Terahertz ambipolar dual-wavelength quantum cascade laser

Terahertz ambipolar dual-wavelength quantum cascade laser L. Lever,∗ N. M. Hinchcliffe, S. P. Khanna, P. Dean, Z. Ikoni´c, C. A. Evans, A. G. Davies, P. Harrison, E. H. Linfield and R. W. Kelsall Institute of Microwaves and Photonics, School of Electronic and Electrical Engineering, University of Leeds, Leeds LS2 9JT, United Kingdom [email protected]

Abstract: Terahertz frequency quantum cascade lasers (THz QCLs) are compact solid-state sources of terahertz radiation that were first demonstrated in 2002. They have a broad range of potential applications ranging from gas sensing and non-destructive testing, through to security and medical imaging, with many polycrystalline compounds having distinct fingerprint spectra in the terahertz frequency range. In this article, we demonstrate an electrically-switchable dual-wavelength THz QCL which will enable spectroscopic information to be obtained within a THz QCLbased imaging system. The device uses the same active region for both emission wavelengths: in forward bias, the laser emits at 2.3 THz; in reverse bias, it emits at 4 THz. The corresponding threshold current densities are 490 A/cm2 and 330 A/cm2 , respectively, with maximum operating temperatures of 98 K and 120 K. © 2009 Optical Society of America OCIS codes: 140.0140 Lasers and laser optics; 140.3070 Infrared and far-infrared lasers; 140.5965 Semiconductor lasers, quantum cascade

References and links 1. M. Tonouchi, “Cutting-edge terahertz technology,” Nat. Photonics 1(2), 97–105 (2007). 2. R. K¨ohler, A. Tredicucci, F. Beltram, H. E. Beere, E. H. Linfield, A. G. Davies, D. A. Ritchie, R. C. Iotti, and F. Rossi, “Terahertz semiconductor-heterostructure laser,” Nature 417(6885), 156–159 (2002). 3. S. Kumar, Q. Hu, and J. L. Reno, “186 K operation of terahertz quantum-cascade lasers based on a diagonal design,” Appl. Phys. Lett. 94(13), 131105 (2009). 4. B. S. Williams, “THz quantum cascade lasers,” Nat. Photonics 1, 517–525 (2007). 5. G. Scalari, C. Walther, J. Faist, H. Beere, and D. Ritchie, “Electrically switchable, two-color quantum cascade laser emitting at 1.39 and 2.3 THz,” Appl. Phys. Lett. 88(14), 141,102 (2006). 6. J. R. Freeman, O. P. Marshall, H. E. Beere, and D. A. Ritchie, “Electrically switchable emission in terahertz quantum cascade lasers,” Opt. Express 16(24), 19,830–19,835 (2008). 7. C. Gmachl, A. Tredicucci, D. L. Sivco, A. L. Hutchinson, F. Capasso, and A. Y. Cho, “Bidirectional Semiconductor Laser,” Science 286(5440), 749–752 (1999). 8. B. S. Williams, H. Callebaut, S. Kumar, Q. Hu, and J. L. Reno, “3.4-THz quantum cascade laser based on longitudinal-optical-phonon scattering for depopulation,” Appl. Phys. Lett. 82(7), 1015–1017 (2003). 9. M. A. Stroscio, M. Kisin, G. Belenky, and S. Luryi, “Phonon enhanced inverse population in asymmetric double quantum wells,” Appl. Phys. Lett. 75(21), 3258–3260 (1999). 10. P. Harrison, Quantum Wells, Wires and Dots: Theoretical and Computational Physics of Semiconductor Nanostructures, 2nd ed. (Wiley, Chichester, 2005). 11. Z. Ikonic, P. Harrison, and R. W. Kelsall, “Self-consistent energy balance simulations of hole dynamics in SiGe/Si THz quantum cascade structures,” J. Appl. Phys. 96(11), 6803–6811 (2004). 12. T. Unuma, M. Yoshita, T. Noda, H. Sakaki, and H. Akiyama, “Intersubband absorption linewidth in GaAs quantum wells due to scattering by interface roughness, phonons, alloy disorder, and impurities,” J. Appl. Phys. 93(3), 1586–1597 (2003).

