CAPÍTULO V. ANÁLISIS DEL COMPORTAMIENTO DE INSOLVENCIA DE
5.1.1. Comportamiento en el termómetro de insolvencia:
case 250µm) is large enough not to hit the electrodes with the focus of the OPCPA. For a first trial experiment, fused silica was used for better comparison with data acquired in the visible during the investigations leading to [114] and [98]. The fused silica substrate used was not IR grade (as it was not available from the supplier used before and rather kept constant for reproducibility). However the wedges before the focus are also made out of non-IR grade fused silica as well to absorb any IR components over 2.6µm that could lead to linear absorption (and damage) at the high focus intensities. In future experiments, that are designed for the IR OPCPA, other dielectrics will be investigated (as discussed in section 6.4).
6.3
Experiment
0 2 4 6 8 10 12 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 wedge position (mm) lock -in v oltage (mV )Figure 6.3: Light-induced current (after trans-impedance amplification) measured while shifting CEP and dispersion by performing a wedge scan.
With the characterized focus spot size (60µm) and the applied output power (between 60 and 210 mW), the intensity assuming 14 fs pulses at a repetition rate of 3 kHz equates to a peak of 350TW/cm2 (≈4.3 V/Å). It should be mentioned however, that due to spatial parts of the OPCPA
output that cannot be focused properly (as indicated in section 7.4.2), the effective field strength could be considerable lower. In the future this could be corrected for by using a hollow core fiber as a spatial filter, like performed in [122]. This way, not only the focusability would be ensured, but also spot sizes and beam pointing could be made constant. Beam pointing before the fiber would result in mere power fluctuations, that are easier to detect and treat statistically in the measurement.
90
6. First prototype experiment: controlling currents in a dielectric using IR few-cycle pulses
The first measurement was to observe how the induced current changes with CEP. Note that the carriers are still injected at any phase, but there is no net polarization after the laser pulse for certain values. To find these, the pair of thin wedges (apex angle measured as 2.8◦) transmitted
by the OPCPA beam scans the optical path length in the material. This changes both CEP and duration of the pulses. Therefore we expect an oscillating current with an envelope (that indicates that longer pulses inject and drive less carriers). Compared to [114], fused silica in the IR will not disperse the pulse as much as in the visible domain. The CEP slip however is not too different. At a center wavelength of 2.1µm the decoherence length is given by Ld = 32.6µm (also see
table B.2), while at 600 nm it decreases to Ld = 15µm. This equals a wedge movement of
655µm to offset the CEP byπ.
In fig. 6.3, we see the result of one of these scans, performed at an input power of 170 mW. The amount of glass that could be introduced was limited by the dimensions of the wedge and the large beam size (needed to avoid unwanted nonlinear effects, see section 5.4). The step size was set to 47.6µm (1000µsteps of the used stepper motor) and 281 position were evaluated. The sig- nal was taken as the mean of 3 consecutive measurements at each point with an integration time of 100 ms. The error bars indicate the standard deviation of these 3 shots. Over this measurement period, CEP drifts of the OPCPA can be neglected, according to section 5.5.
A sine function with quadratic envelope was assumed and fitted to the obtained data. The result is a period of 657µm, which is in excellent agreement to the expectation. The peak voltage was fitted as 1.36 mV.
The chargeQptransferred can be calculated from this as follows: The measured voltage from the
trans-impedance amplifier was taken at an amplification 108 V/Asetting. For a measurement of
1.5 mV therefore, a charge of 1.5 10−11C was induced per second. For a repetition rate of 3 kHz, this means an induced charge of 5 fC (or about 31000 electrons) per laser shot.
Figure 6.3 shows a measured signal that is not centered on zero exactly, but contains an offset of -0.17 mV. This was first attributed to electronic background noise. However, when we blocked the light incident on the sample, the noise floor was only 1.2µV (standard deviation) (and with the cover removed 4.0µV) and centered at zero. The situation changed when the sample was illuminated. Without covers (and without moving the wedge), we observed oscillations on the order of 1 mV. These could be heavily reduced when enclosing the sample by our aluminum cover. We attribute this to radiative noise that is coupled into our measuring circuit, when the light makes the sample a photo-conductive switch. As the lock-in frequency is around 1.5 kHz, where many turbo-molecular pumps work in the lab, we suspect the signal oscillations to be slight variations in their (regulated) rotation speeds. The magnetic fields of these pumps then could have coupled into our current measuring circuit. With a time-constant of 100 ms, the rotation speed variations can easily oscillate in and out of this frequency window.
However, using the shielding the offset could still observed. As it was also constant over the measurement period, it appeared as being part of the measurement. A CEP-flip of the AOPDF, that is not exactly π, would explain the offset. The oscillation in current then does not exactly add up coherently in the lock-in technique. Instead, a part of the signal is not recovered, which