WorksPhysics and Astronomy
Polaronic Effect in High-Harmonic Generation
Abstract
We investigate High-Harmonic Generation (HHG) in the Su-Schrieffer-Heeger (SSH) chain with electron-phonon coupling modeled via the Holstein interaction. The system dynamics are simulated using the tight-binding approximation, with local phonons approximated as quantum harmonic oscillators. Phononic degrees of freedom significantly expand the Hilbert space dimension. This interaction modifies the eigenenergy spectrum by introducing new states within previously existing gaps, enhancing the harmonic yield through additional allowed transitions.
Introduction
High-Harmonic Generation (HHG) spectroscopy is a powerful technique in condensed matter physics for probing ultrafast electron dynamics driven by intense, ultrashort laser pulses. In HHG, a strong laser field accelerates electrons in a material, leading to the emission of radiation containing harmonics of the driving laser frequency [1, 2]. Initially studied in gases–enabling attosecond pulse generation [3]–HHG opened new possibilities for investigating ultrafast dynamics [4]. The first experimental observations of HHG in bulk crystals [5] provided insights into the electronic structure of solids. Subsequent studies revealed that the harmonics generated in solids encode information about the material [6], including its band structure [7], topological phase [8], and other properties. This occurs because the accelerated electrons move through the material according to its electronic bands under the influence of the laser field. Therefore, any modifications to the electronic band structure can affect the electron dynamics and, consequently, alter the HHG spectrum. As a result, HHG has emerged as a promising technique for studying the band structure and dynamical properties of materials [5, 7, 9–23].
In the recent years, there has been a growing interest in the study of topological condensed matter and strong-field HHG theoretically [10, 11, 24–27] and experimentally [28–33]. A key focus is to understand how the HHG spectrum is influenced by additional interactions, such as electron-electron (e-e), electron-phonon (e-ph) interactions, many-body effects, or topological phases.
Significant progress has been made in the study of the many-body HHG, with most discussions centered in the e-e interaction [34] and the topological characteristics of the harmonic spectra [10]. Nevertheless, little attention has been given to the e-ph interactions, despite the fact that several classes of materials that exhibit e-ph interaction such as in crystals [35, 36] or other materials [37–54]. The e-ph interaction is a fundamental aspect of condensed matter physics, playing a key role in determining materials properties and influencing a wide variety of phenomena, from superconductivity to phase transitions.
In this context, the Holstein model [55] is a prototypical framework for describing e-ph interactions [56–59], where electrons are coupled to Einstein phonons (i.e., dispersionless local phonons). A key feature of the Holstein model is that the local e-ph interaction leads to the formation of a bound e-ph state. This results in the creation of a quasiparticle known as a polaron [60].
In the literature, the Holstein model has been widely used, often in conjunction with other models, to describe various types of many-body interactions in condensed matter systems. This combined models provide a quantum mechanical framework for studying rich phenomena such as superconductivity, antiferromagnetism, and charge-ordered phases [61], as well as for simulating these effects in quantum devices [62]. In the context of HHG, the Holstein model has been employed to investigate dynamical field theory and scattering effects [63], as well as phonon-mediated interactions in superconductors [64] and Mott insulators [65]. However, despite these important advances, the specific role of e-ph coupling in HHG spectra remains relatively underexplored, particularly regarding how phonon dynamics may influence ultrafast nonlinear optical responses.
In this work we study the effect of the Holstein polaron [55] on HHG. To this end, we perform numerical calculations within the tight-binding approximation to simulate the coupled electronic and phononic dynamics under a strong laser field. We find that including phonon degrees of freedom causes the Hilbert space to grow exponentially. As a result, we are constrained to simulate only relatively short systems with one electron and a few phonon energy levels. Therefore, we use the Su–Schrieffer–Heeger (SSH) chain [66], a minimal lattice that exhibits a band gap and requires only a few sites per unit cell. The SSH chain has been previously studied in the context of HHG using time-dependent density-functional theory [10], as well as within tight-binding models without e-e interaction [11] and with the Hubbard e-e interaction [12]. However, the effect of different e-ph couplings of Holstein type on HHG in the SSH model has, to best of our knowledge, not been explored. Understanding how phonon-mediated interactions modify HHG spectra can shed light on the ultrafast response of correlated materials and help identify new mechanisms for spectral control. The main body of this work is structured into four sections: (i) The introduction. (ii) The theoretical framework, beginning with the Hamiltonian of the SSH Holstein system and basis construction. We then introduce the methodology used for the HHG spectroscopy. (iii) Simulation results, including electron and phonon dynamics. We analyze the spectra across various values of the e-ph coupling to explore its effect on harmonic generation. (iv) We presents our conclusions with a summary of our main findings.
