Morphological Physics
This section establishes the framework the rest of the document depends on. We describe what Morphological Physics (MP) is, why its mathematical structure makes morphology a first-class input rather than a numerical convenience, and what has been demonstrated empirically versus what remains open. The reader should leave Section 1 with three things in hand: a clear mental model of how MP computes molecular electronic structure from boundary geometry alone, a working understanding of why this approach is structurally well-posed, and a phase-status accounting of which claims rest on demonstrated results, which on framework-projected ones, and which on validation gates currently open. The argument proceeds in three steps. We first describe what MP is and motivate the form of its central equation. We then explain why the framework’s claims about morphology determining electronic structure follow from the equation’s mathematical structure. We close by accounting for what is demonstrated, what is designed, and what is open.
What Morphological Physics computes from what input
The Sturm–Liouville eigenvalue problem is one of the most well-studied equation forms in mathematical physics. Writing a problem in Sturm–Liouville form provides access to roughly two centuries of spectral theory: the eigenvalues are guaranteed real, the eigenstates orthogonal and complete (any function in the appropriate space can be expanded as a sum of them), and the eigenvalues admit a morphological characterization that makes them computable as constrained minimizers of an energy functional. None of these properties is automatic for an arbitrary partial differential equation; all of them follow once the problem is cast in Sturm–Liouville form. The infrastructure built up around this equation form is exactly what Morphological Physics plugs into.
Our task in this subsection is therefore narrower than “solving the molecular electronic structure problem” in full generality. It is to show that the molecular electronic structure problem, for the class of materials OSC design cares about, can be cast in Sturm–Liouville form on the molecular boundary. Once that casting is done, the spectral-theory infrastructure delivers the rest. We motivate the casting first, then write the equation, then state what it solves.
Consider a molecule. Its electrons occupy a region of space with energies determined by the molecule’s nuclear arrangement and the electron-electron interactions. The standard computational approach for computing this electronic structure is density functional theory (DFT), which answers the question by iterating a self-consistent field calculation: posit a trial electron density, build an effective potential from it, solve a one-electron Schrödinger equation in some chosen basis, update the density, and repeat until convergence. The output is the molecule’s ground-state energy and one-electron orbital energies (the HOMO, the LUMO, and their wavefunctions). DFT works because of the Hohenberg-Kohn theorem, which guarantees that the ground-state density determines all properties of the system—but the path to that density is iterative.
We want a different path. For the molecules organic solar cells (OSCs) care about—\(\pi\)-conjugated organic semiconductors such as benzene, Y6, ITIC, PM6—the electronic structure is dominated by a small set of frontier orbitals—the highest occupied molecular orbital (HOMO), the lowest unoccupied molecular orbital (LUMO), and a few orbitals on either side—that are delocalized across the molecular framework. These delocalized orbitals are sensitive to the molecule’s shape: where the conjugation runs, where the rings close, where the molecular ends terminate. They are not particularly sensitive to fine atomic-scale chemistry detail at the nuclei. The shape carries most of the relevant physics.
This observation suggests an alternative formulation. Rather than solving Schrödinger’s equation in three-dimensional free space and trimming the result to the molecule, we can solve a standing-wave equation directly on the molecular region—with the molecular boundary as the wall, and the local electron density as a weighting function that captures how electrons are distributed inside. The mathematical name for this form is the Sturm-Liouville eigenvalue problem:
\[-\nabla^2 \psi(\mathbf{r}) = \lambda \, \rho(\mathbf{r}) \, \psi(\mathbf{r}) \quad \text{on } \Omega, \qquad \psi|_{\partial \Omega} = 0. \label{eq:sl}\]
Here \(\Omega\) is the molecular region (the region of space occupied by the molecule’s electron cloud above a chosen density threshold), \(\partial \Omega\) is the molecular boundary, \(\rho(\mathbf{r})\) is the local electron density (which we treat as a known input rather than something to iterate), and \(\psi, \lambda\) are the eigenstates and eigenvalues we want to find. In plain language: we look for wave functions \(\psi\) that satisfy the standing-wave condition inside the molecule, weighted by where electrons actually are, and that vanish at the molecular edge.
This formulation is the continuous spatial generalization of an idea chemists have used since 1931. In Hückel theory, the \(\pi\)-orbital energies of conjugated molecules are computed from the molecular graph—the topological connectivity of which atoms bond to which—without reference to atomic positions or detailed potentials. The HOMO and LUMO of benzene come out of diagonalizing a small matrix that only knows the ring is a hexagon. The reason this works is that for \(\pi\)-conjugated systems, topology dominates the electronic structure. MP makes the analogous claim in the continuous setting: spatial geometry, not atomic detail, dominates the spectrum for the broader class of organic semiconductor systems.
The structural move that makes this framework different from standard quantum chemistry is the placement of geometry in the equation rather than in the solution. In the standard treatment one writes a Hamiltonian, computes its eigenstates, and observes that the eigenstates carry whatever symmetries are present. Symmetry is recovered as a property of the solution. Here the symmetry—and, as we will see in Section 4, the chirality—is placed at the level of the operator and its domain. Eigenstates inherit it by the spectral theorem, before any numerical computation. Entire classes of approximation error that arise from recovering structure iteratively are avoided when the structure is encoded in the form of the problem.
The S-L formulation does not arrive for free. The full molecular electronic structure problem is a system of nonlinear partial differential equations, and the reason DFT exists as an iterative numerical method is that this system has no closed-form solution in three-dimensional free space. The bridge from the full problem to the tractable S-L form on the molecular boundary passes through a 170-year-old theorem.
In 1851, Bernhard Riemann proved that any simply connected two-dimensional region, no matter how irregular its boundary, can be transformed into a perfect circle in a way that preserves the structure needed to solve partial differential equations. This means any problem solvable on a circle could, in principle, be solved on any shape.
The catch: Riemann’s proof showed only that the transformation exists. It did not provide a way to construct it. For 170 years, this transformation was accessible only for the handful of shapes where it could be guessed by inspection: ellipses, polygons, a few others. For arbitrary boundaries, the transformation remained known to exist but mathematically unreachable.
The morphological framework provides the only constructive solution to this problem for general simply connected domains.
The Morphology Institute’s constructive solution to the Riemann Mapping problem is what allows the intractable full PDE system to be replaced by the tractable S-L formulation on the molecular boundary. The boundary in MP is therefore not an arbitrary modeling choice; it is the geometric object on which the linearization to S-L form is exact. Morphology becomes a first-class input not by analogy but by construction. The full mathematical machinery is documented in the Morphology Institute monograph. What matters for this document is that the S-L formulation we deploy is the consequence of a proved theorem, not a heuristic approximation.
Figure 1 summarizes what MP computes from what input.
+ element types
density ρ(r)
(crudest possible)
at level set
ρ̂ = 0.01 a.u.
frontier orbitals
problem on Ω
for π-systems
Spectrum λᵢ
Coupling matrix μᵢⱼ
solution
Why MP works: morphology determines structure
Section 1.1 wrote the central equation. This subsection states the structural result that makes the equation an engineering foundation, and previews the mathematical apparatus in Appendix A that proves it. The result is that the Sturm-Liouville problem of Equation [eq:sl] produces a spectrum whose ordering, degeneracies, and symmetry character are determined by the molecular boundary alone; the density rescales it but cannot alter its structure. We state this as Theorem 1 and then preview the chain of connections in Appendix A that supplies the proof and places it in the broader mathematical lineage of morphological analysis.
Theorem 1 (Boundary determines spectral structure). Let \(\Omega \subset \mathbb{R}^d\) be a bounded domain with Dirichlet boundary \(\partial\Omega\) and positive density weight \(\rho\). For the weighted Sturm-Liouville eigenvalue problem \(-\nabla^2 \psi = \lambda \rho \psi\) on \(\Omega\) with \(\psi|_{\partial\Omega} = 0\):
The ordering and degeneracy pattern of the eigenvalues are determined by \(\partial\Omega\) and its symmetry group.
The density \(\rho\) enters as a bounded multiplicative weight: smooth changes to \(\rho\) rescale eigenvalues by bounded factors but cannot alter their ordering, their degeneracies, or their symmetry character.
If \(\partial\Omega\) admits no orientation-reversing symmetry, then no eigenstate of the problem admits one either.
