Secondary Metabolites as Bio-Derived Functional Additives in Perovskite Solar Cell Processing: Molecular Mechanisms, Process–Structure Relationships, and Computational Design—A Critical Review

1. Introduction

Since their first report as visible-light sensitizers in 2009 [1], metal-halide perovskite photovoltaics have reached certified research-cell efficiencies of 28.0% for single-junction devices, 30.1% for all-perovskite tandems, and 35.2% for monolithic perovskite/silicon tandems [2], while progress in p–i–n architectures, tandems, and operational durability has accelerated industrial interest [3,4,5]. The remaining translation gap is increasingly governed by reproducible film formation, interface control, manufacturing yield, stability under combined stressors, and environmental management rather than by absorber efficiency alone [4,6,7].

Because solution-processed perovskites form through coupled precursor complexation, solvent coordination, nucleation, intermediate-phase conversion, drying, and thermal treatment, molecular additives can influence both the wet-film process and the final solid-state defect landscape [8,9,10]. A useful review must therefore connect molecular chemistry to process variables rather than classify additives only by biological origin.

1.1. Process Engineering and Commercialization Gap

Perovskite stability is a device-level property that depends on absorber composition, interfaces, transport layers, electrodes, encapsulation, electrical bias, illumination, temperature, humidity, and oxygen exposure [4,5,7,11,12,13]. Surface and grain-boundary defects can promote non-radiative recombination, ion redistribution, and reactions with environmental species; nevertheless, a spectroscopic shift or improved morphology alone does not establish a unique passivation mechanism [11,12].

Scalable coating further introduces rheology, wet-film lifetime, solvent evaporation, coating speed, substrate temperature, and ambient control as coupled variables [6,14]. Additives that perform well in antisolvent-assisted spin coating may fail in blade or slot-die processes if they alter viscosity, drying fronts, phase segregation, or residual-solvent removal.

This review focuses on alkaloids and xanthines, terpenoids, and polyphenols or phenolic acids used or proposed in perovskite photovoltaics. Primary metabolites, proteins, biopolymers, and synthetic analogues are included only when they provide a clearly labeled mechanistic comparator. Capsaicinoids are treated according to structure and reported function rather than assumed to represent all alkaloids.

1.2. Why Secondary Metabolites?

Secondary metabolites provide chemically diverse scaffolds containing carbonyl, phenolic, carboxyl, heterocyclic nitrogen, and hydrophobic motifs. These groups may alter precursor association, coordinate undercoordinated metal sites, form molecule-specific hydrogen-bond networks, regulate crystallization, or affect oxidation and moisture transport [15,16,17,18,19,20,21]. Their value must be evaluated molecule by molecule because natural origin does not guarantee low toxicity, biodegradability, scalable purification, or lower life-cycle impact [7,14,15,16,17].

This article is a structured critical narrative review, not a statistical meta-analysis. Literature was identified through two complementary sources, Scite-indexed scholarly records and CAS SciFinder-n (CAplus/MEDLINE), both searched up to 28 September 2026. SciFinder-n reference searches combined “perovskite solar cell” with family-specific terms for (i) alkaloids and xanthines (alkaloid, caffeine, theophylline, theobromine, capsaicin, berberine, quinine, nicotine; 36 records), (ii) terpenoids and carotenoids (terpene, terpenoid, terpineol, camphor, camphorquinone, lycopene, carotenoid, menthol, limonene, betulin, artemisinin; 141 records), and (iii) polyphenols and phenolic acids (polyphenol, flavonoid, tannic, gallic and caffeic acids, curcumin, catechin, quercetin, myricetin, ferulic acid, resveratrol; 66 records); Scite searches used equivalent terms together with “additive”, “crystallization”, “passivation”, and “stability”. Records were screened by title and abstract. Studies in which the natural compound acted only as a paste or dispersion solvent (e.g., α-terpineol in carbon or TiO₂ pastes), as a reducing agent in nanoparticle synthesis, or in non-perovskite devices were excluded. Recent reviews were used for citation chaining, but mechanistic and quantitative statements were checked against primary studies whenever available. Quantitative entries in Table 4 were extracted from the main text and tables of the primary studies; values available only in Supporting Information are identified as such.

Evidence is classified as: (i) experimentally demonstrated in a perovskite film or device; (ii) supported by perovskite-specific computation; (iii) inferred from a molecular analogue; or (iv) proposed as a testable design hypothesis. Studies were retained when the molecular identity, incorporation route, perovskite composition, and relevant film or device outcome could be resolved. Reports based only on crude extracts or unrelated biological activity were not used to establish perovskite mechanisms.

1.3. Novelty Relative to Previous Reviews

Recent reviews have surveyed biologically derived components, phytochemical additives, natural plant extracts, natural and nature-inspired biomaterial additives, and bio-inspired multiscale strategies for perovskite optoelectronics [15,16,17,22,23]. Han et al. [23], in particular, provide a broad account of natural and nature-inspired biomaterial additives across metal-halide perovskite optoelectronics.

In contrast, the present review deliberately narrows the chemical scope to structurally defined secondary metabolites and selected bio-inspired derivatives relevant to perovskite solar-cell processing. Its novelty is methodological rather than merely taxonomic: it (i) separates experimentally demonstrated, perovskite-specific computational, analogue-inferred, and prospective evidence; (ii) maps donor availability, hydrogen-bonding capacity, redox activity, hydrophobicity, volatility, solubility, and molecular geometry onto the incorporation route; (iii) connects those descriptors to precursor coordination, nucleation, drying, film and interface structure, optoelectronic response, device stability, and performance; and (iv) evaluates translation to scalable coating and sustainability constraints. Primary metabolites, crude extracts, and biopolymers are used only as explicitly labeled mechanistic comparators rather than merged into the secondary-metabolite evidence base.

