1. Introduction

Using photoconductive emitters is one of the most popular ways to generate terahertz (THz) radiation pulses. Main applications of these THz pulses are chemical identification, material characterization, imaging etc [1–4]. Improved efficiency of THz emission will improve the data quality for conventional applications and it can also widen its application in other fields too, like non-linear studies of materials in the THz range. Hence a lot of works have been, and, are being done to improve the efficiency of photoconductive THz emitters [5–8]. Strip line structures of electrodes for these emitters are one of the simplest and oldest designs [9]. They are still being used in modern interdigitated electrode designs for large area emitters [10–12]. In the past, there have been few reports on the study of performance of strip line structures with variation of both electrodes together [9, 13]. However, the asymmetry in electrode widths has not been studied in details so far. In the interdigitated electrode design, a significant part of large area emitters is covered with metallic electrodes. This reduces the total number of photo-generated charge carriers in the emitter after optical excitation and hence it also restricts overall THz emission. Therefore, it is essential to optimize the electrode design to minimize the area covered by the electrodes without affecting the overall THz emission.

In photoconductive techniques for THz emission, a DC electric field is applied to a semiconductor (usually GaAs). Electron-hole pairs are generated within sub-picosecond time using sub-picosecond laser pulses. These newly photo-generated charge carriers are accelerated under the presence of already applied DC electric field in the semiconductor and their acceleration causes THz electromagnetic field radiation. The amplitude of the electric field of the radiated THz pulse depends linearly on applied bias field (E), number of photo-generated charge carriers (N_e and N_h) and carrier mobilities (µ_e and µ_h) [14]. After the photoexcitation equal amounts of electrons and holes will be generated, and they will move in opposite direction since they have opposite charges. Electrons will move towards the anode (positively biased electrode) and holes will move towards the cathode (negatively biased electrode). The mobility of electrons is much higher than that of the holes in GaAs (µ_e/µ_h > 20) [15], the most common semiconductor used as photoconducting medium for THz emission. Hence the major contribution in THz emission comes from the dynamics of electrons. THz emission originates from acceleration of charge carriers inside the photoconducting substrate as well as from propagation/oscillation of surge current fed into the electrodes. THz emission from within the photoconducting substrate should not depend on the electrode widths. However, the motion of carriers excited close to a metallic electrode gives rise to an impulsive THz field. The photoexcited dipole thus interacts with the near-field of the metal electrode and induces local charge displacements in the metal. The resulting THz emission is thus influenced by the mode structure of the electrode which acts as an antenna [16]. In a simplified picture, these modes can be considered as plasmonic excitations located at the Au/GaAs interface of the electrode. THz emission due to the surge current propagation induced by this plasmon excitation depends on the propagation length (l). If the antenna length, which corresponds to the width of strip line electrodes here, is much shorter than the wavelength, the electric field of emitted radiation depends linearly on antenna length, and the THz field E_THz of a dipole antenna at far-field can be written as

where l is the length of short dipole, i is imaginary unit, ω is the frequency of oscillating current fed into the antenna, I₀ is the peak value of the current being fed into the antenna, r is the radial distance of the observation point from the short dipole antenna, θ is the angle made by the observation point on the antenna length [17]. Here θ ~90°, since we are measuring the radiated THz field in a direction perpendicular to the direction of photocurrent just after the carrier excitation. Hence it is expected that strip line emitters with wider electrodes would be more efficient. However, the linear dependence between THz electric field and electrode width is valid only for electrode widths much smaller than the THz wavelength.

Different designs for antenna electrodes have been exploited to optimize the emitted THz pulse. The importance of near-anode excitation is already known and it has been well studied by several groups [18, 19]. Recently there have been several reports where researchers could enhance the THz emission efficiency by adding nanostructured metallizations to the anode structure [5]. This enhancement is associated with a plasmonic resonance of the near infrared excitation. However, so far the strip line width of anode and cathode of a photoconductive emitter has not been studied separately. Here we report a systematic study of the width of strip line electrodes on THz emission efficiency and discuss the influence of the strip width on THz plasmonic excitations in the metallization. Electrode widths have been varied together as well as separately.

2. Experimental details and results

Semi-insulating GaAs (SI-GaAs) was taken as substrate material and 5 nm titanium (Ti) followed by 50 nm gold (Au) was used as electrodes. Electron beam lithography was used to fabricate the structures on the SI-GaAs. Five pairs of strip line electrodes having equal widthsof 2 µm, 5 µm, 10 µm, 20 µm and 50 µm, and two pairs of electrodes having electrode widths of 2 µm - 20 µm and 2 µm - 50 µm were fabricated. The gap between the two electrodes was 10 µm for all structures. Microscope images of symmetric and asymmetric strip line emitters are shown in Figs. 1(a) and 1(b) respectively.

