1. Introduction

Semiconductors excited by femtosecond laser pulses as Terahertz (THz) radiation sources have been frequently used to substitute photo-conductive antennas in THz time-domain spectroscopy (TDS) systems [1,2]. This is because they are inherently robust, compact and simple without the need for external bias. Among all compound semiconductors, InAs, as a narrow band-gap semiconductor, has been identified as the most efficient THz emitter [3,4]. The strong THz emission from InAs is mainly associated with the high electron mobility. For InAs (100), the THz emission is dominated by the photo-Dember effect rather than electrons and holes driven by the surface depletion field or non-linear effects, optical rectification [4–8]. This is due to the large electron-hole mobility difference (the electron mobility is 33000 cm²/Vs and the hole mobility is 460 cm²/Vs), weak surface depletion field for the narrow energy bandgap, large excess energy of photogenerated electrons (1.2 eV for 800 nm excitation wavelength) and short light absorption length (140 nm for 800 nm excitation wavelength). All these properties promote the photo-Dember effect which involves the diffusion of carriers into the bulk of the material and the separation of electrons and holes excited close to the surface. The resulting photo-Dember electric field is screened by doped carriers, favoring low-doped p-type layers [5], although this causes a small downward surface band bending opposing the photo-Dember electric field. Modulation of the photo-Dember electric field by the pulsed laser excitation leads to the THz radiation.

To increase the THz radiation efficiency, approaches to reduce the loss of the THz wave propagation at the emitter-air interface by applying magnetic fields [9–12], lens couplers [13,14] or metal gratings [8] have been pursued. Certainly, enhancing the THz radiation generation efficiency, again by magnetic fields [15], extraction azimuth [16], excitation power [17,18] and excitation wavelength [19] is more promising. In this regard, efforts to directly address the carrier transport processes inside the semiconductor layer are barely found. In this study, we exploit the control of the relevant electron and hole transport through advanced heterostructure design. Inserting a thin n-InGaAs barrier close to the surface of a p-InAs layer blocks the hole diffusion while not affecting much the electron diffusion. The photogenerated electron-hole separation and photo-Dember electric field are increased and, therefore, the THz radiation efficiency.

2. Experimental

The pnp-InAs/InGaAs heterojunction structure was grown by molecular beam epitaxy (MBE) with the metal atomic sources provided by Knudson effusion cells and the group V active species from an As cracker cell. Be and Te thermal effusion cells were used for p- and n-type dopant sources, respectively. The samples were grown on InAs (100) substrates which were degassed at 300°C for 1 hour in the buffer chamber before transferred into the growth chamber. Prior to growth, the substrate temperature was raised to 530°C under As₂ flux to desorb the native surface oxide. The substrate temperature was then decreased to 480 °C for growth with a V/III flux ratio of 15. The growth comprised a 900 nm p-InAs / 10 nm n-In_0.85Ga_0.15As / 100 nm p-InAs heterojunction structure and a 1000 nm p-InAs reference layer. The p-type doping concentration was 5${\times} $10¹⁶ cm^-3 and the n-type doping concentration was 1${\times} $10¹⁷ cm ^-3, determined from Hall-effect measurements of separate calibration samples. The growth rate was calibrated by high energy electron diffraction (RHEED) intensity oscillations. The layer structures and doping profile were confirmed by X-ray diffraction (XRD) and secondary ion mass spectrometry (SIMS).

Time-domain THz spectroscopy was performed in reflection geometry, shown in Fig. 1. A p-polarized mode-locked Ti: sapphire laser with central wavelength of 780 nm, 80 MHz repetition rate, 60 mW power and 100 fs pulse width was used as excitation source. The incidence angle of the pump laser beam to the sample surface normal was 45°. The THz radiation was collected in the direction of the specular reflection and focused on a photoconductive antenna detector using a pair of parabolic mirrors. A delay line was used in the detection path to vary the arrival time of the signal with respect to the optical pulse used for detection. By scanning the delay line and rotating the sample stage, the electric field amplitude and phase of the THz waveform were mapped as a function of time and azimuth. A low pump fluence (<1μJ/cm²) was adopted where the photo-Dember effect for THz generation is most pronounced [20,21].

 

Fig. 1. Time-domain THz spectroscopy setup in the reflection geometry.

Download Full Size | PDF

3. Discussion

Figure 2 shows a scheme of the (a) band structure and (b) layer design of the 100 nm p-InAs top layer/10 nm n-In_0.85Ga_0.15As barrier layer/900 nm p-InAs bottom layer/InAs (100) substrate heterojunction structure. Due to the small thickness and moderate doping level, the n-InGaAs layer is assumed to be fully depleted, leading to two narrow and sharp triangular potential barriers for electrons in the conduction band and a high potential barrier for holes in the valence band. For not too large InGaAs barrier height (Ga content) and thickness, sufficient tunneling and termionic emission are allowed for electrons while the hole transport is blocked. Notably, the n-InGaAs barrier layer is located around, slightly below the light absorption depth. This prevents the generation of too many holes in the p-InAs bottom layer, keeping most of the holes as close as possible to the surface in the p-InAs top layer. The delicate balance of InGaAs barrier height, thickness, doping and location leads to enhanced electron-hole separation, photo-Dember electric field and THz radiation efficiency, as will be shown. Here it is important to note that we also investigated samples with different deeper location of the InGaAs barrier and different lower doping which, however, all showed inferior performance. Therefore, the structure discussed here with intuitively best design proves the usefulness of the hetero-structure method.

 

Fig. 2. Scheme of the (a) band structure with photogenerated carrier transport and (b) pnp-InAs/InGaAs heterojunction structure.