2. Classification of Secondary Metabolites

2.1. Alkaloids, Xanthines, and Capsaicinoids

Alkaloids are nitrogen-containing secondary metabolites with diverse electronic structures. For perovskite processing, the relevant descriptors are not simply the number of nitrogen atoms but their lone-pair availability, protonation state, steric accessibility, adjacent carbonyl groups, hydrogen-bond functionality, and solubility [24,25]. Xanthines such as caffeine and theophylline require separate treatment because carbonyl oxygen atoms are generally more defensible Pb²⁺-binding sites than blanket assignment of every ring nitrogen as a donor [18,19].

Primary studies report molecule-specific outcomes. Caffeine was reported to coordinate Pb²⁺ through its carbonyl groups, increase the crystallization activation barrier, promote preferred orientation, and improve thermal stability; a subsequent UMC–DFT study examined adsorption configurations on MAPbI₃ surfaces [18,26]. Theophylline–BF₄⁻ derivatives combine carbonyl-containing xanthine scaffolds with ionic functionality, while an independent first-principles study evaluated theophylline passivation at MAPbI₃ surface defects [19,27]. Capsaicin is a capsaicinoid with amide, phenolic, and aromatic functionality: it has been introduced through the antisolvent in MAPbI₃ p–i–n devices [20] and blended into SnO₂ quantum-dot electron-transport layers of n–i–p devices, where XPS indicated carbonyl coordination to Sn and Pb whereas the amine interaction was weak [28]; it should therefore not be treated as a generic nitrogen donor. Capsaicin added at 0.1 wt% to the MAPbI₃ precursor was further shown to convert the near-surface region from p- to n-type, forming a homojunction about 100 nm below the surface and reaching 21.88% efficiency in p–i–n devices [29]. Quinoline, the parent aromatic N-heterocycle of the quinoline alkaloids, is included only as a non-metabolite comparator; it has been added directly to precursor solutions [30]. Beyond xanthines and capsaicinoids, the tricyclic alkaloid colchicine catalyzed the direct formation of α-FAPbI₃ in an antisolvent-free one-step process [31], and aminophylline, a theophylline–ethylenediamine salt, has been used as a precursor additive in triple-cation devices [32]. Nucleobases such as adenine and cytosine, occasionally described as alkaloid-based passivators [33], are primary metabolites and are treated here only as comparators. Representative molecular structures of the alkaloid-derived additives, xanthines, and capsaicinoids discussed in this review are shown in Figure 1.

2.2. Terpenoids and Carotenoids: Demonstrated and Prospective Cases

Terpenoids derive from isoprene units and span nonpolar hydrocarbons, oxygenated mono- and diterpenoids, multifunctional triterpenoids, and carotenoid tetraterpenes [34,35,36]. Hydrophobicity, vapor pressure, melting point, surface activity, conjugation, redox behavior, and accessibility of alcohol or carbonyl groups can affect solvent coordination, volatilization, crystallization, interfacial protection, and moisture or oxygen transport.

Direct photovoltaic evidence is available for selected structures. The monoterpene alcohol α-terpineol was used as a precursor-solvent additive and associated with delayed crystallization and improved film quality [37]. Camphor and camphorquinone, described as naturally sourced volatile solid additives, were added to the PbI₂ precursor of a two-step process, with stepwise volatilization linked to crystallization control [38]. Lycopene has been investigated as a carotenoid/tetraterpene-derived additive or protective membrane for oxygen-, ultraviolet-, and environmental-stability control [39,40]. The oxygenated sesquiterpene lactone artemisinin coordinates undercoordinated Pb²⁺ and has been used both as a bulk passivator in flexible devices and as an electron-transport-layer/perovskite interlayer in two-step films [41,42]. Among carotenoids, β-carotene has been applied as a bulk additive [43] and, in a 1.74 eV MAPb(Br_0.2I_0.8)₃ p–i–n cell, both through the antisolvent and as a buried interlayer between the hole-selective monolayer and the perovskite [44], and the xanthophyll astaxanthin as a buried-interface modifier [45]. The lupane triterpenoid betulin, added at 0.001 wt% to a MAPbI₃ precursor, was associated with Pb²⁺ coordination and hydrogen bonding supported by NMR, XPS, and DFT, increasing p–i–n efficiency from 19.14% to 21.15% [46]. By contrast, phytol, limonene, and ursolic or oleanolic acid remain prospective scaffolds; their coordination, segregation, self-assembly, or nucleation effects require perovskite-specific validation.

