Abstract
Up to now, full tunability of waveforms was possible only in electronics, up to radiofrequencies. Here we propose a new concept of producing fewcycle terahertz (THz) pulses with widely tunable waveforms. It is based on control of the phase delay between different parts of the THz wavefront using linear diffractive optical elements. Suitable subcycle THz wavefronts can be generated via coherent excitation of nonlinear lowfrequency oscillators by fewcycle optical pulses. Using this approach it is possible to shape the electric field rather than the slow pulse envelope, obtaining, for instance, rectangular or triangular waveforms in the THz range. The method is upscalable to the optical range if the attosecond pump pulses are used.
Introduction
Generation of fewcycle pulses in terahertz (THz) (0.1–10 THz) range is a subject of intensive research^{1,2,3,4,5}. Large molecules have vibrational and rotational transitions in this region, which makes such pulses important in high resolution spectroscopy, biological sensing, medicine, wireless communication systems etc^{5,6,7}.
Most of the methods of THz pulse generation proposed up to now are based on a nonlinear conversion of optical pulses in nonlinear media^{1,2} using optical rectification in different media such as lithium niobate^{8,9}, semiconductors^{10,11} and organic crystals^{12,13}, or the ionizationbased nonlinearity (so called Brunel mechanism^{14}) which leads to creation of plasma currents in gases^{15,16,17,18} or semiconductors^{1,2} (in the latter case the electric current is created via transition of electrons from valence to conducting bands). The latter idea was further developed by using photoconductive nanoantennas^{5,19}.
Currently, there is a big interest in halfcycle pulses at various wavelengths^{20,21,22,23,24} since they are important in ultrafast spectroscopy^{23,25}, as well as in attosecond science^{26,27}. For instance, they are in certain sense more efficient in excitation of quantum oscillators^{23,28}. Typically, halfcycle pulses contain a single strong burst of electromagnetic field and a long lowamplitude tail of the opposite polarity. This specific feature of subcycle pulses makes them an attractive tool to control electron wavepacket dynamics^{23,28}. Such pulses are sometimes called quasiunipolar^{29}, taking in mind truly unipolar pulses, which, according to some theoretical predictions, is possible to obtain as a result of interaction of light with resonant media^{24,29,30,31,32,33,34,35,36,37,38,39,40}, or as the result of charged particle acceleration^{41,42}.
Subcycle pulses can be especially easily obtained in the THz range^{1,2}, but even in the optical range such pulses are becoming now available^{20,21,22,23,24}. Different methods of unipolar and quasiunipolar pulse generation via accelerated electric charges^{41,42}, halfcycle solitons in nonlinear medium^{30,31,32,33,34,35}, reflection of singlecycle pulse from nonlinear film^{24} as well as in media with different types of nonlinearity^{29,36,37,38,39,40} were recently discussed.
Another important problem in fewcycle pulse generation is pulse shaping. Generation of arbitrary optical waveforms is an important issue in many applications such as coherent control of quantum and nonlinear processes, and lightwave communications^{43}. For THz pulse shaping and phase control different schemes have been proposed, see refs^{44,45,46,47,48,49} and references therein. These methods enable control of THz pulses by introducing dispersion elements^{44}, prism wave plates^{45}, parallelplate waveguides^{46,47} or by the pump pulse shaping^{48,49}.
An alternative recent approach is based on a coherent excitation and control of the free induction decay^{50} in a medium with ultrashort optical pulses^{29,38,39,40}. In this approach, nonlinear dipoles with a resonance at low frequencies (THz or MIR ranges) are to be excited by pairs of ultrashort pulses in optical range, one of them starting the slow THzscale oscillations of the dipoles and another stopping it^{38,39}. Between the optical pulses, a slow oscillator radiates via free induction decay^{50}, which allows to generate subcycle quasiunipolar pulses. Radiation from many of such oscillators located at different spatial positions, being summed with a certain phase delay, resulted in tunable waveforms^{40} appearing at the detector. This method requires the control of spatiotemporal phase delays of partial waves from different oscillators. Methods to control these phases proposed up to now were rather exotic and difficult to realize; for instance, it was suggested to make a thin string of an active medium, illuminated by a spot of light which moves along the string with a superluminal velocity^{51,52}. Such approach, being difficult to realize in practice, provides only limited degree of control over the waveshape of the generated pulse.
Here we introduce a much more affordable approach based on linear diffractive optical elements (DOE), which allows to control the phases of the partial waves coming to the detector from different spatial positions and thus to shape the resulting THz waveform. In contrast to many other approaches, our method allows to control the electric field directly rather than its slow envelope, allowing to obtain exotic waveshapes such as rectangular or triangular ones.
The Setting
The idea of the proposed method is to generate subcycle THz waveforms in a thin nonlinear layer (see Fig. 1) and then phaseshape the partial waves on the way to the detector. We assume that the nonlinear layer possesses a lowfrequency resonance in the THz range. This resonance is excited by a pair of strong fewcycle pulses at the optical frequency. The first pulse, as it is shown in Fig. 2, starts the lowfrequency oscillators whereas the second stops it. In between the pulses, the oscillators radiate via free polarization decay mechanism^{50}. After the layer, we place a diffractive optical element (DOE), which influences the phases of the THz waves coming from different positions of the layer, and thus shapes the THz waveform at the detector.
