Characterization of ErAs:GaAs and LuAs:GaAsSuperlattice Structures for Continuous-Wave TerahertzWave Generation through Plasmonic Photomixing

Shang-Hua Yang1,2& Rodolfo Salas3 & Erica M. Krivoy3 &

Hari P. Nair3 & Seth R. Bank3& Mona Jarrahi1,2

Received: 30 November 2015 /Accepted: 4 February 2016# Springer Science+Business Media New York 2016

Abstract We investigate the impact of ErAs:GaAs and LuAs:GaAs superlattice structureswith different LuAs/ErAs nanoparticle depositions and superlattice geometries on terahertzradiation properties of plasmonic photomixers operating at a 780-nm optical wavelength. Ouranalysis indicates the crucial impact of carrier drift velocity and carrier lifetime on theperformance of plasmonic photomixers. While higher carrier drift velocities enable higheroptical-to-terahertz conversion efficiencies by offering higher quantum efficiencies, shortercarrier lifetimes allow achieving higher optical-to-terahertz conversion efficiencies by mitigat-ing the negative impact of destructive terahertz radiation from slow photocarriers andpreventing the carrier screening effect.

Keywords Photomixers . Plasmonics . Terahertz

Photomixers are widely used for generating continuous-wave terahertz radiation with highspectral purity and broad frequency tunability at room temperature [1–11]. A photoconductivephotomixer is comprised of an ultrafast photoconductor integrated with a terahertz antenna ona photo-absorbing semiconductor substrate. When the photomixer is pumped by two opticalbeams with a terahertz frequency difference, photo-generated carriers in the substrate aredrifted to the photoconductor contact electrodes through an applied bias electric field. The

J Infrared Milli Terahz WavesDOI 10.1007/s10762-016-0255-z

* Mona Jarrahimjarrahi@ucla.edu

1 Electrical Engineering Department, University of California Los Angeles, 420 Westwood Plaza, LosAngeles, CA 90095, USA

2 Electrical Engineering and Computer Science Department, University of Michigan, 1301 Beal Ave,Ann Arbor, MI 48109, USA

3 Electrical and Computer Engineering Department, University of Texas at Austin, 10100 Burnet Road,Austin, TX 78758, USA

induced photocurrent is fed to the terahertz antenna to generate terahertz radiation at the offsetfrequency of the two optical beams. Substrates with high carrier drift velocities and shortcarrier lifetimes are required for efficient operation of photomixers. This is because high carrierdrift velocities offer high photoconductive gains, enabling high power terahertz generation.Moreover, short carrier lifetimes prevent destructive radiation from slow photocarriers [1] andthe carrier screening effect [12], enabling high optical-to-terahertz conversion efficiencies.Unfortunately, most techniques used for realizing short carrier lifetime substrates introducehigh defect density levels in the semiconductor lattice, which can reduce the carrier driftvelocities. Therefore, in order to choose an appropriate photo-absorbing substrate forphotomixers, it is very important to study the tradeoffs between the carrier drift velocity andcarrier lifetime and their impact on the photomixer performance at the desired radiation frequency.

In this work, we investigate the impact of GaAs substrates with epitaxially embeddednanoparticles of rare-earth arsenide (RE-As) compounds, as the short carrier lifetime photo-absorbing substrates, on the performance of plasmonic photomixers pumped at a 780-nmwavelength. By adjusting the deposition of the RE-As nanoparticles embedded inside theGaAs substrate lattice and superlattice geometry, the carrier drift velocity and carrier lifetimecan be controlled [13–16]. Photomixers with plasmonic contact electrodes are used in thisstudy due to their high quantum efficiency and high optical-to-terahertz conversion efficiency[10, 11]. This is because a significantly larger portion of the photocarriers is generated in closeproximity to the plasmonic contact electrodes [17–25]. Therefore, a larger number ofphotocarriers can be routed to the terahertz antenna in a sub-picosecond timescale for efficientterahertz generation and higher optical-to-terahertz conversion efficiencies can be achieved. Inorder to study the impact of the substrate properties on the photomixer performance, identicalplasmonic photomixers are fabricated on ErAs:GaAs and LuAs:GaAs superlattice structureswith different LuAs/ErAs nanoparticle depositions and superlattice geometries and theirperformance is compared with a plasmonic photomixer fabricated on a low temperature (LT)grown GaAs substrate [11].