2.3. Polyphenols and Phenolic Acids

Polyphenols and phenolic acids combine molecule-dependent metal coordination, hydrogen bonding, redox activity, strain regulation, and interfacial packing. Tannic acid has been used as a post-treatment that modifies surface chemistry and has been associated with reduced ion diffusion [21]. Gallic acid acts as a coadditive with SnCl₂ in tin perovskites and has also been developed as a dispersible multidentate chelator [47,48], whereas caffeic acid was reported to regulate mixed Pb–Sn crystallization and limit Sn²⁺ oxidation [49]. Catechin, tea polyphenols, and myricetin (the latter combined with a BF₄⁻ pseudohalide) have similarly been used to suppress Sn²⁺ oxidation in Sn-containing absorbers [50,51,52], resveratrol combines Pb²⁺ coordination with superoxide scavenging [53], and the phenylpropanoid plant sunscreen sinapoyl malate has been applied to the electron-transport material to improve ultraviolet stability [54]. Curcumin has been applied at the buried SnO₂/perovskite interface [55], polyphenol encapsulation has been used for combined moisture stability and Pb containment [56], and redox-active flavonoid interlayers have been reported for strain-relieved Sn–Pb devices [57]. These outcomes remain molecule- and architecture-specific. Outside these three families, a crude aqueous garlic extract attributed mainly to the organosulfur metabolite allicin raised n–i–p efficiency from 22.89% to 24.50%; because the active species in the annealed film was not identified, it is retained only as a comparator [58].

Flavonoids and other phenolic scaffolds remain promising candidates, but antioxidant activity measured in solution is not equivalent to stabilization in a solid perovskite film. Mechanistic validation should include oxidation-state measurements, chemical-location analysis, concentration controls, and device aging under defined atmosphere and illumination.

3. Evidence Framework and Process-Oriented Classification

Bio-derived additives can influence precursor coordination, intermediate-phase stability, wetting, nucleation, grain growth, interface chemistry, ion redistribution, and degradation. These functions depend on molecular identity, purity, concentration, solvent system, incorporation route, and perovskite composition [8,9,10,15,16,17]. Renewable origin is therefore treated as a sourcing attribute rather than evidence of low toxicity, biodegradability, or device compatibility.

3.1. Process-Oriented Classification and Evidence Level

A process-oriented classification should combine molecular family with demonstrated interaction and evidence level (Table 1). This avoids treating biological origin as a mechanism and separates experimentally validated additives from computational candidates and mechanistic analogues [15,16,17].

3.2. Molecular Interactions Governing Bio-Derived Additive Engineering

Bio-derived additives can affect perovskite processing through several non-exclusive mechanisms. Electron-rich donor atoms may coordinate undercoordinated Pb²⁺ or Sn²⁺ sites or alter precursor complexes; hydrogen-bond donors and acceptors may interact with halides, A-site cations, or solvent molecules; hydrophobic scaffolds may modify wetting and water transport; and redox-active groups may influence photo-oxidative pathways. Each assignment depends on molecular structure, concentration, protonation, solvent, surface termination, and incorporation route [9,59,60,61].

Coordination and hydrogen bonding can alter nucleation, intermediate-phase stability, grain growth, surface energetics, and defect-associated electronic states. Aromaticity alone does not demonstrate beneficial electronic coupling, and π–π stacking requires an aromatic organic partner rather than the inorganic [MX₆] framework. Likewise, chelation in solution does not prove defect passivation or lead sequestration in a solid device. Table 2 therefore summarizes potential process effects using conditional language and evidence requirements [59,61,62,63].

4. Molecular Interaction Mechanisms of Secondary Metabolite Additives with Halide Perovskites

4.1. Chemical Interactions between Perovskites and Alkaloid-Type Scaffolds

Alkaloids and related molecular scaffolds form a chemically diverse group of nitrogen-containing compounds. Those discussed here include purine alkaloids such as caffeine and theophylline with N-heterocycles and the capsaicinoid capsaicin (Figure 1). Their structures may incorporate heteroaromatic or partially saturated rings, π-conjugated systems, and polar functional groups such as –OH, –NH, C=O, and –OCH₃; sulfur-containing moieties are also considered in some computational candidates (Section 6.1.2) [64]. These features provide potential hydrogen-bond donors and acceptors, electronically available Lewis-base sites, and polarizable aromatic surfaces that could interact with perovskite precursors, surfaces, grain boundaries, and adjacent transport layers. However, the dominant interaction pathway must be established for each molecule and processing environment, rather than inferred from its classification as an alkaloid or from the number of heteroatoms it contains. [24,59,60]

Halide-perovskite surfaces and grain boundaries may contain undercoordinated Pb²⁺ or Sn²⁺ centers, halide vacancies, antisite defects, interstitial species, and other defect-associated configurations. These sites can contribute to carrier trapping, non-radiative recombination, ion migration, and environmentally induced degradation. [59,60,61,65] The targeted defect is not always the same: first-principles calculations on a PbI₂-terminated (100) surface of a formamidinium-based perovskite identified the Pb-on-I antisite as a readily formed defect and found the interaction of xanthines with iodide vacancies to be energetically unfavorable. [66] Alkaloids and related bio-inspired additives may interact with such regions through molecule-specific combinations of hydrogen bonding, Lewis acid–base coordination involving accessible O-, N-, or S-donor atoms, electrostatic interactions, dispersion forces, and interactions associated with π-conjugated frameworks. [59,60,61,65]

The relative contribution of these mechanisms depends on lone-pair availability, protonation state, hydrogen-bond donor or acceptor character, steric accessibility, molecular orientation, solvent environment, incorporation route, perovskite composition, and the specific interface under consideration. Accordingly, the following subsections distinguish hydrogen bonding (Section 4.1.1) from direct Lewis-base coordination (Section 4.1.2) and weaker noncovalent interactions (Section 4.1.3), and state for each the level of experimental or computational evidence available. Because xanthines are the best-resolved cases, they organize the discussion; capsaicin and quinoline are compared in Section 4.4, and terpenoids and polyphenols are treated in Section 4.5 and Section 4.6.

