There is a recent surge in the research on high-energy, few-cycle THz pulses, which is being driven by strong needs for applications in imaging, security, and nonlinear THz spectroscopy. Currently, many groups use optical rectification in nonlinear crystals (such as ${\mathrm{LiNBO}}_3$ and ZnTe) to generate multi-microjoule, single- and few-cycle THz pulses [1, 2], which have been used to explore new phenomena in nonlinear THz spectroscopy [3]. Large aperture photoconductive antennas (LAPCAs) have also been used as intense THz sources [4], and more recently, the highest optical-to-THz efficiency of $2.0\times {10}^{-3}$ has been reported for an interdigitated LAPCA [5]. The principle of generating THz pulses with LAPCAs is described by the current surge model, where the radiated THz electric field is proportional to the applied bias field [6]. Therefore, high-voltage pulsed fields are desirable for generating high-energy THz pulses from LAPCAs. However, working with pulsed high-voltage power poses significant technical challenges. LAPCAs with interdigitated electrodes overcome many of these challenges by reducing the gap size and, hence, the required applied voltage. However, THz pulses from neighboring electrodes have opposite polarity and destructively interfere in the far field, thus cancelling the THz field. To avoid such destructive interference, a binary shadow mask is typically applied to half of the electrodes. This allows the laser pulse to illuminate and excite only every other electrode, and generates THz radiation through constructive interference in the far field [7, 8]. Consequently, only half of the total area of the antenna is used, and, therefore, the optical-to-THz conversion efficiency is substantially reduced for interdigitated LAPCAs, relative to an LAPCA that does not have the interdigitated structure, but with the same area.

In this Letter, we demonstrate improved optical-to-THz conversion efficiency and THz pulse shaping by placing a binary phase mask on the interdigitated structure of a GaAs LAPCA, avoiding destructive interference while making full use of all electrodes and generating quasi- single-cycle THz pulses. Figure 1 shows the principle of the binary phase mask. The quasi-single-cycle THz pulse results from the overlap in the far field of the half-cycle THz pulses emerging from the masked and unmasked electrodes, which have opposite polarities, but with a relative time delay between them. The time delay between the positive and negative polarity THz pulses is proportional to the thickness of the binary phase mask. The binary phase mask introduces a phase shift between the positive and negative polarity THz pulses, allowing single-cycle THz pulse generation and, hence, higher THz energy pulses [9]. In this Letter, we applied the binary phase mask on an interdigitated structure of an LAPCA. However, this technique can also be applied whenever a periodic illumination is required, such as in the excitation of the recently reported lateral photo-Dember emitter [10].

In our experiments, the antenna is composed of seven identical electrodes spacing with a gap size of $3\mathrm{mm}$. We chose a large gap size in order to work in the THz-field-screening regime so that the output THz pulse was not limited by operating in the space-charge-screening regime. The electrode pattern was fabricated using conventional photolithography. The contacts were made using $\mathrm{Ti}/\mathrm{Au}$. The bias field was provided using an AC voltage amplifier, which is able to generate up to $360\mathrm{V}$ at $2.13\mathrm{kHz}$. The amplified Ti:sapphire laser used in these experiments delivers $500\mathrm{\mu J}$, $60\mathrm{fs}$ pulses at a center wavelength of $800\mathrm{nm}$ with a $10\mathrm{kHz}$ repetition rate. The full width at $1/e$ maximum beam size was $3.4\mathrm{mm}$ on the emitter and we did not use a lens to collimate the output THz beam. The THz electric field was detected by free-space electro-optic sampling in a $[110\mathrm{]}$ ZnTe crystal with a thickness of $1\mathrm{mm}$ [11]. The binary phase mask is composed of three quartz plates that are $3\mathrm{mm}$ wide and $3\mathrm{cm}$ long, in order to cover the electrodes, with different thicknesses used to vary the delay as discussed below.