The proof, with quantified bounds, is in Appendix A. The path through Appendix A is itself a connecting thread we want to preview here, because the same morphological analysis on the energy functional that produces the theorem also produces, in the same appendix and from the same functional, three further results that re-center the framework on physics and probability foundations. We sketch the four threads in the order Appendix A develops them.
The morphological engine.
The Dirac connection and the universe.
The parent structure.
Fisher information and the Cramér-Rao reading.
What’s demonstrated, what’s designed
With the framework on the table, we close Section 1 by accounting for what has been computationally demonstrated, what is designed but not yet implemented, and which acceptance gates remain open. This phase-status discipline runs through the rest of the document; the per-claim accounting appears in Appendix C.
Benzene as the reference demonstration.
Benzene’s HOMO–LUMO gap, computed in MP from atom positions alone with a single-Gaussian-per-atom density approximation—the crudest possible density input, with off-the-shelf untuned Gaussian widths and a 2D cross-section of the inherently 3D molecule—lands at approximately 4.91 eV against a literature value of \(\sim 4.77\) eV: an error of 2.9%. The framework has not been tuned; off-the-shelf parameters were plugged in and the result is already within three percent of the literature value on a problem whose true structure is three-dimensional. This is the empirical signature of morphology dominating chemistry detail: the input contains nothing about the carbon-hydrogen bonds, the \(\pi\)-bonding network, the resonance structure, or any chemistry beyond “six carbons in a hexagon, six hydrogens at the perimeter, single Gaussian per atom.” The framework recovers symmetry directly from the boundary: the \(e_{2u}\) LUMO pair emerges as cleanly degenerate, carrying the \(D_{6h}\) signature of benzene, with no symmetry information supplied as input. The morphological principle finds the symmetry because it is in the geometry. Transition dipoles \(\mu_{ij} = \langle \psi_i | \hat{\mathbf{r}} | \psi_j \rangle\) are computed as direct outputs of the same eigenvalue solve, producing the full coupling matrix without separate calculation—a result we will rely on heavily in Sections 4 and 5.
Y6 and ITIC: the open Phase 0 acceptance gate.
The benzene result establishes accuracy on a small reference molecule. The next acceptance gate, in progress, is the comparison against the published optical gaps of Y6 and ITIC—two OSC active-layer materials with experimentally measured gaps of 1.33 eV and 1.59 eV respectively. This test establishes the framework’s accuracy on the molecules the OSC community actually optimizes against. Implementation is in progress; results will be incorporated into this section when complete.
Multi-scale boundaries for core and \(\sigma\) orbitals.
The current implementation captures the delocalized frontier orbitals at a density cutoff of \(\hat{\rho} = 0.01\) atomic units. This boundary contains HOMO, LUMO, and the few orbitals immediately above and below—the orbitals OSC design cares about. Core orbitals (the 1s electrons localized at each atom) and \(\sigma\)-bonded orbitals live at tighter spatial scales and require correspondingly tighter density cutoffs. The S-L framework handles them as part of the same eigenvalue problem on hierarchical boundaries: same machinery, different cutoffs, with the morphological principle distributing orbitals into their natural domains by characteristic spatial scale. Multi-scale boundary implementation is designed; the current demonstration renders the frontier subset because that is what OSC application physics requires. There is no theoretical obstacle to handling core and \(\sigma\) orbitals; the question is one of implementation scope.
Assembly-scale extension.
The same machinery extends from individual molecules to molecular assemblies. Inter-molecular couplings appear as part of one eigenvalue solve on the assembly’s boundary, with the vectorial structure of the coupling matrix preserving the directional physics that ad-hoc assembly-scale couplings flatten. Implementation is designed; demonstration is the Phase 1 acceptance gate.
Absolute energy anchoring.
One technical note for the careful reader. The current implementation uses a global mass \(m = m_e\) (the electron rest mass) without calibration against a reference molecule, so absolute energy positions of the eigenvalues carry an unknown offset. Energy gaps and spacings are meaningful and validated; absolute positions are unanchored. For OSC design questions—which are about gaps, ordering, transition dipoles, and chiroptical signatures, all relative quantities—this is the correct scope. Two paths to absolute anchoring are viable: reference-molecule calibration (using benzene’s experimentally measured ionization potential at \(-9.24\) eV as a global offset that calibrates all subsequent calculations) and effective-mass tuning per element (fitting element-specific effective masses from a small calibration set, absorbing the boundary-extraction approximation’s effect on absolute scales into element-specific parameters). Either path is a Phase 2 implementation refinement that does not affect any claim made in this document.
Looking forward.
Section 1 has established Morphological Physics as the framework the rest of the document deploys. The S-L formulation factors geometry, density, and morphological structure cleanly. The Morphology Institute’s constructive solution to the Riemann mapping problem justifies the form’s use on arbitrary molecular boundaries. The boundary-determines-structure result makes morphology a first-class input rather than a numerical convenience. Empirical demonstration on benzene confirms the framework’s accuracy at the structural level; remaining acceptance gates are specified in Appendix C.
With the framework on the table, the next question becomes: why has the OSC community optimized around figures of merit that the framework’s structural view exposes as enantiomer-blind? Section 2 traces the historical path that produced this blind spot.
Why the Field Optimized Scalar FoMs
This section accounts for the historical asymmetry between the figures of merit (FoMs) the OSC field has optimized against and the structural properties that govern chirality-mediated performance. The asymmetry is not an oversight. We reconstruct the optimization history—HOMO–LUMO gap through the 2000s, mobility through the early 2010s, morphology metrics through the late 2010s—and show that every FoM the field has adopted is scalar by construction. We then explain why scalar quantities cannot distinguish a molecule from its mirror image, and identify the structural properties that would be needed to make handedness engineerable. We close by surveying the empirical hints—CISS, chiral perovskites, circular dichroism as a characterization signal—that accumulated in parallel with the scalar-FoM program but never coalesced into a design framework. The argument lands the historical claim: the field’s chirality blind spot is structural, not accidental, and lifting it requires a framework that produces structural outputs at scale.
How OSC arrived at its current FoM set
The optimization trajectory across the past 25 years can be traced by which figure of merit was the primary lever in each era.
Through the 2000s and early 2010s, optimization centered on the HOMO–LUMO gap. The bandgap determines which solar wavelengths the active layer absorbs, and matching the gap to the solar spectrum is a precondition for high quantum efficiency. Donor–acceptor energy offsets—the HOMO of the donor relative to the LUMO of the acceptor—set the driving force available for charge separation at the heterojunction. Synthetic chemistry programs through this period explicitly targeted gap tuning: substitution patterns, conjugation length, end-group selection, all aimed at moving the gap into a target window. By the mid-2010s, the gap had been largely optimized for the dominant active-layer architectures, and diminishing returns set in.
The next lever was charge mobility. After photons are absorbed and excitons split into electrons and holes, those carriers must migrate to the electrodes without recombining. Higher mobility yields less recombination loss and therefore higher fill factor and \(J_{sc}\). This produced a wave of work on backbone planarization, side-chain engineering, and crystallinity control—all aimed at increasing the magnitude of the mobility tensor.
By the late 2010s, the community recognized that mobility itself depends on the bulk-heterojunction morphology—the spatial arrangement of donor and acceptor domains in the active layer. New metrics entered the FoM stack: domain size, crystallinity, coherence length, percolation pathway connectivity. These quantified the BHJ morphology in scalar terms, and process optimization (solvent selection, anneal temperature, additive concentration) targeted them.
The standard FoM set today is therefore: the HOMO–LUMO gap (the energetic separation between the highest occupied and lowest unoccupied molecular orbitals, controlling absorption wavelength); the oscillator strength (the magnitude of the optical transition dipole, controlling absorption intensity); the magnitude of the mobility tensor (governing how rapidly charge carriers traverse the active layer); the fill factor (FF, the ratio of the device’s maximum power to the product of its open-circuit voltage and short-circuit current); \(V_{OC}\), the open-circuit voltage; \(J_{SC}\), the short-circuit current density; the power conversion efficiency (PCE); the operational lifetime \(T_{80}\) (the time at which performance drops to 80% of initial); and a set of bulk-morphology metrics (crystallinity, domain size, coherence length, percolation pathway connectivity). Each is a scalar—a single number per molecule, per film, or per device. Synthetic targeting, computational screening, and process optimization all use this set.
