1. Introduction

Crystals of sillenite family such as Bi₁₂SiO₂₀ (BSO), Bi₁₂GeO₂₀ (BGO), and Bi₁₂TiO₂₀ (BTO) are well known as good photoconductors in the visible range of the spectrum. Due to its symmetry (23 - the cubic class without inversion center) sillenite-type crystals also possess the linear electro-optic (Pockels) effect. Both of these features are essential for existence of the photorefractive effect responsible for optical recording in these crystals. Their applications are mainly related with optical data processing and dynamic holographic recording [1] including adaptive holographic interferometry [2,3]. It is also known that such symmetry allows for observation of a linear photogalvanic effect (LPGE) in these crystals [4,5]. This effect manifests itself as generation of electrical current in a homogeneous sample under uniform illumination. Its physical reason is anisotropy of processes of photo-excitation, scattering, and recombination of charge carriers in noncentrosymmetric crystals [4]. Direction of the generated current is exclusively defined by an angle of the linear polarization state in respect to crystallographic axes [4,5]. Since this electric current is related to the momentum relaxation of the excited charge carriers (which is typically of 10⁻¹² – 10⁻¹⁴ s), LPGE has a great potential for applications in ultra-high-speed optoelectronics.

The linear photogalvanic effect was observed in BSO-type crystals more than 30 years ago when a sample of Bi₁₂SiO₂₀ crystal was illuminated by a light beam with slowly modulated linear polarization state [5,6]. However, insufficient attention was paid to investigations of this effect in sillenite crystals (especially in a short-time domain) in spite of the fact that crystals of the other symmetry group 43m (such as GaAs and GaP) were actively studied in a sense of their photogalvanic response to ultra-short light pulses [7–9]. It is worth noting that in the literature the current arising due to the linear photogalvanic effect has also been referred to as a “shift” current [8,9]. Ultrafast induction of the LPGE current (with response time below of 100 fs) was reported in GaAs [8]; in CdS and CdSe [10]; and in multiple quantum wells of GaAs/AlGaAs [11]. In contrast no data concerning the response of sillenite crystals to short laser pulses has been reported to the best of our knowledge.

In this paper we report for the first time the observation of a few-nanoseconds electric pulse generated in a sample of BSO by a laser pulse at wavelengths from 410 to 610 nm. The electric pulse follows the temporal profile of the optical intensity while the directions of the current is changed for the opposite if the input polarization state of the light is rotated by 90°. Observed features allow us to classify the phenomenon as the linear photogalvanic effect.

2. Experimental methodology

2.1. Influence of optical activity

It is known that the density of the photogalvanic current $j_i$ induced in noncentrosymmetric crystals by the polarized light follows the phenomenological expression [4,5]:

Transverse geometry in which the direction of the photogalvanic current flow is orthogonal to the light-beam propagation was used in our study as shown in Fig. 1 . This geometry is more convenient for observation of LPGE current especially when it is induced by light which is strongly absorbed in the crystal. A sample of BSO crystal was cut so that the optically polished faces were ($110$) while two silver electrodes were deposited on the orthogonal ($1\overline{1}0$) faces (see Fig. 1). Sample dimensions are denoted as L, H, and W where the latter is an inter-electrode distance. A linearly polarized light pulse was incident on a polished face of the BSO sample. The polarization state of the incident beam makes an angle θ with the axis [$001$] of the crystal.

 

Fig. 1 Transverse geometry for observation of the linear photogalvanic effect in a sample of the sillenite crystal. Ch1 and Ch2 are two measuring channels of the oscilloscope.

Download Full Size | PDF

For estimations of induced LPGE current we suppose that the intensity profile has the Gaussian shape so that the cross section of the laser beam is circular providing the light spot with diameter of d on the input surface of the sample:

 

Fig. 2 Angle between the polarization plane of incident light and the axis [001] of the BSO sample with L = 1.9 mm for observation of the maximal LPGE current as a function of the light wavelength.