ErAs:GaAs and LuAs:GaAs superlattice structures with different ErAs/LuAs nanoparticledepositions and superlattice geometries are grown by solid-source molecular beam epitaxy(MBE). Figure 1 shows the schematic diagram of an exemplary RE-As:GaAs superlatticestructure. First, a 150-nm-thick undoped GaAs buffer layer is deposited on a semi-insulatingGaAs (100) substrate at 580 °C. Next, the RE-As nanoparticles are incorporated in thesuperlattice structure. The depositions of the RE-As nanoparticles are followed by a 50-/100-Å GaAs layer growth at 530 °C with an As2/Ga beam equivalent pressure (BEP) ratio of

Fig. 1 Schematic diagram of aRE-As:GaAs superlatticestructures

J Infrared Milli Terahz Waves

15 and 3×10−6 Torr BEP of As2. This process is repeated 200/100 times to form a 1-μm-thickRE-As:GaAs superlattice structure. The depositions of the RE-As nanoparticles are monitoredby reflection high-energy electron-diffraction (RHEED) intensity oscillations and reported interms of an equivalent number of RE-As monolayers (ML). The superlattice thickness andperiodicity are measured by high-resolution X-ray diffraction (HR-XRD). Finally, a GaAs caplayer (50/100 Å) is grown on top of the RE-As:GaAs superlattice structure for preventingoxidation of the RE-As nanoparticles. By adjusting the concentration of the RE-As nanopar-ticles and the geometry of the superlattice structure, the carrier mobility and carrier lifetime iscontrolled. For comparison, a LT-GaAs substrate is grown on a semi-insulating GaAs (100)substrate at 250 °C with an As2/Ga BEP ratio of 19 followed by a 10-min post-annealingprocess at 600 °C. The carrier lifetime values of the LT-GaAs, ErAs:GaAs, and LuAs:GaAssubstrates are obtained by time-resolved differential optical pump-probe reflection measure-ments using femtosecond laser pulses centered at an 800-nm wavelength, with an 80-MHzrepetition rate (Table 1). An average pump power of 10 mW, a pump and probe power ratio of10:1, and a beam spot size of 120 μm on the sample are used for these measurements. Themeasured differential photo-reflectance data is fitted to an exponential curve, I0.exp(−t/τ),where I0 is the measured normalized differential intensity, t is the time delay of arrival betweenpump and probe pulses, and τ is the carrier lifetime. Additionally, the resistivity levels of theLT-GaAs, ErAs:GaAs, and LuAs:GaAs substrates are extracted from dark resistance measure-ments (Table 1).

Table 1 Carrier lifetime and resistivity of the LT-GaAs, ErAs:GaAs, and LuAs:GaAs substrates

Material Carrier lifetime (ps) Resistivity (KΩ.cm)

LT-GaAs 0.40 250

ErAs:GaAs (0.15ML/50 Å) × 200 0.67 24.5

ErAs:GaAs (0.1ML/50 Å) × 200 0.67 12.5

LuAs:GaAs (0.5ML/100 Å) × 100 0.80 14.7

LuAs:GaAs (0.25ML/50 Å) × 200 0.74 3

LuAs:GaAs (0.25ML/100 Å) × 100 0.65 12

Fig. 2 Schematic diagram of a plasmonic photomixer fabricated on the RE-As:GaAs substrate

