Coherent Biexciton Transport in the Presence of Exciton–Exciton Annihilation in Molecular Aggregates Rajesh Dutta, Chern Chuang Abstract Biexciton dynamics in molecular aggregates provides a sensitive probe of the interplay between quantum coherence, band structure, and dissipation under strong excitation conditions. We present a theoretical framework for biexciton dynamics in molecular aggregates that explicitly treats populations and coherences across excitation manifolds within a reduced density-matrix formalism. By extending kinetic descriptions beyond the weak-coupling limit, the approach captures the influence of exciton delocalization and exciton-exciton annihilation while remaining computationally tractable within a Markovian description of environmental relaxation. Using this framework, we investigate how the spatial profile and momentum composition of the initial biexciton state govern fluorescence decay and transport. Incoherent initial conditions lead to strongly non-exponential relaxation and time-dependent diffusion driven by nonlinear population kinetics. In contrast, coherently prepared biexciton states exhibit pronounced early-time coherent transport, whose character depends sensitively on whether the initial state is prepared as a standing-wave or traveling-wave superposition of single-exciton modes. Despite nearly identical emission dynamics for J and H aggregate, biexciton transport properties differ markedly due to band structure-dependent interference effect. Our results demonstrate that biexciton dynamics remains strongly influenced by initial-state coherence and momentum composition. Besides initial-state preparation, the coherent-to-incoherent crossover and the diffusive spreading of the exciton density are sensitive to internal conversion processes such as exciton fusion and the decay to the first excited state. The present work establishes initial-state preparation as a key control parameter for many-exciton transport in excitonic systems and provides a general framework for interpreting nonlinear optical experiments beyond population-based descriptions. I Introduction Exciton transport in natural light-harvesting complexes, artificial molecular aggregates, and conjugated polymers has attracted substantial interest over decades, particularly following the recognition of the role of quantum coherence in these systems. 24 ; 10 ; 15 ; 16 Both experimental 24 ; 10 ; 16 ; 15 ; 15 ; 44 and theoretical 33 ; 34 ; 49 ; 2 ; 5 ; 39 ; 20 advances have sought to unravel the mechanisms underlying efficient energy transport, with potential applications in renewable energy technologies such as photovoltaics and solar cells. 11 ; 68 ; 12 ; 71 ; 72 ; 54 ; 21 Most theoretical studies have focused on quantum transport in the single-exciton regime, with the primary aim of elucidating the interplay between static disorder and environmental noise in molecular and materials of different dimensionalities and geometries using dynamical and optical properties. 29 ; 30 ; 55 ; 48 ; 70 ; 43 ; 19 ; 14 ; 13 ; 38 ; 36 ; 7 ; 3 ; 23 In one- and two-dimensional systems, any finite amount of disorder leads to the suppression of transport, whereas environmental noise facilitates transport by mitigating the effects of disorder. 48 However, the single excitonic description is valid when at most one molecule is in the excited state while the remaining molecules are in their ground states, a condition that is naturally satisfied under a weak field. Such weak-excitation conditions are naturally realized under solar irradiation and are relevant for both biological photosynthesis and photovoltaic systems. However, in laser-based experiments the single-exciton approximation can break down, since at high excitation densities or under strong laser fluence multiple excitons may be generated in proximity. The formation of multiexciton states gives rise to exciton–exciton interactions, which can be directly probed through nonlinear optical responses. 57 ; 41 In a seminal study, the nonlinear absorption coefficient of a one-dimensional molecular aggregate composed of coupled two-level systems in the presence of site disorder was evaluated. 58 Experimentally, the first observation of a one to two-exciton transition in J aggregates was reported by Fidder et. al. . 25 Subsequently, an analytic expression for the exciton–exciton scattering amplitude in an infinite, periodic one-dimensional aggregate was derived, providing a theoretical foundation for interpreting the characteristic spectral line shapes of the dominant J-band resonance observed in coherent third-order nonlinear spectroscopy. 1 In the presence of multiexciton states, exciton–exciton annihilation (EEA) becomes an essential, if not dominant, component of the dynamics. 