1. Introduction

As a consequence of the continuous increase of the peak power and the achievable peak intensity of short-pulse solid-state lasers, the improvement of their spatio-temporal quality plays an ever increasing role in high intensity physics and advanced laser technology. Recently, a great amount of experiments confirmed that one key aspect in performing high-intensity laser-matter interactions is the temporal and spatial contrast of the pulses [1–3]. These experiments showed that pre-pulses and amplified spontaneous emission (ASE) of 10⁷–10⁸ W/cm² intensity can considerably influence laser-matter interactions by modifying material properties prior to the arrival of the main pulse [4,5]. Considering that the focused intensity of the present laser systems surpasses 10²² W/cm², and ongoing projects are targeting 10²⁵ W/cm², the necessary intensity contrast is in the 10¹²-10¹⁷ regime. Several nonlinear optical effects have been successfully applied for the improvement of the temporal contrast of high-power systems such as the cross-polarized wave generation (XPW) or nonlinear ellipse rotation (NER) [6–10]. These techniques possess an instantaneous response as they rely on bound-electron dynamics, however, they are only applicable at moderate light intensities. Therefore, they are ideal for decoupling the noise of subsequent amplifier stages in double-Chirped Pulse Amplification (CPA) configurations, but unsuited for final contrast improvement at the output of the system. Up to now the only way of cleaning the high-energy output of such systems has been the application of plasma mirrors (PMs) [11,12]. They can usually enhance the contrast by a factor of hundred, while the best demonstrated performance reached a suppression of 10⁴ at a transmission of 70% [13]. However, it is also known that PMs are less efficient in suppressing the low-intensity coherent pedestal in the close (a few ps) proximity of the pulse which might still disturb sensitive measurements. Recently, we introduced an alternative approach called nonlinear Fourier filtering (NFF) suited for improving the contrast of high-intensity beams of short-pulse ultraviolet (UV) excimer lasers [14,15]. In this work we explore for the first time the applicability of the NFF technique also for the mainstream solid-state CPA systems. Although the ionization dynamics considerably differs in case of the two lasers due to the shorter pulse duration and the three-times longer wavelength, we find that the NFF technique works also well for the near-IR CPA lasers. Remarkably, our measurements reveal that the switching speed of NFF is fast enough for efficient pedestal suppression even less than a picosecond before the arrival of the main pulse.

2. Nonlinear Fourier filtering

In the nonlinear Fourier-filtering technique the separation of the intense main pulse from the noise is based on a spatially selective nonlinear phase shift in the Fourier-plane containing the angular spectrum of the beam, which results in a directional modulation of the beam [14,15]. The experimental realization (see Fig. 1) uses a telescope and a pair of conjugate masks, where the output aperture is placed in the image plane of the input mask. Assuming ideal imaging where an opaque central part of the input mask is imaged onto the central hole of the output aperture – within the frames of geometrical optics – the arrangement completely blocks low-intensity beams linearly propagating through the system. However, for the intense main pulse, the self-generated plasma introduces highly nonlinear phase-modulation across the focal distribution. The phase shift of the high-intensity part of the focal distribution deflects that part of the beam onto the free aperture of the output mask resulting in a significant throughput. In this way, the temporally separated noise becomes spatially separable from the main pulse [14,15].

 

Fig. 1. Schematics of the nonlinear Fourier-filter (based on [15,17]). In the present experiment the following parameters were used: d₁: 7 mm, d: 15 mm, D: 45 mm, t: 2350 mm and k: 1750mm. E = 30 mJ, E1 = 6 mJ.

Download Full Size | PDF

The presented technique relies on the optical Fourier transformation property of lenses, as described e.g. in [16]. In the image system shown in Fig. 1., the L lens performs a Fourier transformation of the input beam at its focal plane, from that the output can be obtained by Fourier back-transformation by a collimation optics. In practical cases it rarely gives back the original spatial distribution of the input. More detailed study of this phenomenon – based on the spatial frequency analysis of imaging [16] – showed that even in the low intensity case finite background emerges in the dark side of the image due to the diffraction of light. The level of this background (related to the spatial contrast of imaging) imposes the primary limitation upon the maximum temporal contrast improvement which is mainly determined by the spatial frequency spectrum of the input mask and by the Optical Transfer Function (OTF) of imaging [16,17]. OTF is dependent on the quality (F-number) of the image system as well as on the spatial coherence properties of illumination [16].