J Infrared Milli Terahz Waves

Figure 2 shows the schematic diagram of the plasmonic photomixer fabricated on the LT-GaAs, ErAs:GaAs, and LuAs:GaAs substrates. Metallic gratings (5/45 nm Ti/Au height, 100 nmwidth, and 200 nm pitch) covered by a 150-nm SiO2 anti-reflection coating are used as theplasmonic contact electrodes. This offers more than 70 % power transmissivity for a TMpolarized optical pump beam at 780 nm through the plasmonic contact electrodes into thephoto-absorbing substrates. Plasmonic anode and cathode contact electrodes are designed tocover a 20×20 μm2 active area each with a 10-μm gap size between the anode and cathodecontact electrodes. This design induces a 2.2-fF capacitance between the anode and cathodecontact electrodes and allows maintaining high carrier drift velocities across the entire plasmoniccontact electrode area [24]. The plasmonic contact electrodes are integrated with a logarithmicspiral antenna, which offers a broadband radiation resistance of 70–100 Ω over 0.1–2 THzfrequency range [26, 27]. The fabricated plasmonic photomixers are centered and mounted on ahyper hemispherical silicon lens to collect and collimate the generated terahertz radiation.

Two distributed-feedback (DFB) lasers (TOPTICA #LD-0783-0080-DFB-1 and TOPTICA#LD-0785-0080-DFB-1) are used to pump the fabricated plasmonic photomixers. The offsetfrequency of the two DFB lasers is tuned in a 0.1–2-THz frequency range by accuratelycontrolling their operation temperatures and driving currents. The outputs of the two DFBlasers are coupled into a semiconductor optical amplifier (TOPTICA BOOSTA PRO 780)through a polarization-maintained fiber. An objective lens is used to asymmetrically focus theamplified optical pump onto the edge of the anode contact electrodes to maximize the inducedphotocurrent. The amplified optical beams from the two DFB lasers are balanced in power toachieve high optical-to-terahertz conversion efficiencies. A small portion of the amplifiedoptical pump is simultaneously monitored on an optical spectrum analyzer by placing a pelliclebetween the output of the semiconductor optical amplifier and the plasmonic photomixers.Finally, the terahertz radiation from the plasmonic photomixers is collimated by a polyethylenelens and detected by a calibrated silicon bolometer from Infrared Laboratories. The experi-mental setup used for characterizing the plasmonic photomixers is illustrated in Fig. 3.

Figure 4a shows the induced photocurrent for the plasmonic photomixers fabricated on theErAs:GaAs and LT-GaAs substrates at an optical pump power of 25 mW. The inducedphotocurrent is proportional to the carrier drift velocity and carrier lifetime. Therefore,assuming that the ratios between the carrier lifetimes are maintained under the 25-mWopticalexcitation and assuming the same carrier lifetime for electrons and holes, the measuredphotocurrent values indicate that higher carrier drift velocities are offered by the ErAs:GaAs(0.1ML/50 Å)×200 substrate compared to the ErAs:GaAs (0.15ML/50 Å)×200 substrate.This can be explained by a lower concentration of scattering centers in the ErAs:GaAssuperlattice structure with lower ErAs deposition. Moreover, the LT-GaAs substrate exhibits

Fig. 3 Experimental setup for characterizing the LT-GaAs plasmonic photomixers

J Infrared Milli Terahz Waves

the lowest carrier drift velocity compared to the ErAs:GaAs substrates. Figure 4b shows theradiation spectra of the plasmonic photomixers fabricated on the ErAs:GaAs and LT-GaAssubstrates at an optical pump power of 25 mW and a bias voltage of 30 V. The terahertzradiation power from the fabricated photomixers at an angular frequency of ω is given by [28]:

Fig. 4 a The induced photocurrent for the plasmonic photomixers fabricated on the LT-GaAs and ErAs:GaAssubstrates at an optical pump power of 25 mW. b The radiation power spectra of the plasmonic photomixersfabricated on the LT-GaAs and ErAs:GaAs substrates at an optical pump power of 25mWand a bias voltage of 30V.c The induced photocurrent for the plasmonic photomixers fabricated on the LT-GaAs and LuAs:GaAs substrates atan optical pump power of 25 mW. d The radiation power spectra of the plasmonic photomixers fabricated on theLT-GaAs and LuAs:GaAs substrates at an optical pump power of 25 mWand a bias voltage of 30 V>

