(a) Scheme of Corbino microwave spectrometer (not to scale) and (b) typical sample with inner and outer diameters and of Corbino disk.
Reproducibility of the microwave connections. The effect of disconnecting the microwave line between sample and calibration measurements is evident from this room-temperature measurement of the impedance of a bulk stainless steel sample. For clarity the “after” data are offset by .
Scheme of the overall setup of the spectrometer. To replace the sample, the glass cryostat is lowered and the stainless steel tube housing the exchange gas is removed. In order to reduce the overall length of high-frequency transmission lines, microwave source and test set are located as close to the top of the cryostat as possible. The test set converts the microwave signal to one of much lower, intermediate frequency (if) which can then easily be transmitted to the (more distant) actual network analyzer.
Design of probe head incorporating two Corbino adapters.
General error model for a microwave reflection measurement.
Temperature dependence of the two-lead dc resistance for several NiCr thin film samples.
Temperature dependence of error coefficients at 10 GHz.
Consistency of “short only” (so) and “three standards” (3s) calibrations: ratio of the impedances obtained by the two procedures (based on exemplary, experimentally determined error coefficients).
Frequency dependence of two NiCr thin films (thickness 10 and 25 nm), measured at 1.7 K.
Frequency dependence of the impedance of a 8 nm thick aluminium film, measured at 3 K. The calibration either employs bulk copper at 3 K (Cu) or the superconducting sample at 1.7 K (SC) as short. The two-lead dc resistance in the inset shows the superconducting transition.
Conductivity spectra of at 3 K: comparison of “short only” calibration (so, SC) using the superconducting sample as short and “three standards” calibration (3s,Al) using bulk aluminum as short, teflon as open, and a NiCr thin film as load. Real and imaginary parts, and , are vertically offset for clarity.
Contributions concerning the reproducibility of the spectrometer. Contact resistance: a bulk aluminum sample was measured at room temperature, removed completely, and measured again (several times). The standard deviation of these complex reflection coefficients is shown. Thermal cycling: The same short standard was measured from 1.7 to 300 K, cooled down again, and remeasured. We calculate the difference of the measured complex reflection coefficients at 1.7 K. This was repeated for five independent measurements of different samples. The average of these differences is presented. For comparison the standard deviation of several measurements (at 3 K) employing the superconducting calibration is shown.
Relative sensitivity with respect to the load impedance assuming a constant sensitivity of 0.001 with respect to the reflection coefficient.
Article metrics loading...
Full text loading...