Experimental Setup

4.3.2.1 Laser Source

For achieving ultrashort laser pulses in the vicinity of the silicon band edge, an optical parametric oscillator (OPO) (Inspire/Newport Spectra-Physics) pumped by the frequency-doubled radiation of a titanium-sapphire ultrashort pulse laser (MaiTai/Newport Spectra-Physics) is used. The OPO emits ultrashort pulses spectrally tunable between 900 nm and 2500 nm, thereby covering the important spectral regions for demonstration of TPA impact in current generation of solar cells. In Fig. 4.5 an autocorrelation trace of an ultrashort pulse emitted by the OPO at 1000 nm center wavelength is shown. The full-width-at-half-maximum (FWHM) of the Gaussian fit to the measured trace equals

Figure 4.5: Autocorrelation trace of an ultrashort pulse emitted by the OPO at 1000 nm center wavelength. The width of the Gaussian fit (red line) equals 223 fs, thus, the FWHM duration of the pulse equalsτFWHM= 223 fs/√

2≈158 fs.

τ_AC = 223 fs, hence, the pulse duration isτ_FWHM = 223 fs/√

2 ≈158 fs.

Pulse duration measurements have been carried out from 900 nm to 1100 nm (upper sensitivity threshold of the applied autocorrelator). Al-thoughτFWHM slightly fluctuates within a range of approximately±10 fs when scanning the OPO wavelengths, no distinct wavelength dependency was observed. Thus, the pulse duration of any OPO wavelength output is approximated by τFWHM≈160 fs.

4.3.2.2 Double-Ring-Resonator Setup

The contribution of TPA to the excess carrier generation in solar cells via ultrashort laser pulses is directly related to the temporal shape of these pulses, especially to their pulse peak intensity (see e.g. Eq. (4.10)).

Hence, experimental evidence of a significant TPA-contribution to theJSC

induced by an ultrashort laser pulse can be deduced from two measure-ments with deviating temporal shape and otherwise identical characteris-tics (e.g. average intensity, wavelength, geometry, direction etc.). If this condition is fulfilled, any variations in the measuredJSCbetween the two measurements can be attributed to TPA.

For this purpose an experimental setup with two ring cavities has been

(a) (b)

Figure 4.6: (a) Schematical drawing of the experimental setup for tem-poral shaping of ultrashort laser pulses by two ring cavities prior to their interaction with a target solar cell. Monitor diode, autocorrela-tor and camera are used to check the functionality of the experimental setup. Published in [33]. (b) Computed temporal pulse shapes emit-ted by the experimental setup representing characteristic situations for temporal alignment of the cavities using the translation stages:

both cavities are aligned to half of the OPO pulse period (top, green,

∆t^1,2_ring=T /2); both cavities are identically misaligned (middle, blue,

∆t¹_ring= ∆t²_ring6=T /2); both cavities are differently misaligned (bot-tom, red, ∆t¹ring 6= ∆t²ring6=T /2). The horizontal axis is normalized to the OPO pulse periodT, the vertical axis to the peak power of the original ultrashort pulses.

developed that is schematically shown in Fig. 4.6a. Incident ultrashort pulses from the OPO are partially transmitted and reflected at a beam splitter. The transmitted beam is delayed with respect to its reflected complement by the roundtrip time ∆tring of a first ring cavity. After this roundtrip, the pulse is again partially transmitted and reflected. The

cavity roundtrip time can be adjusted by a translation stage that varies the cavity length. If the cavity roundtrip time is adjusted to match half of the OPO pulse period (∆tring=T /2), any incident pulse that completed two cavity roundtrips will temporally coincide with the subsequently emitted OPO pulse (2∆t_ring =T), thereby doubling the pulse repetition rate of the pulse train emitted by the OPO (f = 1/∆tring = 2/T = 2frep). By choosing an appropriate beam splitter ratio the respective pulses emitted by the first ring cavity can be equalized in their amplitudes.

As the pulses emitted from the first ring cavity consist of multiple temporally overlapping pulses, any change to the cavity length induces a distortion of the resulting temporal pulse shapes. In order to increase the effect of pulse shape distortion an identical second ring cavity is imple-mented into the setup. In Fig. 4.6b computed temporal shapes emitted by the experimental setup are shown for three characteristic cases: firstly, both cavity roundtrip times are adjusted to half of the OPO pulse pe-riod (top, green, ∆t^1,2_ring=T /2); secondly, both cavity lengths areequally misaligned (middle, blue, ∆t¹_ring = ∆t²_ring 6= T /2); thirdly, both cavity lengths are differently misaligned (bottom, red, ∆t¹_ring 6= ∆t²_ring6=T /2).

From the different pulse peak intensities a variation inJ_SC is expected if there is a significant contribution of TPA.

The experimental setup is completed by an optical lens that focuses the pulse train onto the test cell, thereby reaching average illumination inten-sities in the range of several tens of Wcm⁻², as well as a chopper wheel to enable highly sensitive differential JSC measurements. Transimpedance and lock-in amplifiers (not shown in Fig. 4.6a) are used to keep the target cell in short circuit conditions and capture the differential current increase induced by the chopped ultrashort pulses.

For checking the geometrical alignment of the experimental setup one portion of the focused beam is sampled onto a CMOS-camera. Further-more, a flip mirror can be implemented for a convenient verification of temporal cavity alignment using an autocorrelator. Finally, a monitor diode is used to track laser intensity fluctuations.

In Fig. 4.7 autocorrelation traces for different temporal alignments of the experimental setup are shown. To the left the cavities are temporally misaligned resulting in various pulses measured with the autocorrelator.

Figure 4.7: Autocorrelation traces for different temporal alignment configurations from misaligned (left) to aligned (right), denoted by the arrow on top. The autocorrelation signal merges from multiple pulses into a single pulse when the cavities are aligned. The insets show camera images of the focused pulses for the misaligned and the aligned configuration (after [33]).

When aligning the cavities (from left to right, indicated by the arrow on top) the pulses merge into a single pulse. Furthermore, two insets show camera images of the focused ultrashort laser pulses for temporally misaligned (left) and aligned (right) cavity lengths. Besides the apparently good geometrical alignment, as the individual pulses overlap to a single spot that is close to a Gaussian, fringes appear when the cavities are temporally aligned (note, that they are not appearing in the misaligned case). These are interference fringes resulting from the temporal overlap of multiple pulses and are a strong evidence for a neat temporal alignment of the setup.¹² Likewise, the significant increase of the background noise in the most right autocorrelation trace can be attributed to interference

12The temporal coherence of ultrashort laser pulses, as a measure for their ability to interfere with each other, results from the mode-locking principle that can be envisioned by a single pulse oscillating in the laser cavity. Although the coherence function decays with the number of considered pulse periods, a distinct fringe visibility is expected for the closely spaced pulses interfering in this case [64].

(note the small but distinct peaks next to the main pulse that disappear in the misaligned autocorrelation traces).