#115649 - $15.00 USD

(C) 2009 OSA

Received 12 Aug 2009; revised 23 Sep 2009; accepted 6 Oct 2009; published 19 Oct 2009

26 October 2009 / Vol. 17, No. 22 / OPTICS EXPRESS 19926

13. P. Harrison, D. Indjin, and R. W. Kelsall, “Electron temperature and mechanisms of hot carrier generation in quantum cascade lasers,” J. Appl. Phys. 92(11), 6921–6923 (2002). 14. V. D. Jovanovic, S. Hofling, D. Indjin, N. Vukmirovic, Z. Ikonic, P. Harrison, J. P. Reithmaier, and A. Forchel, “Influence of doping density on electron dynamics in GaAs/AlGaAs quantum cascade lasers,” J. Appl. Phys. 99(10), 103106 (2006). 15. A. Wittmann, Y. Bonetti, J. Faist, E. Gini, and M. Giovannini, “Intersubband linewidths in quantum cascade laser designs,” Appl. Phys. Lett. 93(14), 141,103 (2008). 16. S. Kumar, B. Williams, Q. Hu, and J. Reno, “1.9 THz quantum-cascade lasers with one-well injector,” Appl. Phys. Lett. 88, 121,123 (2006). 17. B. S. Williams, S. Kumar, H. Callebaut, H. Qing, and J. L. Reno, “Terahertz quantum-cascade laser at λ ≈100 µ m using metal waveguide for mode confinement,” Appl. Phys. Lett. 83(11), 2124–6 (2003). 18. M. M. V. Taklo, P. Storas, K. Schjolberg-Henriksen, H. K. Hasting, and H. Jakobsen, “Strong, high-yield and lowtemperature thermocompression silicon wafer-level bonding with gold,” J. Micromechanics Microeng. 14(7), 884–90 (2004). 19. C. Walther, G. Scalari, J. Faist, H. Beere, and D. Ritchie, “Low frequency terahertz quantum cascade laser operating from 1.6 to 1.8 THz,” Appl. Phys. Lett. 89(23), 231121 (2006). 20. S. Kumar, B. S. Williams, Q. Hu, and J. L. Reno, “1.9 THz quantum-cascade lasers with one-well injector,” Appl. Phys. Lett. 88(12), 121–123 (2006). 21. S. Kumar, B. S. Williams, S. Kohen, Q. Hu, and J. L. Reno, “Continuous-wave operation of terahertz quantumcascade lasers above liquid-nitrogen temperature,” Appl. Phys. Lett. 84(14), 2494–2496 (2004). 22. H. Luo, S. R. Laframboise, Z. R. Wasilewski, G. C. Aers, H. C. Liu, and J. C. Cao, “Terahertz quantum-cascade lasers based on a three-well active module,” Appl. Phys. Lett. 90, 112–141 (2007). 23. M. A. Belkin, J. A. Fan, S. Hormoz, F. Capasso, S. P. Khanna, M. Lachab, A. G. Davies, and E. H. Linfield, “Terahertz quantum cascade lasers with copper metal-metal waveguides operating up to 178 K,” Opt. Express 16(5), 3242–3248 (2008). 24. S. Kohen, B. S. Williams, and Q. Hu, “Electromagnetic modeling of terahertz quantum cascade laser waveguides and resonators,” J. Appl. Phys. 97(5), 053106 (2005). 25. E. Anemogiannis, E. N. Glytsis, and T. K. Gaylord, “Determination of Guided and Leaky Modes in Lossless and Lossy Planar Multilayer Optical Waveguides: Reflection Pole Method and Wavevector Density Method,” J. Lightwave Technol. 17(5), 929 (1999).

1.