The basic idea of the Fourier filter is that the central lobe of the focused annular beam in the Fourier plane gains a π phase shift due to the propagation through the self-generated plasma, whereas the sidebands are not affected. Due to the subsequent constructive interference this leads to the creation of some kind of an “inverse” annular beam and to significant diffraction of the energy into the central part of the output aperture [15]. In case of homogeneous illumination of the input mask, by optimization of its d₁/d ratio as high as 56% of the energy of the input beam can be diffracted into the central hole of the output aperture [15] (see Fig. 1.).

For a practical F ≈ 20 image system the spatial contrast of imaging limits the achievable temporal contrast improvement of NFF to ∼10³ [15]. This can be significantly improved (up to 10¹⁰) by the use of an apodized object, whose spatial frequency spectrum is properly matched to the capabilities of imaging. This has been realized in our former experiments by introducing an imaging of low numerical aperture (NA) prior to Fourier-filtering (called pre-imaging) [17]. Due to the “direct amplification scheme” used in short-pulse excimer systems, both the NFF and the pre-imaging can be made as an integral part of the amplifier chain. Integrating this “extended” NFF arrangement into our high-brightness UV excimer laser [18], excellent intensity contrast up to 10¹² was achieved for the output, moreover the ASE at the output of the system was measured to be generated exclusively by the amplifier chain following the temporal filter [18]. With the optimization of the parameters/operational conditions of the amplifier(s) following the NFF, temporal contrasts in excess of 10¹² are calculated to be achievable [19].

3. Experiments

In a “proof of principle” experiment the basic process of NFF; the intensity-dependent diffraction of an IR beam has been demonstrated using an experimental apparatus shown in Fig. 1. In this arrangement the spatial contrast of the imaging does not exceed 10³, because the image system – due to its limited NA capabilities – only partially could manage the rays diffracted on the high spatial frequency components of the input mask. Correspondingly, it limited the maximal achievable temporal contrast enhancement of the correlation-measurements [16,17]. The input beam from a Coherent Hidra Ti:Sapphire laser – carrying E ≈ 30 mJ energy in a 40 fs pulse – was expanded in order to get behind the input aperture of the NFF arrangement a “nearly flat-topped” annular beam with an inner and outer diameter of d₁ = 7 mm and d = 15 mm, respectively. Due to this geometrical attenuation the pulse energy behind the input aperture was E₁ = 6 mJ, which could be further decreased by a variable attenuator. The 1:0.75 imaging system of NFF (k = 1750 mm, t = 2350 mm) was formed by an f = 1000 mm, plano-convex Fused Silica lens of D = 45 mm diameter (marked by L in Fig. 1.). Regarding the size of the annular beam, this corresponds to an effective F-number of F_eff = f/d ≈ 67. However, for the creation of a sharp image of the input mask (where the rays of higher diffraction orders have significant contribution to the sharpness and background noise of the image) the available F-number offered by the full aperture of the image system was F = f/D ≈22. A pulsed gas jet was positioned into the focal plane. The commercial Parker series 9 valve has a straight, cylindrical 1 mm diameter orifice. A supplemental converging nozzle with an exit of 0.7 mm diameter was used to create higher gas density. Similar nozzles were characterized in [20]. Densities in the 1-5·10¹⁸cm^-3 range were measured by x-ray technique previously in [21] and the density profiles were obtained using Abel inversion. The delay between valve opening and the laser pulse, and the distance from the nozzle were varied during the experiments to optimize the effect. Various noble gases were applied, and the phase shift could be optimized either varying the energy of the laser, or by varying the backing pressure. As a result, we experimentally observed the beam profile switching due to the controlled selective phase shift and the corresponding dynamic directional modulation of a TW-class Ti:Sapphire laser pulse, similar to former experiments with a UV excimer laser. In this measurement the spatial distribution of the output was monitored by removing the output aperture. The output plane was imaged onto the sensor of a CCD camera using Fresnel reflections of uncoated wedges for proper attenuation. The beam profiles obtained at the output plane are displayed in Fig. 2 for increasing energy of the annular beam in the range of 0.4-6 mJ. In this experiment we used argon gas of 1 bar backing pressure before the pulsed gas valve.