63 This process is commonly described as a two-step mechanism: first, two excitations interact, resulting in the formation of a higher excited state on one molecule while the other molecule relaxes to the ground state, a process often referred to as exciton fusion. In the second step, rapid internal conversion relaxes the higher excited state to the first excited state. Experimentally, EEA can be characterized using third-order nonlinear spectroscopy as a function of laser fluence, 6 ; 4 ; 18 ; 45 ; 42 ; 56 ; 9 and such EEA signal is commonly modeled as an incoherent, classical reaction governed by a bimolecular rate law. 59 ; 65 ; 26 ; 64 ; 62 ; 66 ; 17 However, when excitons are strongly delocalized and/or at low temperatures, this phenomenological description is no longer adequate. Beyond phenomenological rate-based descriptions, several microscopic theoretical frameworks for EEA have been developed. 27 ; 46 ; 53 In particular, density-matrix–based approaches have been formulated to capture the role of quantum coherence and exciton delocalization. 52 ; 51 ; 8 More recent studies have employed Fermi’s golden rule to describe EEA in molecular aggregates, explicitly accounting for the sign of the dipole–dipole coupling that distinguishes J- and H-type aggregates. 60 Both theoretical analyses and experimental observations have demonstrated that, in H aggregates, coherent suppression of EEA arises from the destructive interference originating from their out-of-phase excitonic wavefunctions. 42 ; 56 Furthermore, by systematically reducing the density-matrix description, effective kinetic equations involving multiple excitation manifolds were derived and successfully applied to describe annihilation dynamics in conjugated polymer systems. 47 ; 28 Despite these advances, a comprehensive theoretical framework that explicitly accounts for coherences within and between different excitation manifolds such as single-exciton, multiexciton, and their cross-manifold coherences—and their impact on the spatiotemporal dynamics of molecular aggregates is still missing. In this work, we pursue two primary objectives. First, we extend the kinetic formulation originally proposed by May 47 to regimes of intermediate and strong coupling, where coherent effects are expected to play a significant role and cannot be neglected. Second, while coherent and incoherent control at the single-exciton level—both temporal and spatial, 31 ; 35 ; 69 ; 37 ; 22 ; 61 has been extensively explored in light-harvesting systems and molecular aggregates, and temporal control of multiexciton processes 47 ; 67 has been addressed theoretically, a corresponding framework for spatial profile dependence and for assessing the role of initial state preparation in multiexcitonic dynamics is lacking. Here, we develop such a theory and apply it to biexciton dynamics in both J- and H-aggregates. Importantly, while we focus here on the homogeneous limit where localization is not a factor, the present approach is general regardless of localization landscape and can be systematically extended to higher multiexciton manifolds. The remainder of this paper is organized as follows. In Sec. II, we introduce the model and Hamiltonian. Section III derives the equations of motion for populations and coherences in a multiexcitonic framework. In Sec. IV, we construct biexciton initial states from which full quantum dynamics are followed. Section V examines the implications of biexciton dynamics across different regimes, with emphasis on experimentally relevant observables such as time-resolved emission and exciton diffusion. Finally, Sec. VI summarizes the paper. II Model and Hamiltonian Under weak optical excitation, the electronic Hilbert space can be restricted to the overall ground state and states containing at most a single excitation per molecule. In this regime, the dynamics is conveniently described in terms of localized molecular excitations that can delocalize via intermolecular Coulomb interactions. In contrast, under strong excitation conditions, EEA becomes significant and suppresses the formation of long-lived multiexciton states. To describe biexciton dynamics in the presence of EEA in a linear molecular with three level systems, 40 we adopt the Hamiltonian proposed by May 47 H_{\mathrm{tot}}=H^{(1)}+H^{(2)}+V^{(12)} (1) where H^{(1)} accounts for the singly excited manyfold, H^{(2)} denotes the third-level excitation manifold, and the cross term V^{(12)} represents the fusion/fission coupling between the two manifolds and is ultimately responsible for EEA. The first excited state Frenkel Hamiltonian can be written as H^{(1)}=\sum_{m}E_{m}B_{m}^{\dagger}B_{m}+\sum_{m\neq n}J_{mn}B_{m}^{\dagger}B_{n} (2) Similarly, second excited state Hamiltonian can be expressed as H^{(2)}=\sum_{m}\varepsilon_{m}D_{m}^{\dagger}D_{m}+\sum_{m\neq n}\mathcal{J}_{mn}D_{m}^{\dagger}D_{n} (3) The interaction term can be given as V^{(12)}=\sum_{m\neq n}\left(K_{mn}D_{m}^{\dagger}B_{n}+\text{H.c.