modal

A bolometer sensor test facility (IBOVAC), which can reproduce ITER-like thermal and vacuum conditions, is available at Max Planck Institute for Plasma Physics. The test facility and sensor calibration procedure are to be described in this section. In previous studies,12,14 spring-loaded pins were employed to establish electrical contact between sensors and cables of the test facility. However, the supplier specified that the spring-loaded contacts reliably operate only up to 200 °C due to the loss of pre-tension of the spring material. Hence, high variance in measured parameters at higher temperatures was associated with the unreliability of the pins. To eliminate this problem and to have more reliable electrical connections at high temperatures, the sensor assemblies and electrical connections have been redesigned as described in this section.

The IBOVAC facility consists of a 2.2 m long and 0.7 m diameter cylindrical vacuum chamber, which can be heated up to 450 °C. The pumping is realized by a two-stage process: first, a mechanical pump is used to reduce the pressure to 10−3 mbar; second, a turbo pump is powered on, which can reduce the pressure down to 10−7 mbar at room temperature. During experiments, the pressure rises due to increased temperature, however, never above 10−4 mbar. This is important for neglecting the effects of the convection and conduction heat transfer of the ambient gas on the calibration results. In addition to the electrical calibration of the sensors at temperatures up to 450 °C, it is possible to perform optical calibration of the sensors with a laser at 405 nm wave length and also to apply pressure shocks to the sensors. For more details, see Ref. 14.

The sensors are installed on a carriage and are placed inside the vacuum chamber. The carriage has a heating element inside it to supplement chamber heating. The temperature of each sensor is measured by placing a K-type thermocouple inside a hole in the ceramic sensor front plate. Three sensor holders, each having four channels electrically connected, can be installed on the carriage, resulting in 12 sensor channels being tested simultaneously.

In each channel, the calibrations are realized by short-circuiting meanders of one of the absorbers (e.g., measurement absorber) and applying a DC voltage (2.5 V) to the meanders of the other absorber, following the procedure described in Ref. 21. In this way, each absorber can be calibrated separately, and the absorbers can be compared with each other. A simplified schematic of the calibration circuit is shown in Fig. 6. Here, M1 and M2 are the meanders of the absorber being calibrated, either measurement or reference absorber, and Rc is the cable resistance. The 2.5 V DC voltage is applied for 3 s, and the circuit current is recorded during this time. The current acquisition is achieved by measuring the voltage change over a 5 Ω resistance. The calculation of sensor parameters using the recorded current is described in Sec. IV.

FIG. 6.

Electrical circuit of sensor meanders during calibrations.

FIG. 6.

Electrical circuit of sensor meanders during calibrations.

Close modal

The control and data acquisition systems are operated using a dedicated setup based on hardware and LABVIEW software from National Instruments. In earlier studies,12,14 only one calibration run was done for each measurement, and data analysis was performed by LABVIEW software. As stated above, in these studies, there was a high scattering in the sensor parameters. In order to address the unreliabilities, the LABVIEW software has been changed so that multiple calibration runs are performed for each measurement. In the current experiments, for each measurement, ten calibration runs are done. In addition, a Python code has been developed for data and statistic analysis, as described in Sec. IV. The LABVIEW software is solely used to acquire the data.

During the experiments, first, a calibration of the sensor is done at room temperature and atmospheric pressure. Then, the vacuum chamber is pumped down, and calibration is done at room temperature and vacuum conditions. Finally, the vacuum chamber and the carriage are heated up to the desired temperatures, and calibrations are performed while the temperature is kept constant.

