Logo Logo
Hilfe
Kontakt
Switch language to English
Detection efficiency and bandwidth optimized electro-optic sampling of mid-infrared waves
Detection efficiency and bandwidth optimized electro-optic sampling of mid-infrared waves
Electro-optic sampling (EOS) is a powerful method for the characterization of electric fields with frequencies in the range of ~ 1-300 THz. For mid-infrared (MIR) radiation (2-20 µm), it can be understood as a two-step process: first, a sub-MIR-cycle visible/infrared gate pulse generates sum and/or difference frequency radiation with the light field under investigation in a nonlinear medium. Second, the newly generated frequencies are detected in a heterodyne scheme with the transmitted gate pulse serving as the local oscillator. Scanning the delay of the gate pulse with respect to the MIR waveform results in a signal proportional to the incident MIR field, with the spectral response depending on the gate pulse duration and phasematching in the detection crystal. The nonlinear frequency conversion on the one hand transfers the detection to the near-infrared spectral range, affording the use of low-noise photodetectors. On the other hand, it limits the detection efficiency and subjects it to a trade-off against bandwidth. Our research group develops high-power ultrashort-pulsed laser sources for field-resolved infrared spectroscopy. Explicitly, nonlinearly post-compressed femtosecond lasers are used both to drive the generation of waveform-stable MIR light for molecular-sample excitation, as well as for obtaining gate pulses for EOS of the full macroscopic sample response. To maximize the sensitivity of our field-resolved spectrometers, this thesis studied the photon detection efficiency of EOS for MIR radiation with wavelengths in the 6-18-μm range in experiment and theory. Three different types of gate pulses were investigated experimentally: first, the EOS detection efficiency was characterized for gate pulses with 1030-nm central wavelength, generated by an Yb-thin-disk oscillator. Limited by multi-photon-absorption-caused damage of the GaSe crystal, with an average gate-pulse power of 450mW, a conversion efficiency of 2% from the MIR into sum-frequency photons was achieved in a 500-μm-thick detection crystal. Accounting for Fresnel reflections at the crystal and losses in the heterodyne detection, up to 0.76% of the incident MIR photons arrived at the balanced diodes. Together with mW-level MIR average powers, this resulted in 13 orders of magnitude frequency-domain intensity dynamic range at 9-μm wavelength for a measurement time of 16 s and a scan range of 3.3 ps. However, phase-mismatch limited the −20 dB spectral width to 1.2 µm. Using a 85-μm-thick GaSe crystal, the full MIR spectrum of the source, spanning from 6.6 to 10.7 µm at −20 dB, was detected, while trading in two orders of magnitude in peak dynamic range. In our research group, the prototype field-sensitive spectrometer with this record detection efficiency and dynamic range is currently being used for fingerprinting real-world biomedical samples, with up to 40 times higher molecular detection sensitivity than commercial Fourier-transform infrared spectrometers. Due to the dispersion of GaSe, the trade-off between detection efficiency and spectral coverage is mitigated for longer-wavelength gate pulses. Using gate pulses centered at 1550 nm wavelength from an Er-fiber laser and a 300-μm-thick crystal, a comparable detection dynamic range and bandwidth as with the 85-μm-thick crystal at 1030 nm was achieved, despite the lower gate pulse power of 120mW. This performance enables high sensitivity spectroscopic measurements, when employing the Er-laser in a dual-oscillator fast-scanning mode (~ 1 kHz scan rate), avoiding low-frequency noise sources. In addition to the broader phasematching bandwidth, choosing a longer gate-pulse wavelength also increases the detection-crystal damage threshold due to reduced multi-photon absorption, allowing for the use of higher gate pulse powers and, consequently, enhancing the nonlinear interaction. This benefit was harnessed in the investigation of limitations to the detection efficiency with 1.9-W gate pulses from a Tm-fiber laser at 1965 nm central wavelength, comparing several EOS crystal thicknesses with respect to detection efficiency and spectral coverage. Traces measured with 100 to 300-μm-thin crystals closely resemble the incident field, spanning from 8.1 µm to 14.2 µm at −10 dB, with a conversion efficiency from the MIR into sum-frequency photons of up to ~ 10%. Using a 500-μm-thick GaSe crystal, more than 20% of the MIR photons from a 3-μm spectral band around 9.3 µm were