Why scalar FoMs miss handedness
Two molecules can have identical chemical formulas, identical bond connectivities, and identical bond lengths—yet be mirror images of one another. The relationship is the same as between a left hand and a right hand: same atoms, same fingers, same joints, but one cannot be superimposed on the other by rotation alone. The two versions are called enantiomers, and the property that distinguishes them is called chirality (from the Greek \(\chi\epsilon\acute{\iota}\rho\), “hand”).
Every measurement in the standard OSC FoM set assigns the same numerical value to a molecule as to its enantiomer. The HOMO–LUMO gap is identical—mirror reflection does not change orbital energies. The oscillator strength \(|\boldsymbol{\mu}|^2\) is identical, since the magnitude of the transition dipole vector is mirror-invariant. The mobility tensor’s eigenvalue magnitudes are identical for the same reason. Fill factor, \(V_{OC}\), \(J_{SC}\), and PCE under unpolarized illumination are integrals over these scalar properties, hence also identical. Domain size, crystallinity, and coherence length are scalar geometric statistics computed by averaging over orientations, hence also identical.
Table 1 pairs each scalar FoM with the structural property it misses—the property that would be needed to discriminate one enantiomer from its mirror.
| Figure of merit | What it captures | What it misses |
|---|---|---|
| HOMO–LUMO gap | Bandgap energy (eV) | Eigenstate symmetry character; chiroptical signatures |
| Oscillator strength | Vector magnitude \(|\boldsymbol{\mu}|^2\) | Vector orientation; sign under enantiomeric inversion |
| Mobility tensor | Magnitudes of eigenvalues | Anisotropy direction; spin-selective transport |
| \(V_{OC}\), \(J_{SC}\), PCE | Performance under unpolarized light | Per-helicity absorption asymmetry |
| Domain size, crystallinity | Statistical size and average alignment | Domain handedness; helicity of the packing |
The blind spot is not an oversight. A scalar quantity cannot, by definition, distinguish a molecule from its mirror image—no synthetic strategy, screening procedure, or process protocol that targets only scalar FoMs can engineer handedness. To make chirality engineerable, the FoM set must include structural quantities: vectorial transition dipoles (with direction, not just magnitude), eigenstate symmetry character (which irreducible representations the orbitals carry), oriented mobility (anisotropy axes, not just eigenvalue magnitudes), and helicity of the morphological packing.
These quantities are precisely what Morphological Physics returns from its boundary-eigenvalue solve. The transition dipole as a vector, the eigenstate’s symmetry under the boundary’s point group, the assembly-scale helicity preference—all are direct outputs of the same eigenvalue computation that produces the spectrum. Until a framework that yielded such structural outputs at scale was available, the field had no path to include them in its FoM stack. With MP on the table, they become engineerable.
The OSC field’s chirality blind spot is structural, not inattention. Self-consistent iteration on a molecular Hamiltonian naturally yields scalar quantities: energies, oscillator strength magnitudes, statistical averages. Vectorial structural quantities (eigenstate symmetry, dipole orientation, helicity preference) exist inside the calculation but are not surfaced as primary outputs, and they do not survive the scale-up to assemblies. A research community optimizes against what its computational machinery produces natively. The figure of merit set follows from the framework, not the other way around. Producing structural quantities natively at every scale is what shifts the FoM set—and that is the structural change MP introduces.
Where the hints came from anyway
The OSC field has not been entirely blind to chirality. Three empirical hints surfaced in parallel with the scalar-FoM optimization program. Each suggested that handedness affects device performance in ways the scalar FoMs could not predict. None coalesced into a unifying design framework. Figure 2 situates them against the design-FoM timeline.
(characterization)
CISS: chirality-induced spin selectivity.
The Naaman group at the Weizmann Institute first reported in 1999 that chiral molecules selectively transmit electrons of one spin orientation over the other. The effect was demonstrated initially in DNA monolayers on metal substrates and extended through the 2010s to chiral organic semiconductors in photovoltaic-relevant architectures. At interfaces where CISS is present, the spin-polarized current suppresses a major recombination pathway—and in principle improves \(V_{OC}\) and operational lifetime. The mechanism was studied in isolation, with experiments targeting CISS interfaces directly rather than incorporating CISS as a lever in the broader OSC optimization stack. There is no published unified design framework that predicts CISS strength from morphology.
Chiral perovskites.
Beginning around 2018 and accelerating since 2020, chiral perovskite1 research demonstrated that chiral organic cations integrated into hybrid perovskite structures produce PCE improvements, spin lifetime improvements, and circular dichroism signatures that achiral perovskite analogues lack. The mechanism involves chirality propagating from the organic cation layer into the inorganic perovskite electronic structure. This work happened in the perovskite research community rather than the OPV community, and the modeling framework that emerged was perovskite-specific. The general phenomenon—handedness as a performance lever in solar cell active layers—was demonstrated, but the design framework remained material-class–specific.
Circular dichroism as characterization.
Circular dichroism (CD) spectroscopy—the differential absorption of left-circularly polarized light versus right-circularly polarized light—has been a standard characterization tool in chemistry since the mid-twentieth century. In OPV research, CD spectra are routinely measured to confirm that a chiral active layer retains its handedness through processing. The technique is deployed as a verification signal (did the chirality survive?), not as a primary design output. There is no published computational framework that takes a target CD spectrum and returns the morphology that would produce it.
Each hint is real, each is published, and each provides partial evidence that chirality affects OSC-class device performance. None individually gives the field a unified design framework. CISS optimization targets interfaces in isolation; chiral perovskite work produces material-specific results; CD spectroscopy is used to characterize after the fact rather than to design before. The empirical evidence has accumulated for fifteen years; the framework that would unify it has not.
Looking forward.
Section 2 has established that the OSC field’s chirality blind spot is a structural consequence of the scalar FoM set the field has optimized against, and that empirical hints have accumulated without unification. Section 3 examines the three hints more closely—what each community has demonstrated, what each has learned, and what specific piece is missing from each—before Section 4 supplies the unification that Morphological Physics provides.
Three Islands of Chirality Work
This section examines the three communities whose work has produced empirical evidence that handedness affects OSC-class device performance. Each is an island—published, real, and bounded by its own framework. We treat each in turn under a common rhythm: what was done, what was learned, what is missing. The pattern across the three is consistent: substantial empirical progress, but no unifying framework that predicts the structural input from morphology with the scale and generality needed for design. Section 4 supplies that unification.
Chiral OSC and CISS
What was done.
The chirality-induced spin selectivity (CISS) effect was first reported in 1999 by Ron Naaman’s group at the Weizmann Institute, who observed that electrons transmitted through self-assembled monolayers of chiral DNA on metal substrates emerged with measurable spin polarization. Over the following two decades, the program extended to chiral peptides, helicenes, metal-organic complexes, and eventually to chiral organic semiconductors in photovoltaic-relevant architectures. David Waldeck’s group at Pittsburgh, working in close collaboration with Naaman, demonstrated CISS in chiral self-assembled monolayers at metal/semiconductor interfaces of direct relevance to OSC device stacks. Other contributors include Yossi Paltiel at the Hebrew University and Helmut Zacharias at Münster, along with several U.S. and European labs developing CISS-based spin valves and spin filters in adjacent device contexts.
What was learned.
The empirical signature is robust across systems. Spin polarizations of 5–25% are routinely measured at chiral organic interfaces, with higher values reported in specific architectures. The downstream mechanism for OSC devices is well-established: spin-polarized carriers suppress bimolecular recombination at the donor–acceptor heterojunction, because recombination requires spin pairing that is statistically disfavored when one spin species dominates. Literature estimates place \(V_{OC}\) gains from CISS interfaces at 30–80 mV (about 3–8% relative) and operational lifetime (\(T_{80}\)) extensions at 2–3\(\times\) over achiral baselines, with the upper end of these ranges contingent on interface engineering details. The depth dependence, temperature dependence, and material dependence of the effect have been characterized; the experimental program is mature.
What’s missing.
