Optical beam deflecting and scanning have found many applications in civil and military systems. The available techniques include mechanical, acousto–optic, electro–optic, all-optical devices and so on.^[1–4] Among them, the mechanical deflectors are limited in developing owing to their low deflecting speed. The acousto–optic devices had been most interesting, but their access speed and resolvable power are still not high enough. Although electro–optic deflectors can scan faster comparatively, the relative phase change induced by electro–optic effect is small in most of the electro–optic materials. Thus it requires a long optical path length in these electro–optic crystals, which is limited by the materials themselves.^[5] In addition, the driving power for the electro–optic devices is mostly large. The polymeric material has been reported to be a promising candidate instead of the electro–optic crystal, because of its compatibility with different substrates, ease of fabrication, low cost, and not so high driving voltage as the electro–optic material, despite a lack of thermal stability and relatively large insertion losses.^[6] Borrowing the method in phased array radar, many phased array optical scanning devices have been implemented.^[7,8] However, none of these approaches are ultrafast enough to apply in the ultrafast light signal scanning/steering. The present popular ultrafast travelling-wave electro–optic deflector employs a quasi-velocity-matching technique to compensate the velocity mis-match between an optical group velocity and a microwave phase velocity, which can operate in a picosecond timescale.^[9] Furthermore, light deflecting controlled with light is also proposed,^[10–12] some of which can easily promote an ultrafast light deflection.

The development of the ultrafast optical deflecting/scanning devices brings an inspiration for the improvement of many optical and photoelectric devices. Take the image converting tube streak camera as an example. The meander-type travelling-wave deflecting plate takes the place of the more conventional standing-wave parallel-plate to acquire both increased deflection sensitivity and frequency bandwidth.^[13,14] To proceed, the fact that space charge effect leads to a trade-off between the temporal resolution and dynamic range when the signal energy is high is becoming a serious obstacle for the further performance improvement of this image converting tube streak camera.^[15] In the early stages, direct deflecting of the optical beam using an electro–optic deflector had been proposed instead of photoelectron beam scanning manner in the image converting tube streak camera.^[16] In the year of 2010, Chris’s group first demonstrated that the serrated light illumination for deflection-encoded recording (SLIDER) concept enables the single shot all-optical solid state camera, which is an example of direct beam scanning.^[17] There are plenty of other applications that can benefit from the development of the optical deflecting devices.^[18–20]

In our previous work, we have investigated the statically deflecting characteristics of a CW signal laser at the wavelength of 1053 nm passing through a multiple-prism deflector under the illumination by a 800 nm CW pump laser.^[21] This work will research an ultrafast beam deflection by arranging the same multiple-prism deflector in the pump probe configuration. Pump probe technique has been successfully and universally applied to investigate the carrier dynamics of the photo-excited materials, where two ultrafast pulses with arbitrary time delay complete the scanning of the dynamic process. By adjusting the time delay between the pump pulse and the probe pulse (denoted as signal pulse in our context), the deflection of the signal light beam can be linearly controlled. The multiple-prism deflector employs an AlGaAs/GaAs waveguide with a serrated gold mask on its surface to realize the discrete deflection of the signal light beam. Study on the transient absorption or transmission of the intrinsic GaAs has shown that the generation of the non-equilibrium free carriers can be recognized as instantaneous as the pulse duration time and their relaxation time can come up to nanosecond scale.^[22,23] Therefore, high response speed and high repetition are allowed in this ultrafast deflection device. In the context, Section 2 gives a theoretical study on the dependences of temporal and spatial resolvable performances on the material dispersion of GaAs including group velocity dispersion and angular dispersion, interface reflection, and interface scattering of the deflector. Section 3 demonstrates the experiments about this ultrafast deflection device. The experimental results show that the temporal resolution is about 10 ps, which is not so much satisfying but really promising in the ultrafast signal scanning, optical communication, light controlling and switching.

The semiconductor material is fabricated by the molecular beam epitaxy (MBE). The undoped waveguide core is a 0.6-μm-thick layer of GaAs, which is sandwiched between a 2-μm-thick upper cladding layer (Al_0.24Ga_0.76As) and 3-μm-thick lower cladding layer (Al_0.24Ga_0.76As) on top of a GaAs substrate. On the waveguide surface, a serrated gold mask is formed by photolithography. The serrated mask is an array of right triangles with two leg lengths of 350 μm and 45 μm, respectively. The number of the triangles N_t is 100. The layering structure of optical deflector is shown in the inset of Fig. 6. The experiments are performed at room temperature.

Fig. 6.

	Figure Option ViewDownloadNew Window
	Fig. 6. The experiment setup for ultrafast light deflection.