}\right) (4) Here, J denotes the coupling between the first excited states ( e ) of different molecules, while \mathcal{J} represents the interaction between the second (higher) excited states ( f ) of different molecules. To account for EEA, the energy difference between the ground and first excited states, E_{m} , is assumed to be comparable to the energy separation between the first and second excited states, \epsilon_{m} . Under this resonance condition, biexciton states can merge by promoting one molecule to the second excited state while simultaneously relaxing the other molecule to the ground state. The coupling K accounts for the formation of second excited states by fusion of first excited states (Fig 1 ). B_{m}^{\dagger} and B_{m} ( D_{m}^{\dagger} and D_{m} ) represent transition operators between zero- (single) and one- (two) excitation manifolds. We also assume that exciton-exciton interactions can be neglected but the general framework remains the same if those were accounted for. Figure 1: Biexciton dynamics in linear chain of three-level systems with nearest neighbor interaction. J is the interaction between first excited state, \mathcal{J} refers to second excited state interaction and K denotes fusion due to exciton-exciton annhiliation. We consider a homogeneous chain of molecules where all the couplings are nearest neighbor and coefficients are real i.e. \begin{split}&\varepsilon_{m}=2E_{m}=2E,\\ &J_{mn}=J\left(\delta_{m,n+1}+\delta_{m,n-1}\right),\\ &\mathcal{J}_{mn}=\mathcal{J}\left(\delta_{m,n+1}+\delta_{m,n-1}\right),\\ &K_{mn}=K\left(\delta_{m,n+1}+\delta_{m,n-1}\right)\end{split} (5) Again our framework is not restricted to homogeneous systems and analysis for disordered systems will be reserved for future work. III Equation of motion In the presence of a dissipative environment, a Markovian master equation for such a system can be expressed in the Lindblad form as follows \frac{\partial}{\partial t}\operatorname{tr}\{\hat{\rho}(t)\,\hat{O}\}=\frac{i}{\hbar}\left\langle[\hat{H},\hat{O}]_{-}\right\rangle-\left\langle\tilde{D}\,\hat{O}\right\rangle\,. (6) Here, an arbitrary operator \hat{O} , constructed from different combinations of transition operators, is introduced. This procedure effectively projects out the relevant diagonal and off-diagonal reduced density-matrix elements, thereby reducing the computational cost associated with solving the full density-matrix dynamics. \mathcal{\tilde{D}}_{e}\hat{O}=\frac{k}{2}\sum_{m}\left([B_{m}^{\dagger}B_{m},\hat{O}]_{+}-2B_{m}^{\dagger}\hat{O}B_{m}\right) (7) \mathcal{\tilde{D}}_{f}\hat{O}=\frac{r}{2}\sum_{n}\left([D_{n}^{\dagger}D_{n},\hat{O}]_{+}-2D_{n}^{\dagger}\hat{O}D_{n}\right) (8) Here, k represents the decay rate from the first excited to ground state and r denotes the decay from second to first excited state. Population of ground, first and second excited states are given by \begin{split}&\Pi_{n}=\langle B_{n}B_{n}^{\dagger}\rangle,\\ &P_{n}=\langle B_{n}^{\dagger}B_{n}\rangle,\\ &N_{n}=\langle D_{n}^{\dagger}D_{n}\rangle\end{split} (9) The completeness relation can thus be written as \Pi_{n}+P_{n}+N_{n}=1 (10) Equation of motion for the diagonal part reads \displaystyle\frac{\partial}{\partial t}P_{m}(t) \displaystyle=-kP_{m}+rN_{m} \displaystyle\quad+\frac{2}{\hbar}\sum_{n}J_{mn}\,\text{Im}[W_{mn}]-\frac{2}{\hbar}\sum_{n}\mathcal{J}_{mn}\,\text{Im}[Z_{mn}] \displaystyle\quad-\frac{2}{\hbar}\sum_{n}K_{mn}\,\text{Im}[R_{mn}]-\frac{2}{\hbar}\sum_{n}K_{nm}\,\text{Im}[R_{nm}] (11) \frac{\partial}{\partial t}N_{m}(t)=-rN_{m}+\frac{2}{\hbar}\sum_{n}\mathcal{J}_{mn}\text{Im}[Z_{mn}]+\frac{2}{\hbar}\sum_{n}K_{mn}\text{Im}[R_{mn}] (12) May and co-workers reduced the full master equation to kinetic rate equations by considering coherence between first excited states W_{mn}=B_{m}^{\dagger}B_{n} , coherence between second excited state Z_{mn}=D_{m}^{\dagger}D_{n} and mixed coherence R_{mn}=D_{m}^{\dagger}B_{n} to decay rapidly, leading to \displaystyle\frac{\partial}{\partial t}P_{m}(t) \displaystyle=-kP_{m}+rN_{m}+\frac{2}{k\hbar^{2}}\sum_{n}J_{mn}^{2}[P_{m}N_{n}-P_{n}N_{m}] \displaystyle\quad-\frac{2}{(k+r)\hbar^{2}}\sum_{n}\mathcal{J}_{mn}^{2}[P_{m}N_{n}-N_{m}P_{n}] \displaystyle\quad-\frac{2}{(k+r/2)\hbar^{2}}\sum_{n}K_{mn}^{2}[P_{m}P_{n}+N_{m}N_{n}+P_{n}N_{m}-N_{m}] \displaystyle\quad-\frac{2}{(k+r/2)\hbar^{2}}\sum_{n}K_{nm}^{2}[P_{m}P_{n}+N_{m}N_{n}+P_{m}N_{n}-N_{n}]. (13) \displaystyle\frac{\partial}{\partial t}N_{m}(t) \displaystyle=-rN_{m}+\frac{2}{(k+r)\hbar^{2}}\sum_{n}\mathcal{J}_{mn}^{2}\bigl[P_{m}N_{n}-N_{m}P_{n}\bigr] (14) \displaystyle+\frac{2}{(k+r/2)\hbar^{2}}\sum_{n}K_{mn}^{2}\bigl[P_{m}P_{n}+N_{m}N_{n}+P_{n}N_{m}-N_{m}\bigr] The nonlinear rate equations describe multiexciton dynamics quite well in the weak-coupling limit where closed-system dynamics can be considered perturbative. Their validity arises not only from the vanishing time dependence of the coherences but also from retaining only the leading (first-order) contributions in J , \mathcal{J} , and K . In this work, we extend May’s approach to incorporate the effects of time-dependent off-diagonal elements originating from four-body correlations of transition operators. 