CISS strength is predicted today from molecular geometry only empirically: synthesize a candidate chiral structure, measure the resulting spin polarization, iterate. Existing theoretical models—Yeganeh–Ratner, Diaz–Mujica, Göhler–Hamelbeck, and others—explain qualitative features of the effect but operate at the per-molecule level, do not extend cleanly to molecular assemblies, and do not take morphology as a design input. A device engineer who wants to know how much CISS-induced \(V_{OC}\) gain to expect from a specific candidate molecular packing has no published method that returns the answer with quantitative accuracy. The CISS community has identified the lever; the design framework that lets the structural input be specified from morphology rather than guessed has not yet been published.
CD spectroscopy and the rotational strength
What was done.
Circular dichroism (CD) spectroscopy is a long-established technique in chemistry. The differential absorption of left-circularly polarized versus right-circularly polarized light produces a wavelength-dependent signal that is signed—positive for one enantiomer, negative for its mirror, and zero for achiral materials. The theoretical foundation was supplied by Leon Rosenfeld in 1928, who derived the rotational strength of an electronic transition as the imaginary part of the dot product of the electric and magnetic transition dipole moments. The Rosenfeld expression has been the standard for computing CD spectra ever since.
For a molecular transition from electronic state \(i\) to state \(j\), the rotational strength is: \[R_{ij} = \operatorname{Im}\bigl(\boldsymbol{\mu}_{ij} \cdot \mathbf{m}_{ij}\bigr), \label{eq:rosenfeld}\] where \(\boldsymbol{\mu}_{ij} = \langle i | \hat{\mathbf{r}} | j \rangle\) is the electric transition dipole (vectorial), \(\mathbf{m}_{ij} = \langle i | \hat{\mathbf{L}} + 2\hat{\mathbf{S}} | j \rangle\) is the magnetic transition dipole (vectorial), and \(\operatorname{Im}(\cdot)\) takes the imaginary part of the dot product.
The rotational strength is the chirality signal at the transition level. The full CD spectrum is the sum of these contributions across all transitions, weighted by appropriate lineshape functions. The expression is signed by construction: enantiomers produce opposite-signed contributions because the magnetic dipole flips sign under spatial inversion while the electric dipole does not.
The key vectorial structure: \(\boldsymbol{\mu}_{ij}\) and \(\mathbf{m}_{ij}\) are both vectors with direction. Their dot product captures their relative orientation, which is what carries the chirality information. A method that returns only magnitudes (oscillator strength \(|\boldsymbol{\mu}|^2\)) cannot reproduce this signal. A method that returns vectorial transition dipoles—as MP does directly from the same eigenvalue solve that produces the spectrum—can.
The computational implementations have been carried by quantum chemistry codes—ORCA, Gaussian, Q-Chem, ADF—which compute CD spectra via time-dependent density functional theory (TDDFT) at the per-molecule level. In OPV research, CD measurements are routinely reported to confirm the handedness of chiral active layers and to track its survival through solution-state and solid-state processing.
What was learned.
The Rosenfeld expression cleanly separates the chirality signal from the magnitude of the optical absorption. Two molecules with identical oscillator strength can produce opposite-signed CD if their electric and magnetic transition dipoles are aligned versus opposed; one molecule can produce zero CD despite strong absorption if the two vectors happen to be perpendicular. The implication is that CD is a vectorial signature, accessible only when both transition dipoles are computed with their direction information preserved. TDDFT computes both and produces per-molecule CD spectra at quantitative accuracy. For chiral OSC materials, this has yielded robust characterization: CD signatures confirm that handedness survives film formation, aggregation in chiral polymers amplifies or suppresses the signal in ways that correlate with morphological packing, and solid-state CD differs systematically from solution-state CD in ways that reveal inter-molecular vectorial couplings.
What’s missing.
TDDFT does not scale. Computing CD spectra for individual molecules at quantum-chemistry-grade accuracy already costs minutes to hours per molecule on modern hardware. Film-scale CD—the assembly of many molecules with their inter-molecular vectorial couplings preserved—is computationally intractable in TDDFT for the system sizes relevant to OSC devices. The standard compromise is to compute per-molecule CD with TDDFT and add ad-hoc empirical corrections for solid-state aggregation; the per-molecule signatures are recognizable in measured solid-state CD, which validates the per-molecule calculation but does not bridge to design. There is no published method that takes a target CD spectrum and returns the molecular morphology or film organization that would produce it. The inverse problem—which is the design problem—has no scalable solution in the current framework.
Chiral perovskites and the recent rise
What was done.
Beginning around 2017–2018, several research groups recognized that incorporating chiral organic cations into hybrid organic–inorganic perovskite structures produced devices with optoelectronic responses distinctly different from their achiral analogues. Among the early demonstrators were Matthew Beard’s group at the U.S. National Renewable Energy Laboratory, Sang-Wook Kim at Yonsei University, and Wei Jiang at the Hong Kong University of Science and Technology. The synthetic strategy was direct: replace the standard organic cation (methylammonium, formamidinium) with a chiral organic cation, often \(\alpha\)-methylbenzylamine derivatives or BINOL-derived structures. The resulting hybrid perovskite inherits handedness from the organic cation layer through cation-induced lattice distortion in the inorganic framework. By 2024, dozens of chiral perovskite compositions had been synthesized and characterized in device-relevant architectures.
What was learned.
Chiral perovskites consistently outperform their achiral analogues on multiple metrics. Spin lifetimes in chiral lead-halide perovskites have been measured 10–100\(\times\) longer than in their achiral analogues. CISS-like effects have been observed at chiral perovskite/electrode interfaces, suggesting that the spin-selectivity mechanism familiar from chiral organic semiconductors operates in hybrid perovskite systems as well. PCE improvements of a few percentage points have been reported across multiple compositions, though attribution to specific mechanisms—helicity recycling versus spin-polarized transport versus defect passivation—remains contested in the literature. The most significant lesson is that handedness is acting as a performance lever in a solar cell architecture with reach commercial-scale interest, demonstrating that the effect is not confined to specialty organic systems.
What’s missing.
The chiral perovskite results were generated within the perovskite research community using modeling frameworks specific to perovskite structures. The cation-induced lattice distortion mechanism, which propagates chirality from the organic cation layer into the inorganic electronic structure, is described in perovskite-specific terms; the mechanistic story does not translate cleanly to all-organic OPV systems even though both are addressing the same underlying question: how does morphological handedness translate to device performance. A chiral OPV researcher cannot directly use the perovskite mechanistic models. A chiral perovskite researcher cannot directly use the CISS-specific models from chiral organic semiconductor work. Each material class has its own framework. The cross-class generalization that would let chirality serve as a unified design variable across solar cell types is not in the literature.
A natural objection arises at this point: if all three communities have produced empirical evidence that handedness affects performance, why is chirality not already part of the OSC design toolkit? The honest answer is two-part. First, the field does design for handedness empirically. The Naaman group has engineered CISS interfaces; the chiral perovskite community has engineered chiral cations; the Yang Yang group at UCLA has built aligned-OPV polarizing cells. Empirical design for handedness exists. Second, the limitation is predictive, not conceptual. Without a scalable predictive framework, the chirality design space is searched by synthesizing a candidate, measuring its response, and iterating. Synthesis is the rate-limiting step in OSC research—often months per candidate—and the number of candidates the field can search by trial-and-error is small compared with the size of the chirality design space. The cross-scale joint design (molecule and assembly and film simultaneously) that the multiplicative-lever story of Section 5 requires is unreachable by serial trial-and-error at any practical timescale. Morphological Physics does not introduce chirality design; it changes the scale and speed at which the chirality design space can be searched, which is what makes the joint-design ceiling reachable in principle.
Three independent research communities have arrived at three different empirical signatures of what Section 4 will reveal to be one mathematical object. Naaman and Waldeck found spin selectivity in chiral organic interfaces; the chiroptical spectroscopy tradition refined the measurement of rotational strength in chiral molecules; chiral perovskite researchers found PCE and spin lifetime gains in handed cation structures. None of these groups coordinated. None had access to a framework that would unify their findings. That three independent investigations converged on three projections of the same underlying phenomenon is itself evidence the phenomenon is real and structural—empirical convergence of this kind is what physics relies on when the unifying framework has not yet been written down. The framework, when it arrives, makes the convergence make sense in retrospect.
Looking forward.