 

Fig. 2. Output spatial distributions of NFF with increasing phase delay (by increasing the energy of the annular beam from 0.4 to 6 mJ) with the use of Ar gas of 1 bar backing pressure.

Download Full Size | PDF

It is seen from the pictures, that the intensity dependent diffraction of the beam is successfully demonstrated in the IR; significant part of the energy carried by the input annular beam is diffracted into the central part of the output. In this first series of measurements the maximum diffraction efficiency of only ∼20-25% was achieved. Here the diffraction efficiency is defined as the energy of the output beam diffracted into the central hole of the output aperture divided by the energy of the annular beam excluding the losses associated with the Fresnel reflections (see later). This is considerably less than the theoretical value of 56% and the experimentally demonstrated 40-45% achieved in the UV at 248 nm [15]. The maximum size of the output aperture was chosen to be somewhat smaller than the calculated ∼5.2 mm inner diameter of the image of the input filter, in order to be completely in the “shadow-part” of the low-intensity image.

Autocorrelation measurements were performed by a 3^rd order cross-correlator (Sequoia, version 5.0) capable of capturing >10⁸ dynamic range and <120 fs temporal resolution. In the measurements of the temporally filtered pulse displayed in Fig. 3(a). and 3b. the output aperture of the NFF arrangement was set to 5 and 4 mm diameter, respectively. Note, that in case of smaller output aperture sizes the reduction of the available cross-section further decreased the overall throughput of NFF (from ∼25% to ∼15%). The beam passing through the aperture was collimated by an additional lens, whose focal length was chosen to have similar fluence on it, as on the focusing optics. During the autocorrelation measurement the delay step size was 17 fs which is finer than the 120 fs temporal resolution of the device and the signal was averaged for three consecutive shots at each delay value. Reference measurements of unfiltered pulses were performed by removing the input mask and not operating the magnetic gas valve. In this measurement the size of the output aperture was chosen to transmit the same energy as the annular beam at the position of the L lens. In this way the Fresnel reflection losses of the numerous components of the NFF system were not included in the calculation of the throughput, since this kind of loss can be avoided by AR coatings. A typical autocorrelation curve-pair (the blue curve is the unfiltered input, the red one is the temporally filtered pulse) is shown in Fig. 3(a). with the use of a 5 mm diameter output filter.

 

Fig. 3. Autocorrelation curves of the input (blue) and the temporally filtered pulses (red) obtained with 5 mm (a) and 4 mm (b) diameter output filters.

Download Full Size | PDF

Clearly, the coherent noise of picosecond duration prior to the main pulse is effectively filtered out; it is only the highest spike in the nearest vicinity of the main pulse, which remained above the ∼10⁻⁷ detection limit of the actual measurement. For this spike the pre-pulse suppression of NFF exceeds 100, which is in good agreement with the predictions of our calculations based on spatial frequency analysis of imaging [16,17], and with our former experimental results obtained in the UV with an image system of similar parameters [15]. The present investigations also confirmed our former theoretical results and experimental observations (see Fig. 2 and 5 in [17]), that the contrast enhancement of NFF is primarily determined by the spatial noise level of imaging of NFF [16,17]. This is known not only dependent on the NA (or F-number) of the image system, but it has also a spatial dependence inside the central hole of the output filter as well (worse contrast is present in the outer part) [17]. For this reason, better pre-pulse suppression is expected by reducing the size of the output aperture. This dependence can be followed on the autocorrelation curves of Fig. 3(b)., made with the use of a 4 mm diameter output filter. Here the suppression rate is increased up to ∼10³, however at the expense of the overall throughput of the filter, which dropped to ∼15%. This observation shows that the maximum contrast improvement of NFF even in this case is mainly limited by the spatial quality of imaging.

We note that in Fig. 3 only the reference measurement is normalized to one, while the measurement of the filtered pulse is normalized to the experimental value of the measured throughput of the NFF. In this way both the time-dependent suppression and the attenuation of NFF can be followed, the contrast improvement is the ratio of the pre-pulse suppression and the throughput for the main pulse of NFF. This kind of separation of the two parameters is based on (or intended by) our assumption, that these parameters are determined by independent processes; the first is by the spatial noise of imaging, the second is by the diffraction efficiency of NFF into the central hole of the actual output filter of a given diameter.