Download Full Size | PDF

It is convenient to simplify Eq. (5) by introducing a dimensionless factor M_S which joint all the factors depending on the material (except for β ₁₄) and geometrical parameters of a particular sample thus completely representing the LPGE current:

The factor M_S is to be calculated for each particular sample in which we are going to estimate the photogalvanic-tensor component β ₁₄. As follows from Eq. (5), in addition to M_S it is enough to measure the current through the sample, the energy of the incident light pulse, and the pulse duration. Note that M_S also depends on the diameter of the incident light spot d because of $Q\left(d,W,H\right)$. Spectral dependence of the M_S factor calculated for BSO sample with dimensions of W = 5.0 mm, H = 6.4 mm, and L = 1.9 mm (which was used in our experiments) and d = 6.5 mm is shown by solid curve in Fig. 3 .

 

Fig. 3 Spectral dependence of the M_S factor for the BSO sample calculated considering wavelength-dependent optimal angle θ ₀(λ) (solid line) and the mean θ ₀ ^av = 22⁰ (squares).

Download Full Size | PDF

Let us estimate how tolerant could be adjustment of the incident polarization angle θ ₀ for measurement of β ₁₄ at the maximal response of the sample. To this end we compare the factor M_S calculated with using the exact value of θ ₀(λ) given by Eq. (6) with that one calculated for an average value of the optimal angle θ ₀ ^av = 22⁰. Spectral dependence of M_S factor calculated with θ ₀ ^av = 22⁰ is shown in Fig. 3 by red squares. As one can seen, the difference in LPGE current expected in two aforementioned cases is smaller than the typical measurement error (≈5%). Therefore, the experimental measurements of spectral dependence of LPGE current in our BSO sample can be carried out in simplified conditions when the incident light polarization angle is fixed.

2.2. Experimental set-up

In our experiments an Optical Parametric Oscillator (OPO) VIBRANT В LD 355-UV with tunable wavelength in the range from 410 to 700 nm was used as a light source. It generates light pulses at the repetition rate from 1 to 10 Hz, which energy varies from 0.1 to 3.5 mJ per pulse depending on the wavelength. Duration of pulses was measured by a photo-receiver Thorlabs DET10A which has the rise time of 1 ns and the fall time of 140 ns. Considering only the rise-up portion of the oscilloscope trace as reliable, we estimated the pulse duration to be about 4 ns (at the level of e ⁻²). The linear polarization state of a light pulse generated by OPO laser was converted into almost perfect circular polarization (with the ellipticity ratio $\le 1.2$) by means of a quarter-wave Fresnel rhomb FRQ as shown in Fig. 4 . A linear polarization state at any required polarization angle θ was selected from the circular polarization by rotation of a polarizer installed after FRQ.

 

Fig. 4 Schematic view of the experimental setup for measurements of LPGE current induced by light pulse illumination: OPO is the optical parametric oscillator; BSO is a Bi₁₂SiO₂₀ crystal of (110) crystallographic cut.

Download Full Size | PDF

The BSO sample cut from the crystal grown in oxygen-free atmosphere was used in this study. The sample orientation is shown in Fig. 1, and its dimensions are L = 1.9 mm, W = 5 mm, and H = 6.4 mm. Two silver electrodes were deposited on the opposite faces of ($1\overline{1}0$). The sample was electrically connected to a digital oscilloscope (with 1 GHz bandwidth) in the differential mode: electric signals produced from each electrode were directed towards two inputs (channels Ch1 and Ch2) of the oscilloscope where they were electrically subtracted one from another. Such a configuration allows us to significantly reduce the influence of electromagnetic interference onto the sensory circuit and increase a signal-to-noise ratio. The oscilloscope was synchronized with OPO laser. Stability of the pulse energy generated by the OPO laser was not very high. To overcome this problem, we monitor the mean energy during each measurement by an additional power meter (not shown in Fig. 4) and simultaneously record to the oscilloscope averaged traces of 100 sequential laser shots.

