1. Introduction

New approaches to ultrafast science such as holography, nanotechnology, biological imaging and fast electron dynamics probing [1] have been constituted based on the extension of the pump-probe techniques to extreme ultraviolet (XUV) and soft X-ray (SXR) regime. With the generation of attosecond XUV pulse, real-time observation of a large range of phenomena involving the dynamics of electrons on the atomic timescale and energy transport in molecules [2] become possible. To achieve this, compact X-ray [3] or high harmonic sources with sufficient intensities and sub-femtosecond durations are necessary.

Several methods for phase-coherent harmonic emission using intense laser pulses have been proposed, including the nonlinear interactions with atoms in rarefied gases [4] and with over-dense plasmas in solid targets [5, 6]. However, the intensities of the APs obtained by laser-gas interaction are very low due to the following two restrictions. One is the power of the pump laser has to be kept low (≤ 10¹⁶ W cm⁻²) because of the phase mismatch effects [7, 8]. The other is the conversion efficiency which is associated with the electron density, is extremely small (10⁻⁷ ∼ 10⁻⁶) [9]. The weak intensity of the emitted attosecond pulses has been one of the main obstacles for applications of the pump-probe techniques. Fortunately, in the scheme of laser overdense plasma interaction, the conversion efficiencies of high-order harmonic generations (HHG) in overdense plasmas were significantly increased [10, 11]. With the much higher electron density and no restriction of the phase mismatch effects (the laser intensities can be over 10¹⁸ W cm⁻²), the APs generated from solid targets are almost several orders of magnitude higher than those from atomic mediums [12, 13]. This feature may provide high harmonics and/or APs of sufficient intensities, fulfilling the requirements in the pump-probe experiments, and furthermore broadly expanding the scope of the applications of APs.

The linear polarization (LP) driving laser employed in harmonic generation at the over-dense plasma surface typically needs an intensity of I_L > 10¹⁵ W cm⁻² for the λ_L = 800 nm laser wavelength [14], which corresponds to a minimum normalized field amplitude $a_L={\left[I_L{\lambda }_L^2/\left(1.37\times {10}^{18}\hspace{0.17em}\text{W}\hspace{0.17em}\mu {\text{m}}^2\hspace{0.17em}{\text{cm}}^{-2}\right)\right]}^{1/2}=0.03$. When an intense ultrashort laser pulse focuses on a solid target, the materials are ionized and converted into plasmas instantaneously. The incident laser pulse reflects on the dense plasma boundary which has an extremely steep interface with vacuum. The interaction process is highly nonlinear when surface electrons are compressed to a thin layer and oscillate as a plasma mirror which leads to harmonic generation. Different reflected harmonic spectra appear when the pump laser intensities vary from weak relativistic region ( $I_L{\lambda }_L^2<{10}^{18}\hspace{0.17em}\text{W}\hspace{0.17em}\mu {\text{m}}^2\hspace{0.17em}{\text{cm}}^{-2}$) to strong relativistic region ( $I_L{\lambda }_L^2\gg {10}^{18}\hspace{0.17em}\text{W}\hspace{0.17em}\mu {\text{m}}^2\hspace{0.17em}{\text{cm}}^{-2}$), which corresponding to two major mechanisms of harmonic generation, coherent wake emission (CWE) [14] and relativistic oscillating mirror (ROM) [15, 16, 11]. The CWE totally dominates the harmonic generation process for a_L < 1, and contributes to a narrow frequency range of the harmonic spectrum, which typically corresponds to harmonic order less than 30. As the laser intensity raises, e.g., a_L ≫ 1, the ROM process becomes the dominant mechanism. The maximum cut-off frequency of the harmonic spectrum is far beyond that from the CWE process, and extends up to ${\omega }_{co}\approx {\gamma }_{\mathit{max}}^3{\omega }_L$, where γ_max is the relativistic factor corresponding to the maximum velocity of the plasma mirror which flies towards the incoming light, and ω_L is the incident laser frequency [17]. Moreover, the harmonic intensity follows the power- law form I(ω) ∝ ω^−p, where ω is the harmonic order and the exponent 5/3 ≤ p ≤ 10/3 depends on the laser pulse parameters [18]. The harmonics are phase locked, and one can obtain a train of extremely short pulses by transforming the high harmonics spectrum into the time domain.

