2025 Volume 38 Issue 6
Article Contents

Jiayue Han, Yu Wang, Wenjie Dou. Mixed Quantum–Classical Approaches to Electron Transmission through Chiral Molecules†[J]. Chinese Journal of Chemical Physics, 2025, 38(6): 833-844. doi: 10.1063/1674-0068/cjcp2509149
Citation: Jiayue Han, Yu Wang, Wenjie Dou. Mixed Quantum–Classical Approaches to Electron Transmission through Chiral Molecules†[J]. Chinese Journal of Chemical Physics, 2025, 38(6): 833-844. doi: 10.1063/1674-0068/cjcp2509149

Mixed Quantum–Classical Approaches to Electron Transmission through Chiral Molecules†

通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Figures(8)

Article Metrics

Article views(104) PDF downloads(0) Cited by(0)

Access History

Mixed Quantum–Classical Approaches to Electron Transmission through Chiral Molecules†

Abstract: The chiral-induced spin selectivity (CISS) effect has attracted widespread interest due to its potential applications in spintronics. However, the influence of electron–vibration coupling on spin polarization has no unified theoretical description. In this work, we revisit the widely used model proposed by Jonas Fransson [Phys. Rev. B 102, 235416 (2020)] with two mixed quantum–classical approaches, namely mean-field (MF) dynamics and surface hopping (SH). Our results show that, in the absence of electron–vibration coupling, the transient spin polarization vanishes for short molecular chains or without spin–orbit coupling (SOC). For longer chains and stronger SOC, the polarization exhibits coherent oscillations reaching up to $ \pm 100\% $. When electron–vibration coupling is included, pronounced spin polarization emerges only at early femtosecond timescales, whose magnitude increases with SOC strength and chain length but decreases as the electron–vibration coupling grows. On picosecond timescales, the polarization decays and fluctuates narrowly around zero, yielding a negligible long-time average. These findings indicate that electron–vibration coupling generally suppresses spin polarization, in contrast to predictions in previous studies. We suggest that alternative or extended theoretical frameworks are required to elucidate the key role of electron–vibration coupling in the CISS effect.

    • Since Louis Pasteur’s pioneering discovery of the connection between molecular chirality and optical activity, chirality has been established as a fundamental concept across the natural sciences [1–4]. Subsequent studies further revealed that it also exerts a profound influence on electronic processes [5]. A prominent manifestation of this influence is the chiral-induced spin selectivity (CISS) effect, in which electrons acquire pronounced spin polarization when transmitted through chiral molecules, polymers, or nanostructures [6, 7].

      The CISS effect was first reported by Ray and co-workers in 1999, who observed a marked asymmetry in electron scattering in the presence of chiral molecules [8]. More than a decade later, Göhler et al. provided direct evidence of spin-selective electron transmission through monolayers of double-stranded DNA [9]. Since then, the effect has been established as a robust experimental phenomenon and explored in diverse contexts, including spintronics [10, 11], chemistry [12–14], and life sciences [15–17].

      Despite the extensive experimental evidence, the theoretical description of the CISS effect remains incomplete. It is generally acknowledged that spin–orbit coupling (SOC) plays a crucial role in the emergence of CISS [18–20]. However, theoretical attempts to reproduce the large polarization observed in spectroscopic experiments often necessitate amplifying the strength of SOC by several orders [21]. This discrepancy persists even when additional mechanisms such as Berry force [22–25], geometric and vibronic effects (e.g., the Jahn–Teller effect) [26–28], redefinition of the figure of merit [29], and interfacial effects [30, 31] are comprehensively considered.

      Another widely discussed hypothesis links the CISS effect to non-Born–Oppenheimer dynamics, with several theoretical studies exploring how electron–vibration coupling may influence spin polarization [32–36]. Among the available models, the most influential is the tight-binding model of a helical molecular chain proposed by Jonas Fransson [37], which extends the framework of the Kane and Mele model [38–40]. Despite its demonstrated reliability and widespread use, studies based on this model have produced questionable results. Fransson himself reported spin polarization up to approximately 35% in the absence of an external magnetic field, which appears surprisingly large for such a simplified model. By contrast, Smorka and co-workers argued that significant spin polarization emerges only in single-trajectory analysis, where vibrational degrees of freedom are fixed at specific initial positions and momenta [41]. This setup effectively corresponds to a zero-temperature limit, which is difficult to reconcile with both the room-temperature manifestation of CISS [42] and the basic requirements for initializing classical degrees of freedom in mixed quantum–classical dynamics [43].