3. Experimental results

3.1. Polarization dependence

Strong electric-current pulses with peak magnitude up to 500 µA were detected in the BSO sample under its illumination by laser pulses at wavelengths from 410 to 610 nm. Shape of the electric pulses and their sign depends on the input polarization state of the light. Few typical examples of detected current pulses are shown in Fig. 5 . Oscilloscope traces at λ = 520 nm were recorded when the average pulse energy of the incident light was 1.1 mJ (a) and 1.0 mJ (b) while it was 0.74 mJ (c) and 0.68 mJ (d) for the case of λ = 480 nm. Light spot at the input crystal surface has slightly elliptical shape (ellipticity ≈1.2) with the average diameter of 2.9 and 3.7 mm (at the level of e ⁻¹) for light at 520 and 480 nm, respectively. The polarization angle was 14°, 104°, 13°, and 103° for traces a, b, c, and d, respectively.

 

Fig. 5 Oscilloscope traces recorded when BSO sample was illuminated by linearly polarized light pulse at wavelengths of 520 nm (a, b) and 480 nm (c, d). The angle of the polarization plane in respect to the [001] axis was: (a) 14°; (b) 104°; (c) 13°; (d) 103°.

Download Full Size | PDF

As it is clearly seen from Fig. 5, the electric current changes its direction of flow for the opposite when the polarization state of the light pulse is switched for the orthogonal. This feature is the typical characteristics of the linear photogalvanic current which also should follow the temporal shape of the light pulse. However, duration of detected pulses (2 ns) is shorter than that of the laser pulse (4 ns). In addition, the positive pulse reaches its extreme 1 ns later than the negative pulse. It is worth noting that the summarized duration of negative and positive pulses is approximately equal to the laser pulse duration.

According to Eq. (5), LPGE current should become zero if the polarization angle is equal to ${\theta }_0\pm$ 45⁰. However, strong electrical pulses were also observed at these angles as shown in Fig. 6 . In this experiment the pulse energy was almost the same as for the case of Fig. 5 (0.71 mJ at λ = 520 nm and 0.98 mJ at λ = 480 nm). The angle of the linear polarization state in respect to the axis [001] was 59° and 58° for the curves (a) and (b), respectively. Note that in this case the diameter of the illuminated beam (2.9 and 3.7 mm for λ = 520 and 480 nm, respectively) was smaller than the inter-electrode distance of the sample.

 

Fig. 6 Oscilloscope traces recorded under sample illumination by light at wavelengths of 520 nm (a) and 480 nm (b) when the polarization angle corresponds to absence of LPGE current. The diameter of the illuminating beam is 2.9 mm (a) and 3.7 mm (b).

Download Full Size | PDF

When the cross section of the illuminating beam at the input face of the sample was increased up to 13 mm, the shape of the sample response on the linearly polarized light became more regular as shown in Fig. 7 in comparison with that shown in Fig. 5. In this case the temporal shift between extremes for positive and negative pulses is much smaller than for pulses shown in Fig. 5. The traces of Fig. 7 were recorded for the laser pulses at the wavelength of 532 nm with the total energy of 0.75 mJ at the polarization angle of 20° and 100°. Note that in this case the sample area was 4 times smaller than the light-beam area, which results in big losses of the light energy and, consequently, in smaller amplitude of induced electric pulses.

 

Fig. 7 Response of the BSO sample on the linearly polarized light beam with the diameter of 13 mm which exceeds the inter-electrode distance of the sample. The polarization angle in respect to the axis [001] is (a) 20° and (b) 100°.

Download Full Size | PDF

Dependence of the amplitude of induced electric pulses as a function of the polarization angle is shown in Fig. 8 . Considering that the shape of induced electrical pulses has both positive and negative parts, which can be comparable one to another for example at ${\theta }_0\pm$ 45⁰ (see Fig. 6), we plot in Fig. 8 the largest magnitude of the response keeping its sign. The experimental data are shown as red squares. These data were obtained when the light beam at the wavelength of 532 nm with the diameter of 13 mm was incident on the sample. Solid line in Fig. 8 shows the theoretical dependence calculated from Eq. (5) using the following parameters λ = 532 nm, ρ = 38 deg/mm, α = 3.5 cm⁻¹, and L = 1.9 mm, which were measured for our particular sample. As one can see from Fig. 8, the polarization angles at which the LPGE current reaches its maxima and minima are in good agreement with the theoretically predicted values. However, there is some discrepancy in absolute values between theoretical and experimental curves, which will be discussed later.