However, when a relativistic LP pulse impinges on a solid target, the plasma boundary would oscillate twice per laser cycle as the ponderomotive force of the LP pulse includes a fast oscillating component, inducing an AP train. While for most applications single AP is preferred. Two methods have been proposed to isolate single AP from the pulse train: using a lambda-cubed (λ³) incident laser pulse [19, 20] and the polarization gating technique [21]. Nevertheless, it is still a challenge to verify both methods for general facilities.

Motivated by the incomparably wide prospects of single APs, we have looked into the possibility of directly producing single APs from the interaction of intense few-cycle circularly pulses (CP) laser with overdense plasmas. Ji et al. found that high harmonics can be produced in the reflected light by the Doppler-effect when the CP laser pulse reflects from the laser-induced one-time oscillating plasma boundary [22, 23]. In present work, we first verify the scheme of single AP generated from CP laser interacting with overdense plasma with a flat foil by 2D PIC simulations in section 2. The generated APs are dispersed in front of the plasma surface with self-focusing and defocusing phenomena, and can hardly be used in most proposed applications. In section 3, by using density-modulated foil targets (DMFT) [24, 25] in the scheme, we find that the angular divergence of the emissions can be controlled by the plasma surface deformation. A single 300 as pulse is obtained with a designated target density distribution. The intensity of the AP is higher than 10²¹ W cm⁻² and has a remarkable feature in the transverse direction. In section 4, an analytical model is employed to follow the spatial distributions of the plasma boundaries in detail. The numerically results agree well with the simulation runs. We also find that the emission directions of high harmonics are both sensitive to the profile of the modulated density (section 5) and to the carrier-envelope phase (CEP) of the incident CP laser pulse (section 6). The emission direction of the generated AP can be controlled in a very wide range in front of the plasma surface by combining the DMFT with an appropriate density profile and a suitable driving laser.

2. Analysis of focusing mirror by 2D PIC simulation

When a relativistic CP laser pulse impinges on a cold-plasma foil, electrons are quickly pushed forward by the steady part of the v × B force. The ions are fixed due to the short interaction period and the much larger mass. An intense space-charge field is created as the electrons are greatly compressed from the surface to inward. This compressed electron layer (CEL) intensively bounces back, and forms a one-time oscillation. A bunch of high-order harmonics will be produced in the reflected light when the CP laser pulse reflects from the laser-induced oscillating plasma boundary. The oscillation is the result of the imbalance between laser light pressure and space-charge separation field force. It is so intense that the maximum speed of the CEL can reach nearly the speed of light, which can produce an ultra-short single AP naturally [22]. However, things make different when we investigate the AP generation by this proposed scheme in 2D geometry. In particular, the plasma boundary will be bended in the case of a laser pulse with a finite width in the transverse (y) direction [26]. The generated radiation fields distribution at the plasma surface may differ in intensity and diffraction direction. These features lead to a process of self-focusing and then defocusing of the reflected harmonics [27], which might be harmful for long-distance transmission of the APs.

In order to study the self-focusing and defocusing features of the deformed plasma boundary in detail, we perform fully relativistic planar 2D PIC simulations by the code VORPAL [28]. The simulation box is 35λ₀ × 30λ₀, corresponding to 10500 × 1500 cells in x and y directions, respectively, where λ₀ is the incident laser wavelength. The high spatial resolution along the laser propagating direction is to ensure the calculation precision on attosecond time scales. The plane target is fully pre-ionized, with ion mass and charge of m_i = 3672m_e and q_i = −e, where m_e and e are mass and charge of the electron, respectively. It is initially located in the region of 28λ₀ < x < 31λ₀ with a density of n₀ = 8n_c, where n_c = m_eω²/4πe² is the critical density for the laser frequency ω. The normalized amplitude of the incident CP laser pulse is $a=a_0{\text{sin}}^2(\pi t/2{\tau }_0)\text{exp}\left(-y^2/r_L^2\right)$. where a₀ = 20, τ₀ = 2T₀ and r_L = 5λ₀. This corresponds to a laser intensity of I₀ = 1.1 × 10²¹ W cm⁻² for the assumed wavelength λ₀ = 1μm. The leading edge of the laser pulse enters the simulation box from the left boundary at the beginning and reaches the target at t = 28T₀. Absorbing boundary conditions are used for both electromagnetic field and particles.