4.1.1. Molecule-Specific Hydrogen Bonding with Halides and Interfaces

Hydrogen-bond donors (–NH, –OH, or protonated N⁺–H) can interact with surface halides. The assignment is molecule-specific: caffeine carries no N–H or O–H donor, theophylline retains one N₇–H donor, and protonation of the free imidazole nitrogen N₉ in theophyllinium salts converts a lone-pair donor into an N⁺–H donor [19]. The clearest case comes from a study of theophylline, caffeine, and theobromine on a formamidinium-based perovskite. An N–H···I hydrogen bond was reported to assist the primary C=O binding to a Pb-on-I antisite in theophylline, with a calculated interaction energy of about −1.7 eV; methylating the N–H (caffeine) weakened it to about −1.3 eV, and the isomer theobromine, whose N–H lies next to the C=O, gave a weaker interaction (about −1.1 eV) and lower device efficiency than the untreated control [66]. These results were obtained by surface post-treatment and cannot be transferred directly to precursor-incorporated caffeine [18].

First-principles calculations on MAPbI₃ indicate that xanthine cations obtained by quaternizing the free imidazole nitrogen bind the surface more strongly than the neutral molecules, with Pb–O and I–H contacts dominating [67]. In [19], salts of the same theophylline cation with BF₄⁻ and Cl⁻ gave different film and device behavior; the calculations reported there address the anion–surface interaction, so the contribution of the cation’s carbonyl and N–H groups was inferred rather than computed. Hydrogen bonding has also been proposed to influence iodide migration at grain boundaries [61], and multifunctional passivants such as trichloromelamine [68] illustrate how coordination and N–H···I interactions may operate together; both are analogies, not demonstrations for the alkaloids discussed here.

4.1.2. Lewis-Base Coordination through O- and N-Donor sites

Alkaloids and related molecular scaffolds do not exhibit uniform affinity for undercoordinated metal sites in halide perovskites. Their coordination behavior depends on lone-pair availability, protonation state, steric accessibility, molecular orientation, competing donor atoms, and the surrounding solvent, surface, or lattice environment. Consequently, coordination should be assigned to a specific functional group and adsorption configuration rather than inferred from the presence of nitrogen alone.

At perovskite surfaces and grain boundaries, coordinatively unsaturated environments may expose Lewis-acidic Pb²⁺ or Sn²⁺ centers and generate electronic states that contribute to carrier trapping and non-radiative recombination. Molecular additives containing electronically available O- or N-donor atoms may coordinate these sites through Lewis acid–base interactions [18,24,25]. When the adsorption geometry is favorable, donation of electron density toward an undercoordinated metal center may decrease the density or electronic activity of defect-associated states. The general donor–acceptor concept is summarized schematically in Figure 2.

The availability of a nitrogen lone pair depends strongly on its molecular environment. Pyridinic nitrogen retains an in-plane lone pair that may participate in coordination, whereas the lone pair of pyrrolic nitrogen contributes to the aromatic π-system and is therefore substantially less available. Protonation can further suppress the Lewis basicity of a nitrogen donor. Accordingly, any proposed N→Pb²⁺ or N→Sn²⁺ interaction must be evaluated with respect to the molecular structure, protonation state, steric environment, surface termination, and adsorption geometry [25,59,62].

Oxygen-containing groups may provide more defensible coordination sites in molecules possessing carbonyl, hydroxyl, or related functionalities. In xanthines such as caffeine and theophylline, carbonyl oxygen atoms are generally the most plausible Pb²⁺-binding sites unless coordination through a ring nitrogen is demonstrated explicitly. These compounds should therefore not be classified generically as nitrogen-donor additives merely because they contain several nitrogen atoms [18,19,24,25].

4.1.3. Dispersion, Metal–π, and Organic π–π Interactions

Alkaloids and related molecular scaffolds containing extended conjugated systems may adsorb at perovskite surfaces and interfaces through dispersion forces, electrostatic contributions, and, under suitable geometrical and electronic conditions, metal–π interactions. These mechanisms depend on molecular orientation, surface termination, intermolecular distance, polarizability, and the surrounding chemical environment. Therefore, the presence of an aromatic ring alone does not establish a specific metal–π interaction.

Metal–π interactions may be considered when an electronically accessible Pb²⁺ or Sn²⁺ center is positioned near the π-electron density of an aromatic system. However, direct experimental evidence for well-defined metal–π contacts in specific alkaloid/perovskite systems remains limited. Such interactions should consequently be described as possible adsorption contributions unless supported by structural, spectroscopic, or computational evidence identifying the corresponding geometry. [60,62]

Conventional π–π stacking requires two aromatic or conjugated organic components arranged at an appropriate distance and relative orientation. The inorganic corner-sharing [MX₆] octahedral framework does not constitute a π-stacking partner. Organic π–π association may nevertheless occur between neighboring additive molecules or between an aromatic additive and an organic interfacial material. Collectively, dispersion, possible metal–π contacts, and organic π–π interactions may influence molecular adsorption, interfacial packing, and local energetics, but they should not be interpreted automatically as Lewis-base coordination or electronic defect passivation. These distinct interaction modes and their principal structural requirements are summarized schematically in Figure 3.