As mentioned above, as spring-loaded pin contacts are unreliable at elevated temperatures and previous studies12,14 showed high fluctuations in the measured parameters, it was necessary to change the electrical connections. Computer aided design (CAD) models of the new sensor assembly are shown in Fig. 7. In Fig. 7(a), the assembly is shown from the bottom side, without showing the cover plate. The sensors are placed and pressed between front and back plates, both made of ceramic. There is a copper foil with a thickness of 100 μm between the front plate and the sensors. This is to increase heat conduction between the sensor and the front plate and to have better thermal contact between them. The back plate has holes that make it possible to reach the contact point of the sensors. There are also copper tracks deposited on the back plate, which make electrical connections from the holes to the sides of the plate. In each hole, the contact point of the sensor is wire-bonded to the metal track on the back plate. At the end of each track, a copper wire is micro-welded. An ODU SPINGTAC® pin is crimped to the other end of the cable. The cables are held together by wire holders. As shown in Fig. 7(b), the sensor assembly is installed on a stainless-steel platform. The platform is installed on the carriage of the IBOVAC test facility. The platform has a ceramic insulator, in which the pin sockets are mounted. The pin sockets are crimped to the cables (not shown in the drawing), which are connected to the feedthrough. The results presented in Sec. V A for the supported membrane sensors are obtained by this assembly.

FIG. 7.

(a) CAD model of the new sensor assembly shown from the bottom (cover plate hidden) and (b) sensor assembly installed on the platform.

FIG. 7.

(a) CAD model of the new sensor assembly shown from the bottom (cover plate hidden) and (b) sensor assembly installed on the platform.

Close modal

During the first tests, self-supporting substrate sensors were damaged by the calibrations at temperatures between 150 and 200 °C (for details, see Sec. V B). To investigate the reasons for this damage in a simpler configuration and to reduce preparation time, it was decided to perform some tests using spring-loaded pin assemblies, and for details of this assembly, see Refs. 1 and 24. The spring-loaded pins can be operated safely up to 200 °C. To reduce the risk of damage to the pins at higher temperatures, the applied voltage for the calibrations was reduced. In addition, the applied voltage is set to zero when there is no calibration being done. This limited time of current flow through the pins and reduced their aging. In this way, the pins could be operated to temperatures up to 300 °C without causing problems in the electrical connections. The results presented in Sec. V B are obtained by this assembly.

Here, the theoretical background and formula used for the calculation of the bolometer sensor parameters are explained, and the computer code developed for the calculations and data analysis is described.

An example of the applied voltage and the recorded circuit current is shown in Fig. 8. As mentioned in Sec. III B, this current fall is used to calculate the calibration parameters of the bolometer sensors, which are the cooling time constant τ, normalized heat capacity κ, and the normalized sensitivity Sn.21 The data fitting process by Python will be explained in Sec. IV B.

FIG. 8.

An example of the recorded data and the exponential fit done by the Python code (blue curve). The residuals are shown in μA at the bottom.

FIG. 8.

An example of the recorded data and the exponential fit done by the Python code (blue curve). The residuals are shown in μA at the bottom.

Close modal

τ is the time-constant of the exponential current fall, which is obtained by doing a fit to the recorded current. Giannone et al.21 have derived the formula for κ, which is also used for the calculation of this parameter in the present study. The derived formula is

κ=RBImax44UcalΔI,
(1)

where RB is the resistance of the bolometer meander at the instance of the calibration (M1 or M2 in Fig. 8), Imax is the maximum current reached by applying the DC voltage Ucal and ΔI is the difference between Imax and steady-state current (Iinf in Fig. 8). Considering Fig. 6, RB can be calculated as

RB=2UcalImaxRC,
(2)

where RC is the external cable resistance.

The sensitivity of bolometer sensors is defined as the ratio of the change in the Wheatstone bridge imbalance due to the absorbed radiation power,

S=ΔVoutP.
(3)

For a Wheatstone bridge, the change in the bridge imbalance can be obtained as

ΔVout=14ΔR1R1ΔR2R2+ΔR3R3ΔR4R4Vin,
(4)

where Vin is the applied voltage (Ucal in bolometer sensor calibrations) and ΔRi is the change in resistance of Ri due to the applied voltage. Since the bolometer sensors are made such that all four resistances have (almost) the same value; without any ohmic or radiation power, the initial bridge imbalance is zero, i.e., ΔVout is simply Vout. Since, during the calibration, one absorber is short-circuited (just two resistances are present, e.g., ΔR2 and ΔR4 are zero), and all the resistances are assumed to have the same value RB, Eq. (4) can be written as