The spatial distribution of electrons at t = 31T₀, as seen in Fig. 1(a), is related to the combination of the ponderomotive force of the laser pulse and the restoring electrostatic force of the ions. The incident laser pulse with a transversally Gaussian profile bends the plasma surface. The red solid line in Fig. 1(b) denotes the space-charge separation field, peaking at the pulse front with the maximum intensity of 33, which plays the key role in switching the electron layer motion and forming the one-cycle oscillation. The intensity of the electrostatic field is much larger than that of the incident laser field, and has a linear profile both in the depletion region and in the compressed layer. Throughout the entire interaction time, the dented electron layer maintains as a Gaussian reflectance mirror. No obvious ripples or bubble structures which induced by multidimensional instabilities in laser-overdense plasma interaction [29, 30, 31] can be found in the plasma boundary, because the timescale for the growth of this Rayleigh-Taylor-like electron instability is in the several fs-range and thus much longer than the reflection process.

 

Fig. 1 (a) The spatial distribution of electrons in uniform density target case at t = 31T₀. The red dotted line is calculated from the analytical model in section 4. (b) The snapshots of the electrostatic field E_x (red solid line) and the incident laser field E_y (blue dashed line) along y = 0 at t = 31T₀.

Download Full Size | PDF

Figures 2(a)–2(c) present the spatial distributions of the reflected laser field magnitude E_y at time t = 34T₀, t = 44T₀ and t = 54T₀, respectively. The incident laser pulse is well reflected and propagates in the −x direction. The reflected pulse is self-focused from the surface due to the curved plasma boundary. The focal spot as intense as ∼ 10I₀ can be reached at t = 44T₀, and the spot size is comparable to the laser wavelength. The self-focusing length is 11.6λ₀ which is in good agreement with the general analytical estimate x_sf = S₀x_RL/2.7 = 11.6 μm [32, 33], where S₀ = n₀/a₀n_c is the ultra-relativistic similarity parameter [34], and x_Rl = πr_L²/λ₀ is the Rayleigh length of the fundamental. The reflected pulse is defocused after the self-focusing rapidly. The whole radiation is simply diffracted away from the optical axis and transmits in diverging directions, as shown in Fig. 2(c).

 

Fig. 2 Spatial distribution of the reflected laser field magnitude E_y at (a) t = 34T₀, (b) t = 44T₀ and (c) t = 54T₀. Temporal profile of the filtered attosecond pulses at the tracing positions of (d) (22λ₀, 0), (e) (16λ₀, 0) and (f) (10λ₀, 0). Notice the different scales on the x and y axis.

Download Full Size | PDF

The single attosecond pulse is generated due to the fact that the plasma surface is accelerated to nearly the speed of light on a short time scale (< 1T₀) [35, 36]. The emission of high-frequency photons as seen at an intermediate distance from the surface occurs only during small fractions of T₀. In frequency space, this leads to high harmonics spectrum which transforms into intense APs in the time domain. To liven up the picture of how the attosecond pulses propagating, the filtered pulses at different distances from the surface are investigated. After ω ≤ 3ω₀ filtering of the low-frequency components of the spectrum which is observed in the position of (22λ₀, 0), as seen in Fig. 2(d), we obtain a single ultra-intense AP with the duration of 356 as and its peak intensity is 1.1 × 10²¹W/cm². Since the harmonics are coherent, laser field at the focal point is strongly enhanced by summing up all the field reflected from different positions of the interaction surface [27, 37]. The maximum intensity of the AP is as high as 4.0 × 10²¹ W cm⁻² at t = 44T₀ and dramatically falls to 4.6 × 10²⁰ W cm⁻² at t = 54T₀, as shown in Figs. 2(e) and 2(f). Due to the extremely broad spectrum of the generated high harmonics, diffraction will exert a major influence on the spectrum propagation. The spectrum changes rapidly because of the different diffraction lengths of the harmonics and different propagation directions. The generated AP radiation will be dispersed in the far field, and can hardly be used in most proposed applications.

