Pulsed High Duty Cycle Operation of λ ~ 8.2μm Quantum Cascade Lasers Tiffany Ko, Zhijun Liu, Claire F. Gmachl Department of Electrical Engineering & Princeton Institute for the Science and Technology of Materials (PRISM) Princeton University, Princeton, NJ 08544
Motivation
•
QEPAS (Quartz-enhanced photo acoustic spectroscopy)1 – Quartz tuning fork (TF) is immersed in gas containing trace chemical analyte. – Light produced by a QC laser, pulsed operated at the resonant frequency (~30kHz) of the TF, is focused between the TF prongs. – Pressure waves caused by the light absorption of the analyte drive the TF generating piezoelectric signal. – Sensitivity level, currently in the parts per billion, is dependent on the average power output of the driving laser source. – Advantages to conventional approaches of PAS : • immunity to ambient acoustic noise Tuning fork next to a mm-graduated ruler.1 • able to analyze extremely small samples (~1mm3) 1A.A.
Quantum Cascade Laser
Optimization parameters for QEPAS-specific QC laser operation:
Quantum cascade (QC) laser emit in the mid-infrared “fingerprint” fingerprint region, making them excellent light sources for trace gas detection systems.
•
¾ Average output power, power efficiency, threshold current at ~30kHz
• • •
Trade-offs •
Pulsed vs. Burst Mode Operation
(
vs.
)
– Pulsed Mode • •
Deviations from the 50% duty cycle cause uneven driving and relaxation times Low frequencies, long period of heating similar to CW operation
– Burst Mode •
•
Allows laser cooling during driving period due to the analyte’s much slower relaxation rate
High average power output vs. high absolute power efficiency – Increased duty cycle Æ increased average power output •
Increased QEPAS detection sensitivity
– Increased duty cycle Æ increased dissipation of electrical power Æ temperature increase Æ laser performance degradation Æ decreased power efficiency •
•
Less efficient use of limited energy resources in a confined system
Operating Frequency – Higher frequencies Æ less heating time Æ cooler active core Æ better performance – Higher frequencies Æ less cooling time Æ more heat retention Æ performance degradation
- M. Weimer, Texas A&M University
1stage = 600Å - Liu et al. IEEE Photon. Technol. Lett., vol. 18 (12), pp. 1347-1349, June 2006.
Kosterev, F.K. Tittel, D. Serebryakov, A. Malinovsky, I. Morozov,Rev. Sci. Instrum. 76, 043 105 (2005)
Results
Experimental p Setup p •
Semiconductor injection laser which utilizes intersubband transitions in the conduction band • 4-well active region Emission wavelength: ~8.2μm • InAlAs/InGaAs/InP 35 stage, “two-phonon resonance” design • Ridge length ~3.4mm CW-RT capable • Ridge width ~ 13.4μm
QC lasers operated in high duty-cycle pulsed mode and in burst mode – Burst mode: 30kHz, 50% duty cycle (d.c.) macro-pulses and variable sub-pulse duration, frequency, and duty-cycle. – Burst-mode operation is aimed at maximizing average output power and efficiency while retaining the 30 kHz and ≤ 50% duty-cycle repetition rate requirement of QEPAS.
• • •
QC lasers c-mounted and on a thermoelectric (TE) cooler held at 16.2oC (~290K) All equipment was GPIB-interfaced with a computer and LabVIEW controlled. Driving signal in pulsed and burst mode: – Agilent high power pulsers (HP8114a)
Measurements: • • •
LabVIEW controlled Gated boxcar averaging of I, V signals Averaging optical power meter
Conclusions • • • •
Peak maximum output power occurs at ~55% duty cycle under pulsed mode operation. Threshold current density increases with increasing duty cycle, but reaches a maximum at ~70% duty cycle during burst mode operation. Burst mode operation has a higher normalized slope efficiency that pulsed mode. Normalized slope efficiency decreases with increasing duty cycle.
Future Directions •
Different driving signal modulation techniques – Sine-square wave – Pulse width modulation
•
Further optimization of parameters
•
QC lasers with different emission wavelengths
– Driving signal macropulse frequency and duty cycle – Device structure, size
Acknowledgements Many thanks to the members of the Mid-Infrared Photonics Group at Princeton University and to Gerard Wysocki and Frank K. Tittel at Rice University for guidance, support, and lively discussion. We would also like to acknowledge partial financial support by MIRTHE (NSF-ERC), DARPA-LPAS, and the Norman D. Kurtz ’58 Fund for Innovation in Engineering Education.
•1