Vout=14ΔRohmRB+ΔRohmRBUcal=ΔRohm2RBUcal,
(5)

where ΔRohm is the change in the resistance due to the applied ohmic power. This can be calculated as

ΔRohm=RBRinf=2UcalImaxRC2UcalIinfRC=2ΔIUcalIinfImax.
(6)

By substituting Eq. (5) in (3), one will obtain

S=ΔRohm2RBPohmUcal,
(7)

where Pohm is the applied ohmic power UcalImax. To make the sensitivity value independent from the applied voltage, Eq. (7) can be normalized to Ucal to obtain the normalized sensitivity as

Sn=ΔRohm2RBPohm.
(8)

As mentioned above, in order to calculate the calibration parameters, a mathematical fit needs to be done on the recorded exponential current fall by the equation shown in Fig. 8. A Python code is developed to read the data of calibration measurements, make the fits, calculate the parameters, perform statistical analysis, and output the data for each sensor and channel. The details of the Python code are described below.

A combination of the ExponentialModel() and ConstantModel() of the Lmfit module25 is employed for fitting the data. An example of the fits is shown in Fig. 8. The outputs of the fit are τ, ΔI, and Iinf, which are used for calculating the calibration parameters.

As described in Sec. III B, ten consecutive runs are done for each measurement. By fitting and calculating the parameters for all of the runs, the code calculates the average and the standard deviation for each output parameter. It was observed that for some measurements, the standard deviation was high, while the recorded data for the runs did not differ significantly from each other. The investigations showed that this is caused by some fits not converging well, which led to artificially high standard deviation. Hence, a filter is implemented in the code to limit the acceptable output parameters and filter the fits that give unrealistic outputs or have large residuals with the recorded data.

The recorded temperature data are fed to the code as well, and the code can correlate the calculated parameters with the measurement temperature.

In this section, the calibration results obtained by the procedure described above are presented. The supported-membrane sensors successfully passed the thermal tests. During the calibrations, a problem with the self-supporting substrate sensors, resulting in the deterioration of the platinum tracks and meanders, was observed. A preliminary investigation was done, which is also presented here.

Several single-channel supported membrane sensors with different HCL thicknesses were calibrated at varying temperatures. The calibration results are shown in Fig. 9. These results are chosen from the reference absorbers of a selection of the calibrated sensors and are representative of all the calibrated sensors. The differences between the measured parameters of reference and measurement absorbers of a sensor are always less than 5%. In Fig. 9, the error bars show the standard deviation of the measured parameters. In most of the measurement points, the standard deviation is less than 1%. Hence, the error bar is smaller than the symbol and is not visible.

FIG. 9.

Calibration results for (a) resistance, (b) cooling time constant, (c) normalized heat capacity, and (d) normalized sensitivity of supported membrane sensors for reference absorbers of selected sensor channels. The error bars show the standard deviation of the measurements, which are typically less than 1%, making the error bars not visible.

FIG. 9.

Calibration results for (a) resistance, (b) cooling time constant, (c) normalized heat capacity, and (d) normalized sensitivity of supported membrane sensors for reference absorbers of selected sensor channels. The error bars show the standard deviation of the measurements, which are typically less than 1%, making the error bars not visible.

Close modal

The linear dependence of the resistance of the meanders to temperature can be observed in Fig. 9(a). This is an expected result since it is known that the resistance of platinum changes linearly with temperature. The measured temperature coefficient of resistance (TCR) is ∼3.2 × 10−3 K−1.