 

Fig. 8 Amplitude of the generated electric signal as a function of the polarization angle θ measured at the wavelength of 532 nm. The signal is reduced to the equal pulse energy. Blue line is a theoretical curve from Eq. (5), and red squares are the experimental data. The diameter of the illuminating beam is 13 mm.

Download Full Size | PDF

3.2. Circular polarization

To understand the reason of unexpected response of the BSO sample on the linearly polarized light at polarization angles at which LPGE current should be null, we carried out experiments with illumination the crystal by circularly polarized light. As known, sillenite crystals do also possess a circular photogalvanic effect [5,6]. However, direction of the current flow induced by the circular polarization always coincides with the direction of a light propagation. Therefore, no crystal response caused by the circular photogalvanic effect was expected in the transverse geometry used in our experiments. Nevertheless, strong electrical pulses of the amplitude comparable with that under illumination by linearly polarized beam (Fig. 7) were detected as one can see in Fig. 9 . The conditions of the experiment were similar to those which were used for measuring the data shown in Figs. 7 and 8: the wavelength was 532 nm; the spot diameter was 13 mm; but the total energy of the incident light pulse was doubled (1.5 mJ) because the polarizer was removed. The ellipticity of the light-beam polarization state was not very high (smaller than 1.2). The cross-section of the illuminated beam was also closed to the circular (with ellipticity ≈1.2).

 

Fig. 9 Oscilloscope traces recorded when the sample was illuminated by the circularly polarized light at the wavelength of 532 nm. The diameter of the illuminating beam is 13 mm.

Download Full Size | PDF

3.3. Spectral dependence

In our study we also measured the spectral dependence of the electric current induced in the BSO sample by nanosecond light pulses. For this end the wavelength of light generated by OPO laser was changed in the range from 410 nm to 610 nm (the line width was 0.3 nm for each generated wavelength) with the step of 10 nm. For every wavelength the incident polarization state was set either at the angles of θ ₀ or of θ ₀ + 90⁰ calculated from Eq. (6). For these polarization angles the induced electric current flows in opposite directions as it is seen in Fig. 5 for wavelengths of 520 and 480 nm reaching either positive or negative extreme. For each wavelength we measured the mean energy of the incident light pulse simultaneously with recording of the average response of the BSO sample. The size of the light beam was varying with the wavelength being minimal at 530 nm (2.9 mm) and increasing in both sides to 4.5 mm at 410 nm and to 3.3 mm at 610 nm. In this experiment small beam size was used because it leads to higher signal-to-noise ratio which is especially desirable for detecting signal at blue and red wavelengths where the output power of the laser is not high. All these data together with measured spectral dispersion of the optical absorption and the literature data of the optical rotatory power of BSO crystals [12,14] allows us to calculate the component β ₁₄ of the photogalvanic tensor by using Eq. (5). Thereafter, the Glass constant G = β ₁₄/α is also calculated. Since the amplitude of induced electric pulses is affected by the non-photogalvanic current which is added to the negative response and subtracted from the positive one (see Fig. 9), it is more realistic to present an average amplitude of the negative and positive pulses as it is done in Fig. 10 .

 

Fig. 10 Spectral dependence of the component β ₁₄ of the photogalvanic tensor (a) and the Glass coefficient G (b) measured for the sample BSO-14.

Download Full Size | PDF

As one can see, the LPGE current is detected in BSO crystal almost in the whole visible diapason of the spectrum. Note that the component β ₁₄ has the meaning of the generated electric current reduced to the same light intensity: ${\beta }_{14}\propto J/I$ (see Eq. (5)). Increase of the LPGE current at short wavelengths (Fig. 10a) is related with increase of the absorption coefficient. Since the Glass coefficient G (in contrast to β ₁₄) is reduced to the same light absorption (G = β ₁₄/α), it represents a photogalvanic sensitivity of the crystal. As one can see from Fig. 10b, the maximal PG sensitivity in the BSO sample is observed at green light (in the vicinity of 520 nm). The Glass coefficient of our crystal (which was grown in oxygen-free atmosphere) is much larger than that of typical Bi₁₂SiO₂₀ crystals (0.5⋅10⁻⁹ cm/V [6], ) but it is smaller than for BSO crystals annealed in vacuum (50⋅10⁻⁹ cm/V at λ = 476 nm [15], ).