Section 3 has documented three islands of chirality work in solar cell research. Each is real, each is published, each provides partial empirical evidence that handedness is a performance lever. None individually unifies the picture. The CISS community has the spin-transport mechanism but lacks the morphology-to-CISS predictor at design accuracy. The CD spectroscopy community has the chirality signal at per-molecule scale but lacks the computational scaling to film and assembly. The chiral perovskite community has demonstrated commercial-relevant performance gains but in a material-class–specific framework that does not translate to OPV. The synthesis across the three is consistent: empirical evidence accumulates, frameworks remain fragmented. Section 4 supplies the unification, showing how Morphological Physics produces the chirality signal as a direct output of one boundary-eigenvalue solve, with the same vectorial coupling matrix carrying the directional information all three islands need at all relevant scales.
What MP Sees Structurally
This section supplies the unification that the three islands of Section 3 have lacked. We argue that the CISS effect, the chirality signal of CD spectroscopy, and the handedness-mediated performance of chiral perovskites are not three separate phenomena requiring three frameworks; they are three observable signatures of one structural fact—that a chiral molecular boundary, treated as the domain of the Sturm–Liouville eigenvalue problem in Equation [eq:sl], produces chiral eigenstates as a direct output. From those eigenstates, the vectorial coupling matrix follows by integration. From the coupling matrix, the chirality signal at every scale follows by the same machinery. The argument proceeds in three steps: how MP recognizes chirality at the molecular scale (the boundary’s symmetry group determines the eigenstate’s symmetry group), how the vectorial coupling matrix carries the directional information that scalar FoMs miss, and how the same eigenvalue machinery extends from a molecule to an assembly to a film without changing form. The section closes by positioning MP against the standard computational chain—both visually and quantitatively—with the comparison that establishes the methodological gap MP fills.
How MP recognizes handedness at the molecular scale
A boundary in space carries a symmetry group: the set of rigid transformations (rotations, reflections, inversions) that map the boundary onto itself. For a hexagonal molecule like benzene, the symmetry group is the dihedral group \(D_{6h}\)—six rotational symmetries and a horizontal mirror plane. For a chiral helical molecule, the symmetry group lacks reflections and inversion by definition; it consists only of proper rotations and the identity. The mathematical statement that a boundary is chiral is exactly the statement that its symmetry group excludes orientation-reversing transformations.
A standard theorem in spectral theory states that eigenstates of a self-adjoint operator on a domain transform under irreducible representations of the domain’s symmetry group. For the Sturm–Liouville problem in Equation [eq:sl], the operator \(-\nabla^2\) is invariant under any rotation or reflection (it is a scalar differential operator), and the weighting \(\rho\) inherits the boundary’s symmetry by construction (the promolecule density carries the symmetries of the molecular geometry). Therefore the eigenstates \(\psi_i\) transform under irreducible representations of the boundary’s symmetry group. The eigenstates’ symmetries are inherited from the domain’s symmetries; they cannot exceed them.
The consequence for chirality is direct. If the boundary’s symmetry group contains no orientation-reversing transformation, neither does the symmetry group of the eigenstates. In plain language: a chiral boundary cannot produce achiral eigenstates. The eigenstates inherit the chirality of their domain by construction. They are chiral not because handedness has been added by a perturbation or computed by a numerical correction, but because the eigenvalue problem itself is posed on a domain that lacks the symmetries that would make achirality possible.
The contrast with density functional theory is structural. In DFT, the Hamiltonian is built from the molecular potential, and chirality enters as a feature of that potential. The eigenstates inherit chirality, but the path from input (atomic positions) through operator (effective potential) to output (orbital wavefunctions) is iterative: the SCF procedure converges on a self-consistent density that happens to be chiral because the input was chiral. The framework treats this convergence as a numerical fact rather than a structural one. MP places the chirality at the source—in the domain of the equation—where it is structurally inevitable.
What is remarkable, and worth pausing on, is that the chirality signal of CD spectroscopy, the spin selectivity of CISS, and the helicity preference of chiral perovskites are the same mathematical object in this framework—entries of one vectorial coupling matrix returned by one eigenvalue solve. Three communities have been examining three projections of one underlying object without recognizing the unity. To make that unity visible is the contribution of the formulation. The engineering payoff that follows in Section 5 is a consequence of the unification, not an independent claim.
Figure 3 renders the structural fact visually. The eigenstate of an achiral boundary is itself mirror-symmetric; the eigenstate of a chiral boundary is itself handed. The figure is schematic—eigenstate lobe patterns are stylized rather than computed—but the structural claim it makes is exact: the symmetry of the domain is the symmetry of the eigenstate.
The vectorial coupling matrix
Once the eigenstates have been computed, downstream observables—absorption strength, exciton transfer rate, CD signal, helicity preference—follow as integrals over the eigenstates. The most important of these integrals is the transition dipole moment.
When light interacts with a molecule, the molecule can absorb a photon by promoting an electron from a lower eigenstate to a higher one. The strength of that promotion—how much light is absorbed at the corresponding frequency—depends on an integral involving the two eigenstates and the position operator. This integral is a vector with three components, one for each spatial direction. Its magnitude tells us how much light is absorbed at that transition. Its direction tells us along which axis the molecule absorbs preferentially. Combined with an analogous integral over the magnetic dipole operator, it gives us the chirality signal via the Rosenfeld expression (Equation [eq:rosenfeld]). The same vector is required by all three communities in Section 3: CISS reads off the orientational structure of these transition dipoles, CD spectroscopy reads off their alignment with the magnetic transition dipoles, and chiral perovskites read off the inter-component dipoles that couple chiral organic layers to inorganic frameworks.
The transition dipole moment from eigenstate \(i\) to eigenstate \(j\) is: \[\boldsymbol{\mu}_{ij} \;=\; \langle \psi_i \, | \, \hat{\mathbf{r}} \, | \, \psi_j \rangle \;=\; \int_\Omega \psi_i^*(\mathbf{r}) \, \mathbf{r} \, \psi_j(\mathbf{r}) \, d^3\mathbf{r}. \label{eq:transition-dipole}\]
Each symbol carries a specific meaning. \(\psi_i\) and \(\psi_j\) are eigenstates from the Sturm–Liouville solve of Equation [eq:sl]—they are the wavefunctions of the molecule’s electronic states at energies \(\lambda_i\) and \(\lambda_j\). The hat on \(\hat{\mathbf{r}}\) denotes a quantum-mechanical operator; in this case \(\hat{\mathbf{r}}\) simply multiplies the wavefunction by the position vector \(\mathbf{r}\). The asterisk on \(\psi_i^*\) denotes the complex conjugate of \(\psi_i\), a technicality required because wavefunctions are generally complex-valued. The domain of integration \(\Omega\) is the molecular boundary defined in Equation [eq:sl]; the volume element \(d^3\mathbf{r}\) indicates that we integrate over every point in space inside \(\Omega\).
The resulting integral has three components (\(\mu_{x,ij}\), \(\mu_{y,ij}\), \(\mu_{z,ij}\)), each computed by integrating the corresponding coordinate against the product of the two eigenstates over \(\Omega\).
The magnitude \(|\boldsymbol{\mu}_{ij}|^2\) is the oscillator strength: how strongly the molecule absorbs light at the frequency of the \(i \to j\) transition. The direction tells the molecule’s response to polarized light: linearly polarized light along \(\boldsymbol{\mu}_{ij}\) is absorbed preferentially. Together with the magnetic transition dipole \(\mathbf{m}_{ij}\), the vector \(\boldsymbol{\mu}_{ij}\) enters the Rosenfeld expression (Equation [eq:rosenfeld]) to produce the rotational strength—the chirality signal at the transition level.
The key fact for design: MP returns the full vector \(\boldsymbol{\mu}_{ij}\) as a direct output of the same eigenvalue solve that produces the spectrum. No separate calculation. No additional iteration. The eigenstates are computed; the dipoles follow by integration.
