model would clearly emerge from the experimental data [31, 32, 33, 20, 34, 35, 36, 37].
However, although many studies highlighted very interesting behaviour, it turns out that this is a very complex phenomenon and no clear consensus has been found in the community yet. Very recently, Allmaras et al. [38] established a detection model for NbN taking into account the timing jitter experimental data. A particularly crucial feature of the detection models is the energy-current relation, which describes the bias current threshold needed at a given photon energy to create a detection event. The measurement of this relation is the primary method of investigation of the detection mechanism. A more detailed introduction of the different models and experimental methods is given in [32].
3.2 Experimental setup
Device under test
The device is fabricated out of a 5 nm thick film of amorphous Mo0.8Si0.2, with aTc= 5K, which is deposited by co-sputtering with a DC and RF bias on the molybdenum and silicon targets, respectively. The MoSi film is deposited on a thermally oxidized silicon wafer and is capped with a 3 nm a-Si layer. X-ray diffraction measurements have been performed, confirming the amorphous nature of the MoSi. The film is patterned into a meandered wire with a width of 170 nm and a gap of 160 nm as shown in the Fig. 3.2, and total surface area of 16×16 µm2 by a combination of EBL and ion beam etching.
The detection efficiency at 1550 nm is 20%. The device has been selected out of tens of other detectors with different widths and fill factor by looking for the highest critical current and widest plateau region in order to have the largest energy range accessible.
16 !m
200 nm
Superconducting nanowires
Figure 3.2: Scanning electron microscope image of the device under test. The white1
dashed circle represents the region where the photons are absorbed. Inset: close-up look.
The detector is mounted in a sorption cryostat reaching 0.75 K and biased with a current source and its critical current is 14.7 µA. The voltage pulses from detection events are amplified by a custom low-noise amplifier cooled to 40 K and by a secondary amplifier at room temperature. The detector is illuminated with unpolarized photons coming from a halogen lamp sent through a grating monochromator, as shown in Fig. 3.3. This provides a continuous spectrum from 750 to 2050 nm. We carefully calibrated the monochromator using laser lines at 632.8, 980.1, 1064.0, 1310.2 and 1550.8 nm. By using the second order of some of these wavelengths, we obtain 9 calibration points, extending up to 2128.0 nm with a 4 nm uncertainty. Appropriate low-pass filters were inserted to avoid crosstalk from higher diffraction orders.
monochromator
variable attenuator
fast
discriminator
cryostat
SNSPDs counter
current source
light
0.8 —> 2.5 K 40 K
room temperature amplifier low noise cryogenic amplifier
Figure 3.3: Schematic of the experimental setup. Broadband light is sent on a grating mirror, which narrows down the spectrum to a specific wavelength depending of the angle.
The filtered light is sent onto the device under test through a variable attenuator to fix the number of incoming photon constant. After a detection event, the electric signal goes through two stages of amplification to be discriminated and counted. The temperature of the device can be adjusted between 0.75 K and 2.5 K.
Discriminator settings
In order to assess that the experimental data have a physical meaning and are not affected by any electronic effect, we must pay attention to the pulse discrimination electronics. By decreasing the wavelength (increasing the energy) of the incident photon, the bias current needed to create an event decreases. As the signal amplitude depends only on the applied bias current, a problem can arise when the detector is operating at such low currents. If the discriminator threshold level is not set correctly, the consequence is that the photon counts cannot be distinguished from the amplifier noise and the shape of the PCR is affected. In Fig. 3.4a, the PCR is plotted as a function of the bias current for different discriminator threshold values at 750 nm photon wavelength. By increasing the threshold value, the bias current needed to overpass it increases and the PCR curves shift to the right. The transition width then represents the noise level of the amplification, which is non-physical and undesired. The discriminator level has to be set as low as possible,
3.2. Experimental setup while still not under the noise level. In addition to this, the transition width becomes steeper, see Fig. 3.4b, where the transition width∆Ib =Ib80%−Ib20% and the bias value Ib50% extracted from Fig. 3.4a are plotted as a function of the discriminator level.
2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.0
Figure 3.4: Tuning of the discriminator settings. (a) Relative photon count rate (photon count rate subtracted by the DCR and divided by the maximum count rate) as a function of the bias current at 0.75 K for an incident photon wavelength of 750 nm. Each color represents one measurement run with a specific discriminator threshold value in millivolts.
Each solid line traces the error function fit for the respective data curve. The leftmost and rightmost curves correspond to 44 mV and 150 mV, respectively. When the discriminator level is too high, more bias current is required in the nanowire to overpass it. The sigmoid shape then represents the electrical noise. (b) Transition width extracted from (a) as a function of the discriminator setting value. The vertical dashed line represent
approximately the maximum reasonable discriminator level.
We verified that none of the curves are affected by scanning the discriminator level and restricting ourselves to the bias currents where the count rate was independent of the threshold level, corresponding to the safe region at the left of the red line in Fig. 3.4b.
During all the measurements, the discriminator level was set to 46 mV. The minimum detectable voltage pulse in the setup occurs at a bias current of approximately 2.5 µA.
A voltage pulse as seen on the oscilloscope for different detector bias currents is shown in Fig. 3.5. The discriminator setting of 46 mV is indicated as a blue dashed line in the figure. We see that even with the lowest pulse amplitude at 2.5 µA, the discriminator level is lower than the pulse amplitude by a significant amount. The second oscillation appearing around 60 ns for currents above 10 µA does not affect our response as the discriminator dead time is set to 80 ns. We also note the high-pass effect of our cryogenic amplification, which reduces the width the typical SNSPD output signal and creates an undershoot. However, this does not affect the shape of the curves that we measured.
0 20 40 60 80 Time (ns)
−200
−100 0 100 200 300
Amplitude(mV)
Disciminator : 46 mV
3µA 45 8
1012 14
Figure 3.5: Oscilloscope traces of pulses for different bias currents. The dashed blue line indicates the discrimination level of 46 mV used during the measurement.