3. Density-modulated foil target

2D PIC simulation confirms that the reflected CP laser pulse can be self-focused. The self-focusing effect will be very useful to generate laser intensity approaching the Schwinger limit [27], however, it is unfavourable to produce a stable-propagating AP. As is well known, the electron dynamics in ultra-relativistic laser-plasma interaction depends on the similarity parameter, rather than on a₀ and n₀ separately [34]. Taking this advantage, we propose a flat foil target with a Gaussian plasma density distribution in the transverse direction, the laser intensity profile is well matched and the self-focusing effect of the generated AP is somehow avoided. As a result, a plan-wave like AP will be generated, which will travel for a long distance without apparent distortion.

As shown in Fig. 3, a CP laser pulse is normally incident on the density-modulated foil target (DMFT). The density profile of the DMFT is defined by

 

Fig. 3 Layout of density-modulated foil target (DMFT). A CP laser pulse is incident on the foil target with a Gaussian plasma density distribution in the transverse (y direction). The transverse density profile of the DMFT (curved black line) is defined by r_d. The maximal electron density is n₀ = 8n_cwhile the cut-off is n_cutoff = 0.2n₀.

Download Full Size | PDF

The advantage of the constant-S surface can nicely be seen in Fig. 4(a), which showing the temporal electron density distribution at t = 31T₀ with gray scale. The plasma boundary is uniformly shifted along the transverse direction and keeping the shape of the compression electron front. The red dotted line is calculated from the analysis model which we will discuss later. The generation of surface plasma waves (SPW) is a vital feature of laser plasma interaction. The SPW propagates long distances along the surface, which causes several multidimensional instabilities, e.g. the Rayleigh-Taylor instability [38, 39, 40]. However, no evidence of any instability is found in our 2D simulations. In the short time-scale of a few-cycle pulse driver, the classical Rayleigh-Taylor instability is not relevant because it presumes ion motion, and the instabilities have no time to develop. It is expected that the plasma surface would remain planar during the interaction period.

 

Fig. 4 2D PIC simulation results for the DMFT case. (a) Spatial distribution of the reflected laser field magnitude E_y (color palette) at t = 36T₀, and the temporal electron density distribution (gray level) at t = 31T₀. The red dotted line is calculated from the analytical model in Section 4. (b) Temporal profile of the generated as pulse after the ω ≤ 3ω₀ frequency filtering. (c) The spectrum of the reflected light observed at (25μm, 0). The red dashed line is the power law predicted by the ROM model. (d) The full width at half maximum (upper panel) and the peak intensities (lower panel) of the APs after filtering out the fundamental and the diploid frequency along x = 25μm.