      To address the above controversy, we adopts the same model Hamiltonian to explicitly investigate the role of electron–vibration coupling in electron transmission. Unlike electron transport setups in the original study, we consider a closed molecular system in which electrons behave as free particles undergoing three-dimensional scattering [44]. This setting provides a controlled platform for isolating the intrinsic contribution of molecular chirality, free from the strong spin–orbit coupling in metal substrates and electrodes, as well as the interfacial effects that often complicate CISS measurements [45–47]. In selecting the simulation methods, fully quantum–mechanical approaches such as nonequilibrium Green’s functions (NEGF) [48, 49], hierarchical equations of motion (HEOM) [50, 51], and quantum master equations (QME) [52, 53] are primarily designed for open quantum systems coupled to external reservoirs and are therefore less suited for the closed-system scenario considered here. Consequently, we adopt a more practical strategy by employing mixed quantum–classical approaches [54–56] to capture nonadiabatic effects arising from electron–vibration coupling, which offer a robust balance between accuracy and computational efficiency.

      This paper is organized as follows: In Sec. II, we introduce the model Hamiltonian adopted in this study. Sec. III provides a brief overview of the key methods, including mean–field (MF) dynamics and surface hopping (SH). In Sec. IV, we discuss the results of spin-resolved dynamics with different parameters. Finally, we conclude in Sec. V.

    II.   MODEL HAMILTONIAN
    • In this study, we adopt the helical molecular chain model proposed by Jonas Fransson in Ref.[37] to investigate the role of electron–vibration coupling in the CISS effect. The molecular Hamiltonian $ H_{{\rm{mol}}} $ consists of three parts:

    • The purely electronic contribution is written as

      The chiral molecule is modeled as a helical structure with $ N $ sites, where $ N = N_{{\rm{laps}}} \times N_{{\rm{sites}}} $. Here, $ N_{{\rm{laps}}} $ is the number of helical windings and $ N_{{\rm{sites}}} $ is the number of sites per winding. Each discrete site is labeled by the index $ j $ and carries a spin degree of freedom $ s \in \{\uparrow, \downarrow\} $. The operators $ c_{js}^{\dagger} $ and $ c_{js} $ represent fermionic creation and annihilation operators on site $ j $ with spin $ s $. Specifically, the first term describes the on-site energy for each site. The equal site-energy limit ($ \epsilon_j=\epsilon_0 $) renders the helical chain homogeneous. The second term accounts for nearest-neighbor hopping with amplitude $ t_0 $, and the third term incorporates spin–orbit coupling between second-nearest neighbors with strength $ \lambda_0 $.

      Each site $ j $ is associated with a position vector $ {\boldsymbol{r}}_j = \left( a \cos\phi_j,\, a \sin\phi_j,\, h_j \right) $, where $ a $ is the radius of the helix, $ \phi_j = 2\pi (j - 1) N_{{\rm{laps}}} / (N - 1) $ is the azimuthal angle, and $ h_j = (j - 1) h / (N - 1) $ is the height of the site along the helix axis, with $ h $ being the total pitch. Unit direction vectors between sites are defined as $ {\boldsymbol{d}}_{j+k} = ({\boldsymbol{r}}_j - {\boldsymbol{r}}_{j+k}) / |{\boldsymbol{r}}_j - {\boldsymbol{r}}_{j+k}| $, and the chiral vector involved in SOC is given by $ {\boldsymbol{v}}_j^{(+)} = {\boldsymbol{d}}_{j+k} \times {\boldsymbol{d}}_{j+2k} $. Here, $ k\in\{+1,-1\} $ is an integer indicating the direction along the chain; $ j+k $ and $ j+2k $ denote the indices of the nearest and next-nearest neighbors of site $ j $, respectively. Unless otherwise stated, we set $ k=+1 $. This cross product $ {\boldsymbol{v}}_j^{(+)} $ encodes the local handedness of the helical chain. Its sign reverses between right- and left-handed configurations, thereby allowing the spin–orbit term $ ({\boldsymbol{v}}_j^{(+)} \cdot {\boldsymbol{\sigma}})_{ss'} $ to capture the intrinsic molecular chirality within the electronic Hamiltonian. For simplicity, we set the helix radius and pitch equal, $ a = h = 1 $.