The light propagation in the system is governed by the wave equation:
where \(P(\overrightarrow{r},t)\) is the polarization of the medium, \(E(\overrightarrow{r},t)\) is the electric field. We consider for simplicity the configurations where all fields are linearly polarized in the same direction, in which case a scalar version of the wave equation can be used.
We suppose furthermore the propagation everywhere except the thin nonlinear layer NM in Fig. 1 to be linear. The layer NM is to possess a lowfrequency resonance Ω_{0} = 2π/T_{0} in THz or MIR range. This can be, for instance, a Ramanactive medium^{53,54}, a set of nonlinearly coupled plasmon resonances in metallic nanoparticles^{55}, nanoantennas^{56,57}, quantum dots^{58}, and others. Of a special interest to date are novel types of materials such as graphene or topological insulators^{59,60,61} which also have strong broadband nonlinear response in the THz range^{60}. The polarization P in the nonlinear layer can be described by:
where Ω_{0} and γ are the resonant frequency and decay rate, respectively, and the function g[E(t)] describes the nonlinear coupling of the oscillators to the pump field. We assume that the frequency of the pump field is in the optical range, that is, far away from Ω_{0}. Because of the nonresonant excitation, the dynamics of populations in the layer can be neglected. We also assume that the molecules/particles are asymmetric and prealigned so that their coupling to the field can be described in lowest order by a quadratic nonlinearity^{38}: g[E] = g_{1}E. Such response is common in nonlinear crystals with a quadratic nonlinearity. However, here we are interested in the lowfrequency response rather than the instantaneous one. Eq. (2), coupled to Eq. (1) describes the light propagation and generation of new frequencies in the nonlinear layer. In particular, the slow oscillators, being excited by ultrashort optical pulses, can radiate after the pump is gone via free induction decay^{50}.
In the next subsections we proceed by consideration of (i) a response of a single oscillator in the layer to the pair of the pulses in Fig. 2 and estimating the nonlinear propagation effects in the layer, (ii) propagation of the partial waves from every of the oscillator to the DOE, (iii) modification of the THz radiation inside the DOE, and (iv) its propagation to the detector as shown in Fig. 1. Furthermore, in the subsequent sections we consider some specific examples of the media and DOEs, determining the resulting THz field amplitude.
Response of a single oscillator
It is natural to start our description from a single nonlinear oscillator. Suppose that an oscillator with the response defiend by Eq. (2) is excited by an ultrashort pulse at the optical frequency, so that the duration of the driving pulse τ_{0} is much smaller than the period of the resonant oscillations in the medium T_{0} = 2π/Ω_{0}. Since we are dealing here with fewcycle pulses, we can neglect the damping of the oscillators in Eq. (2), that is, we assume γ = 0. As it was shown in^{40}, under such excitation Eq. (2) can be integrated to give the lowfrequency response of the oscillator as:
Here the integration is performed over the whole pump pulse duration and beyond. The second term becomes negligible in Eq. (3) if a symmetric temporal shape of the excitation pulse^{40} is assumed. Thus, our simplified equation Eq. (3) shows that after the passage of the pump pulse the polarization of the medium continues to oscillate: P(t) = Π_{0}sin(Ω_{0}t) with \({\Pi }_{0}=\frac{1}{{{\rm{\Omega }}}_{0}}{\int }_{\infty }^{+\infty }g[E(t^{\prime} )]E(t^{\prime} )\cos ({{\rm{\Omega }}}_{0}t^{\prime} )dt^{\prime} \).
Oscillations of the polarization in turn lead to emission due to free induction decay^{50}. In order to generate a quasiunipolar pulse, the oscillations of polarization initiated by the first driver pulse have to be stopped by another (identical) pulse positioned at the distance τ_{p} = T_{0}/2 to the first one (see Fig. 2a). That is, the oscillator makes a half of its own proper oscillation before it is stopped by the “counterpulse”. Figure 2b illustrates the resulting temporal shape of the polarization scaled to \({g}_{1}{E}_{0p}^{2}{/{\rm{\Omega }}}_{0}^{2}\).
The field created by a single oscillator and observed at the distance D is then given by the wellknown solution of the wave equation Eq. (1) in the dipole approximation^{62}:
Figure 3 shows an example of the field produced by an oscillator Eq. (2) under the excitation by two Gaussian pump pulses. Its temporal profile is shown in Fig. 3b and consists of a central low frequeny halfsine wave as well as high frequency oscillations during the action of the pump pulses. Thus, the spectrum of the field shown in Fig. 3a consists of two distant parts, the lowfrequency one corresponding to the halfoscillation at Ω_{0} whereas the highfrequency one corresponds to the cooscillations with the driver pulse. This spectral separation allows us to get rid of highfrequency components by an appropriate lowpass filter, thus keeping only lowfrequency part of the spectrum in Fig. 3a. The spectrum does not contain a dccomponent, i.e. the resulting pulse is not unipolar. However, proper shaping of the filter as described below allows to transform the pulse in Fig. 3b into the quasiunipolar one, with long lowamplitude tails surrounding the central intense burst. The appropriate filter should cut off the frequencies with ω > ω_{cut} ≫ Ω_{0}. That is, its spectral transmission function Tr(ω) of the filter should have the following form:
where Θ is the Heaviside stepfunction, ω_{cut} is the cutoff frequency and a smooth function F[ω] has maximum around Ω_{0}. The function F[ω] is selected so that the waveshape possibly nearest to a single halfwaveshape such as shown in Fig. 2b appears and its particular form depends on the nonlinear coupling function of the specific medium. More detailed description of F(x) is given in the next subsection.