The cooling time constants are presented in Fig. 9(b). It can be concluded that the cooling time constant of the sensor is nearly independent of temperature. This observation was also reported in previous studies.11 The cooling time constant is an indication of the thermal path between the absorber as the source and the silicon structure as the heat sink. Since this path does not change with elevated temperatures, τ is not affected by increasing the temperature. There is a slight change in the values of the sensor with no HCL, which can be associated with a change in contact between the sensor surface, copper foil, and sensor holder at different temperatures. The results also show the drastic effect of the HCL on the cooling time constant. Adding 200 and 350 nm platinum HCLs has reduced τ from 800 to 500 ms and 350 ms, respectively. In a previous study,14 for a sensor with 12 μm absorber thickness and 150 aluminum HCL, a τ value of ∼300 ms was measured. The values presented in the present study are higher, which is partially due to the thicker absorbers. In addition, aluminum’s thermal conductivity at room temperature is approximately three times higher than platinum (237 vs 77 W m−1 K−1). Hence, a 150 nm aluminum HCL can be considered as a 450 nm platinum HCL, which can also describe the lower τ value of older sensors.

Figure 9(c) shows that the normalized heat capacity has a slight linear dependence on temperature, which is much less than TCR, indicating that the heat capacities of the sensors are not dependent on the resistance of the meanders. Although from Eq. (1), it seems that κ is linearly dependent on resistance, and it should be kept in mind that the resistance is also embedded in the other terms of the equation, making the relation between the two non-linear. This is also physically expected since the heat capacity of the sensor is merely a material property and should not be dependent on the resistance of the meanders. The slight increase of κ can be explained by the increase in the heat capacity of the sensor materials (gold, SiN, and platinum) by the increased temperature. The results also show that adding an HCL increases the normalized heat capacity of the sensor. This can mainly be attributed to the cooling time constant. The main definition of the normalized heat capacity is the heat capacity normalized by τ, and a thicker HCL results in lower τ and increased κ.

The results presented in Fig. 9(d) show that the sensitivity of the sensors falls considerably with increased temperature. The temperature-dependence is almost linear and close to TCR, indicating the dependence of the sensitivity to the resistance of the meanders. This dependence can also be seen in Eq. (8), which is used for the calculation of sensitivity. The sensor without HCL has the highest sensitivity, and a thicker HCL results in lower sensitivity. As described in Sec. IV, sensitivity is the measure of change in the output voltage due to input power. By adding a heat conduction layer, the thermal power is lost quickly to the heat sink, and only a smaller part of the applied power will be available for the heating of the meanders. This results in a smaller change in the output voltage and lower sensitivity.

The results of one selected absorber with the thickest HCL from each sensor type are shown in Fig. 10. These represent the behavior observed for the sensors with different HCL thicknesses. The effects of HCL in the sensor parameters are similar to those observed for the supported-membrane sensors.

FIG. 10.

Calibration results for (a) resistance, (b) cooling time constant, (c) normalized heat capacity, and (d) normalized sensitivity of IMM and CSEM self-supporting substrate sensors for reference absorbers of selected sensor channels. The error bars show the standard deviation of the measurement.

FIG. 10.

Calibration results for (a) resistance, (b) cooling time constant, (c) normalized heat capacity, and (d) normalized sensitivity of IMM and CSEM self-supporting substrate sensors for reference absorbers of selected sensor channels. The error bars show the standard deviation of the measurement.

Close modal

As can be seen in Fig. 10(a), the resistances of the sensors increase with the temperature. However, there is a drop in TCR at temperatures higher than 200 °C. The CSEM sensor has a TCR of 2.5 × 10−3 K−1 until 200 °C, while the TCR from 200 to 300 °C is 1.8 × 10−3 K−1. The drop in the TCR of the IMM sensor is higher, changing from 2.9 × 10−3 to 1.1 × 10−3 K−1 for the temperatures lower and higher than 200 °C, respectively.

The cooling time constants are shown in Fig. 10(b). The values are nearly the same for the sensors of both manufacturers. The time constant values are much higher than the supported-membrane type. For the IMM sensors with the same HCL thickness (350 nm Pt), the τ values for supported and self-supporting substrate sensors at room temperature are 365 and 1404 ms, respectively. This shows that membrane thickness and material play an important role in the cooling time constant of the sensors. In addition, the various sensor holders used for the two sensor types, and the resulting difference in the clamping of the sensors and the distance to the heat sink play a role in the difference between τ values.