The experimental setup for ultrafast light deflection is shown in Fig. 6. The delivered ultrashort pulse from the laser has a time duration of 120 fs, a center wavelength of 800 nm and a repetition rate of 1 kHz. The average output power is 3.5 W. The output beam is then split into two portions: one acts as the pump beam with the spot diameter of 8 mm, and the other is coupled into the optical parametric amplifier (OPA TOPAS) to generate a 1000-nm signal beam with a power of 20 mW after being filtered by a triangular prism.

As shown in Fig. 6, the signal beam is coupled into the waveguide core by a cylindrical lens with a focus length of 10 mm. Because of refractive index step at the two side faces of the deflector, the GaAs deflector in fact has a structure of rectangular waveguide in the xoz plane. Therefore, there are many mode spots in the x-axis direction in the GaAs core. To ensure high spatial resolution, a 25-μm-wide slit is perpendicularly placed just in front of the deflector to choose the mode spot near the origin o to investigate, which is not illustrated in Fig. 6. A 25-μm-wide slit in the x direction is horizontally placed just after the waveguide to block the light propagating through the substrate and the light glancing incident on and then reflected by the deflector surface, which only allows the signal propagating through the waveguide core to pass through. A 10× microscope objective is used to image the near-field intensity distribution of signal beam on the end face of the deflector onto a Charge Coupled Device (CCD). Hence the coupling of the signal beam into the waveguide can be checked effectively. The normally incident pump beam is shaped by a cylindrical lens and then illuminates on the upper surface of the deflector. Modulated by the serrated mask, the pump beam instantaneously induces a refractive index change in the GaAs region without the cover of the serrated mask, and thus excites an array of prisms in the waveguide core. The AlGaAs upper cladding layer is transparent to the pump light. The prism array remains latched for a period depending on the relaxation time of photo-injected free carriers in intrinsic GaAs.

At the beginning of the experiment, the pump pulse arrives at the upper surface of the deflector when the signal pulse just begins to exit from its end face. In this case, the signal pulse has no deflection. By gradually decreasing the optical path of the pump pulse, the signal pulse will experience an increasing number of prisms and obtain corresponding deflection. At the moment when the signal pulse propagates through the entire prism array, it will obtain a maximum deflection. Out from the deflector, the deflected signal is focused and coupled onto a CCD camera for recording.

The group refractive index corresponding to the ultrashort signal pulse with a center wavelength of 1000 nm is denoted as n_s. In GaAs, n_s = 3.8. Therefore, T = 57 ps for the deflector sample. Figure 7 gives the deflecting angle of the signal light beam with the increase of the pump fluence when the delay time of the pump pulse is about 60 ps. The result shows that the deflecting angle of the signal light beam is linear to the pump fluence. High deflection sensitivity is often necessary for a light deflector. Therefore, a pump fluence as high as 320 μJ/cm² is applied in the following experiment. Fluence even higher than 320 μJ/cm² causes fast deterioration of the streak image of the signal light on the CCD in our experiment. Possible reasons for this include the serious free carrier absorption effect and the limitation of the CCD dynamic range. From Fig. 7, the maximum deflecting angle is 86.4 mrad. Therefore, with L = 4500 μm and a = 350 μm, the photo-induced refractive index change Δn can be estimated to be ∼0.025 under the pump fluence of 320 μJ/cm². According to Ref. [28], the concentration of the free carrier here is estimated to be 2.5 × 10¹⁸ cm⁻³ based on the photon energy of pump pulse and band structure of GaAs. Correspondingly, the refractive index change is calculated to be 0.016. The differences may be attributed to the uncertainty of the concentration of the photo-induced free carrier. In addition, there is a great difference between the experiment result of 86.4 mrad and the theoretical calculation in Fig. 4, which needs our further efforts to improve the quality of the deflector sample.

Fig. 7.

	Figure Option ViewDownloadNew Window
	Fig. 7. Deflecting angle of signal light beam with the increase of the pump fluence.

Figure 8 depicts the deflecting angle of the signal light beam with different delay time of pump pulse. It can be seen that the signal pulse has no deflection when the time delay is less than zero, for the signal pulse has emitted from the waveguide when the pump pulse works. Afterward, the deflecting angle is linearly increasing with the delay time (or propagating distance along z-axis) until a maximum at last.

Fig. 8.

	Figure Option ViewDownloadNew Window
	Fig. 8. Deflecting angles of the signal light beam with different delay time relative to the pump pulse.