Theory
The Holstein SSH model
Consider the SSH lattice as sketched in the Figure 1, which consists of two sites per unit cell: and . There are two types of electron-electron hoppings: represented by a single line, and represented by a double line. The average distance between ions is , and a small displacement in the ions leads to dimerization, however, this displacement is approximately two orders of magnitude smaller than . We define as the total length of the chain and set the reference frame such that the first site is located at and the last site at . The Holstein SSH Hamiltonian is given by:
Figure 1. Sketch of the model. Sites are represented by black circles. The hoppings and are shown as single and double gray lines, respectively. The average distance between ions is . The e-ph interaction is indicated by dashed gray lines. Local phonons are illustrated in red.
where is the electron kinetic term, represents the phonon energy, and denotes the e-ph interaction. The term is given by:
where and are the electron annihilation and creation operator at site in the th unit cell, and is the total number of unit cells.
Hilbert space
In the Holstein model, each site hosts localized quantum phonon states described by the solutions of the quantum harmonic oscillator [55, 60]. In the sketch shown in Figure 1, these localized phonon states are illustrated in red as quantized energy levels of the harmonic oscillator, while the e-ph hopping is depicted with dashed gray lines. In this manner is:
where is the phonon energy, , are the annihilation and creation operators of local phonons ().
The term that mediates the e-ph interaction is
where is the electron number operator. The inclusion of the e-ph interaction produces a decrease in the ground state energy when a polaron forms, a hypothesis we will examine later in this work.
Hilbert space
The Hilbert space of the e-ph system is constructed as , where and denote the Hilbert space of the electronic and phononic subsystems, respectively. The phonon number per site is truncated to levels for numerical tractability, and we use one electron. For unit cells, , and , and then the total dimension . To limit , we use and (as shown in Figure 1) and assume low-energy phonons dominate low-lying excitations, justifying a small . We verify convergence of spectra with increasing . We write a general state as:
where labels the site in which the electron is localized, and denotes the number of phonons at the th site, is the normalized probability amplitude of the basis element .
Coupling parameters
All reported parameters are given in atomic units (a.u.). We perform simulations using the electron hopping parameters and , which are similar values as in Ref. [66]. Phonons are more easily excited when is small compared to the phononless transition energies and . In this case, a larger cutoff is needed to accurately describe the phononic states, leading to an exponential increase in the basis and matrix size, this is an issue we aim to avoid. Conversely, if is large compared to the phononless transition energies and , then fewer phonons are excited, and the low-energy states closely resemble the phononless case. Therefore, we seek a regime for and that does not require a large while still allowing the low-energy states to differ noticeably from the phononless scenario. To achieve this, we fix and vary , ensuring that is neither too small or too large relative to the phononless transition energies and . Therefore, we select , which is of the same magnitude order as the electron parameters. For the e-ph coupling we initially use the value , but this value is varied to explore the parameter space. For the laser parameters, we set the number of cycles to , the peak vector potential to , and the laser frequency to .
Energy levels
Only a few key eigenstates are populated by the laser, as harmonic peaks mainly arise from ground state transitions to select excited states within the harmonic energy range [12]. Thus, we focus on harmonics up to the 40th order. To improve computational efficiency, we restrict the eigenstate basis to those with eigenenergies within this region of interest. We determine the harmonic order corresponding to an eigenenergy using the relation , where is the laser frequency. We set the laser frequency to around 20 times the energy gap between the ground state and the first excited state for . For each , we compute the energy spectrum, then we use the lowest eigenstates covering up to at least the 40th harmonic order. Figure 2 shows the energy levels for , up
to the 40th harmonic order. We use for , for and for . We notice that for the case the new states appear at higher energies and with we cover the region of interest. Highlighted in blue are the ground state and excited states significantly populated via transitions from it (see Eq. (11)). For , the local phonons can only occupy their ground state, so the phonon creation and annihilation operators cannot induce transitions, and e-ph interaction is suppressed. As a result, this case is equivalent to the phononless scenario. For , the phonon operators allow e-ph interaction, and we observe in Figure 2 that the ground state energy decreases compared to the case. This is a manifestation of the e-ph binding energy discussed earlier. For , the ground state energy is similar to , indicating that the ground state energy has begun to converge with respect to . This observation supports the ansatz introduced earlier, which states that low-energy local phonons play a more significant role in the dynamics of low energy states. The supplementary material includes electron and phonon distributions of the most relevant states for highlighted in blue in Fig.2.