Figure 2 shows the THz pulse shape radiated from an interdigitated GaAs LAPCA with a standard shadow mask, a $510\mathrm{\mu m}$ thick binary phase mask, and no mask applied to the structure. The single-cycle behavior expected with the appropriate phase mask is evident. It should be noted that the pulse width of the generated THz pulse with the phase mask is slightly less than twice that obtained with the shadow mask, which is expected to lead to a doubling in the generated THz pulse energy. This is discussed later in this report. Also, the fact that the THz field radiated from the interdigitated switch without any mask is not zero can be attributed in part to the fact that there is an odd number of antennas (seven in total), and so complete cancellation in the far field is not expected for the unmasked structure (the net field from one antenna will be present with no masking). The inset shows the power spectrum of the radiated THz electric field obtained with a $510\mathrm{\mu m}$ binary phase mask and a traditional shadow mask.

Figure 3 shows the THz pulse shape obtained with a glass phase mask having thicknesses of 0.17, 0.34, 0.51, 0.68, and $1\mathrm{mm}$, and Fig. 4 shows the power spectrum for the THz signals obtained with the 0.17, 0.34, 0.51, 0.68, and $1\mathrm{mm}$ glass phase masks, respectively. Choosing an appropriate thickness for the glass phase mask is crucial. If a thin glass phase mask is used, the positive and the negative half-cycle THz pulses overlap in time. Consequently, the peak THz electric field amplitude is reduced and the THz pulse width is shorter. This phenomenon is particularly pronounced with the 0.17 and $0.34\mathrm{mm}$ thick glass phase masks. For the 0.51, 0.68, and $1\mathrm{mm}$ thick glass phase masks, we do not observe a strong modulation of the peak THz electric field. However, the THz pulse duration continues to increase by delaying the negative half-cycle THz pulse.

These changes in the pulse shape in the time domain introduce two main changes in the frequency domain. Figure 4 shows a shift of the peak of the power spectrum toward higher THz frequencies when reducing the thickness of the glass phase mask from $1\mathrm{mm}$ to $0.17\mathrm{mm}$. For example, the peak of the power spectrum is at $0.46\mathrm{THz}$ with the $0.17\mathrm{mm}$ thick mask, and only $0.23\mathrm{THz}$ with the $1\mathrm{mm}$ thick mask. This shift is a consequence of the reduction of the THz pulse duration as we mentioned earlier. The second observation is the decrease in the maximum amplitude in the power spectrum by reducing the thickness of the glass phase mask from $1\mathrm{mm}$ to 0.17. This reduction is a consequence of the reduction of the peak THz electric field in the time domain. It is important to note that all spectra extend up to $2.5\mathrm{THz}$, regardless of the mask thickness. The oscillations in the power spectrum are the result of the interferences between two physically different THz pulses that we overlap in time using the binary phase mask, which modulates the spectrum with frequency spacing proportional to the time delay between pulses. The maximum energy occurs when the trade-off between shortest separation and maximum field amplitude is optimized.

Next, we compare the THz pulse energy using the different binary phase masks, integrating the power spectrum obtained in Fig. 4. Figure 5 shows the variation of the value of the integrated power spectrum. We observe that the energy, and hence the optical-to-THz conversion efficiency, increases by increasing the mask thickness up to $0.68\mathrm{mm}$. The maximum value obtained with the $0.68\mathrm{mm}$ thick phase mask is 0.95, measured in arbitrary units. In comparison, the integrated value of the power spectrum from a standard shadow mask obtained in Fig. 3 was only 0.54, which is slightly below the value obtained with the $0.34\mathrm{mm}$ thick mask. Therefore, from Fig. 5, we observe a THz pulse energy, and consequently an optical-to-THz conversion efficiency, that is 1.76 times higher than a half-cycle THz pulse generated with a standard shadow mask. In principle, we would expect to see an enhancement in the conversion efficiency close to a factor of 2 using this method, because the pulse duration increases by a factor of 2 in going from a half- to a single-cycle pulse, while the peak intensity is maintained. This is not the case here, because our interdigitated antenna consists of an odd number (seven) of antennas, and, therefore, the shadow mask only throws away $3/7$ of the total energy, and thus one would expect an enhancement in conversion efficiency of $7/4$, consistent with the value of 1.76 observed here.

