Thomas Keating Calibration

Photo-acoustic Thomas Keating power meters are excellent power references in the THz range. A thin film absorbs incoming THz photons, creating heat. The heat causes expansion of a closed air-cell which is finally measured. The thin film can also be heated electrically. 

By providing a defined amount of electrical heat, the device can be calibrated. Details on the calibration can be found at the home page of the manufacturer here. A copy of this file is hosted locally in Google Drive in case the link broke.

A description of the detector is available from the manufacturer. A screenshot of this webpage is hosted locally in Google Drive in case the link broke.

There are two modes of operation of the Thomas Keating power meter: 

• Continuous-wave operation; a modulated (chopped) signal is measured. The Thomas Keating is a power detector. The responsivity is modulation frequency-dependent. Data provided here are for 20 Hz modulation frequency, a duty cycle of 50%, and square wave modulation (“on/off”).

• Measurement of pulses; when measuring pulses that are much shorter than the time constant of the detector, the detector shows an impulsive response, proportional to the pulse energy.  The pulse duration should therefore not exceed 1 ms. The repetition rate of the pulses must be smaller than the inverse time constant of the detector. According to the manual, repetition rates below 40 Hz are fine, however, this is a “soft” boundary. In general, the responsivity will also depend on the repetition rate of the laser. Data given here are for a repetition rate of  1 Hz (the repetition rate of the UCSB FEL) .

DC Characterization

2011-04-15: DC resistance of the film was measured to be 173.91 Ω ± 0.02 Ω for biases between 0.3 and 7 V. 

Continuous-Wave Characterization

We used a frequency generator to provide a square wave bias with 20 Hz modulation and 50% duty cycle. The amplitude was varied between 0.307 and 0.605 V resulting in input powers between 0.542 mW and 2.1 mW. Due to the fairly long time constant of the detector, the output signal is a distorted square wave as illustrated in Fig 2. 

We measured the signal in two ways: 

1. We extracted the peak to peak values using an oscilloscope

2. We used a lock-in to read the RMS value

The determined calibration values are summarized in Table 1. The values are in W/V, i.e. the ratio of input power and detector output voltage for both methods. Note that these values represent the INTERNAL calibration value only. In a THz experiment, the film only absorbs about 50% of the THz signal and parts of the signal may already be lost due to absorption within the window material. The latter may cause a frequency-dependence of the detector at high THz frequencies. 

A calibration of the window absorption can be found in the manual or on the home page of the manufacturer. Losses in the window remain typically small below ~0.7 THz and can be neglected in most cases. 

For 300 GHz, the absorption in the window material is negligible, the external responsivity is 49% of the internal responsivity according to the manual. Calibration values for 300 GHz are shown in Table 1. 

Pulsed Characterization

The signal of the frequency generator was modified to provide short pulses with a repetition rate of 0.99 Hz. We used pulse durations between 0.125 ms and 1 ms and bias voltages on the film between 2.31 and 6.96 V leading to pulse energies between 30.68 µJ and 278.5 µJ. 

The impulsive response of the detector is schematically illustrated in Fig. 3. Note that the shape and duration of this signal does not depend on the pulse duration. It is limited by the speed of the detector itself. 

We measured the values A, B, and C with an oscilloscope. All three quantities (A, B, and C) scale linearly with the pulse energy. We also used a lock-in for detection in addition to the oscilloscope readings. The calibration values are summarized in Table 2. 

The value of B (referenced to zero) is probably more reliable (in the manual also referenced as Vcal), as the negative peak with amplitude A is very sharp and may not be well resolved by some instruments. The fairly high value for lock-in detection is due to the small duty cycle of the pulsed signal, providing a very small voltage even at a high input signal level.

Compared to a measurement performed in 2007, the (internal) peak-peak calibration values differ only by 0.03 %, the Vcal values (B) differ by about 4 %. The pulse energies in that experiment were between 64.8 µJ and 0.58 mJ, the pulse durations were in the µs range (with heating biases around 100 V). 

Written by Sascha Preu

Measurements performed by Sascha Preu, Luis-Claude Brunel and Susumu Takahashi

Ported to new Sherwin Group website & Tables remade by Wei-Hsu (Wish) Lin, 2024-12-29