At lower temperatures, similar to the supported-membrane sensors, the τ values do not change considerably with the increased temperature. However, for both IMM and CSEM sensors, the τ values drop drastically starting from 175 °C and reaching ∼400 ms at 300 °C.

Similar behavior is observed for the normalized heat capacity values. As shown in Fig. 10(c), similar to the supported-membrane sensors, the values increase slightly with increased temperature. This is reversed at 175 °C, at which the values start to drop. The drop in the κ value is higher for the IMM sensor.

The sensitivity values are presented in Fig. 10(d). At room temperature, the values for both sensors are approximately the same and are higher than those of the supported-membrane sensor with 350 nm platinum HCL (1.22 vs 1.02 1/W). This is expected since a higher time constant results in higher sensitivity. Until 175 °C, the sensitivity values drop with increased temperature. As for normalized heat capacity, the behavior change starts from 175 °C, and the sensitivity values increase at the higher temperature. For the IMM sensor, sensitivity increases too much higher values, reaching 3.36 1/W at 300 °C.

The observed behavior for the self-supporting substrate sensors was unexpected since all the delivered sensors were thermally tested by cycling them five times up to 400 °C. However, the results indicate that the sensors can develop problems during the calibrations (applying voltage) at high temperatures. In other words, flowing current through the tracks and meanders at elevated temperatures damages the sensors and changes their properties. The calibrated sensors were opened and investigated after the tests. As shown in Fig. 11, the images of the sensor before and after the tests indicate that the calibration measurements at higher temperatures damage the sensors by changing the structure of the meanders and tracks. This can be observed in the different reflections of light from one meander in the post-testing image of the sensor. As the first observation, it can be stated that due to the flowing current through the platinum meanders and tracks, the adhesion of the layers to the YSZ substrate deteriorates and the layer detaches. However, a detailed investigation of this phenomenon is outside the scope of this study and is a subject for future study.

FIG. 11.

Microscopic images of meanders of a self-supporting substrate IMM sensor with 350 nm platinum HCL before and after calibration tests.

FIG. 11.

Microscopic images of meanders of a self-supporting substrate IMM sensor with 350 nm platinum HCL before and after calibration tests.

Close modal

The results presented above show that the supported-membrane sensors developed by IMM can be safely calibrated up to 300 °C. The sensor parameters change as expected, i.e., τ is approximately temperature-independent, and resistance and sensitivity are linearly dependent on temperature. The uncertainties (standard deviations) of the measurements are low, which is important for reducing the error of the measured radiation power. These sensors were also thermally cycled up to 400 °C by the manufacturer, and there was no damage recorded.

Although the self-supporting substrate sensors were also cycled up to 400 °C by the manufacturers, they showed vulnerability to the calibrations at temperatures higher than 150 °C. Starting from this temperature, the temperature-dependencies of the sensor parameters were inverted, the τ values dropped drastically, and the sensitivities increased. The uncertainties of the measurements were also much higher than the supported-membrane sensors. Post-test investigations showed that the sensors were damaged, in particular in meanders and tracks. Cracks were often found on the sensors or they were completely broken during testing. This can be a sign of incompatibility with the sensor holders and suggests a need for an improved design of sensor holders for this kind of sensor.

Since the grounding of the absorbers of the bolometer sensor in ITER is a requirement (to avoid electrostatic charge build-up on the absorber), the sensors should have a HCL to connect the absorbers to the ground. Yet, the results of Sec. V show that a thick HCL reduces the sensitivity of the sensor drastically. Although cooling the time constant is reduced, analytical studies have shown that the sensor response does not differ at higher frequencies.26 This means that the sensors produced for ITER will have a HCL, but a rather thinner one compared to the HCL thickness of the prototype sensors.