Assuming then that the highfrequency oscillations are filtered out, the emitted radiation E(t) has the form of a quasiunipolar pulse^{38,39,40} and without regard to the lateral tails, can be approximated as the sum of waves of the form Eq. (3) giving a half of a sinusoidal wave:
where t_{k} = t − kτ_{p}, and the effective scaling constant \(\tilde{E}\) is given as: \(\tilde{E}=\frac{{{\rm{\Omega }}}_{0}^{2}{{\rm{\Pi }}}_{0}}{{c}^{2}D}\).
Molecular model
In Eq. (2) we assumed a model of a general lowfrequency oscillator with a quadratic coupling to the IR field. Here we consider a more specific model representing a molecule with the dynamics of the nucleus and the electrons taken into account explicitly^{53,54} (we will call it a Ramanactive medium (RAM) model). The model is classical and consists of two nonlinearlycoupled oscillators with strongly different resonant frequencies, which we call the highfrequency oscillator (HFO) and lowfrequency oscillator (LFO), respectively. The role of HFO and LFO are played by the electrons and nuclei in the molecule^{54}. Let y, x denote the normal coordinates of LFO and HFO oscillations. The potential energy of the nonlinearlybonded oscillators in RAM is given as^{53,54}:
where α and β stand for the elastic constants of the both oscillators, factor γ gives the strength of the nonlinear bonding between HFO and LFO. The equations for the dynamics of RAM under a pulsed excitation take the following form^{53,54}:
Here, M and m are effective masses of the LFO and HFO, E_{p}(t) is the pump electric field, Γ_{e} and Γ_{n} are the damping rates of HFO and LFO, respectively, \({{\rm{\Omega }}}_{0}=2\pi /{T}_{0}=\sqrt{\frac{\beta }{M}}\) is the resonance frequency of LFO (nuclei), \({\omega }_{e}=2\pi /{T}_{e}=\sqrt{\frac{\alpha }{m}}\) is the natural frequency of the HFO (electrons). One can see that this model is to some extend equivalent to Eq. (2). Namely, the displacement of the nuclei in Eq. (8) is driven by the term proportional to x^{2} and hence, due to large ratio of the resonant frequencies of LFO and HFO, the effective equation for LFO can be reduced to Eq. (2).
We suppose that the medium is excited by two Gaussian pump pulses:
separated by the time interval τ_{p} = T_{0}/2. The polarization of the medium for such ultrashort pump pulses can not be adequately described anymore in terms of molecular polarizability and we apply the general definition:
where d is the dipole moment of a single RAM molecule (composed of two coupled oscillators), N is the concentration of molecules and the charges of both oscillators q are equal in magnitude due to charge neutrality of the medium.
Figure 4a shows the spectrum of the pulse obtained using Eq. (4) with the following parameters: Ω_{0} = 10^{13} rad/s, ω_{e} = 10^{15} rad/s, m = 9.1 · 10^{−28} g (electron mass), M = 1.6 · 10^{−24} g (proton mass), electric charge q = 4.8 · 10^{−10} ESU, γ = 10^{8} erg/cm^{3}, Γ_{e} = 10^{14} s^{−1}, Γ_{n} = 0 s^{−1}, E_{0p} = 10^{5} ESU, τ_{0} = 12.5 fs, ω_{0} = 2 · 10^{15} rad/s.
The lowfrequency part of the spectrum obtained with this more detailed model has a significantly lower amplitude but also somewhat different spectrum than the one based on the more simple Eq. (2) (as given in Fig. 3). To find the function F in the filter Tr(ω) making the response at the detector from single oscillator as “unipolar” as possible, we first notice that the envelope of the halfsine wave has the form of (ω/Ω_{0})^{−2} for \(\omega \gtrsim {{\rm{\Omega }}}_{0}\). At the same time, the lowfrequency part of the spectrum in Fig. 4a is decreasing much slower. Based on these arguments, the function F determining the properties of lowpas filter Eq. (5) at low frequencies has the following form:
where the factors μ_{1} and μ_{2} control the maximum of the filter transmittance and its bandwidth respectively. In Eqs (5) and (11) we take ω_{cut} = 10^{14} rad/s and μ_{1} = 25.26, μ_{2} = 13.33. The resulting transmittance function is shown in Fig. 5a.
The waveshape of the pulse from a single RAM molecule according to Eq. (4) after passing through the filter is shown in Fig. 4b. The observer is assumed to be located at the distance D = 10 cm ≫ R. It is seen that the pulse after the filter has the shape of a halfsine wave of the duration T_{0}/2 = 3.14 · 10^{−13} s^{−1} surrounded by weak long tails, which fully compensate the electric area of the central part, giving zero value of the dccomponent. However, it can be easily shown (see Appendix III) that taking into account these long tails results just in a minor modification of the amplitude of the pulse after an arbitrary DOE without changing the main part of its waveform.