As shown in Figs. 7 and 8, the deflection of the signal light beam is proportional to the pump fluence on the deflector, and is also proportional to the delay time between the pump pulse and signal pulse. Therefore, this ultrafast deflector can operate in two ways. First, we can employ the rising edge of a laser pulse, which is on the timescale of T, as the pump illumination to induce a refractive index profile in the deflector, thus realizing the ultrafast deflection of the signal light beam. Second, as mentioned above, make sure that the ultrashort pump pulse arrives at the deflector as soon as the leading part of the ultrafast signal light begins to exit the deflector, and then the time slices of flight across the duration time of ultrafast signal light just represent the delay time between the pump pulse and the corresponding ultrafast signal elements. Therefore, the temporal elements in this ultrafast signal experience an increasing deflection with the time of flight. Both operational styles of the deflector are recognized as instantaneous as the generating time of non-equilibrium free carrier on a timescale of laser pulse duration time.

Figure 9(a) gives the grey levels of the streak images of the signal light beam with different deflections, which are picked up from Fig. 8 with 10-ps time interval. This 10-ps time interval ensures that the streak can be right distinguished from the neighboring one. Therefore, this ultrafast deflection device has a temporal resolution of 10 ps. The flat tops of the curves in Fig. 9(a) result from the limitation of the dynamic range of CCD. It can be seen that the displacement of the sixth curve, representing the streak intensity distribution at t_d = 50 ps relative to the temporal origin, is 194 μm, which is larger than the half width of the deflector of 175 μm. The reasons for this include the following aspects: (i) the magnification of the focusing lens in the experiment is calculated to be ∼2.2, which is inaccurate; (ii) the center of the incident signal light beam deviates from the width center of the deflector; (iii) the serrated gold mask does not strictly fully cover the upper surface of the bar-type deflector in the sample fabrication process, which directly results in a little larger deflection of the signal light beam than that in the ideal situation; (iv) the synchronization error between the signal light pulse and pump pulse is another possible reason.

Figure 9(b) gives the streak image of the signal light beam corresponding to Fig. 9(a). The width of the streak can be achieved. It also can be seen that the light intensity decreases with the increase of the delay time, which may result from the more serious free carrier absorption when the signal light beam propagates a longer pump-modulated region. Consequently, the streak width of the signal light beam decreases from about 100 μm to 73 μm. Based on the theoretical simulations, angular dispersion and interface scattering contribute to the enlargement of signal beam spot size. In the experiment, we have to raise the signal beam power for the purpose of recognizing the streak image with larger delay time, which exactly results in the over-exposure of the first four streaks at the same power. It is known that the formula about the relationship between deflecting displacement X(z) and the propagating distance z is expressed as X(z) = Δnz²/(2a) for a multiple-prism deflector. It should be noted here that the deflecting displacements seems to be proportional to the delay time or the propagating distances, which is due to the fairly small quadratic coefficient Δn/(2a).

Fig. 9.

	Figure Option ViewDownloadNew Window
	Fig. 9. (a) Grey levels of the streak images of the signal light beam with 10 ps time interval. (b) The streak images of the signal light beam acquired by CCD with different time delay between pump and signal pulse.

Here, we omit the time jitter of laser pulses, for the achieved temporal resolution of 10 ps is much larger than the jitter of femtosecond scale. Theoretical temporal resolution of the deflector can be calculated to be 4.2 ps with Δn = 0.016. With T = 57 ps, the theoretical result about the number of the resolvable spots of the deflector M = 13.6. However, the achieved experimental temporal resolution is 10 ps. In other words, we can only distinguish five signal light streaks and M ≈ 5 in the experiment. In addition to the reasons discussed in Section 2.2, the temporal resolution depends on many other factors such as the quality of the waveguide, the quality of the smoothness of two end faces of the deflector, non-uniform spatial distribution of pump intensity, the beginning synchronization error between the signal pulse and pump pulse, the imperfect prism array, and so on.

We have demonstrated an ultrafast beam deflection system with picosecond, even to sub-picosecond, temporal resolution, a large number of the resolvable spots, high response and repetition speed. The theoretical investigations show that material dispersion of GaAs including group velocity dispersion and angular dispersion, interface reflection, and interface scattering of the deflector, impose harmful influences on the temporal and spatial resolvable performances. Thus by properly designing the structural parameters of the deflector, a sub-picosecond scale of temporal resolution can be achieved. The experimental results show that the temporal resolution is about 10 ps, which is not very satisfying but can be improved a lot as the theoretical discussion section has suggested. In fact, we can choose proper semiconductor materials of waveguide deflector depending on the wavelength of the signal light to be deflected. For example, AlGaAs/GaAs is the proper choice for the near infrared signal light with wavelength larger than 870 nm, whose photon energy does not fall into the absorption range of the waveguide core. In the same way, the use of a wider bandgap semiconductor for the waveguide core would enable operation well into the visible spectrum, such as Al_mGa_{1 −m}As with a proper value of m. If m = 0.37 and then λ_g = 650 nm, the signal light wavelength can be larger than 650 nm. Except for the all-optical solid-state streak camera, this configuration can also be used in the optical communication, light controlling and switching, and so on.