    • The vibrational part of the Hamiltonian is expressed as

      The chiral molecule also includes inter-site vibrational modes, where $ R_j $ and $ P_j $ are the classical displacement and momentum associated with the bond between sites $ j $ and $ j+1 $. For a chain with $ N $ sites, there are $ N-1 $ such modes, each representing a classical degree of freedom. For simplicity, all vibrational modes are assumed to have the same frequency.

    • The coupling between electronic and vibrational degrees of freedom is described by

      Electron–vibration coupling enters in two distinct forms. Similar to the electronic Hamiltonian, the first term is spin-independent and modulates the nearest-neighbor hopping with strength $ t_1 $, which represents the standard Su–Schrieffer–Heeger (SSH) type coupling. The second term is spin-dependent and modifies the spin–orbit channel, characterized by the coupling strength $ \lambda_1 $.

    • The Fransson model adopts several simplifying assumptions that make it analytically and computationally tractable but also idealized relative to realistic chiral molecular systems. First, we study electron transmission within a closed chiral system as noted in Sec. I, and therefore omit any explicit electrode–molecule coupling. Second, electron–electron interactions are neglected, reducing the problem to an effective single–electron picture. Third, each site contains only one effective orbital, thus omitting multi-orbital effects and inter-orbital hybridization. Fourth, the chiral molecule is treated as a rigid helical chain, thereby excluding conformational deformations [57]. Finally, all classical vibrational modes share the same vibrational frequency, thereby neglecting complex vibrational spectra and anharmonicity.

      These idealizations inevitably enlarge the gap between the toy model and realistic molecular systems. Nevertheless, they are deliberate choices that allow us to isolate the intrinsic interplay among spin–orbit coupling, chain length, and electron–vibration coupling. Within this controlled minimal framework, the essential mechanisms governing transient spin polarization and its decay can be clearly identified, providing a solid foundation for future extensions toward more realistic, first-principles descriptions.

    III.   METHODS
    • To capture nonadiabatic effects arising from electron–vibration coupling, we employ two widely used mixed quantum–classical approaches: mean–field dynamics [58, 59] and surface hopping [60–62]. Both approaches have been extensively tested for Hamiltonians with SSH-type electron–vibration couplings [63, 64], which justifies their use in the present study. However, standard mean–field dynamics and surface hopping approaches were originally formulated for real-valued Hamiltonians. In our helical molecular chain model, however, the inclusion of spin–orbit coupling renders the Hamiltonian complex-valued, and the Berry force must therefore be taken into account. As discussed below, mean-field dynamics naturally incorporates the Berry force without further modification [65], whereas the surface hopping framework requires explicit corrections [66, 67].

    • In mean–field dynamics, each classical trajectory is driven by the mean force. The classical position $ {\boldsymbol{R}} $ and momentum $ {\boldsymbol{P}} $ evolve according to Hamilton’s equations, while the electronic density matrix $ \rho $ evolves based on the Liouville–von Neumann equation:

      Here, $ H $ represents the total Hamiltonian.

    • In surface hopping, each classical trajectory evolves on a single adiabatic potential energy surface labeled by $ j $. The equations of motion in the adiabatic representation are given by

      where the additional term $ {\boldsymbol{F}}_j^{{\rm{Berry}}} $ denotes the Berry force acting on the nuclei:

      The electronic density matrix evolves according to

      Here, $ V $ is obtained by diagonalizing $ H $, and $ {\boldsymbol{d}}_{jk} $ denotes the derivative coupling between surfaces $ j $ and $ k $.

      Following Tully’s “fewest switches” algorithm, trajectories may hop among different surfaces. The hopping rate from surface $ j $ to $ k $ is

      When a hopping event occurs, the nuclear momentum must be rescaled to conserve the total energy. In this study, we adopt the second method proposed in Ref.[68], where the rescaling direction $ {\boldsymbol{n}}_{jk} $ is defined as

      with $ \eta $ being chosen to maximize $ \Re \left( e^{i\eta} {\boldsymbol{d}}_{jk} \right) $.

    IV.   RESULTS
    • Unless noted otherwise, all simulations employ the default parameter set adopted from Ref.[41]: $\epsilon_0 = -240$ meV, $t_0 = 40$ meV, $t_1 = 4$ meV, $\hbar \omega = 0.4$ meV, $\lambda_0 = 1$ meV, $\lambda_1 = 0.1$ meV, and $k_{\rm B} T = 25$ meV. When exploring parameter variations, we use integer multiples of these baseline values.