4.2. Defect Passivation and Lead-Containment Claims

4.2.1. Passivation of Metal-Related and Halide-Related Defects

Lewis-basic and multifunctional molecules may interact with coordinatively unsaturated Pb²⁺ or Sn²⁺ centers and with halide-related surface or grain-boundary environments. Metal coordination, hydrogen bonding, electrostatic interactions, and local structural reorganization may contribute to these effects. However, the presence of an interaction with the perovskite surface does not, by itself, demonstrate electronic defect passivation. Each mechanism should be assigned to a specific functional group, defect configuration, and adsorption geometry. [59,61,69]

Defects should preferably be described according to their chemical and structural identity, for example, undercoordinated metal sites, halide vacancies, interstitial species, or antisite defects, rather than classified only as generic “cationic” or “anionic” defects. Formal charge states should be specified only when supported by an explicit defect model or appropriate experimental evidence. [11]

Electronic passivation is more convincingly established when molecular interaction evidence is accompanied by complementary optoelectronic measurements. Relevant observations may include enhanced steady-state photoluminescence, longer time-resolved photoluminescence lifetimes, lower trap densities obtained from appropriate electrical measurements, reduced non-radiative recombination losses, increased quasi-Fermi-level splitting, or improved open-circuit voltage.[11,12] Because each individual measurement can also be affected by changes in morphology, crystallinity, energetic alignment, or carrier extraction, a combination of structural, spectroscopic, computational, and device-level evidence is preferable.

Multifunctional organic passivants have demonstrated such improvements in specific perovskite systems [59,68,69]. Nevertheless, their magnitude and molecular origin should not be generalized quantitatively to all alkaloids or related scaffolds without molecule-specific evidence.

4.2.2. Evidence Required for Lead Containment

The presence of Pb²⁺ in high-performance perovskites creates a potential environmental risk when damaged devices release soluble lead species. Because complete Pb substitution often compromises photovoltaic performance, internal sequestration and external containment are being investigated as complementary risk-mitigation strategies [13,63,70].

Multidentate molecules may complex released Pb²⁺ and reduce its mobility, but solution-phase chelation capacity does not by itself demonstrate effective lead sequestration in a damaged photovoltaic device. For alkaloid-derived additives, claims of lead immobilization require direct leaching tests under defined damage, water exposure, and aging conditions. Established device-level containment systems provide the appropriate experimental benchmark [63,70].

Lead immobilization is therefore a promising but insufficiently demonstrated function of alkaloid additives. The supramolecular complex of hydroxypropyl-β-cyclodextrin and 1,2,3,4-butanetetracarboxylic acid (HPβCD/BTCA), a non-metabolite comparator, provides direct device-level evidence for reduced lead leakage [63], whereas comparable leaching data have not been reported for alkaloid additives. Within the secondary-metabolite literature, device-level lead-containment evidence is currently limited to polyphenol-based systems [48,56]. Combining internal coordination with an external containment layer remains a testable design hypothesis [63].

4.3. Protection against Moisture and Photo-Oxidation

Coordination of an additive to an undercoordinated surface metal site can reduce both defect-associated electronic states and local reactivity toward polar molecules. Hydrogen-bond networks may additionally influence water uptake and ionic motion. These effects are supported for several classes of passivants, but the balance between coordination, hydrogen bonding, and hydrophobicity must be demonstrated for each alkaloid. [61,62]

Light and molecular oxygen provide a separate degradation pathway. Bryant et al. [71] demonstrated that photo-generated electrons can reduce O₂ to superoxide, initiating degradation of MAPbI₃ through reactions involving the organic cation. Alkaloid-mediated suppression of this pathway remains plausible where defect passivation or antioxidant activity is demonstrated, but cannot be inferred from metal coordination alone.

4.4. Structure–Interaction Relationships for Alkaloid-Type Scaffolds

Alkaloids constitute a heterogeneous class of nitrogen-containing secondary metabolites. For interaction with halide perovskites, the decisive feature is the electronic availability and accessibility of a lone pair rather than nitrogen hybridization alone. Pyridinic nitrogen is sp² hybridized and retains an in-plane lone pair; pyrrolic nitrogen is likewise sp², but its lone pair participates in aromaticity. Protonation, substitution, steric environment, solvent, and competing O-donor groups must also be considered [24,25]. Table 3 groups representative alkaloid-type scaffolds by their available functional groups, supporting techniques, and the level of evidence reported for each interaction; these assignments must remain molecule- and study-specific rather than generalized across a scaffold class. Quinolines and isoquinolines may combine available N donors with O-containing substituents. In xanthines such as caffeine and theophylline, carbonyl oxygen atoms are the best-supported Pb²⁺-binding sites; N-donor coordination has not been demonstrated for either molecule, and neither N-donor coordination nor hydrogen bonding should be generalized across differently substituted derivatives [18,19,24,25,66]. Protonation of the free imidazole nitrogen, as in theophylline–BF₄⁻ salts, removes that lone pair and creates a hydrogen-bond donor instead; the reported Pb 4f shift for this derivative reflects anion adsorption modeled computationally, so the specific contribution of carbonyl coordination remains inferred rather than calculated [19]. Capsaicin likewise exhibits route-dependent behavior at the absorber and SnO₂/perovskite interfaces [20,28]. For quinoline, included as a non-metabolite comparator, pyridinic N-donation to Pb²⁺ is structurally plausible; however, the reported gains in efficiency and humidity stability [30] have not been assigned to a spectroscopically resolved binding geometry.