The collection of all transition dipoles between all pairs of eigenstates is the molecule’s vectorial coupling matrix. It is structured: rows and columns indexed by eigenstates, entries being three-component vectors. Standard computational practice extracts the magnitudes of this matrix (the oscillator strengths) per molecule via TDDFT, and computes specific cross-products with magnetic dipoles to produce per-molecule CD spectra. TDDFT formally computes vectorial transition dipoles in its underlying calculation, but standard practice reports their scalar magnitudes and projected per-transition CD values rather than the vectorial structure itself. The directional information—which entries of the matrix are aligned with which others, what handedness their relative orientation defines—is therefore available in TDDFT’s intermediate quantities but is not surfaced as a structural object that downstream calculations can read directly. The limitation is one of exposure and composition, not formal capability: TDDFT does not compose vectorial single-molecule outputs into assembly-scale predictions because TDDFT does not reach assembly scale in the first place. In MP, the full vectorial coupling matrix is the primary output at every scale, by construction. Every downstream observable in this document is computed from it.
Same machinery at film scale
Nothing in the Sturm–Liouville formulation requires the domain \(\Omega\) to be a single molecule. The same eigenvalue problem \(-\nabla^2 \psi = \lambda \rho \psi\) on \(\Omega\) with \(\psi|_{\partial \Omega} = 0\) applies to a single molecule, a molecular dimer, an aggregate of dozens of molecules, or a film with periodic boundary conditions. The domain and the density change scale; the form of the equation does not.
At assembly scale, the boundary \(\partial \Omega\) encloses multiple molecular units. The density \(\rho\) inside reflects the morphology of the packing: where individual molecules sit, how they orient, the void regions between them, the degree of crystallinity. The eigenstates \(\psi_i\) become delocalized across the assembly. The vectorial coupling matrix now includes both intra-molecular transitions (electron jumps within one molecule) and inter-molecular transitions (electron jumps from one molecule to another). The inter-molecular entries are what determine exciton transfer rates, charge separation efficiency at donor–acceptor interfaces, and assembly-scale CD signatures.
At film scale, the boundary becomes periodic (the molecular packing repeats), and the density \(\rho\) inherits the film’s morphological structure: alignment direction, helical pitch (if the film is chiral), domain orientation, interfacial sharpness. The eigenstates \(\psi_i\) become film-scale modes that span many unit cells. The coupling matrix captures the film’s directional optical response: per-polarization absorption (the basis for the dichroic ratios reported in aligned-OPV literature), per-helicity absorption (the basis for film-scale CD and helicity selection), and the anisotropy directions that determine angular response. The Rosenfeld expression applied to these film-scale transition dipoles returns the film’s CD spectrum directly—without per-molecule TDDFT followed by ad-hoc film corrections.
The standard computational chain hands off between three frameworks: DFT for the molecular ground state, then TDDFT for the per-molecule excited states, then an empirical coupling model for the assembly, then further phenomenological corrections for the film. Each handoff loses information that does not survive translation between formalisms—particularly the vectorial structure of the couplings, which is reduced to scalar magnitudes at each transition. MP does not hand off. The same eigenvalue machinery applied at the molecular boundary, applied at the assembly boundary, applied at the film boundary, produces the same kind of output at every scale—eigenstates, spectrum, vectorial coupling matrix—preserving the directional information that handedness requires.
Figure 4 compares the two computational paths visually. Table 2 captures the comparison quantitatively.
+ basis set + convergence
to self-consistent density
excited states per molecule
(ad-hoc, hand-fit)
corrections
per-molecule CD, oscillator
strength magnitudes,
fitted film response
+ element types
(Gaussian sum)
at chosen scale
(one shot)
eigenstates, spectrum,
vectorial coupling matrix,
CD spectrum — at the
chosen scale
| Property | Standard practice | Morphological Physics |
|---|---|---|
| Inputs required | Atomic positions, basis set, exchange–correlation functional, convergence criteria | Atomic positions, element type, Gaussian width per element |
| Computation per molecule | Iterative SCF (DFT) + iterative TDDFT for excited states | One eigenvalue solve on the boundary |
| Cost (small molecule) | Minutes to hours | Seconds to minutes |
| Output type | Scalar quantities (gap, oscillator strength magnitudes); CD via separate TDDFT calculation | Vectorial transition dipole matrix, eigenstate symmetries, CD spectrum—all from one solve |
| Scale reachable | Molecule (TDDFT); film-scale couplings are hand-fit | Molecule, assembly, film—same machinery, different boundary |
| Inverse problem | Not posable in current framework | Direct: specify target output, vary boundary parameters in the eigenvalue problem |
Looking forward.
Section 4 has shown how Morphological Physics unifies the three islands of Section 3 as observable consequences of one structural fact: chiral boundaries produce chiral eigenstates, and the same vectorial coupling matrix—returned by one eigenvalue solve—carries the directional information that CISS, CD spectroscopy, and chiral perovskites all read off through their respective measurements. The unification holds at molecule, assembly, and film scale, by the same machinery. With the structural view in place, the engineering payoff comes into focus. Section 5 walks through three multiplicative levers—helicity recycling, CISS-mediated transport, and dark-exciton extension—each with its own published empirical basis and each accessible because MP supplies the unified prediction layer that the standard chain cannot provide.
Three Levers and the Joint-Design Ceiling
This section turns the structural fact established in Section 4—that chiral morphology produces chiral eigenstates as direct outputs of one eigenvalue solve—into concrete engineering. We open by identifying the dominant non-radiative voltage loss limiting third-generation organic photovoltaics: triplet recombination at the donor–acceptor interface, statistically reinforced by the three-fold degeneracy of triplet spin states. The field has no structural lever against this problem in the standard framework, and the matter-only optimization program has reached diminishing returns. Chirality, treated as a structural design variable via Morphological Physics, provides three: CISS-mediated spin polarization that suppresses the triplet recombination pathway, helicity recycling that captures both polarization channels of incident sunlight, and dark-exciton extension that lifts the diffusion-length ceiling. Each lever has independent published empirical support. The MP contribution is that all three are accessible from the same morphological design space and can be jointly optimized, yielding super-linear gains the field cannot reach by serial lever-by-lever optimization. We close with the joint-design ceiling, the framework-projected payoff when matter and light are co-designed against one prediction layer.
The triplet recombination ceiling
Modern non-fullerene-acceptor (NFA) OSC devices have achieved power conversion efficiencies above 19%, but the gap between achieved efficiency and the thermodynamic Shockley–Queisser ceiling remains stubbornly wide. The gap is driven primarily by non-radiative voltage losses at the donor–acceptor interface, and within that loss budget a specific physical mechanism dominates: triplet recombination, statistically amplified by the three-fold degeneracy of the triplet spin manifold. This subsection traces the loss to its mechanism, explains why standard tools have no structural lever against it, and opens the door to chirality as the lever the third-generation problem needs.
The mechanism.
When sunlight is absorbed in the active layer, an electron is promoted from an occupied molecular orbital to an unoccupied one, leaving behind an oppositely-charged hole. The bound electron–hole pair is called an exciton. Excitons carry spin information: each particle has spin \(\pm \tfrac{1}{2}\), and the combined two-spin system can be in either a singlet state (spins antiparallel, total spin \(S=0\)) or a triplet state (spins parallel, total spin \(S=1\)). The singlet is a single state. The triplet has three possible orientations of the total spin vector—\(T_+, T_0, T_-\)—and is therefore three-fold degenerate.
The recombination problem.
Excitons that reach the donor–acceptor interface form a charge-transfer (CT) state: an electron-hole pair with the electron localized on the acceptor molecule and the hole on the donor. The CT state inherits the spin character of the parent exciton. A singlet CT state recombines radiatively (emitting a photon, contributing only the unavoidable radiative limit to V\(_{OC}\) loss in detailed balance). A triplet CT state recombines non-radiatively (dissipating its energy as heat, contributing real V\(_{OC}\) loss). Because the triplet manifold is three-fold degenerate, the statistical weight of the triplet recombination pathway is three times that of the singlet pathway purely from spin counting. In modern NFA-OSC architectures, where the singlet–triplet energy gap of the CT state is typically below \(0.2\) eV, this three-fold degeneracy dominates the non-radiative recombination budget. It is the primary structural source of the V\(_{OC}\) gap that limits third-generation OPV. Figure 5 renders the loss budget as a stacked bar.
Why the standard framework has no structural lever.