In conclusion, we have demonstrated a higher optical-to-THz conversion efficiency from an interdigitated GaAs LAPCA using a binary phase mask rather than the traditional method of using a shadow mask. This discovery is particularly important for research involving high-energy THz radiation and THz pulse shaping experiments, as it directly leads to an enhancement of a factor two of the optical-to-THz conversion efficiency and of the THz pulse energy and the ability to control the temporal profile of the THz pulse. The method is simple to use, uses inexpensive components, and behaves according to expectations.

 

Fig. 1 Schematic diagram showing the principle of a binary phase mask for single-cycle THz pulse generation from an interdigitated photoconductive antenna.

Download Full Size | PDF

 

Fig. 2 THz pulse shape from an interdigitated GaAs antenna at a bias field of $1.2\mathrm{kV}/\mathrm{cm}$ and a fluence of $14\mathrm{\mu J}/{\mathrm{cm}}^2$ with a standard shadow mask, a $510\mathrm{\mu m}$ thick binary phase mask, and no mask. An offset has been introduced for more visibility. The inset shows the power spectrum with a shadow mask and a $510\mathrm{\mu m}$ glass phase mask.

Download Full Size | PDF

 

Fig. 3 Evolution of the THz pulse shape for different glass phase mask thicknesses (0.17, 0.34, 0.51, 0.68, and $1\mathrm{mm}$ thick) of an interdigitated GaAs LAPCA at a bias field of $1.2\mathrm{kV}/\mathrm{cm}$ and a excitation fluence of $14\mathrm{\mu J}/{\mathrm{cm}}^2$.

Download Full Size | PDF

 

Fig. 4 Power spectrum for the 0.17, 0.34, 0.68, and $1\mathrm{mm}$ glass phase masks thicknesses.

Download Full Size | PDF

 

Fig. 5 Variation of the integrated value of the power spectrum obtained in Fig. 5 for the 0.17, 0.34, 0.51, 0.68, and $1\mathrm{mm}$ thick glass phase masks. The horizontal line indicates the integrated power spectrum for the shadow mask for direct comparison.

Download Full Size | PDF

1. F. Blanchard, L. Razzari, H. C. Bandulet, G. Sharma, R. Morandotti, J. C. Kieffer, T. Ozaki, H. F. Tiedje, H. K. Haugen, and F. Hegman, Opt. Express 15, 13212 (2007). [CrossRef]   [PubMed]  

2. A. G. Stepanov, L. Bonacina, S. V. Chekalin, and J. P. Wolf, Opt. Lett. 33, 2497 (2008). [CrossRef]   [PubMed]  

3. L. Razzari, F. H. Su, G. Sharma, F. Blanchard, A. Ayesheshim, H. C. Bandulet, R. Morandotti, J. C. Kieffer, T. Ozaki, M. Reid, and F. Hegman, Phys. Rev. B 79, 193204 (2009). [CrossRef]  

4. D. You, R. Jones, and P. Bucksbaum, Opt. Lett. 18, 290 (1993). [CrossRef]   [PubMed]  

5. M. Beck, H. Schäefer, G. Klatt, J. Demfar, S. Winnerl, M. Helm, and T. Dekorsy, Opt. Express 18, 9251 (2010). [CrossRef]   [PubMed]  

6. J. T. Darrow, X.-C. Zhang, and D. H. Auston, IEEE J. Quantum Electron. 28, 1607 (1992). [CrossRef]  

7. T. Hattori, K. Egawa, S. Ookuma, and T. Itatani, Jpn. J. Appl. Phys. 45, L422 (2006). [CrossRef]  

8. A. Dreyhaupt, S. Winnerl, T. Dekorsy, and M. Helm, Appl. Phys. Lett. 86, 121114 (2005). [CrossRef]  

9. M. Reid and R. Fedosejevs, Appl. Opt. 44, 149 (2005). [PubMed]  

10. G. Klatt, B. Surrer, D. Stephan, O. Schubert, M. Fischer, J. Faist, A. Leitenstorfer, R. Huber, and T. Dekorsy, Appl. Phys. Lett. 98, 021114 (2011). [CrossRef]  

11. Q. Wu, M. Litz, and X.-C. Zhang, Appl. Phys. Lett. 68, 2924 (1996). [CrossRef]