Dispersion, electrostatic, and possible metal–π contributions at perovskite interfaces, and the conditions under which organic π–π stacking may occur, are discussed in Section 4.1.3 and are not established for the scaffolds compared here. Their energetic and electronic contribution has not yet been assessed systematically for indoles, isoquinolines, β-carbolines, and xanthines on relevant perovskite facets, which remains a computational and spectroscopic research need rather than a class-wide demonstrated mechanism [60,62].

4.5. Interaction Mechanisms of Terpenoids and Carotenoids

The available evidence identifies at least four distinct process mechanisms rather than a universal terpenoid effect. In α-terpineol-containing precursor inks, the alcohol functionality and high boiling point were associated with solvent/cation interactions, delayed nucleation, and improved crystallinity [37]. Camphor and camphorquinone act as volatile solid additives; their removal during processing can regulate crystal formation without requiring permanent interfacial coverage [38]. Lycopene instead provides a conjugated hydrophobic carotenoid scaffold associated with oxygen/ultraviolet resistance and interfacial protection [39,40]. A fourth route is direct Lewis-base coordination by oxygenated terpenoids: artemisinin interacts with uncoordinated Pb²⁺ and suppresses the formation of metallic Pb clusters during light soaking [41]. Carotenoids combine hydrophobicity and radical scavenging with, in the case of astaxanthin, carbonyl coordination to Pb²⁺ and Sn⁴⁺ at the SnO₂/perovskite interface [43,44,45], and first-principles work attributes lycopene passivation to a cover-type interaction of its conjugated C=C backbone with the surface [72].

For long-chain alcohols such as phytol or geranylgeraniol, simultaneous surface coordination and hydrophobic grain-boundary anchoring is chemically plausible but has not been demonstrated systematically in perovskite films. Adsorption energies from unrelated O-donor molecules cannot be assigned to these compounds without explicit surface models and comparable computational settings.

These cases demonstrate why incorporation route must be explicit: α-terpineol modifies the wet precursor and annealing history, camphor-derived solids exploit controlled volatilization, and lycopene functions as an additive or protective interfacial layer [37,38,39,40]. Contact angle, surface tension, volatilization profile, in situ crystallization, residual-additive mapping, and electronic characterization are needed to separate solvent, morphological, passivation, and barrier effects.

Polycyclic triterpenoids such as betulin, ursolic acid, and oleanolic acid exhibit self-assembly in several non-perovskite solvent systems [36]. This behavior motivates investigation as an interfacial design concept, and betulin has also been used as a dilute precursor passivator in MAPbI₃ [46]; however, neither observation demonstrates quasi-two-dimensional passivation layers on a perovskite. Terms such as ‘coordinatively adsorbed’ should be reserved for systems with direct structural or computational evidence.

Scale-up assessment requires solubility limits, ink aging, filtration behavior, rheology, surface tension, coating uniformity, volatilization or residual-solvent analysis, and thermal stability. The volatile-additive concept is especially relevant to scalable drying, but evaporation rate, exhaust handling, process window, and residual composition must be established under blade or slot-die coating rather than inferred from spin coating [37,38].

The principal research opportunity is therefore a controlled validation workflow that compares demonstrated α-terpineol, camphor/camphorquinone, artemisinin, betulin, and carotenoid systems with underexplored structures using molecular screening, precursor-compatibility testing, in situ crystallization analysis, film-distribution mapping, optoelectronic characterization, and standardized device aging. Evidence should be assigned to individual compounds, not transferred to the terpenoid class as a whole.

4.6. Interaction Mechanisms of Polyphenols and Phenolic Acids

Polyphenols and phenolic acids offer combinations of phenolic OH, carbonyl, and carboxyl groups. Depending on molecular geometry and protonation, these groups may coordinate undercoordinated Pb²⁺ or Sn²⁺, form hydrogen bonds with halides or organic cations, alter precursor complexation, regulate strain, or participate in redox reactions [21,47,48,49,55,57].

Catechol or galloyl motifs can form multidentate interactions with metal ions, but the binding mode in a perovskite precursor or surface must be established directly. Recent dispersible gallic-acid chelators provide a perovskite-specific multidentate design example [48]. Reported shifts in XPS, FT-IR, Raman, or NMR can support coordination when interpreted with suitable controls; solution-phase metal chelation alone is insufficient.

Hydrogen-bond networks may influence halide motion, local strain, and interfacial chemistry. The magnitude of this effect depends on site density, molecular orientation, and competition with solvent and A-site cations. Computational claims should report the surface, defect state, functional, dispersion treatment, solvation model, and adsorption geometry [61].

In Sn-containing perovskites, antioxidant additives may suppress Sn²⁺ oxidation or related defect formation. Gallic and caffeic acid and redox-active flavonoid interlayers provide perovskite-specific examples, but oxidation-state or redox effects must be distinguished from morphology and strain regulation using XPS, complementary chemical analysis, structural probes, and appropriate molecular controls [47,48,49,57].

Polyphenols can also influence film formation through precursor coordination, solubility, and drying. Concentration-dependent aggregation is a potential limitation because strongly associating molecules may introduce insulating domains, parasitic absorption, or interfacial recombination. Optimal concentrations and solvent compatibility are therefore formulation-specific.

