1. Introduction and background

The quest continues to develop ultrafast phototoconductive (PC) THz sources that can be driven by fiber-optic lasers and components preferably at the 1550-nm (EDFA) or 1030-nm (YDFA) wavelengths. Progress has been steady on PC switches and photomixers for time- and frequency-domain applications, respectively, the most common approach being devices fabricated on InGaAs- or InGaAsP epitaxial layers on InP substrates to reduce the band-gap energy (U_G) below the 1550-nm photon (~0.75 eV) and utilize cross-gap (intrinsic) photoconductivity. THz performance has been achieved using a variety of ultrafast recombination mechanisms mostly based on homogeneous or inhomogeneous distributions of metallic nanoparticles or deep defect levels. This includes As precipitates from low-temperature (LT) MBE growth and anneal [1,2], ErAs nanoparticles from normal-temperature MBE [3,4], Br-irradiated defects [5], Fe-ion-implanted deep levels [6,7], and standard InGaAs layers with the defects located in intervening InAlAs barriers [8]. The problem with all such approaches is critical breakdown field (E_C) and the associated dark current. PC switches and photomixers alike display nearly quadratic dependence of THz power on DC bias, so that high E_C is crucial. From semiconductor physics, E_C tends to vary with band-gap energy superlinearly and studies have yielded universal empirical relationships such as E_C = 1.73x10⁵ (U_G)^2.5 [V/cm, U_G in eV] for low-doped direct-band-gap materials [9]. Hence, the difference between the GaAs E_C (U_G = 1.42 eV) and InGaAs E_C is 4.9x, which is rather close to observed difference in maximum bias voltage between homogeneous GaAs and InGaAs ultrafast PC devices. The addition of epitaxial InAlAs barriers enhances the bias standoff of InGaAs devices, but does not prevent breakdown in the lateral direction [8].

A tempting alternative is to use GaAs with 1030 or 1550-nm drive lasers and utilize sub-band-gap photon absorption mechanisms via the high concentration of defect- or impurity-levels that ultrafast materials generally have. In pulsed-mode, for example, attempts have been made to utilize two-photon absorption and sup-ps recombination via the mid-gap states associated with As-precipitates in low-temperature (LT) GaAs [10,11]. This was then used to demonstrate a PC switch, but the resulting photoconductivity was found to be impractically weak compared to the intrinsic cross-gap effect. In cw mode attempts have been made to overcome the weak 1550-nm absorption by embedding the LT GaAs in a dielectric-waveguide, distributed pin photodiode [12]. But the waveguide length required for strong absorption reduces the electrical bandwidth because of difficulties in velocity matching the photonic and RF waves.

2. Basic characteristics

In this work we demonstrate a much stronger mechanism for driving an ultrafast GaAs PC switch with 1550-nm pulses. The ultrafast material is ErAs:GaAs [13], an ultrafast photoconductor newer than LT GaAs and one that has shown excellent performance both as a photomixer [14] and as a PC switch [15]. In comparative PC-switch studies, it has also been shown unequivocally to be superior to LT GaAs when driven with ~780-nm cross-gap lasers [16]. For the present experiments we tested switches consisting of a 1.0-micron thick, homogeneous 1%-Er-bearing GaAs films grown by molecular-beam epitaxy on semi-insulating GaAs substrates. The PC switch shown in Fig. 1 consisted of a 9x9 micron gap at the center gap of a 3-turn square-spiral antenna. Laser light from a standard 1550-nm EDFA mode-locked laser was focused onto the gap using a fiber-to-free-space coupler and microscope objective. The laser has a pulse width of ~300 fs, a repetition rate of 36.7 MHz, and a maximum average power of 140 mW. THz radiation emanating from the spiral antenna was coupled into free space using a high-resistivity silicon hyperhemispherical lens.

 

Fig. 1 ErAs:GaAs PC switch with 9x9 micron center gap.

Download Full Size | PDF

We first measured the DC photocurrent as a function of bias voltage V_B with a fixed 1550-nm power of 140 mW. As shown in Fig. 2(a) , photocurrent approaches zero as V_B →0, as expected for any photoconductive effect. We then measured the DC photocurrent versus drive power P₀ with the bias voltage fixed at 77 V. As shown in Fig. 2(b), this is concave-down at low P₀ but quasi-linear behavior at higher power. This is in contrast to the quadratic-up behavior displayed by LTG-GaAs switches with 1550-nm drive [11]. The associated current responsivity in Fig. 2(b) at the lowest P₀ is ℜ ≈5 μA/mW, but at the highest P₀ drops to ≈1.0 μA/mW. This latter is only about 4-times less than the ℜ from an identical type of PC switch (same ErAs:GaAs material) measured at the same V_B with a sub-ps pulsed laser source emitting around 780 nm. It suggests that the 1550-nm drive should produce measurable THz power, assuming of course that the bandwidth associated with the new photoconductive mechanism is comparable to that of the traditional intrinsic, cross-gap effect.

 

Fig. 2 (a) Photocurrent vs bias voltage with 1550-nm laser at maximum power (140 mW). (b) DC photocurrent vs average 1550-nm laser power at a fixed bias voltage of 77 V.