Considering the observations described above, the supported-membrane sensors seem to be a more probable choice for ITER. However, there are still planned and ongoing tests on the sensors, the most important one being the neutron irradiation test. The final decision will be made after the completion and evaluation of all tests.

For the calculation of the radiation power, it is important to know the sensor calibration parameters before a measurement is done. Considering that the parameters of the sensors can change during their lifetime (mainly due to thermal, radiation, and neutron irradiation loads), it is important to calibrate the sensors regularly, also at different temperatures, during their operation in ITER. Current commissioning, operation, and maintenance plans in preparation for the final design review of the ITER bolometer diagnostic do foresee such calibration measurements. Additionally, the effect of temperature change during a discharge on the calibration parameters can be compensated by a scheme such as the one presented in Ref. 14.

The final prototype bolometer sensors developed for ITER were described in detail. New sensor holders, based on wire-bonded electrical connections between the sensor and a ceramic plate with metal tracks, were introduced. These were also developed as a practical concept for ITER. The calibration procedure was described, and the data evaluation routine was explained. The results show that the supported membrane sensors can be reliably operated at temperatures up to 300 °C. The resistances and sensitivity values are linearly dependent on temperature, while the cooling time constant is nearly temperature-independent. Self-supporting substrate sensors could only be reliably calibrated up to 150 °C; calibrations at higher temperatures changed the behavior of the sensors, and they were physically damaged. It is presumed that the electrical current flowing during the calibrations affects the adhesion of tracks to the substrate. This phenomenon seems to occur above a certain threshold temperature. A clear conclusion cannot be drawn yet and is the subject of future study.

The results of the current study show that the supported membrane sensors are a more reliable choice for radiation measurements in ITER. However, other tests, such as steam exposure, neutron irradiation, resilience against mechanical loads (vibrations), intense light flashes due to disruption mitigation, and pressure waves, are being performed or planned for the near future. The results of these studies will also contribute to the selection of the bolometer sensors to be used in ITER.

This work was partly supported by Fusion for Energy under Grant No. F4E-FPA-384-SG05. The views expressed in this publication are the sole responsibility of the authors and do not necessarily reflect the views of Fusion for Energy and the ITER Organization. Neither Fusion for Energy nor any person acting on behalf of Fusion for Energy is responsible for the use, which might be made of the information in this publication.

The authors have no conflicts to disclose.

Sina Jahanbakhsh: Conceptualization (equal); Data curation (lead); Investigation (equal); Methodology (equal); Software (equal); Validation (equal); Visualization (lead); Writing – original draft (lead). Jack Davies Hare: Conceptualization (equal); Methodology (equal); Software (equal); Writing – review & editing (supporting). Hans Meister: Conceptualization (equal); Funding acquisition (lead); Investigation (equal); Methodology (equal); Project administration (lead); Supervision (lead); Validation (equal); Writing – review & editing (lead). Christian Ingesson: Funding acquisition (equal); Investigation (equal); Methodology (equal); Writing – review & editing (supporting). Marcin Majewski: Investigation (supporting); Resources (equal); Writing – review & editing (supporting). Florian Penzel: Conceptualization (supporting); Investigation (equal); Resources (equal); Writing – review & editing (supporting). Stefan Schmitt: Conceptualization (equal); Investigation (equal); Resources (equal); Writing – review & editing (supporting). Ulrich Walach: Conceptualization (supporting); Investigation (supporting); Writing – review & editing (supporting). Marc Dubois: Conceptualization (supporting); Investigation (supporting); Resources (supporting); Writing – review & editing (supporting).

The data that support the findings of this study are available from the corresponding author upon reasonable request.