J Infrared Milli Terahz Waves

Prad ¼ RA

2

ηehv

� �2 1

1þ ωτð Þ21

1þ ωRACð Þ2 P1P2 ð1Þ

where RA is the terahertz antenna radiation resistance, η is the photoconductor quantumefficiency, e is the electron charge, hυ is the photon energy, τ is the carrier lifetime, C is thecapacitive loading to the terahertz antenna, and P1 and P2 are the power levels of the beatingoptical beams that are set to be equal for maximum radiation efficiency. As expected, theplasmonic photomixer fabricated on the ErAs:GaAs (0.1ML/50 Å)× 200 substrate offershigher terahertz radiation power levels compared to the other plasmonic photomixers at low

Fig. 4 (continued)

J Infrared Milli Terahz Waves

terahertz frequencies due to its higher quantum efficiency as a result of a higher carrier driftvelocity. A much steeper terahertz radiation power roll-off slope is observed for theErAs:GaAs plasmonic photomixers compared to the LT-GaAs plasmonic photomixer in the0.3–1-THz frequency range. This is because the LT-GaAs plasmonic photomixer is lessimpacted by the destructive radiation from slow photocarriers in this frequency range due toits shorter carrier lifetime value (ωτ<1). The radiation power at higher frequencies (ωτ>1) isdominated by the RC roll-off with an estimated 3 dB roll-off frequency of ∼1 THz. Sinceidentical logarithmic spiral antennas and plasmonic contact electrode designs are used in thefabricated plasmonic photomixers, similar radiation power roll-off slopes are observed for allof the plasmonic photomixers above 1 THz. The dramatic radiation power drop at ∼1.1 THz isdue to water absorption. Despite the steeper radiation power roll-off of the ErAs:GaAsphotomixers, higher terahertz power levels are offered by the ErAs:GaAs substrates comparedto the LT-GaAs substrate above 1 THz. This is because the impact of the high carrier driftvelocities of the ErAs:GaAs substrates on the terahertz radiation power level is more dominantthan the impact of the short carrier lifetime of the LT-GaAs substrate above 1 THz (Eq. 1). Thenegative impact of the carrier screening effect on the photomixer efficiency can be observed bycomparing the radiation power levels from the ErAs:GaAs (0.1ML/50 Å) × 200 andErAs:GaAs (0.15ML/50 Å)×200 plasmonic photomixers. Despite its superior performancein offering higher quantum efficiencies and terahertz radiation power levels at low frequencies,the ErAs:GaAs (0.1ML/50 Å)×200 plasmonic photomixer is more severely impacted by thecarrier screening effect at higher frequencies due to its higher carrier drift velocity. Therefore, itexhibits a higher level of power degradation compared to the ErAs:GaAs (0.15ML/50 Å)×200plasmonic photomixer. The carrier screening effect is less severe for the LT-GaAs plasmonicphotomixer due to its shorter carrier lifetime.

Figure 4c shows the induced photocurrent for the plasmonic photomixers fabricated on theLuAs:GaAs and LT-GaAs substrates at an optical pump power of 25 mW. Assuming that theratios between the carrier lifetimes are maintained under the 25-mW optical excitation andassuming the same carrier lifetime for electrons and holes, the induced photocurrent levelsindicate that higher carrier drift velocities are offered by the LuAs:GaAs (0.25ML/100 Å)×100 substrate compared to other LuAs:GaAs substrates. This can be explained by alower concentration of scattering centers in the LuAs:GaAs superlattice structure with lowerLuAs deposition and thicker GaAs layer. Moreover, the LT-GaAs substrate exhibits a compa-rable carrier drift velocity with the LuAs:GaAs (0.25ML/50 Å)×200 substrate. It should benoted that the carrier drift velocity in LuAs:GaAs substrates is highly affected by the quality ofthe GaAs overgrowth of the LuAs nanoparticles. Defects in the GaAs overgrowth are more ofthe point defect variety and act as both carrier scattering centers and as ultrafast recombinationcenters. However, their density of states is much lower than those of the LuAs nanoparticles.Therefore, these point defects will have a significant effect on the carrier drift velocity, butminimal effect on the carrier lifetime. Carrier lifetime is dominated by recombination at theLuAs nanoparticles because the point defects are bleached out at high excitation densities.Figure 4d shows the radiation spectra of the plasmonic photomixers fabricated on theLuAs:GaAs and LT-GaAs substrates at an optical pump power of 25 mW and a bias voltageof 30 V. The plasmonic photomixer fabricated on the LuAs:GaAs (0.25ML/100 Å)×100substrate generates the highest terahertz radiation power at low terahertz frequencies due to itshighest quantum efficiency. As expected, the plasmonic photomixers fabricated on the LT-GaAs and LuAs:GaAs (0.25ML/50 Å)×200 substrates generate similar terahertz power levelsat low terahertz frequencies due to their similar carrier drift velocities. Similar to the