47 To close the resulting set of equations and keep the dynamics tractable, four-body correlators are factorized into pairwise transition operators or two-site correlation within a mean-field type approximation. 67 This procedure leads to the following simplified equation of motion for the off-diagonal elements. \displaystyle\frac{\partial}{\partial t}W_{mn}(t) \displaystyle=i\frac{(E_{m}-E_{n})}{\hbar}-kW_{mn}+\frac{iJ_{nm}}{\hbar}[(\Pi_{m}-P_{m})P_{n}-(\Pi_{n}-P_{n})P_{m}] (15) \displaystyle+\frac{i}{\hbar}\sum_{k\neq{n,m}}[J_{km}(\Pi_{m}-P_{m})W_{kn}-J_{nk}(\Pi_{n}-P_{n})W_{mk}] \displaystyle+\frac{i}{\hbar}\sum_{k\neq{m}}\mathcal{J}_{mk}(Z_{mk}W_{mn}+R_{mn}R^{*}_{km})-\frac{i}{\hbar}\sum_{k\neq{n}}\mathcal{J}_{kn}(Z_{kn}W_{mn}+R^{*}_{nm}R_{kn}) \displaystyle+\frac{i}{\hbar}\sum_{k\neq{m}}K_{mk}(R_{mk}W_{mn}+R_{mn}W_{mk})-\frac{i}{\hbar}\sum_{k\neq{n}}K_{nk}(W_{mn}R^{*}_{nk}+R^{*}_{nm}W_{kn}) \displaystyle+\frac{i}{\hbar}\sum_{k\neq{m,n}}K_{km}(\Pi_{m}-P_{m})R_{kn}-\frac{i}{\hbar}\sum_{k\neq{m,n}}K_{kn}(\Pi_{n}-P_{n})R^{*}_{km} \displaystyle\frac{\partial}{\partial t}Z_{mn}(t) \displaystyle=i\frac{(\epsilon_{m}-\epsilon_{n})-(E_{m}-E_{n})}{\hbar}-(k+r)Z_{mn}-\frac{i}{\hbar}\sum_{k\neq{m}}J_{mk}(Z_{mn}W_{mk}+R_{mk}R^{*}_{nm}) (16) \displaystyle+\frac{i}{\hbar}\sum_{k\neq{n}}J_{kn}(Z_{mn}W_{kn}+R_{mn}R^{*}_{nk})+\frac{i}{\hbar}\mathcal{J}_{nm}[(P_{m}-N_{m})N_{n}-(P_{n}-N_{n})R_{mk})] \displaystyle+\frac{i}{\hbar}\sum_{k\neq{m,n}}[\mathcal{J}_{km}(P_{m}-N_{m})Z_{kn}-\mathcal{J}_{nk}(P_{n}-N_{n})Z_{mk}] \displaystyle+\frac{i}{\hbar}\sum_{k\neq{m,n}}[K_{mk}(P_{m}-N_{m})R^{*}_{nk}-K_{nk}(P_{n}-N_{n})R_{mk}] \displaystyle-\frac{i}{\hbar}\sum_{k\neq{m}}K_{km}(R^{*}_{nm}Z_{mk}+Z_{mn}R^{*}_{km})+\frac{i}{\hbar}\sum_{k\neq{n}}K_{kn}(Z_{kn}R_{mn}+R_{kn}Z_{mn}) \displaystyle\frac{\partial}{\partial t}R_{mn}(t) \displaystyle=i\,\frac{\left(\epsilon_{m}-E_{m}-E_{n}\right)}{\hbar}\,R_{mn}-\left(k+\frac{r}{2}\right)R_{mn} (17) \displaystyle-\frac{i}{\hbar}\sum_{k\neq m}J_{mk}\left(R_{mk}W_{mn}+R_{mn}W_{mk}\right) \displaystyle-\frac{i}{\hbar}\sum_{k\neq m,n}J_{nk}\,R_{mk}\left(\Pi_{n}-P_{n}\right)+\frac{i}{\hbar}\sum_{k\neq m,n}\mathcal{J}_{km}\,R_{kn}\left(P_{m}-N_{m}\right) \displaystyle-\frac{i}{\hbar}\sum_{k\neq n}\mathcal{J}_{kn}\left(R_{mn}Z_{kn}+Z_{mn}R_{kn}\right) \displaystyle+\frac{iK_{mn}}{\hbar}\left[(P_{m}-N_{m})P_{n}-(\Pi_{n}-P_{n})N_{m}\right] \displaystyle+\frac{i}{\hbar}\sum_{k\neq{m,n}}K_{mk}(P_{m}-N_{m})W_{kn}-\frac{i}{\hbar}\sum_{k\neq{m,n}}K_{nk}(\Pi_{n}-P_{n})Z_{mk} \displaystyle-\frac{i}{\hbar}\sum_{k\neq{m}}K_{km}(R^{*}_{km}R_{mn}+Z_{mk}W_{mn})-\frac{i}{\hbar}\sum_{k\neq{n}}K_{nk}(R^{*}_{nk}R_{mn}+Z_{mn}W_{kn}) These equations are the primary results of this paper and are employed in combination with the equation of motion of the population terms. In the following we shall examine a number of representative cases in various parameters regimes. IV Initial coherent biexciton states To showcase the utility of this framework, we consider a one-dimensional molecular aggregate consisted of three-level systems in standard Frenkel exciton manners depicted in Fig. 1 . We are interested in the spatiotemporal development of biexcitons that may result from realistic microscopy experiments. 32 ; 50 ; 73 ; 7 Analyzing such trajectories on a lattice model can potentially correlate insights from time-resolved microscopy, typically carried out in the low fluence limit, with those obtained in more traditional EEA experiments where the focus is the transient non-exponential component of the fluorescence emission signal. For the purpose of studying microscopy, we start with two normalized single-exciton wavepackets \sum_{m}|\phi^{(i)}_{m}|^{2}=1,\qquad i=1,2, (18) where the site index runs m=1,\dots,N_{mol} . The spatial overlap between the two single-exciton wavepackets is defined as S=\langle\phi^{(1)}|\phi^{(2)}\rangle=\sum_{m}\phi^{(1)*}_{m}\phi^{(2)}_{m}, (19) which characterizes the degree of spatial and momentum overlap between the two excitons. Frenkel excitons are composite bosons subject to a hard-core constraint that forbids double occupancy of a site. This constraint can be enforced explicitly by using an antisymmetrized two-exciton wavefunction. The anti-symmetrized and normalized biexciton wavefunction is given by \Psi_{mn}=\frac{\phi^{(1)}_{m}\phi^{(2)}_{n}-\phi^{(1)}_{n}\phi^{(2)}_{m}}{\sqrt{2\left(1-|S|^{2}\right)}}, (20) where m and n label the lattice sites occupied by the two excitons. The normalization factor ensures \sum_{m,n}|\Psi_{mn}|^{2}=1. (21) This biexciton wavefunction represents a pure biexciton state composed of two indistinguishable excitons prepared in the wavepackets \phi^{(1)} and \phi^{(2)} . For vanishing overlap S\approx 0 , the state reduces to two well-separated excitons, whereas finite overlap introduces exchange-induced quantum coherence between the excitons. Such composite wavepackets can be thought as idealized excited state absorption contribution in a pump-probe microscopy measurement. To characterize observables associated with single-exciton operators, we