1. School of something FACULTY OF OTHER Generation of continuous wave terahertz radiation from Fe-doped InGaAs and InGaAsP Reshma A. Mohandasᵃ, Joshua R. Freemanᵃ, Mark C. Rosamondᵃ, Osama Hatemᵃ Siddhant Chowdhuryᵃ, Lalitha Ponnampalamᵇ, Martyn Ficeᵇ, Alwyn J. Seedsᵇ, Paul. J. Cannardᶜ, Michael. J. Robertsonᶜ, David. G. Moodieᶜ, A. Giles Daviesᵃ, Edmund H. Linfieldᵃ and Paul Deanᵃ a) Institute of Microwaves & Photonics, University of Leeds, Leeds, LS2 9JT, United Kingdom b) University College London, London, WC1E 6BT, United Kingdom c) CIP Technologies, Adastral Park, Martlesham Heath, Ipswich, Suffolk, IP5 3RE, United Kingdom email: J.R.Freeman@leeds.ac.uk elram@leeds.ac.uk

2. Outline Theory of Photomixing Why Fe doped InGaAs and InGaAsP? Experimental set-up Results Summary

3. Theory: Photomixers

4. Modulation in conductance results in THz generation THz coupled out using an antenna collimated by Si lens For an ideal Photomixer  Very low carrier lifetime  High resistivity

5. Low Temperature Gallium Arsenide (LT-GaAs) is the most commonly used material  Bandgap corresponding to ~ 800 nm,  High resistivity, short carrier life-time  Lasers are quite expensive Why InGaAs and InGaAsP?

6. Low Temperature Gallium Arsenide (LT-GaAs) is the most commonly used material  Bandgap corresponding to ~ 800 nm,  High resistivity, short carrier life-time  Lasers are quite expensive Indium Gallium Arsenide (InGaAs)  Bandgap corresponding to ~ 1550 nm  1550 nm – Telecommunications wavelength regime  Well developed optical amplifiers and components  Poor resistivity Indium Gallium Arsenide Phosphide (InGaAsP)  Band gap corresponding to ~ 1300 nm (lattice matched for InP)  Poor resistivity Why InGaAs and InGaAsP?

7. Methods adopted to increase resistivity in semiconductors  Low temperature growth  Annealing  Ion implantation  Ion irradiation  Ion doping¹ 1. C.D.Wood et al, APL 2011 Why InGaAs and InGaAsP?

8. Fe ions doped into InGaAs and InGaAsP using Metal Organic Chemical Vapor Deposition (MOCVD)  Fe ions incorporated into active layer  Creates deep acceptors  Better surface quality and uniform doping In, Ga, As and P ratio adjusted, bandgap shifted from 1300 nm to 1550 nm M. Silver et al, IEEE, 1995 Fe: In ₀. ₇₀ Ga ₀. ₃₀ As ₀. ₈₇ P ₀. ₁₃

9. n-InP and n-InGaAs/n-InGaAsP cap layers protect the surface from oxidation n-InP n-InGaAs Fe: In₀.₆₂Ga₀.₃₈As Fe-InP InP Substrate Structure of Fe doped InGaAs and Fe doped InGaAsP n-InP n-InGaAsP Fe: In₀.₇₀Ga₀.₃₀As₀.₈₇P₀.₁₃ Fe-InP InP Substrate Processing Method Concentrated hydrochloric acid removes the n-InP layer Con HCL

10. 3-turn self complementary logarithmic spiral antenna was chosen Active region was 11.3 µm × 11.3 µm 3 pair of fingers with finger size ~ 0.2 µm and gap size ~ 1.6 µm e-beam lithography was used to define the pattern, Ti/Au of 10 nm/150 nm thickness used for metallization Processing Method Fe doped InGaAs 10 µ m Fe doped InGaAsP

11. Two InGaAs and three InGaAsP wafers were chosen Fe doping concentration (× 10¹⁶cm¯³) 1.0 9.5 10.0 Fe doping concentration (× 10¹⁶cm¯³) 0.5 5.5 Fe doped InGaAsFe doped InGaAsP Experimental set-up

13. Results: Bandwidth InGaAs Bandwidth from the two InGaAs wafer Amplitude and bandwidth low in higher doping Charge carriers trapped by acceptors