      For both mean–field dynamics and surface hopping, initial displacement $ R_j $ and momentum $ P_j $ are sampled from the Boltzmann distribution. Each mode is initialized with zero mean, $\langle R_j \rangle = \langle P_j \rangle = 0$, and standard deviation $\sqrt{k_{\rm B} T / \hbar \omega}$ for both $ R_j $ and $ P_j $. The electron is initially localized at the first site, with equal spin-up and spin-down probabilities. The spin-resolved populations $ N_\uparrow(t) $ and $ N_\downarrow(t) $ at the last site are obtained by averaging over 15,000 trajectories.

      The mechanism of electron transmission is illustrated schematically in FIG. 1, where chiral–induced scattering selectively favors one spin orientation, even though the system is initialized without net spin polarization. The transient spin polarization is defined as

      In addition to the time-dependent $ {SP}(t) $, it is useful to define a time-averaged spin polarization, which is particularly relevant for comparison with steady-state experimental measurements. For a generic time-dependent observable $ O(t) $, the average over time is

      where $ T $ is the total simulation time. Here, $ O(t) $ has already been averaged over all trajectories at each time $ t $. Applying this definition, the time-averaged electronic populations at the last site, $ \langle N_\uparrow \rangle $ and $ \langle N_\downarrow \rangle $, are computed separately. The corresponding time-averaged spin polarization is then

      Here, the angle brackets denote combined trajectory and time averages.

      It is important to note that $ {SP}(t) $ itself is a ratio and not a directly observable quantity. Consequently, averaging $ {SP}(t) $ over time does not yield a meaningful result. Instead, one should first compute the time-averaged electronic populations and then construct $\langle SP \rangle$ according to Eq.(16), which provides a physically well-defined and experimentally relevant measure of spin polarization.

    • In this subsection, molecular vibrations are neglected, and the electronic dynamics is governed only by Eq.(7). With identical on-site energies and uniform couplings, the helical chain is translationally homogeneous and free of static disorder. The spin-resolved dynamics therefore differ only through phase accumulation induced by chain length and SOC, leading to phase-locked oscillations of the spin polarization rather than irreversible decay.

      FIG. 2 presents the spin-resolved electronic populations and the corresponding spin polarization for three chain lengths: $N = 1 \times 3$, $2 \times 3$, and $3 \times 6$. For the shortest chain ($N = 1 \times 3$), the spin-up and spin-down populations at the last site remain nearly identical throughout the simulation (FIG. 2(a)), leading to negligible spin polarization (FIG. 2(b)). This indicates that very short molecular chains cannot generate spin polarization. When the chain length increases to $N = 2 \times 3$ and $3 \times 6$, the spin-resolved populations begin to deviate in time, even though their overall envelopes remain similar (FIG. 2(c, e)). The spin polarization arises from a relative phase shift between the spin-up and spin-down populations and oscillates between approximately $ -100\% $ and $ +100\% $ (FIG. 2(d, f)). In longer chains, polarization emerges later, reflecting the increased propagation time required for the electron to reach the last site.

      FIG. 3 illustrates the influence of SOC strength in the absence of molecular vibrations. The chain length is fixed at $N = 3 \times 6$, while the SOC parameter $ \lambda_0 $ is varied as $ 0\times $, $ 1\times $, and $2\times $. When SOC is absent ($\lambda_0 = 0$), the spin-up and spin-down populations remain nearly identical throughout the simulation (FIG. 3(a)), leading to negligible spin polarization (FIG. 3(b)), even though the molecular chain is sufficiently long. This demonstrates that SOC is essential for generating spin polarization. As SOC increases to $\lambda_0 = 1 \times$ and $2\times$, the spin-resolved populations diverge markedly in time (FIG. 3(c, e)), giving rise to pronounced spin polarization (FIG. 3(d, f)). At early simulation times, a stronger SOC not only enhances the magnitude of polarization but also prolongs the period during which the polarization remains near saturation. At longer times, the polarization eventually oscillates between $ -100\% $ and $ +100\% $.

      Taken together, the results without electron–vibration coupling reveal a consistent picture: when the chain is short or SOC is absent, the spin-up and spin-down populations remain nearly identical, yielding no spin polarization. In contrast, increasing either the chain length or the SOC strength leads to enhancement of transient spin polarization, which eventually oscillates between $ -100\% $ and $ +100\% $.