 

Fig. 1 Microscope images of the strip line electrodes with (a) symmetrical widths of 20 µm and (b) asymmetrical widths of 2 µm and 20 µm. The gap between two electrodes is 10 µm for both emitters. Strip line width will act as antenna length in these emitters.

Download Full Size | PDF

Current vs voltage characteristics of all seven emitters were studied under pulsed excitation at ~50 mW power. A mode locked Ti:Sapphire laser with wavelength centered around 800 nm, 100 fs pulse width and 76 MHz pulse repetition rate was used for pulsed excitation. Results are shown in Fig. 2 for bias range of −10 V to 10 V. There is no significant change in photo-current with variation of electrode widths and the magnitude of the photo-current is almost symmetrical for positive and negative biases. This implies that even the narrowest strip line structure is able to provide the average current required to the emitter without significant potential drop in the electrode line.

 

Fig. 2 Current-voltage characteristics of all the emitters under pulsed illumination. Photocurrent does not depend significantly on the electrode widths.

Download Full Size | PDF

Finally THz emission efficiency of all the emitters was studied using a standard THz time domain setup with electro-optic detection by a 2 mm thick ZnTe crystal. Parabolic mirrors of 2 inch diameter and 4 inch focal length were used to collimate and focus the THz beam. The same Ti:sapphire laser was used to pump the emitters. Lock-in detection was used to extract the signal from noise. Electronic chopping was performed by applying a square wave bias (with 10 V amplitude) on the emitters. THz pulses emitted by five symmetrical emitters with varying electrode widths are shown in Fig. 3(a). The THz amplitude increases as the width of both electrodes increases from 2 µm to 50 µm and with almost linear dependence on electrode width for the width range of 2 µm to 20 µm (see Fig. 3(b)). The observed dependence is much more pronounced than what Zhang had reported in 1996 [9].

 

Fig. 3 (a) THz electric field from strip line emitters with symmetrical electrode widths of 2 µm, 5 µm, 10 µm, 20 µm and 50 µm. THz field amplitude increases with the electrode width. In (b) Electric field at the peak of THz pulse is plotted with anode width. Saturation-like curve fitting the experimental data is a guide to eyes.

Download Full Size | PDF

Zhang has observed only 8% increase in THz electric field upon changing the electrode width from 5 µm to 10 µm, whereas we are observing ~80% increase for the same set of electrode widths. This discrepancy is probably due to the large difference in electrode gaps. In [9], the electrode gap was 4 mm and an unfocused laser beam was used to pump the emitters. Hence it is expected that THz emission was taking place due to acceleration of charge carriers inside the semiconductor throughout the length of 4 mm and the device was acting as a large area emitter. In the present case the electrode gap is much smaller than the THz wavelength and hence it is expected that a major contribution in THz emission is due to current surge in the metal electrodes. This current surge is caused by photoexcited carriers incident at the electrode, thus inducing local charge displacements at the metal surface. The resulting impulsive excitation launches a plasmon-like perturbation propagating at the Au/GaAs interface. The induced charge oscillation results in the emission of a THz pulse from the electrode which acts as an antenna. According to Eq. (1), this THz emission scales linearly with electrode width (antenna length) for narrow electrodes. From Fig. 3(b), it is clear that initially THz emission increases linearly with the electrode (in particular anode) width but at wider widths it becomes sublinear. The observed sublinear behavior is attributed to the electrode width becoming comparable with the THz wavelength in the material. In fact, assuming a modal refractive index of the THz plasmon in the range 2 – 3, a 50 µm wide metal electrode will exhibit a roundtrip frequency of 1 – 1.5 THz [20]. In particular, making electrodes wider than ~50 µm is not going to increase their efficiency significantly.