construct the one-particle reduced density matrix from two-particle description by tracing over one exciton, \rho^{(1)}_{mn}=\sum_{q}\Psi_{mq}\Psi^{*}_{nq}. (22) Substituting the anti-symmetrized biexciton wavefunction yields \rho^{(1)}_{mn}=\frac{\phi^{(1)}_{m}\phi^{(1)*}_{n}+\phi^{(2)}_{m}\phi^{(2)*}_{n}-S\,\phi^{(1)}_{m}\phi^{(2)*}_{n}-S^{*}\,\phi^{(2)}_{m}\phi^{(1)*}_{n}}{1-|S|^{2}}. (23) The diagonal elements of the reduced density matrix, P_{m}=\rho^{(1)}_{mm}, (24) represent the average exciton population at site m , while the off-diagonal elements describe single-exciton quantum coherence. The trace of the reduced density matrix satisfies \mathrm{Tr}\,\rho^{(1)}=\sum_{m}\rho^{(1)}_{mm}=2, (25) reflecting the presence of two excitons in the biexciton state. As we are interested in the dependence of initial spatial profiles, in the context of optical microscopy, we assume each of the constituent single exciton as a standing wave under a Gaussian envelope, whose wavefunctions can be written as \phi^{(i)}_{m}=\frac{\sqrt{2}}{\sqrt{w\sqrt{\pi}\left(1+e^{-k_{i}^{2}w^{2}}\right)}}\cos(k_{i}m)\,\exp\!\left[-\frac{(m-m_{0i})^{2}}{2w^{2}}\right], (26) where m_{0i} denote the spatial centers of the wavepackets, w is their spatial width, and k_{i} are the corresponding central momenta. Similarly, considering each of the initial wavepacket as traveling gaussian wave, the single exciton wavefunction can be expressed as \phi^{(i)}_{m}=\frac{1}{\sqrt{w\sqrt{\pi}}}\exp\!\left[-\frac{(m-m_{0i})^{2}}{2w^{2}}+ik_{i}m\right] (27) V Result and discussion We are interested in the features of biexciton dynamics in time-resolved fluorescence emission and correlate them with those in its spatio-temporal development. The former can be defined in terms of decay from the first and second excited states as follows I\left(t\right)=\sum_{n}[kP_{n}(t)+rN_{n}(t)] (28) Total exciton number is defined as N_{tot}\left(t\right)=\sum_{n}[P_{n}(t)+2N_{n}(t)] (29) While a common coarse-grained metric of the latter is the mean squared displacement (MSD) that can be obtained using the following equation \mathrm{MSD}(t)=\frac{\sum_{m=1}^{N_{\mathrm{mol}}}\bigl(m-\bar{m}(t)\bigr)^{2}\left[P_{m}(t)+2N_{m}(t)\right]}{\sum_{m=1}^{N_{\mathrm{mol}}}\left[P_{m}(t)+2N_{m}(t)\right]}, (30) \bar{m}(t)=\frac{\sum_{m=1}^{N_{\mathrm{mol}}}m\left[P_{m}(t)+2N_{m}(t)\right]}{\sum_{m=1}^{N_{\mathrm{mol}}}\left[P_{m}(t)+2N_{m}(t)\right]}. (31) Time derivative of MSD provides the time dependent diffusivity in one dimension as follows D_{t}=\frac{1}{2}\frac{d}{dt}[\text{MSD}(t)] (32) which in the diffusive regime becomes the diffusion constant. V.1 Incoherent initial biexciton states We begin our analysis by using an incoherent rate-equation framework Eq. ( III ) to describe experimentally accessible quantities including time-resolved emission and exciton density diffusion. Furthermore, we assume the initial state of the dynamics to be a fully localized, double Kronecker delta \phi^{(1)}_{m}=\delta_{n-1,m} and \phi^{(2)}_{m}=\delta_{n+1,m} . Although such description is inadequate for describing coherent effect and a completely localized initial double excitation is generally not experimentally realizable in molecular aggregate context, it offers a simplified approach to elucidate the competition among different processes in multilevel molecular aggregate systems. Figure 2: For a linear chain of three-level systems with nearest neighbor interaction with J=\mathcal{J}=-0.1 and decay from first excited state k=0.1 semi-log plot of time dependent emission, mean squared displacement and time dependent diffusivity are plotted, with varying decay rate r from second excited state in panels (a) to (c) at fixed K=1 and with varying fusion K at fixed decay rate r=1 in panels (d) to (f). Initially biexcitons are placed symmetrically with respect to the center of the 21-site molecular aggregates. The dynamics is propagated with population-only kinetic equations. In Fig. 2 , the time-dependent emission intensity, MSD, and time-dependent diffusivity are shown for different fusion and recombination rates initialized from the double Kronecker delta state.Although the dynamics is governed by incoherent rate equations, the emission exhibits a pronounced non-exponential decay. This non-exponential behavior arises from the nonlinear coupling between exciton and biexciton populations, together with transport-induced time-dependent decay rates, rather than from quantum coherence effects. Further scrutinizing the results obtained from the incoherent description, we note that the interconversion between the populations P_{m} and N_{m} is controlled by the fusion rate K , while the non-radiative decay of the fusion product is described by the recombination rate r . The resulting relaxation is inherently non-exponential, reflecting the coexistence of multiple inter-converting populations and the spatial profile-dependent nature of EEA. At early times, the high exciton density of the first-excited state