4. Discussion

In solid-state (e.g. in Ti:Sapphire) lasers besides the ASE, a coherent temporal pedestal associated with the CPA scheme – superimposed on the flat, several nanosecond long, ASE-related background – forms a ps (from several to 100 ps) triangular pedestal approaching the 10⁻³-10⁻⁵ relative intensity background level. As the ASE sensitively depends on the actual gain of the amplifiers and also on the coupling between them, while the coherent part including sharp peaks are induced by parasitic reflections or non-perfect chirp management of the CPA system, the actual shape of the temporal pedestal is a unique characteristic feature of the light sources.

It is known that the contrast improvement in a single-stage PM determined by the ratio of the plasma and cold reflectivities is moderate; in many cases does not exceed two orders of magnitude considerably, however, it can approach 3 and even 4 orders of magnitude with the use of special AR coatings [13,22]. This is sufficient to suppress the ASE background and the eventual pre-pulses down (or close) to the presently available detection limit [23–25]. On the other hand, multi-photon ionization of solid dielectrics may ignite the fast formation of the reflecting electron plasma in a rather broad intensity range (from a few TW/cm² up to 1000 TW/cm²) where the ionization rate exponentially increases (see e.g. Figure 2 in [26]). Consequently, ionization occurs in PMs in a broad dynamic range, thus allowing a part of the low-intensity pedestal to contribute to the build-up of the reflecting plasma. Therefore, the actual shape of the pedestal can have an influence on the switching time as well as on the density gradient of the plasma interacting with the main pulse, which can deteriorate the reflectivity down to 30-40%. Under ideal conditions with controlled plasma build-up the reflectivity can reach up to 96% [27]. Furthermore, in case of PMs spatial and temporal integrating effects might also play a role in the switching mechanism. As the mirror surface is generally located outside the focal plane, slight inhomogeneities in the spot distribution can further smear out the switching point on the intensity scale (see [28–30]). The detection of the optimal time of the trigger point is also not obvious without performing an autocorrelation measurement making the setting time consuming. These difficulties altogether frequently lead to too early switching and thus to an incomplete removal of the coherent temporal pedestal in the close vicinity of the main pulse, as pointed out in [31–33].

As far as the possible use of NFF for IR CPA systems is concerned, one has to consider that the dominant (most intense) part of the noise of CPA systems is coherent (in contrast to the incoherent ASE in the UV). For this reason the use of NFF for such systems is straightforward, due to the most pronounced spatial contrast capabilities of imaging for coherent illumination [16,17]. At the same time, in the IR the spatial resolution is lower for the same F-number of the imaging system due to the longer wavelength, resulting in a more blurred image of the edges of the input mask diffracting more background noise onto the output aperture of the system.

Unlike PMs – which rely on reflection off a dense solid plasma at critical density – the NFF technique relies on an intensity-dependent phase shift of the order of π accumulated during propagation through a dilute gas plasma with a density orders of magnitude below the critical density. The temporally filtered pulse passing through the central hole of the output aperture is the result of an interference of the outer rings of the focal distribution and that part of the central lobe which experience defocusing due to the intensity-dependent phase shift. However, due to the low angular frequency of diffraction associated with the central lobe, changes in its shape or size induces a very smooth modulation in the back-transformed plane, resulting in negligible modification of the distribution on the opening of the output aperture and have practically no effect on the focusability of the transmitted beam. Furthermore, the beam is transmitted through the arrangement, which is much less sensitive to inhomogeneities than reflection off a plasma surface. All these effects lead to an effective spatial filtering and thus to a virtually distortionless transmitted beam, as was already observed in the UV [15,17]. In those measurements no effect of the plasma defocusing was observable as well. Furthermore, photoionization in noble gases is much more sensitive to the light intensity than in solids. For example, in the case of krypton gas the intensity range where the photoionization rate changes 12 orders of magnitude spans only a factor of ∼50 (4-200 TW/cm²). This results in a ∼20 times steeper switching dynamics compared to fused silica greatly reducing the influence of the temporal pedestal to the switching process. This is also supported by the different intensity-dependent switching dynamics of PM and NFF. Based on our available measurement data for PMs, the tangent of the slope of the throughput-intensity curves is around 1 (on a logarithmic scale), while in our former studies ∼3 is obtained for NFF [17] allowing faster switching. The fast switching manifests in Fig. 4., where the change in the wave number Δk is displayed in function of the intensity in three noble gases of 1 mbar pressure for 500 fs pulses of a KrF excimer laser at 248 nm used in our former experiments [14,15,17,18] as compared with 50 fs long Ti:Sapphire laser pulses centred at 800 nm utilized in the present work. The required π phase shift for optimally switching the transmission can be obtained by integrating Δk along the propagation.