As said, the form of the filter Eq. (5) is taken to smooth out the oscillations at the edges of the pulse. To illustrate this in more details we plot in Fig. 5b the pulse shape from a single RAM molecule Eq. (4) after passing through a simple lowpass step filter Tr(ω) = Θ[ω_{cut} − ω], with the same value of ω_{cut}. One can see that, although the highfrequency oscillations at the edges of the central halfsine wave are indeed filtered out, the emitted pulse still possesses sharp peaks at the edges. These peaks make the phasedelay control procedure we consider later more involved. Therefore we apply a filter with relatively narrow transmission bandwidth at low frequencies which turns these sharp peaks into the lowamplitude long tails in Fig. 4b.
Estimation of the generated THz field and of the nonlinear sideeffects
Even before we start considering the details of propagation it is reasonable to estimate the amplitude of the THz wave at the detector produced by the whole layer as well as the nonlinear propagation effects. For this, we assume the shape of the nonlinear layer to be a disk of the radius R = 1 cm, thickness h_{L} = 0.1 mm and concentration of oscillators N ≈ 5 · 10^{22} cm^{−3}. To achieve a reasonable THz field strength, the pump field in the RAM model described in the previous section should be taken to be 10^{5} ESU which corresponds to around 1 TW/cm^{2} intensity. The pulse with such large radius and intensity should have the energy of about 50 mJ. Nevertheless the big radius (much larger than the wavelength of THz radiation) is absolutely necessary for the following shape control. The total number of oscillators is N_{0} = NπR^{2}h_{L} ≈ 1.5 · 10^{21}. The amplitude at the detector can be estimated as a sum over all the oscillators as \({E}_{\Sigma } \sim {N}_{0}\tilde{E}\) which gives for these parameters around ≈300 ESU ≈100 kV/cm, which is comparable to the amplitudes generated by other methods.
We remark that the width of our layer is much less than THz wavelength so that we can neglect the THz diffraction in the layer. Besides, the THz wave is still too weak to demonstrate nontrivial nonlinear propagation effects on this length scale. The nonlinear propagation effects of the pump pulse can be also easily estimated and are still negligible for these parameters. Namely, for such a large pulse diameter we have the characteristic nonlinear distance n_{2}ω_{0}I/(cS) ≈ 100 m (here n_{2} is the nonlinear coefficient which we take as n_{2} = 3 × 10^{−16} cm^{2}/W for SiO_{2} for simplicity, S is the area of the beam and I is the intensity which is in our case around 1 TW/cm^{2}). The absorption of the pump wave can be estimated as the following: the part of the excited atoms is estimated as μ ≈ (Ω_{0}/ω_{0})^{2} ~ 10^{−4}. Therefore the energy μN_{0}ℏΩ_{0} absorbed from the pump should be around ≈0.5% of the pump energy and thus the absorption of the pump can be also neglected in the first approximation.
Propagation through DOE
The pulse in Fig. 3 obtained for Eq. (2) or alternatively from Eqs (7–10) actually gives the emission from a single oscillator. In the previous section we made a raw estimation of the generation efficiency by a simple summation of impacts of all oscillators without taking into account diffraction effects.
In the following we sum the impacts of all oscillators in a more rigorous way including all the elements in Fig. 1. We consider the further propagation to be linear and therefore described by a standard diffraction theory^{62}. As a first step, we calculate the field just in front of the DOE (see Fig. 1). We assume that the distance between the DOE and the layer is significantly larger than the transverse size of the layer. We denote the entrance plane of DOE as z = z_{0} (while z = 0 corresponds to the surface of the layer) and \({\overrightarrow{r}}_{\perp }\) is the vector in this plane. The wave equation Eq. (1) can be integrated to give the field \(E({\overrightarrow{r}}_{\perp },z={z}_{0},t)\) as^{62}:
where \(\tilde{D}\) is the distance between the center of the medium layer and point \({\overrightarrow{r}}_{\perp }\): \(\tilde{D}=\sqrt{{z}_{0}^{2}+{\overrightarrow{r}}_{\perp }^{2}}\), the integration is performed over the whole volume of the nonlinear medium (NM) and the unit vector \(\overrightarrow{u}^{\prime} \) is pointing from the center of the layer to the point \({\overrightarrow{r}}_{\perp }\). Let us assume that we consider the points at the entrance plane of DOE close to the optical axis, i.e. the term \(\frac{\overrightarrow{r}^{\prime} \overrightarrow{u}^{\prime} }{c}\) in Eq. (12) can be neglected. The wavefront in the farfield close to the optical axis can be approximated by a plane wave, what allows us to use scalar fields in Eq. (12).