enhances fusion, rapidly transferring the population from the two-exciton manifold P_{m} into the fusion channel N_{m} , leading to a rapid initial decay of the total excitation signal. As the system evolves, spatial spreading reduces local excitation density, suppressing further fusion, and slowing the effective decay rate, which manifests as pronounced curvature in \ln[I(t)] . This interplay is strongly modulated by the lifetime of the fusion product. For small recombination rates r , the N_{m} population persists and temporarily stores excitation density, allowing continued spatial redistribution and resulting in a larger MSD and enhanced time dependent diffusivity D_{t} . In contrast, larger values of r rapidly remove fusion products, limiting the time available for spatial spreading and reducing both the MSD and D_{t} . Similarly, increasing the fusion strength K accelerates the depletion of the two-exciton population P_{m} , suppressing long-range transport despite the absence of quantum coherence. 61 For small K , fusion is inefficient at initial times, allowing the two-exciton population to persist and spread spatially, which leads to rapid MSD growth and initially increasing D_{t} . Increasing K enhances early-time annihilation, producing an early maximum in D_{t} followed by monotonic decay. V.2 Coherent initial biexciton states Figure 3: Same system specifications as in Fig. 2 but for initial biexciton state corresponding to two exciton gaussian wavepackets with k_{1}=\frac{\pi}{N_{mol}+1} and k_{2}=\frac{2\pi}{N_{mol}+1} with the width w=3 , described in main text. The dynamics is propagated with fully coherent description. The inset highlights the short-time MSD and D_{t} . To develop experimentally relevant biexciton initial states and examine their impact on the interplay between quantum coherence and EEA dynamics, we first investigate biexcitons initially prepared as standing-wavepackets (Fig. 3 ). The corresponding wave vectors are chosen such that these states represent optically bright transitions from the band edge of the single-exciton manifold to the biexciton manifold in a J-aggregate. 58 By varying the width of the wavepacket w , the initial state can be tuned from a localized, effectively incoherent regime in small w to a fully delocalized regime in large w . We consider that the population and coherence are initially restricted to the first excited states as a biexciton. The inclusion of coherence in the equation of motion [Eq. ( 15 ) to Eq. ( 17 ))] together with a coherently prepared biexciton wavepacket produces pronounced transient features (Fig. 3 ) that are absent in the incoherent kinetic description. The initial biexciton state is prepared as an anti-symmetrized product of two spatially extended coherent single-exciton wavepackets both centered at the middle of the chain, which gives rise to nonzero off-diagonal site coherences in the reduced density matrix. These coherences encode phase relations between different sites, leading to transient coherent spreading that suppresses fusion and recombination, giving rise to shoulders and curvature in the emission decay as well as the enhanced early-time MSD growth. (Insets of Fig. 3 , where the time frame is the same as in Fig. 2 .) While incoherent rate equations predict purely diffusive spreading, the inclusion of quantum coherences and phase-correlated initial wavepackets reveals a pronounced ballistic transport regime at early times (MSD \propto t^{2} and D_{t}\propto t ). Non-trivial temporal and spatial profiles last until after coherences decay, and the dynamics cross over to a purely kinetic regime similar to that described by rate equations. Since the biexciton population and its associated coherences decay slowly for small r , population–coherence feedback loops persist over long times, leading to oscillatory transport and emission dynamics. The coherent-to-incoherent transition and the duration of diffusive spread of time-dependent exciton density are sensitive to the exciton fusion rate ( K ) and the decay to the first excited state ( r ). This demonstrates that, even within a Markovian framework, coherent state preparation fundamentally reshapes the interplay between transport, manifold-crossing, and annihilation, yielding relaxation dynamics that cannot be inferred from population kinetics alone. Furthermore, as we follow the trajectories in Fig. 3 to longer times, the wavepackets reach the open boundaries of the chain and bounce back, evidenced by the shrinking of MSD and negative D_{t} values. This can be considered as a crude approximation to disorder where the excitons are forced to localize within a infinite square well. On the other hand, no such information can be deduced from the time-resolved fluorescence signal. This again demonstrates the limitation of the latter as a means to gauge transport. V.3 J- vs H-aggregates Figure 4: Same setup as in Fig. 3 with r=1 . Panels (a) to (c) are for fixed K=1 and in panels (d) to (f) K=0.1 . Here