J Infrared Milli Terahz Waves

ErAs:GaAs plasmonic photomixers, the LuAs:GaAs plasmonic photomixers show a muchsteeper terahertz radiation roll-off slope compared to the LT-GaAs plasmonic photomixer inthe 0.3–1-THz frequency range. Unlike the ErAs:GaAs plasmonic photomixers, the impact ofthe high carrier drift velocities of the LuAs:GaAs substrates on the terahertz radiation powerlevels is less dominant than the impact of the short carrier lifetime of the LT-GaAs substrateabove 1 THz (Eq. 1). This is because of the lower carrier drift velocities of the LuAs:GaAssubstrates compared to the ErAs:GaAs substrates for the ErAs/LuAs nanoparticle depositionsand superlattice structures analyzed in this study. Therefore, the plasmonic photomixer fabri-cated on the LT-GaAs substrate generates higher terahertz radiation power levels above 1 THzcompared with the plasmonic photomixers fabricated on the LuAs:GaAs substrates. Similar tothe ErAs:GaAs plasmonic photomixers, the LuAs:GaAs plasmonic photomixers with highercarrier drift velocities are more severely impacted by the carrier screening effect at higherfrequencies. Therefore, similar radiation power levels are offered by the LuAs:GaAs (0.25ML/50 Å)×200 and LuAs:GaAs (0.25ML/100 Å)×100 plasmonic photomixers above 1 THz.

In summary, the impact of ErAs:GaAs and LuAs:GaAs superlattice structure properties onthe performance of plasmonic photomixers pumped at 780 nm optical wavelength is analyzed.The results suggest that higher concentrations of scattering centers are introduced in thesuperlattice structures with higher ErAs/LuAs nanoparticle depositions, which lead to lowercarrier drift velocities and quantum efficiencies. However, the measured carrier lifetimes donot show a significant dependence on the ErAs/LuAs nanoparticle depositions and superlatticegeometries analyzed in this study. As such, ErAs:GaAs and LuAs:GaAs substrates with lowErAs/LuAs depositions can offer higher optical-to-terahertz conversion efficiencies comparedto LT-GaAs substrates. The results show that the carrier drift velocity is the most importantfactor that needs to be increased to achieve higher optical-to-terahertz conversion efficienciesat low terahertz frequencies. However, the carrier lifetime becomes an equally important factorthat needs to be reduced to achieve higher optical-to-terahertz conversion efficiencies at higherfrequencies as the terahertz radiation oscillation cycles become comparable with the carrierlifetimes. Therefore, depending on the utilized ErAs/LuAs nanoparticle depositions andsuperlattice geometries, ErAs:GaAs and LuAs:GaAs plasmonic photomixers could outperformLT-GaAs plasmonic photomixers at the desired terahertz radiation frequencies.

Acknowledgments The authors gratefully acknowledge the financial support from Presidential Early CareerAward for Scientists and Engineers (# N00014-14-1-0573 and # W911NF-09-1-0434), NSF CAREER Award (#N00014-11-1-0096), and ONR Young Investigator Award (# N00014-12-1-0947).

References

1. S. Preu, G. H. Dohler, S. Malzer, L. J. Wang, and A. C. Gossard, J. Appl. Phys. 109, 061301 (2011).2. E. R. Brown, Int. J. High Speed Electron. Syst. 13, 497 (2003).3. J. Bjarnason, J. Chan, A. W. M. Lee, E. R. Brown, D. C. Driscoll, M. Hanson, A. C. Gossard, and R. E.

Muller, Appl. Phys. Lett. 85, 3983–3985 (2004).4. E. Peytavit, S. Lepilliet, F. Hindle, C. Coinon, T. Akalin, G. Ducournau, G. Mouret, and J.-F. Lampin, Appl.