The recombination problem is fundamentally about spin. Two paths to suppression are physically available: lift the singlet–triplet degeneracy of the CT state (chemistry—shift the gap through molecular engineering), or bias the spin populations entering the CT state so that the singlet pathway dominates (physics—engineer the carrier transport for spin selectivity). The first path has been pursued for fifteen years through molecular engineering (push–pull architectures, fluorinated substituents, asymmetric acceptor designs) with steadily diminishing returns; the singlet–triplet gap can only be tuned so much without compromising other device-relevant properties. The second path, biasing spin populations through structural design, has had no framework in the standard computational toolkit. Selecting electron spin requires a chiral medium, and chirality has not been a computable design variable.
The chirality opening.
A chiral interface that produces spin-polarized electron transport addresses the triplet recombination problem directly. When the electron arriving at the donor–acceptor interface carries a dominant spin orientation, the spin pairing required to form a triplet CT state is statistically suppressed. The three-fold degeneracy stops being a free energy reservoir for non-radiative loss. The V\(_{OC}\) recovers. Operational lifetime extends, because triplet-derived reactive intermediates are the dominant degradation pathway in NFA-OSC and suppressing triplet formation suppresses the degradation channel as well.
Disentangling the design variables
Before walking through the three engineering levers, we pause for the conceptual scaffolding the reader needs. The terms spin, chirality, enantiomer, racemic mixture, and handed morphology do not all mean the same thing, and the engineering implications of each differ.
Spin is a property of electrons.
Each electron carries an intrinsic angular momentum of \(\pm \tfrac{1}{2}\) in units of \(\hbar\), independent of any property of the molecule it occupies. In a closed-shell molecule—the ground-state configuration of essentially every OSC active material—every occupied orbital contains one spin-up and one spin-down electron, and the total spin is zero. Spin is a property the electron carries; the molecule’s role is to provide the environment in which spin dynamics play out.
Molecular chirality is a property of nuclear geometry.
A molecule is chiral if no improper symmetry operation (mirror reflection, spatial inversion, improper rotation) maps the molecule onto itself. The two mirror-image forms of a chiral molecule are called enantiomers. Two enantiomers have identical chemical formulas, identical bond connectivities, identical bond lengths, and identical energies in any achiral environment. They differ only in handedness.
Enantiomeric composition is set by synthesis.
Standard organic synthesis from achiral starting materials produces a racemic mixture—50% of each enantiomer. The two enantiomeric components produce opposite-signed chirality effects (opposite CD signals, opposite CISS direction), which average to zero in any bulk measurement. Producing a single enantiomer (an enantiopure sample) requires asymmetric synthesis (chiral catalysts or chiral starting materials) or post-synthesis resolution. The chirality-based OPV literature uses enantiopure samples by design, paying the synthetic cost to access the handedness signal.
Assembly chirality is independent of molecular chirality.
This is the design variable that most concerns the engineer. Achiral molecules can pack chirally—helical column stacks of planar aromatics, twisted bilayer arrangements, screw-axis ordering in molecular crystals—producing chiral morphologies built from symmetric building blocks. Conversely, a racemic mixture of chiral molecules typically packs into an achiral film because the two enantiomers cancel at the assembly scale. The four combinations are summarized in Figure 6.
CISS couples chirality to spin.
The chirality-induced spin selectivity effect is the bridge from chirality (a property of nuclear geometry) to spin (a property of electrons). At a chiral interface, the geometry of the medium influences the spin orientation of carriers passing through. The mechanism is not yet fully settled in the theoretical literature, but the empirical fact is robust: handed structures produce spin-polarized currents.
What MP returns at each cell.
Because MP’s input is the boundary of the molecular or assembly density at the chosen scale, the chirality of the eigenstates follows from the geometry of the boundary regardless of whether the underlying molecules are chiral. For the achiral-molecule-with-chiral-packing case (the bottom-left cell of Figure 6), MP returns film-scale handed eigenstates from the boundary alone. For the enantiopure case (bottom-right), MP returns molecule-scale and film-scale handed eigenstates compounded. For the racemic case (top-right), MP returns sign-cancelled outputs, correctly capturing the empirical fact that bulk handedness effects vanish. For the achiral-both case (top-left), MP returns achiral eigenstates, correctly capturing conventional OSC behavior. The framework handles all four cases by the same machinery, with only the boundary input changing.
Lever one: CISS-mediated transport (the headline)
This subsection treats CISS-mediated spin polarization as the headline lever, because it directly addresses the triplet recombination ceiling of Section 5.1. CISS converts the structural handedness of an interface (boundary geometry) into a statistical bias in electron spin transmission, which directly suppresses the dominant non-radiative voltage loss in third-generation OPV.
The mechanism.
Electrons passing through a chiral medium emerge with their spin partially aligned along the direction of motion—one helicity preferred over the other by the medium’s handedness. Implemented at the donor–acceptor heterojunction or at the electrode contact, this spin filtering means that the electron arriving at the charge-transfer formation site is statistically more likely to carry a particular spin orientation. When that orientation matches the hole’s spin, a singlet CT state forms (radiative recombination, in detailed balance). When it does not, the system must form one of the three triplet sub-states—but the spin-pairing probability for triplet formation is reduced when one spin species dominates the electron population. The three-fold degeneracy of the triplet pathway loses its statistical advantage. The dominant V\(_{OC}\) loss channel in modern NFA-OSC is suppressed at the source.
The design variable.
For the engineer, the controllable parameter is the structure of the chiral interface: which molecules sit at the donor–acceptor heterojunction (or at the electrode contact), in what orientation, with what packing handedness. For systems using enantiopure chiral active-layer materials (chiral fullerenes, chiral perylene diimide derivatives, cation-modified perovskites), the design variable is the chemistry of the chiral component combined with its interfacial alignment. For systems using achiral active-layer materials, the design variable is purely the packing morphology—handed columnar stacks at the heterojunction, twisted bilayer arrangements, screw-axis ordering in the contact region. Both routes are accessible to the engineer through processing control: alignment fields, surface templates, anneal protocols, choice of additives that promote one packing motif over another.
What MP returns.
Given the boundary morphology of the heterojunction (as specified by the engineer’s processing choices), MP returns the vectorial coupling matrix at the interface. The components of this matrix that link orbital states across the heterojunction—the inter-component dipoles that determine charge transfer kinetics—carry the directional information from which the spin polarization magnitude is computed. The CISS strength is a structural output of the same eigenvalue solve that returns the gap, the oscillator strength, and the optical response. The engineer specifies a candidate morphology; MP returns the expected spin polarization and its consequence for V\(_{OC}\) and \(T_{80}\).
Quantitative payoff.
Published CISS measurements report spin polarizations of 5–25% at chiral organic interfaces (Naaman group, Waldeck group, and several U.S. and European labs). The V\(_{OC}\) gain consequence in NFA-OSC architectures incorporating CISS interfaces has been estimated at \(30\)–\(80\) mV (\(\sim 3\)–\(8\%\) relative V\(_{OC}\), corresponding to a roughly proportional PCE improvement). The operational lifetime gain is more dramatic: triplet-derived reactive intermediates are the dominant degradation pathway in NFA-OSC, and structurally suppressing triplet formation is projected to extend \(T_{80}\) by \(2\)–\(3\times\) over achiral baselines. The upper end of these ranges is contingent on interface engineering details that MP’s prediction layer would let the engineer optimize directly rather than discover by candidate synthesis.
(triplet pathway suppressed)
spin-polarized transmission
Lever two: helicity recycling
The second lever addresses a different physical channel: the absorption of polarized sunlight. A chiral active layer absorbs one circular polarization preferentially over the other—but on its own this is a loss rather than a gain, because half of unpolarized incident light passes through. Helicity recycling is the architectural addition that converts the absorption preference into a net gain: a back mirror reflects the unabsorbed helicity, the reflection flips its polarization, and the now-preferred helicity is absorbed on the second pass.
The mechanism.
Sunlight is unpolarized—equal parts left-circularly polarized (\(\sigma^-\)) and right-circularly polarized (\(\sigma^+\)). A chiral active layer with handedness preference absorbs (say) \(\sigma^+\) at high cross-section and transmits \(\sigma^-\) at low cross-section. Without a back mirror the \(\sigma^-\) component passes through and is lost; net absorption is approximately \(50\%\) of available light. With a back mirror placed at the optical path length for constructive interference, the transmitted \(\sigma^-\) reflects, and reflection from a mirror flips the helicity: \(\sigma^- \to \sigma^+\). On the second pass through the active layer the now-\(\sigma^+\) light is absorbed at the layer’s preferred cross-section. Both helicities are eventually absorbed; net absorption approaches \(100\%\).