Next, to correctly describe the propagation of the field Eq. (12) through the DOE we assume that the Fresnel number N_{F} of DOE is large:
where λ_{0} is the characteristic wavelength of the lowfrequency oscillations, h is the thickness and L_{⊥} is the characteristic transverse size (e.g., radius) of DOE. This condition ensures that the diffraction in DOE is yet negligible, and the action of the DOE can be described by the multiplication of the field in \(({\overrightarrow{r}}_{\perp },\omega )\) space by a transmission function:
which represents the standard transfer function of the Fresnel diffraction for large Fresnel numbers N_{F}^{62}. Here the delay function \(\tau ({\overrightarrow{r}}_{\perp })\) is related to the DOE thickness \(h({\overrightarrow{r}}_{\perp })\) as:
where n(ω) is the refractive index of the DOE. The transmission function Eq. (13) connects the field Eq. (12) just before the DOE with the field just behind the DOE. In this approximation, the DOE introduces a spatiallyvarying delay to the wavefront.
The dispersion inside the DOE can significantly influence the THz waveform. Nevertheless, the material of DOE is not obliged to have resonances near Ω_{0} (in contrast to the nonlinear layer), and thus a material with a flat dispersion can be chosen. Therefore, in the following consideration the dispersion in DOE is not taken into account. Nevertheless, since our THz pulses are broadband it is reasonable to set limits on the dispersion up to which the modifications in the waveshape of the THz pulse are still negligible. For this we must assume that the DOE width h ≪ L_{D}, where \({L}_{D}={\tau }_{{\rm{\Sigma }}}^{2}/{\beta }_{2}\) is the dispersion length of the material for our typical pulse, τ_{Σ} is the duration of the generated THz waveforms, β_{2} = ∂^{2}k/∂ω^{2} is the groupvelocity dispersion, k is the wavenumber. Thus, the dispersion can be neglected if the groupvelocity dispersion β_{2} in the vicinity Ω_{0} is: \({\beta }_{2}\ll {\tau }_{{\rm{\Sigma }}}^{2}/h\).
The field at the observation point
The shape of the resulting waveform at the distant observation point is determined by the superposition of halfcycle pulses passing through the different parts of the DOE. In the Fraunhofer approximation such superposition is given by the following integral (up to a phase factor which we skip for clarity):
where \(\overrightarrow{r}\) is the vector pointing to the observation point, \(\overrightarrow{u}=\overrightarrow{r}/\overrightarrow{r}\) is the unit vector pointing in the same direction and the vector \({\overrightarrow{r}}_{\perp }\) lies in the initial plane, which is now located just behind the DOE.
Summarizing the equations (6), (12), (13) we find for the field just behind the DOE (without regard to the lateral tails):
where N_{0} is the total number of oscillators in the layer. The field at the observation point Eq. (14) attains the form:
Eq. (15) implies that one can control the waveshape of the field at the detector by controlling of \(\tau (\overrightarrow{r})\), that is, by imposing certain delay in the arrival of the partial halfsine waves.
Shaping THz Waveforms with DOEs
Optical wedge
We now consider particular DOEs which allow to shape the resulting THz radiation. We start by considering the simplest 1D case (1D DOE) In this case we take the DOE in the form of a transparent wedge (see Fig. 1), the thickness of which varies linearly in the transverse direction (along laxis with l = 0 corresponding to the wedge apex) according to the expression:
where α is the apex angle of the wedge. We assume that the wedge is transparent in the THz range and neglect the dispersion as discussed before. Assuming that the THz field is detected in farfield at the observation angle ϕ (see Fig. 1), we obtain the following expression for the total time delay in dependence on the transverse lcoordinate:
where n is the refractive index of the wedge at Ω_{0}.
Based on Eq. (15), the electric field measured at the detector without regard to the weak lateral tails can be calculated as:
with a scaling factor E_{0} and
where L is the length of the layer, D is the distance from the wedge to the observer, ρ_{L} is the linear density of medium oscillators and the parameter η is given as:
The ananlytical solution of Eq. (16) is given in the Appendix I. Particularly, it is worth considering the value of the observation angle ϕ, given by the simple law of refraction. As can be easily seen, parameter η then turns to zero. From Appendix I one can see that for such case the field at the observation point represents just a halfsine wave times the total number of oscillators in the layer N_{0}, since all partial waves after DOE propagate inphase in this direction. However, for other values of the observation angle ϕ in the case when:
the field at the observation point attains the form of a rectangular quasiunipolar pulse with the amplitude:
According to Eq. (17) E_{m} is dependent on ϕ and the refractive index n. The shapes of the resulting pulses in this case are illustrated in Fig. 6. In Fig. 6a we consider three different exemplary values of n. The apex angle of the wedge was chosen as α = π/6 and the angle ϕ was assumed to be ϕ = π/6. As the refractive index increases, the pulse duration also increases, whereas the amplitude decreases accordingly. In Fig. 6b we study the dependence of the pulse width on the observation angle ϕ for n = 5.0. Similar to Fig. 6a the duration of the pulses increases with the angle ϕ of the observer.
Twodimensional DOE
Here we extend our method to more complex twodimensional geometries shown in Fig. 7. We again excite the system with two pulses as shown in Fig. 2 separated by the delay τ_{p} = T_{0}/2 and the resulting emission is passed through a DOE.