Download Full Size | PDF

The incident laser field in vacuum runs in the positive direction, while the reflected field is translated backwards in the negative direction. The spatial distribution of the laser field E_y at t = 37T₀ is depicted in Fig. 4(a). The incident laser pulse is better reflected from the overdense plasma layer compared to the unoptimized case in Fig. 2(a). Evidently, no self-focusing or defocusing phenomenon occurs in front of the planar surface. These distributions look similar to the 2D simulation results reported by Ji et al. [22], where they employed a CP laser pulse with sufficiently flat intensity distribution across the focal spot (Super-Gaussian) to irradiate an uniform plane target. This is another possibility to keep the relativistic S-parameter constant alongside the surface. High-order harmonics are contained in the trailing edge of the reflected pulse. As the reflected fields stably propagate backwards, the profile of the plan-wave-like AP is well maintained. Figure 4(c) shows the on-axis radiation spectrum in the log-log scale. The decay in peak harmonic power levels with harmonic number ω is well represented across the range by I(ω) ∝ ω^−8/3, which is in agreement with the similarity analysis and the PIC simulation spectrum [17]. After ω ≤ 3ω₀ frequency filtering, a single ultra-short light pulse is obtained, as seen in Fig. 4(b). The AP has a duration of 215 as and its peak intensity reaches 9.1 × 10²⁰ W cm⁻². The conversion efficiency is 4 orders of magnitude larger than those in laser-atom interactions [9]. After filtering out the diploid frequency components of the spectrum which are observed at different y positions along x = 25λ₀, we can see that the durations of the single APs vary slightly around 300 in the region of |y| ≤ 4λ₀. The corresponding intensities are plotted in the lower panel of Fig. 4(d), where a remarkable feature is that the intensity curve shows a flatten profile within the range of |y| ≤ 2λ₀ and drops linearly beyond that region. The peak intensity is ∼ 1.4 × 10²¹ W cm⁻² and is slightly higher than the incident pulse intensity. One should be noticed once again, that the DMFT case yields a clear advantage for the attosecond pulse generation.

4. Analytical model

In order to have a better understanding of the different electron dynamics between uniform density target case and DMFT case, we propose an analytical model as shown in Fig. 5 [26]. When a CP pulse impinges on an overdense plasma layer, the electrons at the plasma vacuum interface are greatly compressed and accelerated forward by the ponderomotive force of the incident laser pulse, creating an intense space-charge field with the immobile step-like ion background. The momentum of the CEL is determined by the imbalance between the pondero-motive and space-charge forces, which implies an uniform electron density n_p0 and a thickness of l_s = x₂ − x₁ equalling the skin depth ${\lambda }_s=c\sqrt{m_e/4\pi n_e{e}^2}$. In view of the simulation result shown in Fig. 1(b), we assume that the electrostatic field peaks at the CEL left surface (x = x₁) with the intensity of E_x0 = 4πn₀e(x₁ − x₀), and has a linear profile both in the depletion region (x₀ < x < x₁) and in the compressed layer (x₁ < x < x₂). The ponderomotive force experienced by the electrons can be written as

 

Fig. 5 Schematic drawing of the interaction model. The few-cycle incident laser pulse (red solid line) acts on the left side (x₁) of the moving compressed electron layer (blue line). The ions (black thick line) are fixed with an initial density n_i. The electric field (red dash-dot line) has a linear profile both in the depletion region (x₀ < x < x₁) and in the compressed layer (x₁ < x < x₂).

Download Full Size | PDF

Equation (6) can be solved numerically with the corresponding initial conditions x(0) = 0 and u(0) = 0. The plasma boundary profiles of the uniform density target case and the DMFT case calculated from the analytical model are plotted in Fig. 6(a). The incident laser pulses and the targets density profiles are the same as the corresponding simulation parameters employed in section 2 and section 3, respectively. The spatial distribution of the CEL along the transverse in the DMFT case (red solid line) is uniformly shifted in the region of |y| ≤ 6.4λ₀, but presents a Gaussian shape in the uniform density target case (blue dotted line). The numerically results fit nicely with the simulation results, as seen in Fig. 1(a) and Fig. 4(a). Furthermore, we trace the momentum of the on-axis electron according to the exact calculating of Eq. (6), as seen in Fig. 6(b). It is clear that, the electron is first accelerated forward by the laser ponderomotive force, and then intensively bounced back at t = 3T₀ due to the intense space-charge separation field force. The maximum shifting distance of the electron is less than 0.7λ₀, it is tiny comparing to the spot size of incident laser (r_L = 5λ₀). The speed of the plasma boundary can reach nearly the speed of light in 1T₀, and high harmonics are generated relaying on the Doppler-effect when the CP laser pulse reflects from the one-time oscillation plasma boundary.