upconverted. Further increasing the crystal thickness resulted in saturation of the depletion, explained by temporal walk-off and reduced peak powers due to dispersion. The overall number of detectable MIR photons of ~ 6.4% from within the detection crystal an interaction time window, together with mW-level MIR powers, lead to a peak intensity dynamic range > 10^14, with twice the detection bandwidth as for the 1030-nm gate pulses in the efficiency-optimized configuration, thus spanning ~ 5 μm at −20 dB. Despite the MIR depletion upon detection, the EOS signal scaled linearly with the field strength for average photon numbers between 10^3 and 10^17 per second within our measurement accuracy, because enough MIR photons stay available for nonlinear interaction. The multi-percent-level conversion efficiency allows for characterization of waveforms with an average of 22 photons inside the detection crystal in a 2.2-ms-long integration time window per temporal element. The combination of sensitivity, dynamic range and spectral coverage finds application e.g., in broadband vibrational spectroscopy, where the minimum detectable concentration is only a factor of ~ 4 higher than what would be possible when detecting all incident MIR photons. Furthermore, the detection bandwidth allows for the simultaneous measurement of multiple molecular species with spectrally wide-spread absorption lines. Employing the high detection dynamic range, a further study in the frame of this thesis concerned the use of EOS as a highly sensitive characterization technique for the stability and reproducibility of the MIR waveform and, therefore, for the control over optical fields. These capabilities were demonstrated by measuring the temporal fluctuations of the EOS trace, resulting in a record-low timing jitter of < 10 as over billions of pulses. A theoretical model simulating the chain of nonlinear processes from the laser frontend to EOS detection confirmed the measured values, identifying intensity noise of the modelocked oscillator front-end as the main source of the remaining MIR waveform instabilities. These jitter values were 3 orders of magnitude above the field fluctuations expected from a shot noise- limited driving pulse train.
Lasers, Nonlinear Optics, Electro-Optic Sampling, Mid-Infrared Generation, Field-Resolved Spectroscopy
Hofer, Christina
2021
Englisch
Universitätsbibliothek der Ludwig-Maximilians-Universität München
Hofer, Christina (2021): Detection efficiency and bandwidth optimized electro-optic sampling of mid-infrared waves. Dissertation, LMU München: Fakultät für Physik
[thumbnail of Hofer_Christina.pdf]
Vorschau
PDF
Hofer_Christina.pdf

5MB

Abstract

Electro-optic sampling (EOS) is a powerful method for the characterization of electric fields with frequencies in the range of ~ 1-300 THz. For mid-infrared (MIR) radiation (2-20 µm), it can be understood as a two-step process: first, a sub-MIR-cycle visible/infrared gate pulse generates sum and/or difference frequency radiation with the light field under investigation in a nonlinear medium. Second, the newly generated frequencies are detected in a heterodyne scheme with the transmitted gate pulse serving as the local oscillator. Scanning the delay of the gate pulse with respect to the MIR waveform results in a signal proportional to the incident MIR field, with the spectral response depending on the gate pulse duration and phasematching in the detection crystal. The nonlinear frequency conversion on the one hand transfers the detection to the near-infrared spectral range, affording the use of low-noise photodetectors. On the other hand, it limits the detection efficiency and subjects it to a trade-off against bandwidth. Our research group develops high-power ultrashort-pulsed laser sources for field-resolved infrared spectroscopy. Explicitly, nonlinearly post-compressed femtosecond lasers are used both to drive the generation of waveform-stable MIR light for molecular-sample excitation, as well as for obtaining gate pulses for EOS of the full macroscopic sample response. To maximize the sensitivity of our field-resolved spectrometers, this thesis studied the photon detection efficiency of EOS for MIR radiation with wavelengths in the 6-18-μm range in experiment and theory. Three different types of gate pulses were investigated experimentally: first, the EOS detection efficiency was characterized for gate pulses with 1030-nm central wavelength, generated by an Yb-thin-disk oscillator. Limited by multi-photon-absorption-caused damage of the GaSe crystal, with an average gate-pulse power of 450mW, a conversion efficiency of 2% from the MIR into sum-frequency photons was achieved