the initial biexciton states are standing wave (SW) packets with k_{1}=\frac{\pi}{N_{mol}+1} and k_{2}=\frac{2\pi}{N_{mol}+1} and traveling wave (TW) packets with k_{1}=\frac{2\pi}{N_{mol}+1} and k_{2}=\frac{N_{mol}\pi}{N_{mol}+1} for J and H aggregate, respectively. An interesting question that arises is whether one can distinguish the spatio-temporal dynamics of biexcitons given the exciton band structure. As is well known, for an incoherent initial condition in the single-exciton manifold, J- and H-aggregates cannot be distinguished through population dynamics alone. To separate their effects, one must instead examine coherences or related optical properties such as absorption and emission spectra. Similarly, for incoherent biexciton initial conditions, J- and H-aggregates cannot be distinguished using time-resolved emission or MSD, as both probes only the population dynamics. To compare J- and H-aggregates, instead of MSD, we consider relative MSD, defined as \text{MSD}(t)-\text{MSD}(0) , since the initial spread of exciton density differs for J- and H-aggregates due to their different momentum components. In Fig. 4 the distinction between J- and H-aggregates is determined not only by the sign of the excitonic coupling but also by the momentum composition of the optically bright biexciton state. For J-aggregates, the initial biexciton is constructed from the lowest two standing-wave exciton modes k_{1}=\frac{\pi}{N_{mol}+1} and k_{2}=\frac{2\pi}{N_{mol}+1} , whereas for H-aggregates it involves high-momentum modes near the top of the exciton band at k_{1}=\frac{2\pi}{N_{mol}+1} and k_{2}=\frac{N_{mol}\pi}{N_{mol}+1} . 58 Although these states are all optically bright and exhibit nearly identical emission dynamics, their transport properties differ markedly. The MSD and the time-dependent diffusivity exhibit a pronounced sensitivity to the momentum composition and phase structure of the initial biexciton state. This sensitivity results in enhanced spatial spreading in J-aggregates and suppressed transport in H-aggregates, particularly under traveling-wave initialization. This behavior can be understood from the population current appearing in Eq. ( III ), which is proportional to the imaginary part of the inter-site coherences. For a homogeneous molecular aggregate with nearest-neighbor coupling J , the exciton dispersion ( 2J\cos k ) results a group velocity ( 2J\sin k ). Since the population current is directly proportional to imaginary part of the coherences, it is therefore linked to both the magnitude and sign of 2J\sin k , and hence to the phase gradients encoded in the coherences. For standing-wave initial conditions, the biexciton state contains equal contributions from +k and -k components. These opposite momenta generate currents of equal magnitude but opposite sign, leading to cancellation of the net current at t=0 . Consequently, the initial imaginary part of the coherences vanishes, and there is no net population current at the time of initial state preparation. Since there is no initial imaginary component of the coherences, the dynamically generated imaginary part that develops during the evolution is not sufficient to distinguish between J- and H-aggregates. Consequently, the transport dynamics under standing-wave are significantly less sensitive to the sign of the coupling and to the momentum ( k ) component compared to the traveling-wave case, resulting in similar transport behavior for J- and H-aggregates. In contrast, traveling-wave initial conditions carry a definite momentum and thus possess a finite imaginary component of the coherences and a nonzero population current. The subsequent dynamics therefore become sensitive to both the sign of the coupling and the value of the momentum of the initially prepared optically bright states. In J-aggregates, the biexciton is predominantly formed from low-order exciton modes located near the bottom of the exciton band, where the dispersion is steep. The large slope of the dispersion corresponds to a large group velocity, so phase gradients efficiently generate population currents. Traveling-wave preparation thus produces a substantial imaginary component of the intersite coherences, resulting in enhanced spatial spreading and a larger MSD. In contrast, the H-aggregate biexciton contains significant contributions from modes close to the upper band edge (large k states), where the dispersion becomes locally flat and the group velocity is strongly reduced. Because the population current scales with 2J\sin k , these band edge components generate only weak current-carrying coherences. Even under traveling-wave initialization, the phase evolution produces only a small imaginary part of the coherences and a correspondingly weak net population current. As a result, spatial transport is suppressed and the MSD remains small. As a result, transport is suppressed in H-aggregates. These differences highlight the role of coherence and band-dispersion effects in biexciton transport, which