4. Discussion

As seen from Figs. 5 and 7, the pulse of induced electric current changes its direction of flow for the opposite after the input polarization state of the laser pulse is rotated by 90°. This is very specific feature which does not belong to the most of known photoelectric phenomena. There are only three photoelectric effects in which the direction of generated electric current depends on the polarization state of the incident light: optical rectification (OR) [16], linear photogalvanic effect (LPGE) [4,5], and optical orientation of dipolar centers (OODC) [17,18]. In sillenite crystals all these effects are described by the same third-rank tensor of Eq. (2). Consequently, their polarization dependencies should be similar like it is given by Eq. (5) and shown in Fig. 8. However, the shape of the electric current generated by the laser pulse owing to these effects is different one from another. The current due to OR effect is proportional to the time derivation of the light pulse envelope [16]. In contrast, the LPGE current repeats the shape of the laser pulse [4]. The situation with the OODC effect is more complicated: for laser pulses which are longer than the characteristic relaxation time, the current is proportional to the time derivation of the light pulse but for pulses much shorter than the relaxation time, the electric current follows the shape of the light pulse during its duration [17]. The OODC effect was recently observed in the same BSO sample, and corresponding relaxation time was found to be 3.5 µs at the light intensity of 0.26 W/cm² [18]. Much higher intensity was used in our experiments (≈1 MW/cm²) so that the relaxation time of dipolar centers at such high intensity can be much shorter but it is hardly believable if it can be smaller by three orders of the magnitude because of the nonlinear dependence of the relaxation time on the light intensity (see Eq. (8) from Ref [18].). Therefore, the OODC effect can also result in the electrical current repeating the shape of the laser pulse.

Estimation of the OR effect in BSO crystals shows that it should be at least 3-4 orders of value smaller than the photogalvanic effect for nanosecond laser pulses. The optical rectification is not related with real electron transitions. Therefore, it is hardly possible that OR effect can possess any spectral features. However, as one can see from Fig. 10, the spectral dependence of the observed current possesses clearly visible spectral features at the wavelengths of 440, 480, 515, 550, and 580 nm. Moreover, the spectral features observed in our particular BSO sample (see Fig. 10b) are almost coincides with the features (430, 465, 500, and 540 nm) observed in spectral dependence of the LPGE current measured in vacuum-reduced Bi₁₂SiO₂₀ crystals using light which polarization state is modulated at low frequency (< 100 Hz) [19]. Similarity of both spectra should be considered as an impressive argument in favor of the photogalvanic origin of observed current pulses taking also into account that the OODC-current at slow modulation is much lower than the LPGE current [18]. Consequently, we can conclude with a high degree of probability that electrical pulses induced in a Bi₁₂SiO₂₀ crystal grown in the oxygen-free atmosphere by properly polarized nanosecond laser pulses are caused by the linear photogalvanic effect.

Temporal shape of the observed electric pulses could be explained if we consider that (i) a photocurrent independent from the light polarization may be induced in the sample owing to non-uniform illumination and/or sample inhomogenities, and (ii) electrical coupling of the sample with external circuit is capacitive rather than resistive.