Introduction

The terahertz frequency range of the electromagnetic spectrum extends from ∼ 1 − 10 THz, and is one of the least developed despite the range of potential applications, including gas sensing, astronomy, medical and security imaging, telecommunications, and non-destructive testing [1]. This is primarily because of the lack of affordable and compact sources: the power generated by electrical sources, such as transistors and Gunn diodes, rolls off considerably at high frequencies, whilst the electronic bandgaps in semiconductors are not small enough to generate light in the terahertz range. Quantum cascade lasers (QCLs) are intersubband devices, in which the initial and final radiative states both lie in the same band. As such, they can be used as sources of electromagnetic radiation where the photon energy is considerably smaller than the bandgap. Since its realization in 2002 [2], the THz QCL has seen considerable development, with devices having been reported operating at temperatures up to 186 K [3], and from 0.84–5.0 THz (see reference [4] for a detailed review). Such devices emit almost exclusively either at a single frequency, or over a small range of Fabry-Perot modes, with only a limited tunability (within typically ∼ 100 GHz) achieved by varying the device bias. For spectroscopic applications, however, there remains a clear need for the development of electrically-tunable THz QCLs that operate over a broad wavelength range. Electrically-switchable dual-wavelength operation has been achieved in THz QCLs previously using bias-controlled selective injection into a large well, but this required the use of an external magnetic field [5]. In an alternative approach, two THz QCLs were grown monolithically, one on top of the other, to form a single ridge waveguide [6]. However, this inevitably reduces the performance, as the modal overlap with the gain medium is reduced. In the midinfrared frequency range an electrically switchable two-color design has been realized in which

#115649 - $15.00 USD

(C) 2009 OSA



100

100 5

60

Energy (meV)

Energy (meV)

5 80 4 1’

3 2

40

80

4 1’

60

3 2

40

1

20 40

50

60

70 80 Growth axis (nm)

(a)

1

20

90

100

40

50

70 80 60 Growth axis (nm)

90

100

(b)

Fig. 1. Conduction band diagrams under (a) forward bias of 8.8 kV/cm and (b) reverse bias of 10.0 kV/cm, with the squared magnitude of the wavefunctions illustrated. The radiative transition is from state 4 into state 3 under both biases. The epitaxial layer thicknesses, in nm, are: 4.6, 17.0, 4.4, 7.5, 2.3, 8.0, 1.7 & 7.0, where the Al0.15 Ga0.85 As barriers are shown in bold font, the GaAs wells are in plain font, and the central 5.6 nm of the underlined well is doped at 5 × 1016 cm−3 to provide a sheet doping density of 2.8 × 1010 cm−2 .

both positive and negative applied biases result in emission at different frequencies [7]. Critically, this device employed a highly diagonal transition, but until very recently [3], terahertz devices have not been reported with such diagonal transitions, rather only devices that exploit much larger optical matrix elements. Here, we present the modeling and experimental realization of a two-color THz QCL design that employs an almost vertical transition, and is switchable by reversing the bias across the device. It uses the same active region for both emission frequencies and requires no external magnetic field. 2.

Description of the device

Our device is based on a resonant-phonon depopulation scheme [8], in which the lower radiative state is separated from the injector of the next period by approximately the energy of a longitudinal-optical (LO)-phonon (36 meV in GaAs). The polar LO-phonons give rise to fast electron scattering [9], and carriers are extracted from the lower radiative state by LO-phonon emission. The active region, shown in Fig. 1, consists of four wells, of which the three narrower wells are optically active and the widest one is used to depopulate the lower radiative state. The two barriers separating the three optically active wells have different thicknesses, and this results in two different anti-crossing energies. There are therefore two intersubband transitions which have significant optical matrix elements, and can be exploited to generate THz radiation. Under what we have termed forward bias, the bandstructure is engineered so that the radiative transition exploits the smaller of the two active region anti-crossing energies, as shown in Fig. 1a. Reversing the bias allows the radiative transition to exploit the larger of the two anti-crossing energies, as shown in Fig. 1b. Although the bandstructure plots shown in Fig. 1 correspond to opposite polarity biases, we may still label the wavefunctions according to a common theme. The radiative transition is between states 4 and 3. State 5 is at a higher energy than the injector state of the the previous period (state 1’), and as such it will have a very small population. However, there is still a large matrix element between this and state 4, and this leads to an absorbing transition. It is important that the system is engineered so that the the difference in energy between the 4 → 3 and 4 → 5 transitions is sufficient to avoid re-absorption of the emitted radiation under both forward and

#115649 - $15.00 USD

(C) 2009 OSA



reverse bias. 3.

