SPECTROSCOPY: Femtosecond spectroscopy of QCLs reveals carrier dynamics
Because of their mid- to far-IR emission wavelengths, high output powers, and high wall-plug efficiencies, quantum-cascade lasers (QCLs) have been used for applications in trace-gas sensing, breath analysis, and environmental detection since they were invented in 1994.
Because of their mid- to far-IR emission wavelengths, high output powers, and high wall-plug efficiencies, quantum-cascade lasers (QCLs) have been used for applications in trace-gas sensing, breath analysis, and environmental detection since they were invented in 1994. However, QCLs also enable the study of quantum phenomena in semiconductor structures such as electron transport, relaxation, and light-matter interactions.
At the Center for Advanced Studies in Photonics Research (CASPR) at the University of Maryland Baltimore County (UMBC; Baltimore, MD), Anthony Johnson's group have coupled femtosecond mid-IR pulses into a room-temperature 4.8 µm QCL to investigate carrier dynamics (see www.laserfocusworld.com/articles/366067), with some surprising results.1
The pump-probe experiment
Using a QCL from Claire Gmachl's group at Princeton University (Princeton, NJ), ultrafast mid-IR time-resolved pump-probe spectroscopy is used to study QCL carrier dynamics.2 Femtosecond near-IR pulses from a 76 MHz Ti:sapphire oscillator seed a 250 kHz regenerative amplifier, which in turn pumped an optical parametric amplifier (OPA). Tunable femtosecond mid-IR pulses are obtained by difference-frequency generation (DFG) of the signal and idler beams from the OPA using either a 1-mm-thick silver gallium sulfide crystal (3 to 7 µm wavelength) or silver gallium selenide crystal (7 to 12 µm wavelength).
In the pump-probe setup, 4.8 µm mid-IR pulses, resonant with the QCL lasing transition, have a temporal width of 140 fs and a repetition rate of 250 kHz. A strong pump beam is coupled into the QCL to study the carrier gain dynamics—the pump beam either depletes the electron population in the upper lasing level (subband) or excites electrons from the lower lasing level. A second, weaker probe beam is used to monitor the evolution of the change of gain as a function of the delay between probe and pump. This pump-probe signal is directly related to the occupation probability of the carriers in the intersubband and provides information about the transport of electrons through the injector and active regions of the QCL, ultimately revealing information about the transit time of electrons across the device.
Ultrafast carrier dynamics
The time-resolved pump-probe transmission signal of the QCL biased just below threshold at 28 V (0.8A) indicates at least two different relaxation time constants of 200 fs and 3 ps. Although lasing has not begun, population inversion between the upper and lower lasing levels has been initiated. Thus, the fast gain recovery time of 200 fs is likely due to the depopulation of the lower laser level by longitudinal optical phonon scattering, which is typically less than 500 fs. The electrons are still in the active region of the QCL, but at energy levels below the lower laser level. The longer recovery of the gain is due to the transport of these electrons through the injector region and back into the upper laser level (in the downstream active region). This corresponds to a measured recovery of 3 ps (see figure).
A mid-IR tunable source (top) is used to investigate carrier dynamics of quantum-cascade lasers (QCLs). Data is shown for the room-temperature, time-resolved pump-probe transmission of a 4.8 µm QCL at a pulsed bias (250 kHz, 100 ns) of 28 V (0.8 A), just below lasing-threshold bias (31 V, 0.85 A), with degenerate pump and probe beams of 140 fs duration at a wavelength of 4.8 µm synchronized to the pulsed bias (bottom); gain recovery is on the order of 3 ps (black fitting curve). (Courtesy of UMBC)
Additional experiments indicate a slower recovery time with decreased bias due to the dramatic increase of the upper lasing state lifetime, because the phonon-assisted nonradiative decay in a QCL far below threshold is much slower than the photon-driven stimulated emission near or above threshold. The researchers plan on testing this hypothesis with temperature dependent pump-probe experiments. At zero bias, the group observed an increase in the probe signal as opposed to a decrease of the probe signal due to gain depletion by the pump pulse under high bias. This indicates that the pump pulse excites electrons from the lower lasing level to the upper lasing level; without the presence of the pump pulse, the electrons are more likely to remain in the lower lasing state at zero bias. Under zero-bias conditions, when the probe pulse enters the QCL after the arrival of the pump pulse, it is amplified instead of being absorbed.