Crystal illumination with absolutely uniform light intensity is hardly achievable. Even for the expanded beam of 13 mm in diameter, the intensity variations on the sample were up to 30%. Consequently, diffusion current J_dif should be induced due to non-uniform generation of photoexcited charge carriers. Evidently, this current does not depend on the light polarization. However, non-uniformity of illumination is not enough for explanation of all features observed in our experiment. In the case of symmetrical location of the laser spot in respect to the electrodes, induced diffusion current from both sides should compensate each other. Nevertheless, we did not find such a location in our experiment when the sample was illuminated by linearly polarized light at the angle at which LPGE should be zero. This can be explained by inhomogemeity of sample parameters relevant to photo-excitation of charge carriers especially along the direction orthogonal to the sample faces with electrodes. As known [20], variations of the photoconductivity are typically observed in ingots of BSO crystals grown by the top-seeded solution-growth method. These variations relate to gradients of the optical absorption and trap density of the crystal being mainly directed along any radius of the crystal ingot. Irrespective of particular pulling direction (either [100] or [110] axis is usually used for sillenite-crystal growth) a sample cut similar to ours should manifest the non-uniformity along the [$1\overline{1}0$] crystallographic axis which coincides with the direction of the current flow in our experiments.

In the case of partial illumination of the sample area, capacitive type of electrical coupling between BSO sample and external circuit is evident. However, the coupling is of the same type even in the case of sample illumination by expanded light beam (13 mm in diameter, for example). The reason is non-ohmic electric contacts of sillenite crystals due to surface barriers [21]. Height of the barrier in vacuum-annealed BSO crystals (which are very similar to our sample from the point of view of the photoconductivity) is higher than 0.5 eV [21]. Therefore, density of the saturation current of reverse-biased contact in such a sample is about of 10⁻² Acm⁻² which is close to the current densities observed in our experiments. Consequently, application of several tens of mV is needed to provide conductive current through both contacts. These values are much higher than the pick voltages observed in experiments with expanded laser beam (Fig. 7) so the capacity of the contacts rather than their resistance determines the current flow out of the crystal.

In the case of capacitive coupling, an external current J_ex is a result of charging (discharging) of the coupling capacitor C_C due to voltage V_in(t) induced by a photocurrent J_ph generated in the illuminated domain of the sample. In terms of common electronics, the illuminated domain can be considered as a current source placed in parallel to a RC-circuit with equivalent resistance R_ph of the illuminated part of the sample and equivalent capacitor C_ph of the same part. As known, temporal shape of voltage V_in(t) generated in such a circuit depends on the relation between the duration of the current pulse J_ph(t) and the product of $R_{ph}{C}_{ph}$: the voltage closely follows the shape of the current if ${\tau }_p>>R_{ph}{C}_{ph}$ but it is proportional to the integral of the current when ${\tau }_p<<R_{ph}{C}_{ph}$. In its turn, the temporal shape of the external current J_ex(t) depends on the particular value of the product of $R_{osc}{C}_C$ where $R_{osc}$is the input resistance of the oscilloscope: it repeats the shape of V_in(t) when ${\tau }_p<<R_{osc}{C}_C$ but it approaches to the derivative $d{V}_{in}\left(t\right)/dt$ with diminishing of $R_{osc}{C}_C$ product.

The product of $R_{ph}{C}_{ph}$ is nothing else as the Maxwell relaxation time τ_M of the illuminated part of the crystal, which depends on the light intensity. Numerical estimations for BSO crystal show that τ_M is about 0.1 ns for the laser pulse energy of 1 mJ at the wavelengths in the blue-green region of the spectrum and the laser beam diameter of 3 mm. The coupling capacitor C_C is estimated as 5 pF for this diameter of the laser beam, which leads to the product of $R_{osc}{C}_C$to be smaller than 1 ns. With these parameters of the electric circuit one can expect that V_in(t) induced by the LPGE current almost repeats its temporal shape but J_ex(t) resembles the derivation $d{V}_{in}\left(t\right)/dt$. Therefore, the response registered by the oscilloscope is a two-lobe pulse with both positive and negative parts as it is shown in Fig. 6. In the case of the expanded beam (d = 13 mm) the relaxation time τ_M increases more than 20 times, which leads to V_in(t) being proportional to an integral of the induced current pulse. After differentiation in external circuit, the external current J_ex(t) reproduces the temporal shape of the LPGE current. Influence of the Maxwell relaxation time on the shape of the registered electric pulse is also supported by the data shown in Fig. 11 .