 

Fig. 6 Numerical results from the analytical model. (a) Spatial positions of the plasma boundary of the uniform density target (blue dotted line) and the DMFT (red solid line) at t = 3T₀. (b) The displacement (blue solid) and speed −β_x (red solid) of the on-axis electron surface layer.

Download Full Size | PDF

5. Control of the attosecond pulse emit direction

The incident laser pulse is well reflected from the plasma boundary with a constant ultra-relativistic S-parameter in the transverse direction. Attosecond peaks can be obtained and the intensity is comparable to the incident laser pulse. In order to interpret the changes of the reflected pulses with the relativistic S-parameter alongside the layer surface, we varied the transverse density profile parameter r_d from 1λ₀ to 7λ₀ with the other parameters unchanged, so that

In Fig. 7(a), the distribution of the electron-layer surface driven by the laser pulse looks like a convex lens in case of r_d = 3λ₀, thus the reflected radiation spread as a relevant curved diffusion. Surprisingly, the reflected radiation has been split into two separate impulses. Each of them is moving in its own direction. The less intense impulse (1) propagates along the center axis, and the more intense impulse (2) is directed towards the lower left-hand direction. The maximum intensity value is as high as 0.6I₀ at t = 42T₀, and falls slowly by the influence of diffraction. Tracing this impulse, it follows near 10° direction with respect to normal incidence direction, and also has a divergence of ∼ 10°. When r_d decreases to 2λ₀, the reflected impulse direction increases to ∼ 15°. The impulse direction increases while the density profile parameter r_d decreases, which can be explained by the direct relationship between the reflected radiation and the electron-layer surface deformation.

 

Fig. 7 Snapshots of the electron densities and filtered impulse intensities for the case of (a) r_d = 3λ₀ and (b) r_d = 7λ₀. The electron densities at the time of t = 31T₀ are drawn with gray gradient. The red dotted lines are the positions of the plasma boundaries calculated from the analytical model. Numbers (1) and (2) in subgraph (a) indicate the two intense impulses in the reflected radiation. Notice that the intensities of the impulses are normalized to I₀.

Download Full Size | PDF

Being exposed to the laser, the plasma surface is curved and shaped appropriately. The reflected high harmonics which consist of several pulses propagating in different directions can be focused to high intensities if r_d > r_l. Two snapshots of the impulse intensities at the time of t = 42T₀ and t = 47T₀ for the target profile r_d = 7λ₀ are shown in Fig. 7(b). The reflected harmonics focus to the center axis and enhance the intensities of the APs. The focal distance is much further comparing to the radiation focusing from the planar surface in section 2. This is due to the smaller curvature of the plasma surface in this case. The intensity gain of this work is about proportional to a₀, which is not as strong as that in the coherent harmonic focusing (CHF) [27]. But, comparing to the CHF, our simulation conditions are more realistic and achievable. To obtain higher intensity gain, it is needed to maximize the solid angle and to minimize the focal distance which will make the maximum use of the temporal focusing caused by the relativistic surface effects [32]. These two features can be optimized by modulating the DMFT profile parameters of this work.

6. Effect of the incident pulse phase offset

The laser-plasma interaction is phase sensitive due to the short incident pulse duration. Thus, as the driver pulses we employed in present work come to a few-cycle laser pulse regime, the effects of phase offset between laser field and carrier envelope should be taken into account [19, 43]. Simulations are performed to study this effect, where the phase of the incident pulse ϕ is changed from 0 to 2π by every π/2 in the DMFT scheme with r_d = 3λ₀ and r_d = 7λ₀ separately. Then we measured the filtered radiation pulse intensities with different pulse phases..