    • In this subsection, we simulate spin dynamics with electron–vibration coupling using mean-field (MF) dynamics and surface hopping (SH) approaches. They produce nearly identical results across all parameter regimes, indicating that both reliably capture the essential features of spin polarization induced by electron–vibration coupling, including its onset, amplitude, and time evolution.

      Including electron–vibration coupling converts thermal nuclear fluctuations into time-dependent modulation of the hopping and SOC amplitudes, i.e., dynamic off-diagonal disorder. The resulting phase randomization across trajectories produces rapid dephasing of $ \rm{SP}(t) $, which explains the transient growth at early times followed by decay toward a vanishing long-time average. Therefore, we check spin polarization signals at both the femtosecond and picosecond timescales.

    • First, we show short-timescale results with total simulation time set to 250 fs. In a closed, unbiased, and time-reversal-symmetric Hamiltonian, the net spin polarization must vanish at equilibrium or upon long-time averaging. Nevertheless, our protocol of preparation and detection is intrinsically nonequilibrium: one-way initialization at the first site and readout at the last site. In this setting, spin-dependent phase is accumulated, so the two spin channels reach the last site with different arrival times and amplitudes. The ensuing transient imbalance between spin-up and spin-down populations produces a finite but short-lived spin polarization.

      FIG. 4 compares the spin-resolved populations and corresponding spin polarization for three molecular chain lengths: $N = 1 \times 3$, $1 \times 5$, and $2 \times 5$. For the shortest chain (FIG. 4(a)), the spin-up and spin-down populations evolve almost identically, yielding negligible spin polarization (FIG. 4(b)). As the chain length increases to $1 \times 5$ and $2 \times 5$ (FIG. 4(c, e)), clear differences between spin-up and spin-down populations emerge, especially during the early-time dynamics. In longer chains, the arrival of population at the last site is delayed and the amplitude is reduced, which in turn postpones the onset of spin polarization. Once established, however, the polarization reaches finite values and persists over an extended time window before eventually decaying to zero (FIG. 4(d, f)).

      FIG. 5 shows the influence of spin–orbit coupling when molecular vibrations are included. The chain length is fixed at $N = 1 \times 5$. To vary the SOC strength, we jointly scale the parameters $ (\lambda_0,\lambda_1) $ by factors of $ 1\times $, $ 2\times $, and $ 3\times $ relative to their default values. For the $ 1\times $ case (FIG. 5(a, b)), a transient spin polarization develops at early times and then relaxes toward zero. The initial rise is associated with a slight temporal and amplitude mismatch between the spin-up and spin-down populations induced by electron–vibration coupling. As the SOC scaling increases to $ 2\times $ and $ 3\times $ (FIG. 5(c–f)), the separation between the spin-up and spin-down populations becomes more pronounced. Consequently, at a given time the transient spin polarization is larger for stronger SOC. In all cases, the polarization gradually decays toward zero within a few hundred femtoseconds.

      FIG. 6 examines the effect of electron–vibration coupling strength on spin polarization dynamics in the presence of molecular vibrations. The chain length is fixed at $N = 1 \times 5$, while the coupling parameters $ t_1 $ and $ \lambda_1 $ are scaled by factors of $ 1\times $, $ 3\times $, and $ 5\times $ from top to bottom. As $ t_1 $ and $ \lambda_1 $ increase, the electronic population reaches the last site more rapidly (FIG. 6(a, c, e)), indicating that stronger electron–vibration coupling accelerates electron transmission. However, this acceleration comes at the cost of reduced spin polarization. As shown in FIG. 6(b, d, f), the amplitude of spin polarization decreases systematically with increasing $ t_1 $ and $ \lambda_1 $, and its decay to zero becomes faster. This trend is opposite to that reported by Fransson, where stronger coupling was argued to enhance spin polarization [37].

      In summary, when electron–vibration coupling is included, pronounced spin polarization emerges at early times but subsequently decays to zero. Increasing the chain length prolongs the duration of noticeable spin polarization by enlarging the spatial and temporal separation between the arrival of spin-up and spin-down populations at the last site. Strengthening the SOC enhances the transient polarization by increasing the spin-dependent phase accumulation during propagation along the helical chain. In contrast, stronger electron–vibration coupling accelerates electron transmission while simultaneously reducing both the magnitude and lifetime of spin polarization due to enhanced dephasing. Overall, chain length and SOC enhance transient spin polarization, whereas electron–vibration coupling suppresses it through dynamic disorder and loss of coherence.

