Radiation of relativistic electrons created in tunnel ionization of atomic gases by laser beams of extreme intensity N.V. Makarenko, S.V. Popruzhenko Abstract We consider tunnel ionization of atomic argon in a femtosecond laser pulse of intensity 10 21 − 10 22 ​ W / cm 2 10^{21}-10^{22}{\rm W/cm}^{2} aiming to investigate the relativistic dynamics and radiation of photoelectrons released from their parent ions inside the laser focus. Radiation of such electrons accelerated along the laser field propagation direction appears to have moderate power but can be considerably enhanced by a collision with a relatively weak counter-propagating laser pulse. Using the theory of laser-induced tunneling in atoms and ions and that of nonlinear Thomson scattering, we demonstrate that angular distributions and spectra of emitted photons can serve as a probe of the peak intensity in the focus. The angular distribution of emitted radiation in the plane of polarization and propagation of the ionizing laser beam appears narrow and peaked around an intensity-dependent angle, making this ionization setup a source of collimated XUV radiation. I Introduction The recent development of the new generation of femtosecond laser sources with peak power in the interval 1-10 petawatts (PW) has opened a way to systematic laboratory studies of laser-matter interactions at intensities 10^{22}{\rm W/cm}^{2} and higher. Presently, several laser facilities 28 ; 50 ; 15 ; 10 ; 51 ; 49 are already capable of delivering pulses of 3-5PW power. The peak power of 10-15 PW at a pulse duration of 20-30 fs looks realistic for the near future. Projects of multi-beam sources 29 ; 19 open a potential possibility to reach the sub-exawatt level of power. For a review of the current state of the art of the high-power laser sources and of the future projects, see recent publications 13 ; 11 . These advances in laser technologies are expected to make the intensity domain 10^{22}-10^{24}{\rm W/cm}^{2} accessible for regular experiments soon. At such extreme intensities, the nonperturbative regime of light-matter interaction becomes dominant, and the electromagnetic field strength and its distribution inside the focal volume are the key characteristics that determine probabilities of nonlinear phenomena of classical and quantum electrodynamics. Therefore, reliable and versatile diagnostics of the laser focus are in high demand for forthcoming experiments on the interaction of extremely strong electromagnetic fields with charged particles, atoms, molecules, and plasmas. Several methods to measure the peak intensity of laser radiation and its distribution in the focus have been recently proposed and discussed in the context of multi-PW installations, see e.g. 14 ; 23 ; 48 ; 45 ; 5 ; 46 ; 36 and references therein. The one based on the observation of multiple tunnel ionization of heavy atoms 4 ; 47 ; 22 ; 5 ; 6 ; 25 suggests a way to calibrate the peak intensity with precision on the level of \approx 20-50\% by observing the highest charge states of a given atomic specie. Numerical calculations made for argon, krypton, and xenon 5 ; 6 ; 25 demonstrated its potential accessibility. Although such relatively low accuracy withdraws the method from the list of high-precision protocols of measurements customary to nonlinear optics and spectroscopy, it remains sufficient for rough estimates of the conditions achieved in the focal spot of an extremely intense laser beam. The scheme of intensity calibration through the observation of multiple ionization of heavy atoms is now under consideration at several multi-PW laser installations, and the process of tunnel ionization is included into particle-in-cell codes used for simulations of ultrafast plasma dynamics in the field of intense laser radiation, see the recent publication 25 for further details and references. Except for possible applications in diagnostics, ionization of deeply bound states in laser fields of ultrahigh intensity can be used as a source of seed electrons for laser–plasma accelerators 27 ; 24 ; 7 and for initiation of quantum-electrodynamical cascades of elementary particles 42 ; 2 . The post-ionization dynamics of photoelectrons and the spectra of emitted high-energy photons enter as input data in simulations of these multi-particle phenomena. In this paper, we address an effect directly connected with ionization in the focus of extremely powerful laser beams: electrons liberated from ions through tunneling will accelerate in the same electromagnetic field, which caused the ionization, and emit electromagnetic waves predominantly in the direction of the laser pulse propagation. For a given atomic species, spectra and angular distributions of this radiation will depend on the peak laser intensity, on its distribution within the focus, and on the laser pulse duration. This therefore provides an additional opportunity for probing the electromagnetic field structure in the focal volume. It is essential that this ionization-induced radiation can be analyzed in the same experiment designed to study multiple ionization and the