Simulation results

Our device was designed by simulating the performance of many candidate structures, over a range of fields, under both forward and reverse bias using a semi-classical rate equation approach [10, 11]. Scattering rates for electrons were calculated for the following processes: LOphonon [10], acoustic phonon [10], electron–electron [10], alloy disorder [12], ionized impurity [12], and interface roughness [12]. State populations were calculated self-consistently using an energy balance method to find the electron temperatures [13]. Optical absorption and gain were then calculated according to the method in reference [14]. The lineshapes were assumed to be Lorentzian, with linewidths, Γi, j , for optical transitions from state i to j calculated according to µ ¶ 1 1 + (1) Γi, j = h¯ τi τ j where τi is the lifetime of state i, and τ j the lifetime of state j. The lifetime of a given state was calculated as the inverse of the summation of the scattering rates from that state to all other states. The calculated full width at half maximum (FWHM) linewidths at 25 K were 1.7 meV and 2.2 meV for the radiative transitions under forward and reverse biases, respectively. This is in good agreement with the value of 2 meV reported in reference [2].1 The emission and absorption frequencies in the three optically active wells are controlled by the anti-crossing energies of the three states belonging to those wells, and hence by the thicknesses of the two active region barriers. In principle, one might expect that it is possible to have the forward and reverse bias emission frequencies as far or closely spaced in frequency as desired. However, one must consider that there is only a finite range over which THz QCLs have been demonstrated: the lowest frequency resonant phonon design reported emits at 1.9 THz [16], and the highest has not exceeded 5 THz. At frequencies below 2.5 THz, device performance becomes limited by free carrier losses (which increase as λ 2 ) and by the difficulty of achieving selective injection [4]. At high frequencies, waveguide losses increase considerably as one approaches the Reststrahlen band, and additionally a non-radiative pathway via LOphonon emission opens up from the upper radiative state into the lower one. It is also important to avoid re-absorption of the emitted radiation, particularly owing to the transition 4 → 5. This effectively places a limit on how closely spaced the two emission frequencies can be. Under forward bias, the photon energy is considerably less than the energy of the 4 → 5 absorption peak, as shown in Fig. 2a. Under reverse bias however, it can be seen from the wavefunctions in Fig. 1b that the photon energy is much closer to the energy gap between states 4 and 5. It is therefore important to ensure that these two transitions are sufficiently separated in energy that re-absorption of the emitted radiation does not occur. At the design (reverse) bias, E5 − E4 ≈ 13 meV (3.2 THz), and the photon energy is ∼ 17 meV (4 THz). These energies are sufficiently far apart to avoid re-absorption, given the linewidths calculated from equation 1, and the gain and absorption peaks can be clearly resolved, as shown in Fig. 2b. 4.

Fabrication

The following laser structure was grown on an undoped GaAs substrate using molecular beam epitaxy (MBE): a 250-nm undoped GaAs buffer layer, a 300-nm undoped Al0.50 Ga0.50 As etch1 It should be noted that we have neglected inhomogeneous linewidth broadening owing to fluctuations of the quantum well dimensions on length scales that are large compared to the in-plane electron mean free path. This is consistent with the approach taken in reference [15], in which experimental values for linewidth broadening were reproduced considering only lifetime broadening of the intersubband transitions.