In case of r_d = 3λ₀, the generated impulses in three phase offsets π/2, π, 3π/2 are shown in Figs. 8(a)–8(c), respectively. The direction of the peak intensity presents a clockwise rotation while increasing the phase offset. When the phase is changed by π, the deflected light shows an inversion symmetry about the incident laser direction, e.g., the distributions in Fig. 8(a) and 8(c). All the reflected radiations split into two impulses. The transmission directions of the less intense impulses in three phase offsets change slightly alongside the normal direction, which is because the motion of the reflecting surface around the center position is almost irrelevant to the phase offset of the incident laser. The divergence and the FWHM width of the reflected radiation with ϕ = π/2 are larger than those in ϕ = π, which leads to a little bit smaller peak intensity. The symmetrical structures also can be found in the self-focus process with r_d = 7λ₀ as shown in Figs. 8(d)–8(f). With these finer snapshots, we can find that the focused radiations are also composed of two impulses. The main reflected pulse (1) is transmission on the center axis, and the direction of the less intense pulse (2) is phase sensitive. Two impulses overlap with each other in ϕ = 0 or ϕ = π, which enhance the peak intensity.

 

Fig. 8 Filtered radiation intensity distributions at t = 42T₀ with different pulse phases for the cases of r_d = 3λ₀ ((a)–(c)) and r_d = 7λ₀ ((d)–(f)).

Download Full Size | PDF

It is well known, the direction of the resulting AP is difficult to be controlled, with the features we discussed above. The reflected AP from the plasma surface has a strong dependence on the plasma density profile in the transverse direction, and is also sensitive to the phase offset of incident laser pulse. Thus, the direction of the reflected radiation can be controlled by combining the DMFT with an appropriate density profile and a suitable driving laser.

7. Conclusion and Discussion

In the case of atomic harmonics, ionization gives rise to the saturation intensity, which limits the intensities of the APs. Fortunately, the theoretical treatment of the generation process of harmonics from laser-overdense plasma interaction reveals that there is no fundamental limitation in using the highest possible laser intensities available. The attosecond pulses are directly generated by reflecting from the plasma surface, therefore, the higher the laser intensity the stronger the nonlinear response of the medium. When a CP laser pulse impinges on the plasma surface, high harmonics can be produced in the reflected light relying on the Doppler-effect. An ultra-intense single attosecond light pulse can be naturally generated when the few-cycle relativistic CP laser pulse reflects from the laser-induced one-time oscillating plasma boundary. We investigated this scheme by 2D PIC simulations and analysis model. The spatial distributions of the compressed electron layer calculated from the analytical model agree well with simulation runs.

The incident laser pulse with a transversally Gaussian profile bends the plasma boundary. The reflected pulses are focused down to a small spot by the naturally curved surface and rapidly defocused after the self-focusing. The whole radiation is simply diffracted away from the optical axis and transmits in diverging directions, which can hardly be used in most proposed applications. We presented that the reflected high harmonics can be controlled by employing a density-modulated foil target. When the target density distribution fits that of the laser intensity profile, the incident laser pulse is better reflected from the overdense plasma boundary with no self-focusing or defocusing phenomena in front of the planar surface. This feature is extremely useful since the optical filters have to be placed inside the far-field in a real experiment. The peak intensity of the AP generated from the center part of the plasma boundary is as high as 1.4×10²¹W cm⁻². The intensity curve shows a flatten profile within the range of |y| ≤ 2λ₀. and drops linearly beyond that region in the transverse direction. The FWHMs vary slightly around 300 as in the region of |y| ≤ 4λ₀. This could be a valid solution to get an intense single AP with good contrast ratio by combining DMFT with the use of an optical filter.

The prepulse plays an important role in ultra-intense laser plasma interaction, as reported by Zheng et al. [44], one can control the emission attosecond pulses by changing the target surface profile with proper preplasma. In our results, the emission directions of high harmonics are also sensitive to the profile of the modulated density and to the phase offset between the laser field and the carrier envelope. By modifying transverse density profiles and phase offsets, the reflected impulses are defocused in given directions or focused down to a spot with different focal lengths. With these features, the emission direction can be controlled in a very wide range in front of the plasma surface by combining the DMFT with an appropriate density profile and a suitable driving laser. These results show very encouraging scientific potentials in a lot of fields.