Phys. Lett. 99, 223508 (2011).5. E. Peytavit, C. Coinon, and J.-F. Lampin, J. Appl. Phys. 109, 016101 (2011).6. J. Mangeney, A. Merigault, N. Zerounian, P. Crozat, K. Blary, and J. F. Lampin, Appl. Phys. Lett. 91, 241102

(2007).7. H. Ito, T. Yoshimatsu, H. Yamamoto, and T. Ishibashi, Appl. Phys. Express 6, 064101 (2013)8. H. Ito, F. Nakajima, T. Furuta, K. Yoshino, Y. Hirota and T. Ishibashi, Electron. Lett. 39, 1828 (2003)

J Infrared Milli Terahz Waves

9. F. Nakajima, T. Furuta, and H. Ito, Electron. Lett. 40, 1297–1299 (2004)10. C. W. Berry, M. R. Hashemi, S. Preu, H. Lu, A. C. Gossard, M. Jarrahi, Appl. Phys. Lett. 105, 011121

(2014).11. S.-H. Yang, M. Jarrahi, Appl. Phys. Lett. 107, 131111 (2015)12. G. C. Loata, M. D. Thomson, T. Löffler, H. G. Roskos, Appl. Phys. Lett. 91, 232506 (2007).13. M. Griebel, J. H. Smet, D. C. Driscoll, J. Kuhl, C. A. Diez, N. Freytag, C. Kadow, A. C. Gossard, K. von

Klitzing, Nature Mater. 2, 122-126 (2003).14. R. P. Prasankumar, A. Scopatz, D. J. Hilton, A. J. Taylor, R. D. Averitt, J. M. Zide, A. C. Gossard, BAppl.

Phys. Lett. 86, 201107 (2005).15. E. M. Krivoy, H. P. Nair, A. M. Crook, S. Rahimi, S. J. Maddox, R. Salas, D. A. Ferrer, V. D. Dasika, D.

Akinwande, S. R. Bank, Appl. Phys. Lett. 101, 141910 (2012).16. C. Kadow, S. B. Fleischer, J. P. Ibbetson, J. E. Bowers, A. C. Gossard, J. W. Dong, C. J. Palmstrom, Appl.

Phys. Lett. 75, 3548-3550 (1999).17. S.-G. Park, K. H. Jin, M. Yi, J. C. Ye, J. Ahn, and K.-H. Jeong, ACS Nano 6, 2026 (2012).18. S. Liu, X. Shou, and A. Nahata, IEEE Trans. Terahertz Sci. Technol. 1, 412 (2011).19. S.-G. Park, Y. Choi, Y.-J. Oh, and K.-H. Jeong, Opt. Express 20, 25530 (2012).20. M. Jarrahi, IEEE Trans. Terahertz Sci. Technol. 5, 391-397 (2015).21. N. T. Yardimci, S.-H. Yang, C. W. Berry, M. Jarrahi, IEEE Trans. Terahertz Sci. Technol. 5, 223-229 (2015).22. S.-H. Yang, M. R. Hashemi, C. W. Berry, M. Jarrahi, IEEE Trans. Terahertz Sci. Technol. 4, 575-581 (2014).23. C. W. Berry, N. Wang, M. R. Hashemi, M. Unlu, M. Jarrahi, Nature Communications 4, 1622 (2013).24. C. W. Berry and M. Jarrahi, New J. Phys. 14, 105029 (2012).25. C. W. Berry, M. R. Hashemi, M. Jarrahi, Appl. Phys. Lett. 104, 081122 (2014).26. Y. Huo, G. W. Taylor, and R. Bansal, J. Infrared, Millimeter, Terahertz Waves 23, 819–839 (2002).27. E. R. Brown, A. W. M. Lee, B. S. Navi, and J. E. Bjarnason, Microwave Opt. Technol. Lett. 48, 524–529

(2006).28. C. W. Berry, M. Jarrahi, J. Infrared, Millimeter and Terahertz Waves 33, 1182-1189 (2012).

J Infrared Milli Terahz Waves