#115649 - $15.00 USD

(C) 2009 OSA



200

0

0

-1

Gain (cm )

-1

Gain (cm )

200

-200

-400

-200

-400 1

2

3 Frequency (THz)

(a)

4

5

1

2

3 Frequency (THz)

4

5

(b)

Fig. 2. Simulated gain (or negative absorption) at 25 K. (a) Under a forward bias of 8.8 kV/cm, the gain peak at 2.5 THz from the radiative transition 4 → 3 is far from the absorbing transition 4 → 5. (b) Under a reverse bias of 10.0 kV/cm, the gain peak is at 4.0 THz, and the 4 → 5 absorption peak is at 3.2 THz. These peaks are sufficiently separated to be clearly resolved, and hence avoid the emitted radiation being re-absorbed.

stop layer, a 75-nm n+ (5×1018 cm−3 ) GaAs contact layer, 190 periods of the active region (giving 10 µ m of gain medium), and a second n+ (5×1018 cm−3 ) 50-nm GaAs contact layer. The devices were processed into a gold-gold double-metal waveguide configuration using a procedure similar to that described in references [17] and [18]. First, a quarter of the wafer, along with a quarter of a similarly sized and shaped n+ single-side polished GaAs wafer, were coated with titanium/gold (30 nm/600 nm) using an electron-beam evaporator. The two quarter wafers were then bonded together using thermocompression bonding at 300 ◦ C and 10 MPa, before being cleaved into individual 5 × 6 mm pieces. The undoped GaAs substrate of the epitaxially grown wafer was then fully removed using a sequence of lapping and selective etching in a solution of ammonia and hydrogen peroxide (19:1 by volume), with the etch stop layer being removed using concentrated (50%) hydrofluoric acid. Photolithography was used to define the contacts, with the top non-alloyed titanium/gold contacts (20 nm/200 nm) deposited prior to etching the laser cavities. These top contacts were protected with photoresist during wet-etching in a solution of sulphuric acid, hydrogen peroxide, and water (1:8:1 by volume), and the structures were etched to within a few tens of nanometers of the lower metal layer. The laser cavities were finally cleaved into devices, indium mounted at 180 ◦ C onto gold-coated copper blocks, and gold wire bonded. Devices were mounted onto the cold finger of a continuous-flow helium cryostat, and electrically and optically characterized using a 2.5-µ s pulse width at a repetition rate of 10 kHz. For measurement of the output power of the devices, the terahertz radiation was collected using an f /0.8 parabolic mirror and coupled into a helium-cooled silicon bolometer (Thomas Keating 1840) using a second f /1.0 parabolic mirror. For spectral analysis, the radiation was collimated using three f /1.0 parabolic mirrors, and coupled into a Bruker 1FS66-V FTIR spectrometer equipped with a helium-cooled silicon bolometer. In each case, the driving pulses were modulated at a frequency of 168 Hz to allow lock-in detection, improving the sensitivity. The measurements were performed in an atmosphere of dry nitrogen, in order to minimize absorption by water vapor.

#115649 - $15.00 USD

(C) 2009 OSA



C

Current (A)

urrent (A)

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

0.0

1.6

20

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

2.0

2.2

24

1.6

4.2 K

0.16

22 20

0.14

0.10 10 0.08 8 0.06

6

0.04

4

18 Voltage (V)

0.12

12

Optical Output

Voltage (V)

85 K

(mW)

70 K

1.2

16 1.0

14

O p

55 K

14

12

0.8 4.2 K

10

25 K

8

55 K 70 K

6

0.02

0

0.00

0.4

85 K

4 2

0.6

40 K

Op

16

100 K

0.2

113 K 10

2

0

100

200

300

400

C

500

600

0

0.0 0

100

200

300

400

2

urrent Density (A/cm )

(a)

ut (mW)

1.4

40 K

ut

25 K

tical

18

500

600

700

2

Current Density (A/cm )

(b)

Fig. 3. Current–voltage characteristics and optical power for a range of heatsink temperatures, under (a) forward bias, and (b) reverse bias. The device was 1.66 mm long and 175 µ m wide.

5.