in a 500-μm-thick detection crystal. Accounting for Fresnel reflections at the crystal and losses in the heterodyne detection, up to 0.76% of the incident MIR photons arrived at the balanced diodes. Together with mW-level MIR average powers, this resulted in 13 orders of magnitude frequency-domain intensity dynamic range at 9-μm wavelength for a measurement time of 16 s and a scan range of 3.3 ps. However, phase-mismatch limited the −20 dB spectral width to 1.2 µm. Using a 85-μm-thick GaSe crystal, the full MIR spectrum of the source, spanning from 6.6 to 10.7 µm at −20 dB, was detected, while trading in two orders of magnitude in peak dynamic range. In our research group, the prototype field-sensitive spectrometer with this record detection efficiency and dynamic range is currently being used for fingerprinting real-world biomedical samples, with up to 40 times higher molecular detection sensitivity than commercial Fourier-transform infrared spectrometers. Due to the dispersion of GaSe, the trade-off between detection efficiency and spectral coverage is mitigated for longer-wavelength gate pulses. Using gate pulses centered at 1550 nm wavelength from an Er-fiber laser and a 300-μm-thick crystal, a comparable detection dynamic range and bandwidth as with the 85-μm-thick crystal at 1030 nm was achieved, despite the lower gate pulse power of 120mW. This performance enables high sensitivity spectroscopic measurements, when employing the Er-laser in a dual-oscillator fast-scanning mode (~ 1 kHz scan rate), avoiding low-frequency noise sources. In addition to the broader phasematching bandwidth, choosing a longer gate-pulse wavelength also increases the detection-crystal damage threshold due to reduced multi-photon absorption, allowing for the use of higher gate pulse powers and, consequently, enhancing the nonlinear interaction. This benefit was harnessed in the investigation of limitations to the detection efficiency with 1.9-W gate pulses from a Tm-fiber laser at 1965 nm central wavelength, comparing several EOS crystal thicknesses with respect to detection efficiency and spectral coverage. Traces measured with 100 to 300-μm-thin crystals closely resemble the incident field, spanning from 8.1 µm to 14.2 µm at −10 dB, with a conversion efficiency from the MIR into sum-frequency photons of up to ~ 10%. Using a 500-μm-thick GaSe crystal, more than 20% of the MIR photons from a 3-μm spectral band around 9.3 µm were upconverted. Further increasing the crystal thickness resulted in saturation of the depletion, explained by temporal walk-off and reduced peak powers due to dispersion. The overall number of detectable MIR photons of ~ 6.4% from within the detection crystal an interaction time window, together with mW-level MIR powers, lead to a peak intensity dynamic range > 10^14, with twice the detection bandwidth as for the 1030-nm gate pulses in the efficiency-optimized configuration, thus spanning ~ 5 μm at −20 dB. Despite the MIR depletion upon detection, the EOS signal scaled linearly with the field strength for average photon numbers between 10^3 and 10^17 per second within our measurement accuracy, because enough MIR photons stay available for nonlinear interaction. The multi-percent-level conversion efficiency allows for characterization of waveforms with an average of 22 photons inside the detection crystal in a 2.2-ms-long integration time window per temporal element. The combination of sensitivity, dynamic range and spectral coverage finds application e.g., in broadband vibrational spectroscopy, where the minimum detectable concentration is only a factor of ~ 4 higher than what would be possible when detecting all incident MIR photons. Furthermore, the detection bandwidth allows for the simultaneous measurement of multiple molecular species with spectrally wide-spread absorption lines. Employing the high detection dynamic range, a further study in the frame of this thesis concerned the use of EOS as a highly sensitive characterization technique for the stability and reproducibility of the MIR waveform and, therefore, for the control over optical fields. These capabilities were demonstrated by measuring the temporal fluctuations of the EOS trace, resulting in a record-low timing jitter of < 10 as over billions of pulses. A theoretical model simulating the chain of nonlinear processes from the laser frontend to EOS detection confirmed the measured values, identifying intensity noise of the modelocked oscillator front-end as the main source of the remaining MIR waveform instabilities. These jitter values were 3 orders of magnitude above the field fluctuations expected from a shot noise- limited driving pulse train.