 

Fig. 11 Response of the BSO sample on the illumination at different wavelengths by linearly polarized light pulses at the polarization angle corresponding to the maximal negative electric current: (a) λ = 600 nm, E ₀ = 0.31 mJ, θ ₀ = 112°; (b) λ = 590 nm, E ₀ = 0.39 mJ, θ ₀ = 112°; (c) λ = 520 nm, E ₀ = 0.68 mJ, θ ₀ = 104°; and (d) λ = 480 nm, E ₀ = 1.01 mJ, θ ₀ = 103°. The diameter of the illuminating beam is 3 mm.

Download Full Size | PDF

It is seen from Fig. 11 that at wavelengths of 470-540 nm the negative peak of the electric current is reached earlier than at wavelengths of 590-610 nm. The reason of this difference is smaller photoconductivity in the latter region because of both smaller pulse energy and smaller quantum efficiency of charge excitation. Photoconductivity diminishing results in increasing of τ_M. It is easy to show from the above consideration of V_in(t) formation that the position of the external-current-peak extreme is inversely proportional to τ_M, which explains the results shown in Fig. 11.

Situation with the registered response in the case of the diffusion current is more complicated because J_dif can be not zero even after finishing of the laser pulse end: it slowly diminishes with the characteristic time of photoconductivity decay. Therefore, the kinetics of photocurrent stimulated by nanosecond laser pulses is to be known. Observation of the time evolution of photoconductivity excited in a BSO crystal by picosecond light pulses have shown a fast decay of photocurrent (smaller than 1 ns) followed by a slower one (with time constants ranging from 30 ns to 12 μs) [22]. Supposing that our BSO sample has similar photoconductivity kinetics, the diffusion current is a sequence of fast pulse and slow diminishing tail with the fall time of the fast pulse being as short as 1 ns. Consequently, all conclusions made for the external response induced by the LPGE current are also valid for the diffusion current. Therefore, the temporal shape of the sample response on the linearly polarized laser light should be considered as a mixture of the LPGE- and the photo-diffusion current pulses with both pulses having proper two-lobe shape at high enough level of the sample illumination.

Another mechanism should be mentioned, which can affect both the time evolution and magnitudes of the photoinduced current pulses. The capacitor charging in the above mentioned scheme is nothing else but creation a space charge (SC) due to spatial non-iniformity of LPGE- and/or photo-diffusion currents. The process of space-charge formation is used for hologram recording in photorefractive crystals, including sillenite crystals [1]. In particular, recording of holograms in BSO crystals by nanosecond and picosecond laser pulses was experimentally demonstrated [22–24]. The space-charge electric fields created by a sequence of initial laser pulses (which is used for observation of the sample response in majority of our experiments) can be stored, at least partially, up to arrival of the successive pulse. In this case a drift current due to stored SC-fields and the photoconductivity induced by new arriving laser pulse will be generated with the sign of this current being opposite to the signs of the LPGE and/or diffusion currents. In the beginning of the laser pulse the drift current can temporarily exceed the LPGE current if dependencies of the sample photoconductivity and LPGE on the light intensity are different.

There is one more factor affecting the shape of the electrical pulse, which is the unmatched impedance of the BSO sample with the electric cable connecting the sample with the oscilloscope. Its influence is defined by both the RC-product of the sample and parameters of the external circuit. Unmatched impedance results in damped oscillations which are observable after finishing of the light pulse as it is seen, for example, for all traces of Fig. 11 and Fig. 5b.

5. Conclusion

In this work, for the first time, the linear photogalvanic effect is observed in bismuth silicon oxide crystal illuminated by nanosecond light pulses from the visible part of the spectrum (410 – 610 nm). We have found that the electric signal induced by a short laser pulse consists of two components: one of them depends on the polarization state of the light pulse but another does not. The latter component can be diminished by increasing of uniformity of the illumination and decreasing inhomogeneities of the crystal. It was shown for the former component that the direction of the current electric current is changed for the opposite if the linear polarization state of the laser pulse is changed for orthogonal. Spectral features of the BSO-sample response on the short laser pulses almost coincide with these observed under continuous illumination with polarization modulation at low frequency. Obtained results show that sillenite crystals are very prospective for development on their base different devices of ultra-fast optoelectronics.