Figure 2. The energy levels for phonon cutoffs . Blue indicates strong transitions from the ground state. Parameters used , and phonon parameters , (all in a.u.).
High-Harmonic Generation
To perform the high-harmonic simulation, we add a laser interaction term to the Hamiltonian. We consider a classical laser field polarized along the chain direction, treated within the dipole approximation in the length gauge. Then the Hamiltonian is:
where is presented in Eq. (1), is the electric field amplitude and is the position operator of electrons that we calculate as
where (a.u.) is the average distance between ions. We simulate the system over the time interval , using the electric field defined by:
where is the laser frequency, is the number of field cycles and is the field amplitude. Using the Hamiltonian in Eq. (6) we solve the time-dependent Schrödinger equation:
We express the general state as a linear combination of the eigenstates of as follows
The time dependence is contained in the electric field amplitude , while the braket forms a time invariant term that mediates the transition between states, which we define as the transition matrix , and we also define the ground state transition vector as:
Where we have defined due to the ground state transitions are more important because of the initial condition.
The previous expansion in the time-dependent Schrödinger equation in Eq. (9) leads to:
where is the eigenvalue of the eigenstate , and . The term inside the sum mediates the transition between the states and . We solve these ordinary differential equations using the RK4 method, with the ground state , as the initial condition, thus obtaining .
To compute the emitted harmonic yield, we employ a semi-classical approach where the radiated light is proportional to the electron acceleration. We calculate the electron position expectation value, apply a three point discrete derivative to calculate the acceleration, Then we use a Hanning window and perform a Fast Fourier Transform (FFT) to obtain the electron oscillation frequency. Finally, we take the absolute square and compute the base 10 logarithm as follows:
where is the frequency obtained from the FFT. We normalize the yield to the intensity of the fundamental harmonic using , where we subtract the fundamental harmonic yield due to the use of the logarithmic scale.
Results
Density Distribution
We begin by analyzing the electron and phonon clouds using the computed . Figure 3 displays their distributions alongside the electric field profile (dashed red line). Panel (a) shows the electron density, calculated as:
Figure 3. (a) The electron density distribution in Eq.14 during the simulation. The dashed red line shows the electric field shape. (b) Phonon density distribution. We used the same coupling and phonon parameters as in Fig. 2, and the laser parameters were , , and (a.u.).
with . In the figure, we calculate the position of the electron and phonons at the th site using . We observe in the figure that the electron cloud follows the shape of the electric field. Figure 3.(b) shows the phonon density distribution, which we calculate using the previous equation, but substituting the electron number operator with the phonon number operator, . We observe that the phonon density distribution follows the electron density distribution, which is a consequence of the Holstein interaction.
Harmonic spectrum
Figure 4.(a) shows the harmonic spectrum for the first 40 harmonic orders and . For the case , the Hilbert space has dimension , and we use states in the simulation. For , and we use . For , and we use . As we increase the new states
appear at higher energies, therefore is enough for and . Figure 4(b) shows the eigenstates in function of and the harmonic order. For , we observe the highest peak at the 1st harmonic. The harmonic yield then decreases until around the 13th harmonic. After the 13th harmonic, the yield increases again, reaching a maximum around the 20th harmonic. Beyond this point, the harmonic yield decreases. When observing the eigenstates for in Figure 4(b), there is a state around the 20th harmonic, this corresponds to the first excited state, which causes the increase in the harmonic yield around this region.
Figure 4. (a) The normalized harmonic yield as a function of the harmonic order for . (b) The most relevant eigenstates based on . The same coupling and laser parameters were used as in Fig.3.
We observe that for the results are converged because the yield data overlap with those for and , which agrees with the ansatz stated before. For , we see that around the 13th harmonic order, the harmonics generated are around two orders of magnitude larger compared to . This increase in the harmonics is due to new states that emerge from the e-ph interaction. Additionally, the inclusion of the e-ph coupling decreases the ground state energy, which also shifts the harmonic order of the excited states. This shift can be observed in Figure 2 by comparing the harmonic order labels on the right. Moreover, this change in the harmonic order of the excited states also shifts the plateau of the harmonic yield around the 20th harmonic.