Download Full Size | PDF

To verify the designed heterostructure and doping, XRD spectra and SIMS depth profiles are shown in Figs. 3(a) and (b) for both, the heterojunction structure and reference layer. Clear pendellösung fringes together with a narrow InAs (004) reflection are observed in the XRD spectrum of the heterojunction structure while only the narrow InAs (004) reflection is observed for the reference layer. The pendellösung fringes are due to the InAs top layer separated by the thin InGaAs layer from the InAs bottom layer. The measured XRD spectra agree well with the simulated ones indicating the designed layer structures. The simulations are based on the theoretical approach developed by Takagi and Taupin for elastically deformed crystals, which constitutes a generalization of the dynamical X-ray diffraction theory [22,23]. The broad feature for the heterojunction structure originates from the InGaAs layer with 15% Ga content. The SIMS depth profiles of Ga, In and As also agree with the designed InAs/InGaAs heterojunction structure, including the dopant distribution of Be and Te.

 

Fig. 3. (a) XRD spectra and simulation of the pnp-heterojunction structure and p-reference layer; (b) SIMS depth profiles for the heterojunction structure.

Download Full Size | PDF

Figures 4(a) and 4(b) present the time-domain and frequency-domain THz radiation spectra for the heterojunction structure and reference layer. The amplitude of the transient THz emission in Fig. 4(a) is increased by 70% for the heterojunction structure compared to that for the reference layer. The line shapes coincide, indicating the same generating mechanism of the THz radiation associated with the photo-Dember effect. This is further manifested by the same spectral bandwidth shown in Fig. 4(b). The relatively narrow band width around 1.4 THz is attributed to the large Femtosecond laser excitation spot diameter of 4 mm, which has also been reported previously [24,25]. A similar THz emission intensity enhancement for the pnp hetero-structure is also observed at higher frequencies for a smaller excitation laser spot.

 

Fig. 4. (a) Time-domain spectra and (b) frequency-domain spectra of the pnp-heterojunction structure and p-reference layer.

Download Full Size | PDF

Figure 5 shows the azimuth angle dependence of the THz radiation amplitude for the heterojunction structure and reference layer. The angle dependent part stems from the non-linear optical refraction whose contribution to the total amplitude is low for both structures. This indicates the dominance of the photo-Dember effect equivalently for the heterojunction structure and reference layer. The non-linear optical refraction accounts for 5% and 2% for the heterojunction structure and reference layer, respectively. From the weak angle dependent part and difference of the total amplitude, the enhancement of the photo-Dember effect by around 70% due to the n-InGaAs barrier layer is most clearly evidenced.

 

Fig. 5. Azimuth angle dependence of the THz radiation amplitude for the pnp-heterojunction structure and p-reference layer.

Download Full Size | PDF

The blocking of the hole transport by the n-InGaAs barrier layer is independently shown by the surface photovoltage (SPV) measurements carried out by Kelvin probe force microscopy (KPFM) [26–28]. KPFM measures the contact potential difference (V_CPD) between the PtIr metal coated probe tip and the semiconductor surface, i.e., the difference of the work functions ($\emptyset $_tip−$\emptyset $_surface). Under light illumination, the SPV is obtained by the change of V_CPD. Though the SPV originates from the change of the surface band-bending, which is not related to the photo-Dember effect governing the THz radiation, it provides information about the distribution of the photogenerated carriers which screen the surface depletion field. The magnitude and sign of the SPV provide the amount and type of carriers accumulated at the surface with the negative sign for electrons and positive sign for holes [29–32]. To exclude morphological influences, atomic force microscopy (AFM) measurements were carried out before the KPFM measurements to determine the root mean square (rms) roughness which is below 1 nm for both samples.

Figure 6 shows the KPFM CPD images and time traces (line scans from top to bottom in the images) for the (a, b) heterojunction structure and (c, d) reference layer. Illumination is by a 310 nm UV LED with the light off/on incident marked. The SPV is -20 mV for the heterojunction structure and -43 mV for the reference layer. The negative sign implies electron accumulation, in accordance with the downward energy band-bending for p-type layers. Assuming unhindered motion for the electrons, the smaller magnitude of the SPV for the heterojunction structure reveals that relatively more holes stay close to the surface and, hence, confirms the blocking of the hole transport due to the InGaAs barrier layer.

 

Fig. 6. KPFM CPD images and time traces for the (a, b) pnp-heterojunction structure and (c, d) p-reference layer. The light off/on incidence is indicated.

Download Full Size | PDF

4. Summary

A pnp-heterojunction structure with a thin n-InGaAs barrier layer embedded in a p-InAs (100) thin film has been proposed to enhance the THz emission of p-InAs under ultrafast pulsed laser excitation due to the photo-Dember effect by affecting the photo generated free carrier transport. The designed layer structure and doping profile were verified by XRD and SIMS analysis. The THz radiation intensity increased by 70% with the inserted n-InGaAs barrier layer and azimuth angle dependent time-domain THz spectroscopy measurements confirmed the enhancement of the photo-Dember effect. The inserted n-InGaAs barrier layer blocked the diffusion of photogenerated holes into the InAs bulk while still allowing sufficient electron tunneling to enhance the electron-hole separation and, therefore, the photo-Dember electric field. The blocking of the hole transport was further confirmed by surface photovoltage measurements in Kelvin probe force microscopy. This demonstrates that it is possible to enhance the THz emission from semiconductors by the application of heterostructures. Detailed theoretical and experimental work is now required to find the optimum heterostructure and doping design for maximized enhancement of the THz radiation efficiency.