subsequent laser acceleration of photoelectrons. Here we analyze this ionization-induced radiation in view of the aforementioned perspective. We show that the co-propagating motion of photoelectrons, although accelerating them up to ultra-relativistic energies, suppresses radiation considerably so that only a few high-energy photons will be emitted per single atom. A relatively weak counter-propagating probe pulse is needed to enhance the emitted energy. The spectral and angular structure of this signal and its sensitivity to the time delay between the two pulses will also be addressed in our calculations. The paper is organized as follows. In the next section, we formulate the statement of the problem, introduce basic equations and discuss parameters of the laser pulses, of the atomic target and approximations used in our calculations. Section III presents analysis of the radiation spectra with and without the probe counter-propagating pulse. The last section delivers conclusions and an outlook. II Statement of the problem and basic equations We consider radiation of electromagnetic waves by fast electrons born in the laser focus through ionization of argon ions at low pressures (see the sketch in Fig.1). Argon is chosen for the following reasons. Firstly, noble gases are commonly used in strong-field ionization experiments, see e.g. 4 ; 47 ; 22 , for they are easy to store and to operate with in a chamber. Secondly, ionization potentials of the inner shells of these atoms are well adjusted to probe laser intensities in different intervals of the wide 10^{21}-10^{24}{\rm W/cm}^{2} domain. In particular, the 1s shell of argon is depleted due to field ionization at intensity \approx 4\cdot 10^{21}{\rm W/cm}^{2} 5 . The concentration of atoms in the interaction area is assumed to be as low as n_{at}\approx 10^{12}-10^{13}{\rm cm}^{-3} to eliminate the effects of space charge 5 and the macroscopic currents, which can be induced by ionization of atomic outer shells in the environment of the focal spot. Figure 1: Sketch of the interaction setup: a focused laser beam with the wave vector {\bf k} ; a diluted atomic gas jet crossing the laser focus (red dots); trajectories of electrons released in multiple tunnel ionization of atoms (blue lines); emitted photons (yellow curling lines). The gas is ionized by a laser beam focused up to the waist size w_{0}\approx 2-4\lambda with \lambda\approx 1\mu{\rm m} being the wavelength, which corresponds to the frequency \omega_{L}\approx 1.9\cdot 10^{15} sec -1 . For such moderate focusing, the approximation of a Gaussian beam works acceptably well so that we can use it to estimate the focal Rayleigh length as z_{R}=\pi w_{0}^{2}/\lambda\approx 30\lambda for w_{0}=3\lambda . This gives for the effective focal volume V_{\rm eff}\approx\pi w_{0}^{2}z_{R}\approx 10^{3}\lambda^{3}=10^{-9}{\rm cm}^{-3} (1) with \approx 10^{3}-10^{4} atoms inside. For tighter focusing, w_{0}\approx\lambda , the Gaussian form is no longer a good approximation for the description of a stationary laser beam 9 . In this case one of many known exact solutions (see 43 ; 26 and references therein for examples) can be used to set the field distribution up. Multi-PW lasers of type 28 ; 19 ; 29 ; 15 are expected to deliver pulses of \tau=20-30 fs duration. As an example, the current configuration of the Apollon laser facility 49 delivers 22-fs long pulses at peak power \approx 2 PW. In this case, the parameter kw_{0}\approx 20 is smaller than \omega_{L}\tau\approx 40-60 , so that solutions found for stationary beams are also approximately valid to describe pulsed fields 26 . Within this approximation, we use the field \begin{split}{\bf E}&=\frac{{\bf E}_{0}}{\sqrt{1+z^{2}/z_{R}^{2}}}\exp\left(-\frac{r^{2}}{w_{0}^{2}\left[1+z^{2}/z_{R}^{2}\right]}\right)\\ &\times\cos\left(\omega_{L}t-kz-\frac{kr^{2}}{2z\left[1+z^{2}/z_{R}^{2}\right]}+\arctan\left(\frac{z}{z_{R}}\right)\right)\\ &\times f(\omega_{L}t-kz),\\ {\bf H}&=-c\cdot\operatorname{rot}\int^{t}{\bf E}(t^{\prime})\,dt^{\prime}\end{split} (2) of a linearly polarized Gaussian beam with the electric field peak amplitude E_{0} and a sinusoidal time envelope f of total duration \tau=2M\pi/\omega_{L}\equiv MT_{L} f(\omega_{L}t-kz)=\begin{cases}\sin^{2}\!\left(\frac{\omega_{L}t-kz}{2M}\right),&\omega_{L}t-kz\in[0,2M\pi],\\ 0,&\omega_{L}t-kz\notin[0,2M\pi]~.