 

Fig. 4. Intensity-dependent development of Δk in Ar (a, d), Kr (b, e) and Xe (c, f) gases of 1 mbar pressure. The upper row displays the curves on a strongly magnified scale revealing the contribution of the optical Kerr-effect.

Download Full Size | PDF

There are several mechanisms which can change the refractive index, thus contributing to the phase shift of a high-intensity beam in a noble gas:, where ω denotes the central angular frequency of the light pulse and c is the speed of light in vacuum. The two most dominant effects in the intensity range relevant to our experiments are the optical Kerr-effect and photoionization-induced free-electron plasma. The Kerr-effect caused by light-induced anharmonic oscillations of bound electrons [34] causes an instantaneous refractive index shift which is linearly proportional to the light intensity: $\mathrm{\Delta }n = {n_2}I$, where n₂ denotes the nonlinear refractive index. The influence of the Kerr-effect can only be made visible at lower intensities by magnifying the vertical scale by two orders of magnitude as shown in the upper insets (Fig. 4(d), (e), (f)). We note that a linear increase of the wave number due to the Kerr effect appears in Fig. 4 as an exponential shape on the semi-logarithmic scale. For the calculations in the IR the precise values of [35] were utilized, in the UV due to the lack of reliable data we estimated the effect of dispersion by multiplying the IR values by a factor of three based on the results obtained for n₂ of argon in the deep UV [36]. At higher intensities photoionization sets on causing a rapid decrease of the wave number. The index of refraction of a dilute plasma can be approximated by $n = 1 - \frac{{\omega _p^2}}{{2{\omega ^2}}}$, where ${\omega _p}$ denotes the plasma frequency which is linearly proportional to the square root of the free-electron density. Therefore, the formation of ionized plasma decreases the refractive index by $\mathrm{\Delta }n ={-} \frac{{\omega _p^2}}{{2{\omega ^2}}}$. The photoelectron density was calculated according to the Keldysh-PPT theory [37,38].

In the herewith presented experiments ionization-initiated phase shift is appropriate for NFF, where the necessary, π phase shift is introduced to the central, most intense lobe of the focused radiation during the propagation in a plasma of length L. This condition can easily be defined by the $\frac{\lambda }{L} = \frac{{\omega _p^2}}{{{\omega ^2}}} = \frac{{{n_e}}}{{{n_c}}}$ equation, as can be expressed either with the ratio of the frequencies or of the n_e electron density and the n_c critical density of the plasma.

The evolution of Δk strongly resembles an ideal step-like function, whose switching point is different for the two considered wavelengths and for the three gases. The main difference between the – otherwise similar – curves is that the build-up of ionization is slightly shifted towards smaller intensities when switching from Ar to Kr and to Xe. The amplitude of the change of the Δk function is also dependent on the wavelength of radiation (three times larger in the IR) and proportional to the gas pressure. The ionization process according to the intuitive multi-photon description requires ∼10-12 photons in the IR but only 3-4 in the UV. This is the reason why Δk is considerably steeper for the IR pulses than for the UV. This, however, makes the alignment of the necessary π phase shift (by the control of the intensity of the central lobe in the Fourier plane) even more sensitive compared to the UV case. This pronounced sensitivity and the possibility of non-optimum phase shift could be one of the reasons of the limited diffraction efficiency observed in our IR experiments. Note, that a non-optimum relative-phase shift could be resulted on both sides of the switching slope of the Δk: at too low intensities the phase shift introduced to the central lobe is too low, however, if the intensity is too high, even the side-lobes of the diffraction pattern start to ionize, tending to decrease the phase difference between the central and side lobes. This reduces the proper operation range of the intensity to approximately one order of magnitude. On the one hand, this requires precise alignment of the switching point of NFF, on the other hand it is an inherent feature of nonlinear devices of fast switching characteristics. What makes the setting practically easy is that the switching point is automatically placed onto the leading edge of the main pulse corresponding to optimal for contrast enhancement, if the output beam distribution switches from annular to flat-topped and the throughput reaches its maximum. This requires only a simple visual check and energy measurement. The excellent switching performance of NFF is supported by Fig. 5., where the enlarged central portion of the autocorrelation curves of Fig. 3(b). is shown both on logarithmic and on linear scales. Keeping in mind that the actual temporal resolution of the autocorrelator is ∼120 fs, only qualitative statements can be derived for the temporal features of the 40 fs pulse. However, it is clearly seen on the figure, that the steepest part of the autocorrelation curve of the filtered pulse is that portion, where switching of NFF takes place. Moreover, the filtering procedure removed some part of the leading edge of the main pulse, seemingly leading to even shorter pulse duration.