In order to understand how different values of the e-ph interaction parameter affect the harmonic yields, we plot the normalized harmonic yield for various values in Figure 5. We also highlight with dots the magnitude order of the ground state transition vector (ranging from black to white according to , where white indicates a high value and black a low value). The vertical dotted red line indicates the value used for Figure 3 and Figure 4. We observe that as decreases from zero, the harmonic yield is enhanced around the 13th harmonic, while
Figure 5. as a function of the e-ph coupling . The points shown in gray scale correspond to states with a high . The vertical red line indicates the used in Figure 4.
the harmonics above the 20th harmonics are diminished. This can be explained by the fact that as decreases, the ground state transition value of the state near the 13th harmonic changes from gray to white, indicating that this state becomes increasingly relevant. As this state gains relevance, the harmonic yield in this region increases, explaining the enhancement around 11th and 15th harmonics. Moreover, the decrease of the harmonic yield above the 20th harmonic is also related to this state. As decreases, the low energy states become dominant, and the probability amplitude concentrates more on these dominant states. This leads to a diminution in the population of the states above the 20th harmonic, causing the harmonic yield in this region to diminish.
Conclusions
In this work, we studied the effect of the Holstein polaron on HHG. From the Hamiltonian, we identified an e-ph binding energy by comparing the phononless case () with the case including phonons (). From the simulation results, we examined the electron and phonon density distribution and observed that the electron cloud follows the shape of the electric laser field, as expected. Additionally, we found that the phonon cloud follows the movement of the electron cloud, indicating a binding between the electron and the phonons.
The inclusion of phononic degrees of freedom introduces new energy levels accessible form the ground state can transition to during the simulation. These phonon induced states become increasingly relevant as the e-ph coupling strength increases and decreases. Simulations show that the harmonic yield is significantly modified by these new states, increasing by up to two orders of magnitude in regions where they emerge. As a result, emission from higher-energy states is reduced due to the dominant population of low-energy states, leading to a decreased harmonic yield near the first excited state of the phononless case.
Acknowledgment
We gratefully acknowledge funding by the German Research Foundation through IRTG 2676/1 “Imaging of Quantum Systems”, project no. 437567992, and CRC 1477 “Light-Matter Interactions at Interfaces,” Project No. 441234705.
Data Availability
All data generated and the code used in this study are available in GCA’s Gitlab repository at https://gitlab.uni-rostock.de/zr2304/data-and-code-of-polaronic-effect-in-high-harmonic-generation.
Supplementary material
Relevant Eigenstates
In the simulation, the most relevant state is the ground state, as the system starts from this state. Additionally, the states with strong transition from the ground state are important. We identify these states by selecting those with the highest values of , where the log is to sort the result by order of magnitude.
The inclusion of phonons leads to more complex eigenstates, each characterized by a distinct electron and phonon distribution. Figure 6.(a) shows the ground state, where we plot both the electron and phonon distributions. To this end, we calculate the expectation value of the product of the phonon number operator and the electron number operator at different sites:
Figure 6. Electron distribution relative to the phonon distribution for the four most relevant states. and indicate the -th and -th site, respectively. (a) Ground state (). (b) First excited state (). (c) . (d) . Panels (b), (c) and (d) correspond to the states with the highest values of . The same coupling and phonon parameters were used as in Figure 2.
where and indicate the th and th site, respectively, and refers to the th eigenstate. This quantity illustrates how the electron distribution correlates with the phonon distribution in the eigenstate . For the ground state, we observe that electrons and phonons are more likely to be together near the center of the lattice. The three eigenstates with the higher values of are as follows: the first excited state () in Figure 6.(b), seventh () in Figure 6.(c), and twelfth () in Figure 6.(d). These states are highlighted in blue in Figure 2.
References
- [64]the retarded phonon-mediated interaction, PHYSICAL REVIEW B 94, 10.1103/PhysRevB.94.224519 (2016).
- [65]S. Sota, S. Yunoki, and T. Tohyama, Density-matrix renormalization group study of third harmonic generation in one-dimensional mott insulator coupled with phonon, Journal of the Physical Society of Japan 84, 054403 (2015), https://doi.org/10.7566/JPSJ.84.054403.
- [66]W. P. Su, J. R. Schrieffer, and A. J. Heeger, Solutions in polyacetylene, Phys. Rev. Lett. 42, 1698 (1979).