Proceeding analogously to the previous section we determine the field at the observation point as:
where E_{0} is a scaling factor, ρ_{S} is the surface density of medium oscillators, \({t}_{k}(r,\phi )=t\frac{D}{c}\tau (r,\phi )k{\tau }_{p}\), R is the disk radius, D  distance from the DOE to the observation point (we may assume that the field is measured in the focal plane of a lens placed parallel to DOE) and τ(r, φ) is the delay due to DOE. If we take the observation point at the symmetry axis of the DOE, τ(r, φ) is given, to an inessential constant value, by:
where the factor h(r, φ) describes the relief depth of the DOE and n is the refractive index of DOE at Ω_{0}.
Assuming first the spiral phase plate (SPP) having the relief of the form:
that is, phase but not radialdependent, we obtain the following time delay function τ(r, φ) corresponding to the phase plate:
with the constant A:
Taking for definiteness the following parameter values: n = 3, α = 10 μm/rad we can estimate the value of parameter A as A ~ 0.1 ps/rad.
Figure 8 shows the results of a numerical calculation of the field Eq. (18) for various values of the parameter A that corresponds to different relief depths of the phase plate. As one can see, the variation of the relief depth accessible with the DOE allows to produce rectangular quasiunipolar pulses with a tunable duration ranging from hundreds femtoseconds to tens picoseconds.
By varying the delay function we can gain even more control over the waveshapes of the THz response. For instance, Fig. 9 shows the results of calculation of Eq. (18) for a more complex case of a helical axicon with
In particular, depending on relative values of parameters A and κ we can obtain the continuous transition from the rectangular to triangular quasiunipolar pulses (see Fig. 9 as well as Appendix II for more details).
Conclusion
To summarize, we have proposed a new method for generation of subcycle THz pulses with unusual waveshapes such as rectangular or triangular ones. Our method is based on generation of halfoscillations of THz radiation in a nonlinear medium with a slow response with a pair of fewcycle pump pulses, and tuning the phase delay between different spatial parts of the resulting response using linear diffractive optics. According to our estimations, for high enough input powers in the tens of mJ range the amplitude of the generated field up to ≈100 kV/cm can be obtained. Although the threshold for THz generation is relatively low — in our simulations we needed intensities of 1 TW/cm^{2} or less, high pulse energies are required to achieve spatially broad pulses. The beam diameter must be in every case much larger that the THz wavelength and thus must be at least in mm range. More generally, one can ask if we can produce any required waveshape by tuning the phase delay h(r, φ) of DOE. This more general question is beyond the scope of this paper but there are no obvious reasons why it could not be possible. Moreover, by tuning the DOE one might be able to produce tunable THz pulses from more convenient and less broadband THz responses than considered here.
Thus, the results of the present work extend the possibility to generate customary waveshapes to the terahertz range. Such customary shapes of the electric field (rather than the slow envelope) was up to now a prerogative of electronics. We believe that such new possibility will be useful in ultrafast optical information processing. Furthermore, the method can be upscaled to the optical range if the strong attosecond pump pulses are used.
References
 1.
Roskos, H. G., Thomson, M. D., Kress, M. & Löffler, T. Broadband THz emission from gas plasmas induced by femtosecond optical pulses: From fundamentals to applications. Laser Photon. Rev. 1, 349–368 (2007).
 2.
Reimann, K. Tabletop sources of ultrashort THz pulses. Rep. Progr. Phys. 70, 1597 (2007).
 3.
Bartel, T., Gaal, P., Reimann, K., Woerner, M. & Elsaesser, T. Generation of singlecycle THz transients with high electricfield amplitudes. Opt. Lett. 30, 2805–2807 (2005).
 4.
Babushkin, I., Skupin, S. & Herrmann, J. Generation of terahertz radiation from ionizing twocolor laser pulses in Ar filled metallic hollow waveguides. Opt. Expr. 18, 9658–9663 (2010).
 5.
Lepeshov, S., Gorodetsky, A., Krasnok, A., Rafailov, E. & Belov, P. Enhancement of terahertz photoconductive antenna operation by optical nanoantennas. Laser Photon. Rev. 11, 1600199 (2017).
 6.
Jepsen, P. U., Cooke, D. G. & Koch, M. Terahertz spectroscopy and imaging–Modern techniques and applications. Laser Photon. Rev. 5(1), 124–166 (2011).
 7.
Parrott, E. P. & Zeitler, J. A. Terahertz timedomain and lowfrequency Raman spectroscopy of organic materials. Applied spectroscopy 69(1), 1–25 (2015).
 8.
Fülöp, J. A. et al. Efficient generation of THz pulses with 0.4 mJ energy. Opt. Expr. 22, 20155–20163 (2014).
 9.
Nugraha, P. S. et al. Demonstration of a tiltedpulsefront pumped planeparallel slab terahertz source. Opt. Lett. 44, 1023–1026 (2019).
 10.
Fülöp, J. A. et al. Highly efficient scalable monolithic semiconductor terahertz pulse source. Optica 3, 1075–1078 (2016).
 11.
Blanchard, F. et al. Terahertz pulse generation from bulk GaAs by a tiltedpulsefront excitation at 1.8 μm. Appl. Phys. Lett. 105, 241106 (2014).
 12.
Vicario, C. et al. Generation of 0.9mJ THz pulses in DSTMS pumped by a Cr:Mg_{2}SiO_{4} laser. Opt. Lett. 39, 6632–6635 (2014).