1.
K. F.
Mast
,
J. C.
Vallet
,
C.
Andelfinger
,
P.
Betzler
,
H.
Kraus
, and
G.
Schramm
, “
A low noise highly integrated bolometer array for absolute measurement of VUV and soft x radiation
,”
Rev. Sci. Instrum.
62
,
744
750
(
1991
).
2.
E.
Iborra
,
A.
Sanz‐Hervás
, and
T.
Rodríguez
, “
A new design of a semiconductor bolometer on rigid substrate for fusion plasma diagnostics
,”
Rev. Sci. Instrum.
64
,
1714
1717
(
1993
).
3.
P. T.
Lang
and
K. F.
Mast
, “
Photoresponse of a miniaturized ultrabroad-band low-noise metal-film bolometer detector array
,”
J. Opt.
27
,
25
(
1996
).
4.
S. P.
Langley
, “
The bolometer and radiant energy
,”
Proc. Am. Acad. Arts Sci.
16
,
342
358
(
1880
).
5.
L. C.
Ingesson
,
B.
Alper
,
B. J.
Peterson
, and
J.-C.
Vallet
, “
Chapter 7: Tomography diagnostics: Bolometry and soft-x-ray detection
,”
Fusion Sci. Technol.
53
,
528
576
(
2008
).
6.
F.
Reimold
,
M.
Wischmeier
,
M.
Bernert
,
S.
Potzel
,
D.
Coster
,
X.
Bonnin
,
D.
Reiter
,
G.
Meisl
,
A.
Kallenbach
,
L.
Aho-Mantila
, and
U.
Stroth
, “
Experimental studies and modeling of complete H-mode divertor detachment in ASDEX Upgrade
,”
J. Nucl. Mater.
463
,
128
134
(
2015
).
7.
R. A.
Pitts
,
X.
Bonnin
,
F.
Escourbiac
,
H.
Frerichs
,
J. P.
Gunn
,
T.
Hirai
,
A. S.
Kukushkin
,
E.
Kaveeva
,
M. A.
Miller
,
D.
Moulton
 et al, “
Physics basis for the first ITER tungsten divertor
,”
Nucl. Mater. Energy
20
,
100696
(
2019
).
8.
L.
Giannone
,
D.
Queen
,
F.
Hellman
, and
J. C.
Fuchs
, “
Prototype of a radiation hard resistive bolometer for ITER
,”
Plasma Phys. Controlled Fusion
47
,
2123
(
2005
).
9.
A.
Gusarov
,
S.
Huysmans
,
L.
Vermeeren
,
E. R.
Hodgson
, and
M.
Decréton
, “
In situ in-reactor testing of potential bolometer materials for ITER plasma diagnostics
,”
Fusion Eng. Des.
82
,
1179
1184
(
2007
).
10.
H.
Meister
,
L.
Giannone
,
L. D.
Horton
,
G.
Raupp
,
W.
Zeidner
,
G.
Grunda
,
S.
Kalvin
,
U.
Fischer
,
A.
Serikov
,
S.
Stickel
, and
R.
Reichle
, “
The ITER bolometer diagnostic: Status and plans
,”
Rev. Sci. Instrum.
79
,
10F511
(
2008
).
11.
H.
Meister
,
T.
Eich
,
N.
Endstrasser
,
L.
Giannone
,
M.
Kannamüller
,
A.
Kling
,
J.
Koll
,
T.
Trautmann
,
ASDEX Upgrade Team
,
P.
Detemple
, and
S.
Schmitt
, “
Optimization of a bolometer detector for ITER based on Pt absorber on SiN membrane
,”
Rev. Sci. Instrum.
81
,
10E132
(
2010
).
12.
H.
Meister
,
M.
Kannamüller
,
J.
Koll
,
A.
Pathak
,
F.
Penzel
,
T.
Trautmann
,
P.
Detemple
,
S.
Schmitt
, and
H.
Langer
, “
Reliability issues for a bolometer detector for ITER at high operating temperatures
,”
Rev. Sci. Instrum.
83
,
10D724
(
2012
).
13.
H.
Meister
,
H.
Langer
, and
S.
Schmitt
, “
An alternative geometry for bolometer sensors for use at high operating temperatures
,”
Fusion Eng. Des.
112
,