Radical-scavenging activity is a plausible complementary stabilization mechanism, particularly under illumination and oxygen exposure. Lycopene and redox-active flavonoid studies strengthen this design hypothesis, but antioxidant capacity measured by solution assays does not quantify protection inside a perovskite film [39,40,57]. Direct validation requires controlled-atmosphere aging, reactive-oxygen-species probes, oxidation-state measurements, and comparison with non-antioxidant structural analogues [71,73].

4.6.1. Evidence Requirements for Interfacial Electronic Effects

Conjugated polyphenol scaffolds may modify interfacial dipoles, strain, or packing when located at a perovskite/transport-layer interface. Curcumin at the buried SnO₂/perovskite interface and redox-active flavonoid interlayers provide direct device examples [55,57]. Claims of improved charge transfer or π–π stacking still require energy-level measurements, chemical-location evidence, and an identifiable aromatic organic partner; electronic coupling with the inorganic framework should not be described as π–π stacking.

4.6.2. Process Engineering Considerations

For translation beyond spin coating, polyphenol studies should report solubility, viscosity, surface tension, contact angle, drying window, additive retention, and batch purity. Purified standards, concentration series, and scalable coating trials are needed to distinguish genuine molecular effects from uncontrolled changes in ink formulation.

5. Process–Structure–Optoelectronic Property Relationships

5.1. Incorporation Route and Precursor Chemistry

The incorporation route determines which process stage an additive can influence. Precursor additives can change metal–solvent coordination, complex lifetime, supersaturation, and intermediate phases; antisolvent additives act during rapid solvent extraction and nucleation; post-treatments primarily modify the exposed surface and near-surface interface [9,18,20,21,30]. These routes should not be compared as mechanistically equivalent. Figure 4 maps the additives reviewed here onto the process stage at which they are introduced and the variables they can influence: most demonstrations enter through the precursor ink, whereas the coating, annealing, and encapsulation stages remain comparatively unexplored.

Concentration is equally important. Insufficient loading may not measurably change precursor or surface chemistry, whereas excessive loading can retard conversion, induce insulating residues, alter stoichiometry, or create unfavorable energy-level offsets. Solubility limits and chemical identity should be reported in molar units where possible, together with solvent composition, solution aging, filtration, and deposition conditions.

5.2. Crystallization and Film Structure

Caffeine, capsaicin, quinoline, caffeic acid, α-terpineol, and camphor-derived additives illustrate distinct pathways for crystallization control [18,20,30,37,38,49]. Caffeine and α-terpineol act from the precursor solution and retard aspects of nucleation or crystal growth; capsaicin is introduced through the antisolvent and becomes enriched near grain boundaries and the top interface; quinoline alters precursor-mediated growth and orientation; caffeic acid modulates nucleation in mixed Pb–Sn absorbers while limiting Sn²⁺ oxidation; and camphor/camphorquinone use controlled volatilization to modify the crystallization trajectory. Colchicine is particularly relevant to scalable deposition because it steers direct α-FAPbI₃ formation without an antisolvent quench, with devices retaining 83.6% of their initial efficiency after more than 1500 h at 85 °C [31].

Grain size, roughness, and diffraction intensity are useful descriptors but cannot independently establish electronic quality. A defensible process assignment requires time-resolved or in situ measurements of wet-film evolution, statistically defined morphology, crystallographic orientation, chemical mapping, and matched control films processed under identical conditions.

5.3. Defects, Recombination, and Interfacial Energetics

Evidence for passivation should converge across chemical and optoelectronic measurements. XPS or vibrational shifts can indicate a changed local environment but do not uniquely establish coordination geometry. Increased photoluminescence may reflect reduced non-radiative recombination, altered extraction, morphology, or dielectric screening. Strong assignments should combine spectroscopy with time-resolved photoluminescence, quasi-Fermi-level splitting or electroluminescence, trap-sensitive measurements, and device-level voltage losses [11,12,61,62].

Tannic acid provides a surface-treatment example in which multiple hydroxyl and carbonyl functions, interfacial-energy changes, and a cross-linked network were associated with reduced recombination and ion diffusion [21]. Curcumin-modified SnO₂/perovskite interfaces, capsaicin-modified SnO₂, and flavonoid interlayers broaden the evidence for interface-selective defect, charge-transfer, strain, and redox regulation [28,55,57]. Such multifunctionality is advantageous, but individual contributions should be separated through molecular controls, concentration series, and location-sensitive characterization.

5.4. Charge Transport, Ion Migration, and Impedance

Electrical interpretation requires restraint. EIS-derived resistances and capacitances depend on bias, illumination, frequency window, perturbation amplitude, temperature, and equivalent-circuit choice. SCLC-derived trap densities depend explicitly on thickness, dielectric constant, contact selectivity, and assumed trap-filled-limit voltage. Reviews should therefore compare control and treated samples only when measurement conditions and analysis assumptions are matched [62,65,68,69].

Ion-migration claims should be supported by complementary evidence such as bias-dependent transients, temperature-dependent measurements, operational hysteresis, compositional mapping, or validated modeling. A longer photoluminescence lifetime alone is not proof of suppressed ion migration.

5.5. Device Performance and Stability

Table 4 compares representative primary studies using control-to-treated performance and explicit aging conditions wherever they were available. The values are not a meta-analysis because device architecture, active area, perovskite composition, encapsulation, scan direction, and aging protocol differ. Stabilized output and champion scan efficiency are distinguished when the source reports both.