Since the mobility of electrons is much higher than that of holes in GaAs, and for that reason they contribute more in THz emission, a larger number of electrons will be reaching the anode than holes reaching the cathode during the first few picoseconds after the pulse excitation. Therefore the anode width is expected to be more important than the cathode width. To verify this argument, emitters with asymmetric electrode widths of the strip linestructures were tested. The emitter was pumped with the same optical power of 50 mW under two different bias conditions, once the wider electrode as anode and once the narrower electrode as anode. For electronic chopping unipolar square wave bias was applied on the emitters. Results are compared with those for symmetric strip line emitters with 2 µm and 20 µm wide electrode emitters in Fig. 4(a). We observe that the THz emission from the strip line emitter with asymmetric electrode width is higher when a wider (20 µm) electrode is chosen as anode, and the THz signal was almost the same as the THz signal from the emitter where both electrodes are 20 µm wide. Whereas when a narrower (2 µm) electrode was chosen as anode, then the THz emission was less and it was almost the same as the THz emission from a symmetric emitter with 2 µm width of each electrode. Similar result was seen for an electrode pair of 2 µm and 50 µm width (see Fig. 4(b)). It is also clear that not only the THz pulse amplitude but the pulse shape too does not depend on cathode geometry. Thus we observe that the THz emission is not affected if the cathode width is decreased from 20 µm to 2 µm, whereas it depends almost linearly on the anode width for the range of 2 µm to 20 µm. This result can be applied for the fabrication of large area emitters with interdigitated electrode design. Since a wider anode gives more THz emission and a narrower cathode does not harm the emitter performance, it is advisable in the interdigitated electrode design to have one set of alternate electrodes wide to act as anode and the other set of alternate electrodes narrow to act as cathode.

 

Fig. 4 THz electric field from strip line emitters with symmetric and asymmetrical electrode widths. The THz pulse amplitude and the pulse shape is similar for the similar anode widths independent of the cathode widths.

Download Full Size | PDF

In Fig. 5 we plot the FFTs of the THz pulses emitted from the emitters with symmetrical electrode. There is no clear evidence of resonance peaks corresponding to the strip line width. In the earlier works by Matsuura et. al. [21] and Tani et. al. [22] too, they could observe the enhancement of THz emission with the dipole length but the peak position in the FFT was shifted possibly due to factors like dipole width and ohmic loss in the antenna. In the strip line geometry, dipole width is actually strip line length which is too large as compared to the dipole length (strip line width) which may be diminishing the expected resonance behavior of the dipole length. To verify this, we simulated radiation efficiency spectra of 10 µm, 20 µm and 50 µm wide strip line structures for two different lengths, 10 µm and 800 µm using CST microwave software. Our simulation method was similar to the one used by Wallauer et al. [16]. At the excitation spot an electrical dipole was placed to drive the current oscillation in the strip line. The simulation considers THz fields radiated in all the directions. Although the measured THz spectrum will depend also on many other parameters other than the strip line resonance behavior, like spectrum of the excitation current impulse and detector response, the simulation can tell how prominent resonance peaks of strip line width one should expect.

 

Fig. 5 FFT of THz pulses from symmetric strip line emitters with varying electrode widths from 2 µm to 50 µm.

Download Full Size | PDF

Simulated results for 10 µm and 800 µm long strip line are plotted in Figs. 6(a) and 6(b) respectively. We attribute the absence of any clear cut peaks corresponding to the strip line width in Fig. 5. to two facts, 1.) for longer strip line structures resonance peaks are not very prominent, as observed by the simulated results in Fig. 6(b), and 2.) we are limited by our detector response for the higher THz frequencies. Resonance peaks of 5 µm and 2 µm wide electrode will correspond to much higher THz frequencies than our detection limit and hence they cannot be compared with the measured FFTs.

 

Fig. 6 Simulated radiation efficiency of different strip line geometries using CST Microwave software. (a) For smaller strip line length (10 µm) resonance peaks corresponding to strip line widths are more prominent. But in (b), when strip line length is longer (800 µm) resonance peaks are not so prominent. Periodic oscillations in the spectrum are due to Fabry-Perot modes arising from the reflections from the surfaces of the substrate cuboid considered in the simulation.

Download Full Size | PDF

3. Conclusion

In summary, in strip line structures of photoconductive THz emitters, the width of the anode is important. The THz emission efficiency increases strongly with anode width for the range of 2 µm to 50 µm, whereas the cathode width, whether it is 50 µm or 20 µm or 2 µm, does not matter much for THz emission. The results show that for strip line emitters with rather small gap the emission mediated by the metallization plays an import role in addition to the direct emission of accelerated charge carriers in the semiconductor material. The electrode-mediated THz emission is attributed to plasmonic oscillations induced by photoexcited electrons arriving at the anode. These results will likely help to improve the efficiency of interdigitated large area emitters. In interdigitated large area emitters, a significant portion of emitter area is covered with electrodes, which now can potentially be reduced by reducing the cathode width such that a higher fraction of pump power can contribute to THz emission.