14. Bandwidth from the three InGaAsP wafer Amplitude and bandwidth low for higher doped Results: Bandwidth InGaAsP

15. Low doped InGaAs and InGaAsP wafer Water absorption lines plotted Results: Bandwidth InGaAs & InGaAsP

16. Results: Photocurrent Decrease in photocurrent as doping increased Due to the trapping of photocarriers by acceptors

17. Quadratic dependence with the bias Decrease in signal amplitude as doping increased Results: Amplitude

18. InGaAs Fe doping concentration (× 10¹⁶cm¯³) Bandwidth (THz) 0.5>2.4 5.52.0 InGaAsP Fe doping concentration (× 10¹⁶cm¯³) Bandwidth (THz) 1.0>2.4 9.52.0 10.0<2.0 Summary Bandwidth from InGaAs Bandwidth from InGaAsP

19. CW terahertz generated from Fe:In₀.₆₂Ga₀.₃₈As and Fe:In₀.₇₀Ga₀.₃₀As₀.₈₇P₀.₁₃ Fe doped on InGaAs and InGaAsP to increase resistivity Bandwidth was >2.4 THz from low doped InGaAs and InGaAsP Signal amplitude decreased as doping increased in InGaAs and InGaAsP An optimum doping level required for better bandwidth and power Summary

20. School of something FACULTY OF OTHER Progress on CW locking of THz QCLs Joshua Freeman Reshma Mohandas Siddhant Chowdhury Yingjun Han Edmund Linfield Giles Davis School of Electrical Engineering FACULTY OF ENGINEERING

21. Outline  NIR sideband generation in the QCL waveguide  THz injection into QCL  Coherent detcetion  FTIR/bolometer results

22. Sideband QCL locking scheme Mixing takes place in the QCL cavity 2 nd order non-linearity of the GaAs f NIR f NIR - f THz f NIR + f THz

23. Sideband QCL locking scheme NIR lasers tuned to f Laser1 – f Laser2 = f THz

24. QCL with NIR waveguide Metal-metal QCL waveguide Wet-etched ridge: 55 um & 60 um wide Device lengths around 2 mm Wafer grown at Cambridge (V426)

25. Intra-cavity sidebands at Leeds Resonance around 1550nm Free-space coupling with microscope objectives

26. Sideband efficiency 60 um wide ridge55 um wide ridge Sidebands measured over a wide range of wavelengths Narrow ridge expected to exhibit peak at longer wavelength Maximum efficiency observed: -48dB

27. Sideband summary  Sidebands on 1550 nm measured at Leeds  Maximum efficiency: -48dB  Strategies for improving signal strength:  Vertical sidewalls (ICP currently being commissioned)  Longer ridges (initial results not encouraging)  AR coating for 1550nm  Improved active regions

28. CW THz injection locking Results from March: L1: DBR L2: DBR

29. CW THz injection locking L1: RIO fixed wavelength L2: SANTEC tunable TSL-710 Optical power ~30mW on Tx/Rx Toptica InGaAs emitter and receiver QCL: Single plasmon waveguide, 2THz, CW, 750mA THz path length is controllable.

30. Laser joint linewidth New laser diode pair: ~20MHz drift Previous laser diode pair: ~400MHz drift/jitter

31. Results: Coherent detection QCL is in CW THz emitter voltage is modulated Injected frequency is scanned Receiver current is measured on a lock-in THz delay is not changed so the signal can be negative. Only phase-locked signals detected.

32. Results: Coherent detection  Frequency is scanned again. The phase is also measured.  The phase-change across the peak shows a shift of π

33. Results: Coherent detection This is repeated for different QCL modes

34. Is the QCL locked?

35. Voltage change of QCL There is a significant change in voltage when the resonant frequency is reached (repeat scans are shown):

36. QCL current The signal disappears below threshold (the QCL threshold is about 725mA) The QCL mode red-shifts with increasing current (this is normal)

37. FTIR characterisation Same conditions for QCL and emitter as before. The detection is now sensitive to all frequencies (not just those which are phase-locked)

38. Spectrum of QCL Spectrum of QCL with no injection 750mA, 15K, CW. Single mode at 2014 GHz

39. Evidence of mode-pulling  The injected THz frequency is tuned across the QCL mode at 1997 GHz  The emission is ‘pulled’ to that mode

40. Stability  Mode pulling not always observed owing to alignment difficulties.  Alignment is critical.  Without coherent detection: no way to know if the injection is aligned  Mode hopping observed during FTIR scans

41. Summary: THz injection locking  Evidence of THz injection locking observed  Joint linewidth of laser diodes is critical  Calculations based on power ratio suggest QCL is locked  Locking range not current measureable (expected to be around 10MHz)  Further measurements planned