    • We also present long-timescale results with total simulation time extended to 50 ps. In the long-time limit of this finite, closed, and unbiased system, although significant polarization emerges very early, time-reversal symmetry enforces the vanishing average spin polarization.

      FIG. 7 presents the long-timescale dynamics obtained with the default parameter set. Over 50 ps, the electronic population does not converge to a steady state but instead exhibits oscillatory behavior characteristic of a closed system. As shown in FIG. 7(a), the MF method maintains oscillations with nearly constant amplitude, whereas the SH method produces gradually damped oscillations due to the decoherence introduced by stochastic surface hops. The corresponding spin polarization dynamics are displayed in FIG. 7(b). The instantaneous values fluctuate around zero within a narrow band throughout the simulation. As illustrated in FIG. 7(c, d), both MF and SH yield identical time-averaged spin-resolved populations of 0.117 for spin-up and spin-down electrons, resulting in a vanishing time-averaged spin polarization.

      Furthermore, FIG. 8 shows the statistical distribution of spin polarization over the entire picosecond-scale simulation, with both MF and SH results included. The histograms indicate that the polarization values remain narrowly centered around 0% and exhibit an almost perfectly symmetric distribution with respect to positive and negative values. This statistical picture is in full agreement with the time-domain analysis in FIG. 7: although transient spin polarization fluctuates, its symmetric distribution ensures that the time-averaged spin polarization is essentially close to zero.

      We have also performed long-timescale simulations for all other parameter combinations considered in the preceding subsections. As expected, the details of the electronic dynamics vary with system parameters, leading to differences in the oscillation amplitudes. Nevertheless, the qualitative behavior of the spin polarization remains the same: it fluctuates within a small range around zero, and its time average is always zero. For this reason, we only present the default case which is representative of the general behavior here.

      Our simulations yield nearly vanishing time-averaged spin polarization, as confirmed both by direct averaging and by the statistical distribution analysis. This outcome stands in clear contrast to experimental observations of sizable and robust spin polarization [44]. The discrepancy suggests that the present model, even though it incorporates electron–vibration coupling, does not fully capture the essential mechanisms of CISS. In particular, the role of electron–vibration coupling in sustaining long-lived spin polarization remains controversial, and the absence of additional amplification or stabilization channels may be responsible for the suppression observed here. Identifying and addressing these missing ingredients will be crucial for reconciling theoretical predictions with experimental results in future studies.

    V.   CONCLUSION
    • In this work, we systematically investigated spin polarization dynamics in a closed helical molecular chain. When the electron–vibration coupling is neglected, the purely electronic regime shows coherent oscillations between $ -100\% $ and $ +100\% $. When electron–vibration coupling is included, strong but short-lived polarization emerges at the early stage before decaying rapidly. Longer simulations confirm that the spin polarization fluctuates narrowly around zero, leading to negligible time-averaged values. This relaxation behavior closely parallels charge and exciton diffusion in finite, unbiased media: transient currents or density imbalances may arise after a localized preparation, but in the absence of driving or reservoirs, time-reversal symmetry implies that the system relaxes to a stationary state with zero net flux [69, 70].

      Our findings also underscore the limitations of the present framework. Although the model captures short-time spin-dependent dynamics mediated by SOC and electron–vibration coupling, it fails to reproduce the finite, robust spin polarization consistently observed in experiments. This suggests that the description of electron–vibration coupling within this model is insufficient and that broader limitations remain in capturing the full CISS phenomenon. Future theoretical developments should therefore aim to modify or extend the current framework, incorporating additional mechanisms such as electron–electron interactions [71–74]. Several new models have recently been proposed to address these challenges [75, 76]. Looking ahead, advancing beyond model systems will require first-principles nonadiabatic molecular dynamics [77–80] driven by high-accuracy electronic-structure methods (e.g., noncollinear time-dependent density functional theory (TDDFT) [81] and relativistic electronic-structure methods [82]). Such developments are expected to eventually bridge theoretical simulations with experimental observations.

    Figure (8)  Reference (82)

Catalog

    /

    DownLoad:  Full-Size Img  PowerPoint
    Return
    Return