 13.
Lu, J. et al. Efficient terahertz generation in highly nonlinear organic crystal HMBTMS. Opt. Expr. 26, 30786–30794 (2018).
 14.
Babushkin, I. et al. Terahertz and higherorder Brunel harmonics: from tunnel to multiphoton ionization regime in tailored fields. J. Mod. Opt. 64, 1078–1087 (2017).
 15.
Kim, K. Y., Glownia, J. H., Taylor, A. J. & Rodriguez, G. Terahertz emission from ultrafast ionizing air in symmetrybroken laser fields. Opt. Expr. 15, 4577–4584 (2007).
 16.
Kim, K. Y., Taylor, A. J., Glownia, J. H. & Rodriguez, G. Coherent control of terahertz supercontinuum generation in ultrafast lasergas interactions. Nature Photon. 2, 605–609 (2008).
 17.
Babushkin, I. et al. Tailoring terahertz radiation by controlling tunnel photoionization events in gases. New J. Phys. 13, 123029 (2011).
 18.
Berge, L., Skupin, S., Köhler, C., Babushkin, I. & Herrmann, J. 3D Numerical Simulations of THz Generation by TwoColor Laser Filaments. Phys. Rev. Lett. 110, 073901 (2013).
 19.
Lepeshov, S. et al. Boosting Terahertz Photoconductive Antenna Performance with Optimised Plasmonic Nanostructures. Sci. Rep. 8, 6624 (2018).
 20.
Krausz, F. & Ivanov, M. Attosecond physics. Rev. Mod. Phys. 81, 163 (2009).
 21.
Wu, H.C. & MeyerterVehn, J. Giant halfcycle attosecond pulses. Nature Photon. 6, 304–307 (2012).
 22.
Xu, J. et al. Terawattscale optical halfcycle attosecond pulses. Sci. Rep. 8, 2669 (2018).
 23.
Hassan, M. T. et al. Optical attosecond pulses and tracking the nonlinear response of bound electrons. Nature 530, 66–70 (2016).
 24.
Arkhipov, M. V. et al. Generation of unipolar halfcycle pulses via unusual reflection of a singlecycle pulse from an optically thin metallic or dielectric layer. Opt. Lett. 42, 2189–2192 (2017).
 25.
Ramasesha, K., Leone, S. R. & Neumark, D. M. Realtime probing of electron dynamics using attosecond timeresolved spectroscopy. Annu. Rev. Phys. Chem. 67, 41–63 (2016).
 26.
Corkum, P. B. & Krausz, F. Attosecond science. Nature physics 3(6), 381 (2007).
 27.
Calegari, F. et al. Advances in attosecond science. J. Phys. B 49(6), 062001 (2016).
 28.
Arkhipov, R. M. et al. Unipolar subcycle pulsedriven nonresonant excitation of quantum systems. Opt. Lett. 44, 1202–1205 (2019).
 29.
Arkhipov, R. M. et al. Generation of unipolar pulses in nonlinear media. JETP Lett. 105, 408–418 (2017).
 30.
Bullough, R. K. & Ahmad, F. Exact Solutions of the SelfInduced Transparency Equations. Phys. Rev. Lett. 27, 330 (1971).
 31.
Kalosha, V. P. & Herrmann, J. Formation of optical subcycle pulses and full MaxwellBloch solitary waves by coherent propagation effects. Phys. Rev. Lett. 83, 544 (1999).
 32.
Kaplan, A. E. & Shkolnikov, P. L. Electromagnetic â€œbubblesâ€ and shock waves: unipolar, nonoscillating EM solitons. Phys. Rev. Lett. 75, 2316–2319 (1995).
 33.
Song, X., Yang, W., Zeng, Z., Li, R. & Xu, Z. Unipolar halfcycle pulse generation in asymmetrical media with a periodic subwavelength structure. Phys. Rev. A 82, 053821 (2010).
 34.
Vysotina, N. V., Rosanov, N. N. & Semenov, V. E. Extremely short dissipative solitons in an active nonlinear medium with quantum dots. Opt. Spectr. 106, 713–717 (2009).
 35.
Leblond, H. Halfcycle optical soliton in quadratic nonlinear media. Phys. Rev. A 78, 013807 (2008).
 36.
Kozlov, V. V., Rosanov, N. N., Angelis, C. D. & Wabnitz, S. Generation of unipolar pulses from nonunipolar optical pulses in a nonlinear medium. Phys. Rev. A 84, 023818 (2011).
 37.
Leblond, H. & Mihalache, D. Models of few optical cycle solitons beyond the slowly varying envelope approximation. Phys. Rep. 523, 61–126 (2013).
 38.
Arkhipov, R. M., Arkhipov, M. V., Babushkin, I., Belov, P. A. & Tolmachev, Y. A. Generation of unipolar optical pulses in a Ramanactive medium. Laser Phys. Lett. 13, 046001 (2016).
 39.
Arkhipov, A. V. et al. Fewcycle pulsedriven excitation response of resonant medium with nonlinear field coupling. Laser Phys. Lett. 13, 126001 (2016).
 40.