are completely absent in incoherent rate-equation descriptions. VI Conclusions In this work, we have developed an extended theoretical framework for describing biexciton dynamics in three-level molecular aggregates that goes beyond phenomenological rate-equation models by explicitly incorporating populations, single and biexciton manifold coherences, and cross-manifold coherences within a single dynamical description. By extending May’s kinetic approach 47 into the regime of intermediate and strong inter-site coupling, we have shown that coherences between the single- and double-excitation manifolds cannot, in general, be neglected and play a decisive role in determining both the temporal relaxation and spatial transport of multiexciton states. The resulting equations of motion provide a transparent yet flexible platform for analyzing exciton–exciton annihilation in systems where delocalization, coherence, and dissipation coexist. A key finding of this study is that the nature of the initial biexciton state and its spatial extent, momentum composition, and phase coherence could reshape the subsequent dynamics. In the case of incoherent initial conditions, the dynamics is governed by nonlinear population kinetics arising from the coupling between single and double excitation manifolds. This results in a strong non-exponential emission decay reflecting the interplay between fusion and recombination. Even in this incoherent limit, the spatial spreading of the exciton density cannot be described by a simple bimolecular rate law due to the presence of exciton annihilation which couples the single and double excitation. When coherently prepared biexciton states are considered, qualitatively different and novel dynamical features emerge. Coherent initial wavepackets generate pronounced transient dynamical features in both emission and transport, including early-time ballistic spreading and a subsequent slowdown of the dynamics, arising from the suppression of effective annihilation and decay compared with incoherent predictions. These effects originate from nonvanishing coherences in the purely single-exciton and biexciton manifolds, as well as cross-coherences between the two manifolds of the reduced density matrix, which encode phase correlations between spatially separated sites and give rise to coherence-induced transport flux. As these coherences decay, the system crosses over to a kinetic regime, demonstrating how coherent and incoherent dynamics are continuously connected within the present framework rather than representing distinct limiting cases. Another important conclusion of this work is the identification of clear signatures that distinguish J- and H-aggregates at the biexciton level. While population-based observables such as time-resolved emission remain largely insensitive to the sign of the excitonic coupling and momentum composition to the bright biexciton states under coherent initial state preparation, transport observables remain strongly sensitive to the momentum content and phase structure of the biexciton wavepacket, especially for initial states prepared as traveling wavepackets. For coherently prepared states, J-aggregates exhibit enhanced spreading driven by initial net population current arising from low-momentum exciton modes with steep dispersion and finite group velocities, whereas H-aggregates show suppressed transport because band-edge modes with locally flat dispersion generate weak current-carrying coherences. This demonstrates that biexciton transport provides a powerful probe of aggregate type and band structure that is inaccessible through time-resolved emission measurements alone. Beyond the specific results presented here, the theoretical formulation developed in this work opens several avenues for future investigation. The present approach can be systematically extended to include higher multiexciton manifolds, enabling fluence dependence in nonlinear spectroscopic experiments. Incorporating non-Markovian environmental effects could be another important extension, since memory induced system-bath correlations may sustain long-lived coherences and alter annhilation pathways in ways that cannot be captured within a purely Markovian description. In summary, we have established a coherent description of biexciton dynamics that unifies transport, annihilation, and optical relaxation within a single theoretical framework. Our results highlight quantum coherence not only as a temporal effect, but also shaped by the spatial structure of the wavepacket that can be harnessed to alter multiexciton processes in complex excitonic systems. By demonstrating how coherence, band structure, and initial-state preparation jointly control biexciton transport and annihilation, this work offers a microscopic basis for designing materials and excitation protocols that reduce annihilation losses while preserving efficient energy transport or probing microscopic details that are otherwise challenging to interrogate.