\end{cases} (3) The integer M gives the total number of optical cycles in the pulse. In our calculations, we set M=16 which gives a full width at half maximum \tau_{1/2}\approx 25 fs. In the following, we assume the electric field of the beam ( 2 ) to be polarized along the x axis. While the electric field is orthogonal to the wave vector {\bf k} , the magnetic field defined as ( 2 ) has some small longitudinal component which however makes only a small effect on the dynamics and radiation of photoelectrons. II.1 Tunnel ionization In this subsection, we introduce the formalism used for description of the ionization step. In contrast to other sections of this paper where the CGS system is applied, here we use atomic units e=m=\hbar=1 with e and m being the elementary charge and electron mass correspondingly. In these units, the electron charge equals -1, and the speed of light c=1/\alpha\approx 137 . We apply the non-relativistic model of tunnel ionization of atomic levels proposed by Smirnov and Chibisov (SC) 41 for the static field and by Perelomov, Popov and Terentiev (PPT) 30 ; 32 ; 34 ; 38 for the low-frequency field. The tunnel limit is validated by the smallness of the Keldysh parameter 17 ; 35 \gamma_{\rm K}=\frac{\sqrt{2I_{p}}\omega_{L}}{E_{0}}~, (4) with I_{p} being the ionization potential. In infrared laser fields of wavelengths \lambda\simeq 1\mu{\rm m} ionization of atomic levels except for a few outer electrons, always proceeds in the tunnel regime, \gamma_{K}\ll 1 . For intensities greater than 10^{18}{\rm W/cm}^{2} this condition is well secured with \gamma_{\rm K}\simeq 10^{-2}-10^{-3} or even less 5 . Relativistic effects on the probability of ionization are negligible up to intensities \approx 10^{23}{\rm W/cm}^{2} at least 5 ; 39 . Then the magnetic field of the wave does not have an effect on the tunneling process itself, and the rate of ionization is determined by the instant electric field amplitude E({\bf R}_{j},t) , which also depends on the position of the j -th atom {\bf R}_{j} . Figure 2: Distributions in the number of electrons N_{e} released through tunneling per interval \Delta t=T_{L}/15 with the rate defined by Eq.( 5 ) from N_{\rm at}=100 ions {\rm Ar}^{8+} randomly distributed in the interaction volume. The ion charge is shown by color (see the color coding bars in the right upper corner). The laser field time envelope ( 3 ) is shown by a black solid line. Time is given in laser periods. Panel (a) corresponds to intensity {\cal I}=10^{21} W/cm 2 , panel (b) – to {\cal I}=10^{22} W/cm 2 . For the lower intensity, the 1s shell is not ionized. We model ionization within the single-electron approximation as a sequential probabilistic process. The standard definition of the sequential ionization process assumes that electrons are being removed in the order of increasing ionization potentials. Recent studies 25 show that this is not always the case; however, deviations from the sequential channel averaged over space and time remain rather minor, and we do not account for them here. The probability for the outer electron with orbital and magnetic numbers l and m to be removed from its atom (ion) with residual charge Z_{c} (in these notations, Z_{c}=1 for a neutral atom and {Z_{c}}=N for A^{(N-1)+} ion) during the time interval dt is given by the SC or PPT static rate 41 ; 31 ; 32 ; 35 \begin{gathered}dW(t)=w(I_{p},\nu,l,m;E(t))dt~,\\ w=2^{2\nu+1}C^{2}_{\nu l}B_{lm}I_{p}F^{1+|m|-2\nu}(t)\exp\left(-\frac{2}{3F(t)}\right)~.\end{gathered} (5) Here F(t)=\frac{\sqrt{{\bf E}^{2}(t)}}{(\rm 2I_{p})^{\rm 3/2}} (6) is the reduced electric field scaled by the characteristic field E_{\rm ch}=(2I_{p})^{3/2} of the atomic level (note that the electric field depends also on the atomic position {\bf R}_{j} , \nu=\frac{Z_{c}}{\sqrt{2I_{p}}} (7) is the effective principal quantum number, C^{2}_{\nu l}=\frac{2^{2\nu-2}}{\nu\Gamma(\nu+l+1)\Gamma(\nu-l)} (8) is the squared asymptotic coefficient of the bound state single-electron wave function calculated in Hartree approximation 35 and the factor B_{lm}=\frac{(2l+1)(l+|m|)!}{2^{2|m|}|m|!(l-|m|)!} (9) reflects the angular structure of the bound state. For details of the derivation of the rate formula ( 5 ), as well as for its applications and generalizations we refer the reader to reviews 35 ; 38 and references therein. Application of the SC or PPT models for the calculation of ionization events meets two restrictions. Firstly, at relatively low intensities, when the outer shells are ionized, the sequential channel can be deeply overtaken by the nonsequential one mediated by recollisions 20 ; 12 ; 8 . Because of this, at intensities below \simeq 10^{17}{\rm W/cm}^{2} and in the field with polarization close to linear the tunnel rates ( 5 ) become inapplicable for the calculation of ion yields. Although we consider much stronger fields here, moderate intensities, where the recollision mechanism of ionization is relevant, will be reached on the front edge of the pulse or in the pre-pulse, leading to the removal of several electrons with the lowest ionization potentials. Secondly, with the field amplitude growing, the regime of ionization can switch from the tunneling to the over-the-barrier mode 3 ; 35 ; 18 . Then the SC and PPT formulas again become inapplicable and overestimate the rate by several times for F=F_{*}=0.1...0.2 and higher fields, see examples in 35 . The necessity of new formulas 3 ; 44 ; 18 extrapolating the tunnel rates to the over-the-barrier domain is determined by the effective value of the reduced field F_{\rm eff} required to deplete the level. The latter is determined by the level quantum numbers and by the pulse duration, which lowers F_{\rm eff} for smoother time envelopes. We have checked that (a) for the pulse parameters used in our calculations F_{\rm eff}\approx 0.05