579
586
(
2016
).
14.
F.
Penzel
,
H.
Meister
,
L.
Giannone
, and
D.
Zhang
, “
Calibration parameter drift compensation of metal resistive bolometers operating in a thermal varying environment
,”
Fusion Eng. Des.
121
,
100
104
(
2017
).
15.
A.
Gusarov
,
S.
Huysmans
,
L.
Giannone
, and
H.
Meister
, “
Radiation stability of resistive bolometers intended for operation in the ITER environment
,”
Fusion Eng. Des.
86
,
1200
1203
(
2011
).
16.
H.
Meister
,
M.
Willmeroth
,
D.
Zhang
,
A.
Gottwald
,
M.
Krumrey
, and
F.
Scholze
, “
Broad-band efficiency calibration of ITER bolometer prototypes using Pt absorbers on SiN membranes
,”
Rev. Sci. Instrum.
84
,
123501
(
2013
).
17.
H.
Meister
,
S.
Schmitt
,
I.
Szenthe
,
A.
Szakál
,
H.
Albrecht
, and
F.
Gillemot
, “
Irradiation tests of bolometer sensor prototypes for ITER
,”
Fusion Eng. Des.
161
,
111995
(
2020
).
18.
E. R.
Müller
and
F.
Mast
, “
A new metal resistor bolometer for measuring vacuum ultraviolet and soft x radiation
,”
J. Appl. Phys.
55
,
2635
2641
(
1984
).
19.
K. F.
Mast
,
H.
Krause
,
K.
Behringer
,
A.
Bulliard
, and
G.
Magyar
, “
Bolometric diagnostics in jet
,”
Rev. Sci. Instrum.
56
,
969
971
(
1985
).
20.
A.
Murari
,
M.
Cecconello
,
L.
Marrelli
, and
K. F.
Mast
, “
Analysis of the calibration methods and error propagation for the sensitivity S and the cooling time constant τc of the gold metal foil bolometers
,”
Rev. Sci. Instrum.
75
,
2692
2699
(
2004
).
21.
L.
Giannone
,
K.
Mast
, and
M.
Schubert
, “
Derivation of bolometer equations relevant to operation in fusion experiments
,”
Rev. Sci. Instrum.
73
,
3205
3214
(
2002
).
22.
A.
Huber
,
K.
McCormick
,
P.
Andrew
,
P.
Beaumont
,
S.
Dalley
,
J.
Fink
,
J. C.
Fuchs
,
K.
Fullard
,
W.
Fundamenski
,
L. C.
Ingesson
 et al, “
Upgraded bolometer system on jet for improved radiation measurements
,”
Fusion Eng. Des.
82
,
1327
1334
(
2007
).
23.
R.
Reichle
,
J.
Fuchs
,
R.
Giannella
,
N.
Gottardi
,
H.
Jackel
,
K.
Mast
,
P.
Thomas
, and
P.
Van Belle
, “
Bolometer for ITER
,” in
Proceedings of the International Workshop on Diagnostics for ITER
(
Plenum Press
,
Varenna
,
1996
).
24.
H.
Meister
,
M.
Bernert
,
W.
Biel
,
M.
Han
,
L. C.
Ingesson
,
K.
Mukai
,
F.
Penzel
,
B. J.
Peterson
,
R.
Reichle
,
M. L.
Reinke
 et al, “
Bolometer developments in diagnostics for magnetic confinement fusion
,”
J. Instrum.
14
,
C10004
(
2019
).
25.
M.
Newville
,
T.
Stensitzki
,
D. B.
Allen
, and
A.
Ingargiola
, “
LMFIT: Non-linear least-square minimization and curve-fitting for Python
,”
2014
.
26.
E.
Müller
,
G.
Weber
,
F.
Mast
,
G.
Schramm
,
E.
Buchelt
, and
C.
Andelfinger
, “
Design of a four-channel bolometer module for ASDEX upgrade and tore supra
,” Technical Report No. IPP 1/224,
Max Planck Institute for Plasma Physics
,
1985
.