Download Full Size | PDF

3. THz measurements

We proceeded with the THz power measurements starting with a broadband, calibrated LiTaO₃ pyroelectric detector with a 0.01-in black polyethylene window to block 1550-nm leakage and thermal IR radiation. The experimental results for broadband THz power vs V_B and P₀ are plotted in Figs. 3(a) and 3(b), respectively. The vertical scale in both plots is rms (lock-in amplifier readings). Correcting for the rms reading, we obtain an equivalent peak-to-peak reading of 520 mV (confirmed on an oscilloscope). The pyroelectric detector has a calibrated, broadband external responsivity of ≈5000 V/W between 0.1 and 1.0 THz. So the maximum power measured from the switch is ≈105 μW. This is comparable to the broadband THz power measured from an identical type of switch (same ErAs:GaAs material) at the same V_B, but driven with 25 mW of average power from the 780-nm sub-ps pulsed laser source. Hence, the new 1550-nm-driven photoconductive mechanism is about 5-times less efficient in terms of THz-to-laser power ratio. The dependence of THz power on V_B and P₀ is also somewhat different than the 780-nm-driven switch. The bias-dependence in Fig. 3(a) is close to quadratic (see fit curve) and power-dependence is weaker, P_THz ≈(P₀)^1.6. The 780-nm performance is usually the opposite with P_THz varying close to quadratic with P₀ and slower with V_B.

 

Fig. 3 (a) AC signal (rms) from THz pyroelectric detector vs bias voltage with a constant 1550-nm laser power of 140 mW. (b) AC signal (rms) from same detector vs 1550-nm average laser power at a constant bias of 77 V.

Download Full Size | PDF

As a rough estimate of the bandwidth of the 1550-nm-driven PC switch, we carried out power measurements using a set of Schottky-diode zero-bias rectifiers mounted in rectangular waveguide and operating in three distinct bands centered around 92 (W-band), 415, and 675 GHz. These rectifiers act as band-limited filters with very sharp low-frequency turn-on (waveguide cutoff) and more gradual high-frequency rolloff. This enables a discrete estimate of the THz switch power spectrum knowing the external responsivity of the rectifiers and their noise equivalent bandwidth. The data is plotted in Fig. 4 , normalized to the signal from the lowest-frequency rectifier.

 

Fig. 4 AC Signal from THz pyroelectric detector vs frequency at a constant bias of 77 V and constant 1550-nm laser power of 140mW.

Download Full Size | PDF

The bandwidth is obtained by fitting the discrete spectrum to a single-pole Lorentzian function, S(f) = A/[1 + (2πfτ)²]⁻¹, where A is a constant and τ is the photocarrier lifetime. This has been found to be a good fit to the THz power spectrum of PC switches whose photocarrier lifetime is significantly longer than the RC electrical time constant – a likely condition in our case since the gap capacitance of the switch is << 1 fF. For the experimental data in Fig. 4, the best fit to the data occurs when A = 1.08 and τ = 0.45 ps, This corresponds to a −3-dB frequency-domain bandwidth of B = (2πτ)⁻¹ = 354 GHz, which is comparable to the bandwidth deduced from 780-nm time-domain measurements for the identical type of switch (same ErAs:GaAs material and antenna) with a 780-nm femtosecond laser [17]. However the laser pulse in our experiments (300 fs) is considerably longer than that used at 780 nm, so that the fundamental bandwidth of our switch could be even higher than 354 GHz.

4. Physical interpretation

All of the results presented here are consistent with the new 1550-nm-driven PC switch mechanism being extrinsic photoconductivity rather than the traditional intrinsic (cross-gap) photoconductivity. Extrinsic photoconductivity is distinguished by a transition from a localized-impurity or defect energy level to the closest energy band (conduction or valence), and then subsequent unipolar photocarrier transport (electron or hole) within that band [18]. It is well known in doped GaAs and has long been utilized to make high-power PC switches operating at the ~10-ps time scale [19]. Through growth conditions discovered in Ref [20], the present PC switch material contains ErAs in the form of crystalline nanoparticles, and these nanoparticles are associated with a very large density of energy levels near the middle of the GaAs bandgap. This explains the sub-ps electron-hole photocarrier lifetime in intrinsic operation, and should explain the fast extrinsic operation through a large capture cross section for electrons or holes, as the case may be. The ErAs nanoparticules have also been found to display sub-band-gap absorption that reaches a peak strength around λ = 2.5 μm, either through a particle-plasmon [21], or quantum-dot resonance.

From the work presented here, we can only speculate on the exact absorption mechanism. However, we know with certainty that it creates photocarriers, which in turn exhibit good electrical transport (i.e., good mobility) and the sub-picosecond lifetime necessary to generate useful levels of THz radiation in photoconductive switches. In GaAs this would favor electrons over holes because of their superior band transport. In any case, the absorption coefficient is likely much weaker than the cross-gap value around 780 nm, which is typically ~10⁴ cm⁻¹. And this would partially explain the 4-times lower external current responsivity and 5-times lower laser-to-THz conversion efficiency of the 1550-nm-driven switch. But lower absorption has a beneficial aspect which is more gradual photocarrier and thermal generation with depth than normally occurs in GaAs photoconductive devices. This should help improve the reliability and maximum drive power, which are often limited by electric and/or thermal stress at the surface of planar photoconductive devices. And the fact remains that 1550-nm photons are much more affordable than 780-nm photons, and much easier to route and control via the wide variety of active and passive components available from the fiber-optic telecomm industry.

5. Conclusion

We have shown that an ErAs:GaAs photoconductive switch can demonstrate useful levels of THz power when driven by an ultrafast 1550-nm fiber-mode-locked laser. The external responsivity and THz generation efficiency are lower than those in the same switch driven by 780 nm sub-ps pulses, but the absolute THz power level is comparable. The likely mechanism for the 1550-nm excitation is extrinsic n-type (electron) photoconductivity from the ErAs-nanoparticles to the conduction band, although more research is necessary to prove this unequivocally.