Pakhomov, A. V. et al. Alloptical control of unipolar pulse generation in a resonant medium with nonlinear field coupling. Phys. Rev. A 95, 013804 (2017).
 41.
Bessonov, E. G. On a class of electromagnetic waves. Sov. Phys. JETP 53, 433–436 (1981).
 42.
Bessonov, E. G. Conventionally strange electromagnetic waves. Nucl. Instr. and Meth. A 308, 135–139 (1991).
 43.
Weiner, A. M. Ultrafast optical pulse shaping: A tutorial review. Opt. Commun. 284, 3669–3692 (2011).
 44.
Mendis, R., Nag, A., Chen, F. & Mittleman, D. M. A tunable universal terahertz filter using artificial dielectrics based on parallelplate waveguides. Appl. Phys. Lett. 97, 131106 (2010).
 45.
Kawada, Y., Yasuda, T. & Takahashi, H. Carrier envelope phase shifter for broadband terahertz pulses. Opt. Lett. 5, 986–989 (2016).
 46.
Gingras, L. & Cooke, D. G. Direct temporal shaping of terahertz light pulses. Optica 4, 1416–1420 (2017).
 47.
Gingras, L., Cui, W., SchiffKearn, A. W. & Ménard, J. M. Active phase control of terahertz pulses using a dynamic waveguide. Opt. Expr. 26, 13876–13882 (2018).
 48.
Nugraha, P. S. et al. Efficient semiconductor multicycle terahertz pulse source. J. Phys. B 51, 094007 (2018).
 49.
Lu, J. et al. Tunable multicycle THz generation in organic crystal HMQTMS. Opt. Expr. 23, 22723–22729 (2015).
 50.
Allen, L., Eberly, J. H. Optical Resonance and TwoLevel Atoms (John Wiley & Sons Inc., New York, 1975).
 51.
Arkhipov, R. M. et al. Generation of Extremely Short Pulses upon Excitation of a Resonant Medium by a Superluminal Light Spot. Opt. Spectr. 124, 536–540 (2018).
 52.
Ziguleva, D. O. et al. Laser beam deflector based generation of fewcycle electromagnetic pulses in a circular nonlinear medium. Opt. Commun. 424, 170–176 (2018).
 53.
Platonenko, V. T. & Khokhlov, R. V. On the Mechanism of Operation of a Raman Laser. Sov. Phys. JETP 19, 378–381 (1964).
 54.
Akhmanov, S. A., Nikitin, S. Yu. Physical Optics (Nauka, Moscow, 2004; Clarendon, Oxford, 1997).
 55.
Ginzburg, P. et al. Nonlinearly coupled localized plasmon resonances: Resonant secondharmonic generation. Phys. Rev. B 86, 085422 (2012).
 56.
Novotny, L. & van Hulst, N. Antennas for light. Nature Photon. 5, 83–90 (2011).
 57.
Biagioni, P., Huang, J.S. & Hecht, B. Nanoantennas for visible and infrared radiation. Rep. Prog. Phys. 75, 024402 (2012).
 58.
Bera, D., Qian, L., Tseng, T. K. & Holloway, P. H. Quantum dots and their multimodal applications: a review. Materials 3, 2260–2345 (2010).
 59.
Yao, X., Tokman, M. & Belyanin, A. Efficient Nonlinear Generation of THz Plasmons in Graphene and Topological Insulators. Phys. Rev. Lett. 112, 055501 (2014).
 60.
Hartmann, R. R., Kono, J. & Portnoi, M. E. Terahertz science and technology of carbon nanomaterials. Nanotechn. 25, 322001 (2014).
 61.
Chen, S. et al. Broadband optical and microwave nonlinear response in topological insulator. Opt. Mater. Expr. 4, 587–596 (2014).
 62.
Born, M., Wolf, E. Principles of optics (Pergamon Press, 1980).
Acknowledgements
Russian Science Foundation, project 171901097 (theoretical consideration in sections 1.3–1.5 and analysis of pulse shaping by DOE in section 2). Russian Foundation for Basic Research, 190200312a (demonstration of quasiunipolar pulse generation in RAM in section 1.2). I.B., A.D. and U.M. are thankful to Deutsche Forschunggemeinschaft (DFG), projects BA 4156/42, MO 85019/2, as well as Germany’s Excellence Strategy within the Cluster of Excellence PhoenixD (EXC 2122, Project ID 390833453). The publication of this article was funded by the Open Access Fund of the Leibniz Universität Hannover.
Author information
Affiliations
Contributions
A.V.P. provided analytical theory, R.M.A. and M.V.A. performed numerical simulations, I.B., A.D., U.M. and N.N.R. analysed the results and participated in the discussions, article formulation and writing. All authors reviewed the manuscript.
Corresponding author
Ethics declarations
Competing Interests
The authors declare no competing interests.
Additional information
Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary information
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Pakhomov, A.V., Arkhipov, R.M., Arkhipov, M.V. et al. Unusual terahertz waveforms from a resonant medium controlled by diffractive optical elements. Sci Rep 9, 7444 (2019). https://doi.org/10.1038/s4159801943852w
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s4159801943852w
Further reading

Population difference gratings created on vibrational transitions by nonoverlapping subcycle THz pulses
Scientific Reports (2021)
Comments
By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.