Results and discussion

Figure 3 shows the measured current–voltage characteristics and the optical power as a function of current under both forward and reverse bias. The simulated alignment fields under forward and reverse bias are 8.8 kV/cm and 10.0 kV/cm, respectively. Given that the active region of the device is 10-µ m thick, we would expect the device to operate at 8.8 V and 10.0 V, assuming this voltage is uniformly dropped across the active region. However, the devices use non-alloyed contacts, and therefore exhibit Schottky barriers, as has been reported previously for THz QCLs [19]. This can be seen in the current–voltage curves, which show relatively high operating voltages compared to those predicted theoretically, and a voltage offset at low current density. A feature can be seen in the current–voltage curves at applied voltages of around 12 V and 17 V for forward and reverse bias, respectively. Under reverse bias, there is a sharp increase in the current density at 16 V, corresponding to the opening of a parasitic current channel [20], where 36 meV is dropped per period and the injector of one period becomes aligned in energy with the excited state of the large well of the next period down-field (i.e., state 2 in Fig. 1). This parasitic channel closes at around 18 V, and very little additional current flows until an applied bias of around 21 V is reached, when the injector state comes into alignment with the upper radiative state. This parasitic feature is more pronounced under reverse bias than forward bias, owing to the larger photon energy, which gives a larger change in the applied bias between the parasitic and operating fields. The emission spectra are shown in Fig. 4; we observe laser action at ∼ 2.3 THz under forward bias and ∼ 4.0 THz under reverse bias. Under forward bias there is a small discrepancy between the frequency of the simulated peak gain and the measured emission spectrum. This may be explained either by considering that the simulated gain peak in Fig. 2a is quite broad, and a cavity-mode is selected at 2.3 THz, or by considering that there is some uncertainty associated with the precise thickness of the layers in the MBE growth of the 10-µ m-thick active region. The threshold current density at 4.2 K was 490 A/cm2 under forward bias and and 330 A/cm2 under reverse bias. This is broadly similar to (in fact, generally smaller than) what has been observed in similar devices [21, 22, 23], indicating that no significant loss of performance is associated with the dual-frequency emission. The maximum operating temperatures were 98 K under forward bias and 120 K under reverse bias (obtained using slightly shorter pulse widths

#115649 - $15.00 USD

(C) 2009 OSA



Frequency (THz)

Frequency 2.15

2.20

2.25

2.30

2.35

3.90

16.8 V

(a . u )

(a.u)

Optical Output

16.2 V

u

u

O tp t

3.95

4.00

4.05

4.10

4.15

23.6 V

cal

Opti

(THz)

2.40

16.0 V

22.8 V

22.4 V

22.2 V

22.0 V

21.8 V

15.8 V 21.4 V

71

72

73

74

75

76

77 -1

Wavenumber (cm )

(a)

78

79

80

81

130

131

132

133

134

135

136

137

138

139

140

-1

Wavenumber (cm )

(b)

Fig. 4. Emission spectra at 25 K. (a) Forward bias emission centered at 2.3 THz, and (b) reverse bias emission centered at 4.0 THz. Very little Stark shift is observed in the emission frequencies, which is characteristic of vertical intersubband transitions.

than were used during the other measurements; 0.7 µ s and 0.2 µ s, respectively). Furthermore, as well as a higher operating temperature, the reverse bias also gives a larger dynamic range and a higher output power. There are several reasons why we can expect improved performance in reverse bias (i.e., at shorter wavelengths). The facet reflectivity of metal-metal waveguides increases at longer wavelengths, and therefore a smaller fraction of the terahertz radiation will be emitted at 2.3 THz [24]. As both emission frequencies are subject to identical doping owing to the common active region, and given that free carrier loss scales as λ 2 , it is expected that the losses in the active region will be higher at 2.3 THz than at 4.0 THz. Assuming bulk material behavior, we find waveguide losses of 46 cm−1 and 26 cm−1 at the two frequencies by determining the complex propagation constants of the waveguide according to the method in reference [25], using complex refractive indices of the different layers from the Drude model. Additionally, selective injection becomes more challenging as the upper and lower radiative levels become closer in energy, resulting in a smaller population inversion under forward bias, in comparison to that expected under reverse bias. 6.

Conclusion

A two-color electrically-switchable THz QCL operating at 2.3 THz and 4.0 THz has been demonstrated. The device functions under both positive and negative biases, and uses the same active region for both emission frequencies. This has significant advantages over previous twocolor THz QCLs: first, it does not require an external magnetic field; second the optically active part of the device occupies the full height of the waveguide rib, maximizing the overlap with the guided mode.

#115649 - $15.00 USD

(C) 2009 OSA