The design variable.
Three engineering parameters set the lever: the active layer’s dichroic ratio (how strongly it prefers one helicity, controlled by the morphology), the back-mirror placement (the optical path between absorber and mirror, tuned for constructive interference at the absorption wavelength), and the active-layer thickness (chosen against the dichroic ratio for optimal first-pass plus second-pass absorption). All three are processing parameters within the engineer’s control.
What MP returns.
Given the chiral active layer’s morphology (its boundary at film scale), MP returns the helicity-resolved absorption cross-section as part of the vectorial coupling matrix at film scale. The dichroic ratio is computed structurally—the engineer can solve the inverse problem (given a target dichroic ratio, find the packing morphology that produces it) rather than measuring after candidates have been synthesized.
Quantitative payoff.
Literature estimates from chiral-photonic simulation work and aligned-OPV experimental results (Yang Yang group at UCLA, Greenfield group, and others) place the relative PCE gain from helicity recycling at approximately \(10\%\) for typical NFA-OSC architectures with experimentally realized dichroic ratios. The mechanism is independent of CISS: chiral active layers without back mirrors do not benefit; the gain requires the architectural addition.
(set by morphology)
Lever three: dark-exciton extension
The third lever addresses the lifetime of the exciton between absorption and charge separation. In aligned chiral films, the lowest-energy exciton can be optically dark—symmetry-forbidden from radiative decay back to the ground state. Dark excitons live longer than bright excitons; longer lifetime translates to longer diffusion length, which directly controls the active-layer thickness over which excitons can be harvested. The lever’s downstream observable is \(J_{SC}\) improvement.
The mechanism.
A bright exciton is one whose transition dipole moment back to the ground state is non-zero; it can emit a photon and return on a timescale set by the dipole strength. A dark exciton has zero (or near-zero) transition dipole to the ground state by symmetry: radiative decay is forbidden. The exciton must instead decay through other channels—non-radiative relaxation, charge separation at an interface, or intersystem crossing—which are typically slower than radiative decay. Lifetime extends. The exciton diffusion length scales as \(L_D = \sqrt{\tau D}\), where \(\tau\) is the lifetime and \(D\) is the diffusion coefficient. Extended \(\tau\) at unchanged \(D\) gives extended \(L_D\), and an extended \(L_D\) permits a thicker active layer that absorbs more light without losing exciton-to-charge conversion efficiency.
The handedness connection.
In aligned chiral films, the eigenstates carry the film’s handedness through their nodal structure. The lowest-energy exciton—the one created first after absorption and the one that needs to diffuse to a charge-separating interface—can have a transition dipole moment that vanishes by the film’s symmetry. The bright/dark assignment of each exciton is a structural consequence of the alignment morphology, not an empirical accident.
The design variable.
The engineer controls the alignment of the active layer through processing: alignment direction (set by mechanical rubbing of the substrate, applied magnetic or electric field during anneal, growth on an aligned template) and alignment quality (degree of order in the resulting film). Active-layer thickness is then optimized against the extended \(L_D\): a film that is too thin does not harvest all the available absorption; one that is too thick adds parasitic optical losses.
What MP returns.
MP predicts the bright/dark assignment of each eigenstate from the aligned-film morphology. The diffusion length follows from the radiative-decay-suppressed lifetime and the diffusion coefficient. The engineer’s optimization problem becomes solvable structurally: solve for the alignment morphology that maximizes dark-exciton \(L_D\).
Quantitative payoff.
Literature estimates of \(L_D\) extension in dark-exciton systems range from \(2\times\) (modest alignment) to \(10\times\) (optimized alignment). For a baseline OSC \(L_D\) of \(\sim 20\) nm, the extension brings \(L_D\) into a range where the active-layer thickness can be increased substantially. The \(J_{SC}\) gain in optimized thickness configurations is approximately \(10\)–\(20\%\) relative, with the upper end requiring high-quality alignment and well-tuned thickness against the extended diffusion length.
LD ~ 20 nm
LD ~ 40–200 nm
The joint-design ceiling
Sections 5.3 through 5.5 introduced three levers, each with independent published evidence and each accessible from Morphological Physics’s unified prediction layer. This subsection treats them together: first stacked sequentially against an unchanged matter-side morphology to estimate the achievable ceiling under serial optimization, then jointly to show the super-linear ceiling that emerges when morphology is designed against all three structural levers simultaneously.
Sequential composition: the matter-only ceiling plus serial chirality additions.
The matter-only Phase 2 design ceiling—PM6:Y6 baseline plus process knob optimization plus material substitution plus interfacial coupling design plus custom donor–acceptor pair design—is estimated at roughly \(12\times\) BIPV LCOE parity. Adding the three chirality levers serially, each optimized against a morphology held fixed from the previous step, multiplies the LCOE gains: CISS contributes through V\(_{OC}\) recovery and \(T_{80}\) extension; helicity recycling contributes through PCE gain via two-helicity absorption; dark-exciton extension contributes through \(J_{SC}\) gain via extended diffusion length. The serial composition lands the achievable ceiling at approximately \(20\times\) parity. Each lever’s individual contribution is literature-grounded; the serial stacking is the framework’s projection from independent rung-by-rung gains.
Joint composition: super-linear gains from cross-terms.
The same morphological choice produces all three structural effects simultaneously. A handed columnar stack at the donor–acceptor heterojunction produces CISS-mediated spin polarization and gives the active layer dichroic absorption preference and produces a dark-exciton manifold from the aligned eigenstates. Optimizing the morphology against any one of these effects in isolation produces a morphology that is suboptimal for the other two; jointly optimizing against all three captures cross-terms that the serial procedure necessarily loses. The joint-design ceiling sits at approximately \(25\)–\(35\times\) parity, with the range reflecting uncertainty in cross-term magnitudes that joint design has not yet been computationally explored at scale. Figure 10 shows the four-rung headroom ladder.
the super-linear region vs. rung 3)
The honest provenance accounting.
The cross-terms that drive the joint super-linearity are framework-projected, not measured. No published OSC device incorporates all three levers in a jointly-optimized morphology; the field cannot do this design without a unified prediction layer, and Morphological Physics at assembly scale is still in implementation. The validation gates are explicit: Phase 0 is the Y6/ITIC small-molecule comparison against literature gaps (in progress); Phase 1 is the assembly-scale demonstration of CISS prediction (designed); Phase 2 is chiral validation against a published CD spectrum (designed); Phase 2+ is full joint-design implementation with inverse-problem solvers (Phase 2+). The serial \(\sim 20\times\) ceiling is achievable through serial optimization with literature-grounded methods; the joint \(\sim 25\)–\(35\times\) ceiling requires the framework’s full deployment. Appendix C documents the validation status per claim.
The factoring of geometry from chemistry that Sturm–Liouville form provides has an engineering consequence beyond the mathematical one. It separates the engineering disciplines that contribute to OSC design. Atomic identity and atomic placement are the chemist’s domain; molecular packing and film morphology are the engineer’s. In the standard computational framework these design surfaces are entangled in the calculation—changing the packing requires re-running the quantum chemistry that started with the atoms. In MP, the same eigenvalue machinery applied to a different boundary returns the engineer’s answer directly. Two communities that have historically had to work in serial can now work in parallel against a shared prediction layer. The framework does not introduce new chirality engineering—empirical chirality engineering exists today. What it introduces is the prediction layer that makes existing chirality engineering scale-tractable, and the chemist–engineer collaboration genuine.
Looking forward.
Section 5 has shown that the triplet recombination ceiling limiting third-generation OPV—the dominant non-radiative voltage loss the standard framework cannot address structurally—is unlockable by treating chirality as a structural design variable. CISS-mediated spin polarization directly suppresses the triplet pathway. Helicity recycling captures both polarization channels of unpolarized sunlight. Dark-exciton extension lifts the diffusion-length ceiling. Each lever has independent published empirical support; the Morphological Physics contribution is that all three are accessible from one morphological design space, jointly optimizable, and yielding super-linear payoff against the matter-only Phase 2 ceiling. The validation gates that establish the framework’s predictions are explicit and documented in the appendix. The conclusion synthesizes what we have argued, what remains open, and what the next steps require.