 

Fig. 5. Enlarged central temporal portion of the autocorrelation curves of Fig. 3(b). both on logarithmic (a) and linear intensity scale (b).

Download Full Size | PDF

Important to mention, that in the present (and in all of our former) studies the contrast limit of the nonlinear switching was never reached; either the limited spatial contrast of imaging [15], or our detection limit [17] determined the experimentally demonstrated contrast improvement of NFF. This fact leaves room for further improvements.

The superior switching dynamics of NFF makes NFF as an alternate candidate for fast temporal cleaning of high-power pulses of CPA systems, if the reason of the presently limited throughput is found and the practical limits of the energy scalability of NFF are identified.

Note, that in our considerations the throughput of NFF was identified/defined as the efficiency of the basic process of NFF (the so-called diffraction efficiency); as the final energy in the central hole of the output divided by the energy of the annulus.

On the present experimental campaign, the creation of an annular beam by geometrical filtering, and the Fresnel reflection of uncoated refractive optics introduce additional loss of energy. The reason, that in our considerations these losses were disregarded is that the generation of an annular beam can also be realized without loss of energy by the use of a conical optics (axicon), and reflection losses can be avoided by AR coatings.

Peak power scalability is an important aspect for contrast enhancing methods. In order to gain basic insight to the scalability of NFF, we need to have a closer look at the focal spot size formed by the diffraction, since the phase shift needs to be introduced in the Fourier plane: $x = Cf\lambda /d$, where C is a constant in the order of unity characteristic to the beam shape, f denotes the focal length, while d and x are the beam diameters on the focusing optics and in the focus, respectively. After multiplying both sides by the peak power (P) and taking the square of the equation we get the following:

5. Conclusions

In a proof-of-principle experiment the basic process of NFF; the intensity-dependent diffraction of an IR radiation was demonstrated, which is shown to be used for efficient temporal cleaning of TW-class Ti:Sapphire laser pulses. The experimentally observed ∼10³ background suppression was attributed to the F ≈ 22 imaging capabilities of the simple, lens-based image system of the present experimental setup.

High-dynamic-range autocorrelation measurements confirmed that the virtually instantaneous, intensity-dependent and high-order switching characteristics of NFF makes possible to efficiently suppress the coherent pedestal from the foot of the main pulse, even if its duration is below 100 fs. In agreement with the results of calculations, the diffraction-based switching process is proven to be activated around 10¹⁴ W/cm² which results in a clean and sharp leading edge for the main pulse, which could not be resolved by our available temporal measurement apparatus. By using helium as nonlinear medium the working intensity can be increased up to ∼10¹⁶ W/cm² which together with the linear scaling of the length of the arrangement with input peak power allows NFF arrangements of practical size for multi-TW class lasers. Considering these facts, successful application of NFF to solid-state CPA systems can have even more pronounced importance than in case of excimer-based systems; offering significant increase of the temporal contrast of such systems; including the possibility of complete removal of the ps coherent pedestal.

The switching mechanism of the NFF setup considerably differs from those of conventional PM arrangements resulting in disparate characteristics, at the same time it is capable of reaching figure of merit values similar to the conventional PM devices. Furthermore, the gas medium of NFF induces no contamination at all, and recovers fast after the pulse requiring no challenging transport of pristine surface into the beam path for every pulse. This makes this method ideal for high-intensity laser systems with kHz repetition rate. Therefore, NFF adds a valuable new item to the toolbox of contrast improvement of high-intensity lasers with multi-TW peak power. As a next step, we will proceed with further improving the contrast enhancement by optimizing the imaging performance and studying the details of power scalability of the technique.