The expanded evidence matrix distinguishes precursor or bulk modifiers [37,38,39,48], interface-selective treatments [28,40,55,57], and external encapsulation [56]. These routes address different rate-limiting processes and should not be pooled into a single additive-performance distribution. Among the newer studies, camphorquinone provides operational MPPT evidence under ISOS-L-1 [38], whereas catechin-Q is evaluated under ISOS-D-1I dark storage [57]; the protocols answer different stability questions.

5.6. Process Compatibility and Scale-Up

Translation beyond spin coating requires formulation and coating data: additive purity, solubility, viscosity, surface tension, contact angle, coating speed, substrate temperature, drying window, residual composition, and batch-to-batch reproducibility [6,14]. Standardized stability assessment should follow the ISOS consensus framework and report encapsulation, atmosphere, humidity, temperature, spectrum, electrical condition, active area, sample count, and retention criterion [75].

The most credible scale-up workflow is iterative: molecular screening; precursor-compatibility testing; in situ film-formation analysis; film and interface characterization; statistically powered device validation; standardized aging; and life-cycle or techno-economic assessment. This sequence links molecular design to manufacturable process windows rather than extrapolating directly from adsorption energy to PCE.

6. Theoretical Modeling and Rational Design of Secondary Metabolite Additives

Computational methods can connect molecular structure with precursor association, surface adsorption, defect states, charge redistribution, ion migration, and environmental interactions. Their principal value is comparative candidate ranking and mechanistic testing, not direct prediction of device efficiency. The integrated computational–experimental workflow is summarized in Figure 5.

For additive engineering, useful molecular descriptors include donor-atom accessibility, adsorption energy, charge transfer, dipole moment, hydrogen-bond capacity, polarizability, logP, and solubility. Descriptor values are meaningful only when calculated with consistent models and validated experimentally.

6.1. Electronic-Level Modeling of Additive–Perovskite Interactions

6.1.1. Defect Physics as the Basis for Additive Design

First-principles studies explain why many intrinsic bulk defects in lead-halide perovskites are relatively shallow, whereas surfaces, interfaces, and grain boundaries can introduce chemically reactive or electronically active sites. Additive design should therefore specify the modeled termination, defect, charge state, and chemical environment rather than refer generically to ‘the perovskite surface’ [11,59,65].

Undercoordinated metal sites and halide-related defects provide possible targets for coordination or hydrogen bonding. The relevant state depends on composition, surface reconstruction, and operating environment; simplified neutral-slab models may not represent charged defects or realistic precursor-covered surfaces.

6.1.2. Adsorption Energies and Donor–Metal Coordination

Adsorption energy is a useful comparative descriptor when the same surface, coverage, functional, dispersion correction, basis, and solvation treatment are used. It should not be compared quantitatively across incompatible computational protocols.

For xanthine-derived additives, modeling should test all plausible donor sites and conformations. Experimental and computational work on caffeine and theophylline derivatives supports carbonyl-dominated interaction with Pb²⁺ and illustrates how functionalization, surface termination, and defect identity can change adsorption and interfacial electronic structure (Figure 6) [18,19,26,27,64,67].

Theophylline–BF₄⁻ derivatives provide a relevant theoretical–experimental example: calculated interaction and charge redistribution were evaluated alongside device response. Such correlations are informative but do not prove that adsorption energy alone controls PCE [19].

Candidate derivatives containing sulfur or other polarizable donors may strengthen adsorption [64], but stronger binding is not automatically beneficial. Screening must also exclude deep gap states, unfavorable band offsets, poor solubility, and irreversible precursor complexation.

The preferred candidate is therefore the molecule that balances binding, electronic benignity, process compatibility, and stability—not necessarily the molecule with the most negative adsorption energy.

Charge-density difference, projected density of states, electrostatic potential, and Bader or related charge analyses can clarify whether adsorption creates an interfacial dipole, redistributes charge, or introduces shallow or deep states. These descriptors require consistent sign conventions and clearly defined reference densities. Figure 7 illustrates the interpretation of charge-density redistribution within a defined additive–perovskite model.

From a materials engineering perspective, such modeling can help prioritize functional groups before synthesis, although reductions in development time or cost have not yet been quantified.

6.1.3. Charge Transfer and Interfacial Energy Alignment

Strong adsorption is useful only when the resulting surface electronic structure remains compatible with charge transport. Computational predictions should therefore be validated against spectroscopy, photoluminescence, mobility, and device measurements rather than interpreted as direct performance forecasts [11,18,19].

At the process level, molecular coordination can alter precursor speciation and crystallization as well as the final surface. Explicit-solvent or cluster models are needed when the proposed mechanism occurs in the wet film; a vacuum surface calculation alone cannot establish solution-phase behavior.

Within the selected model, projected density of states (PDOS) and Bader charge analyses can indicate interfacial dipole formation, band-edge shifts, and the emergence of shallow versus deep states.

A robust workflow combines adsorption geometry, energetic comparison, electronic-state analysis, and experimental validation. Uncertainty should include alternative adsorption sites, surface terminations, molecular conformers, and computational method. Representative adsorption configurations and their corresponding comparative adsorption energies are shown in Figure 6.

6.2. Modeling Environmental Stability and Degradation Mechanisms

6.2.1. Water Diffusion and Grain Boundary Effects

Water, oxygen, heat, light, and ionic redistribution interact during perovskite degradation. Atomistic modeling can compare adsorption, reaction pathways, and diffusion barriers, but isolated-water calculations cannot reproduce a humid, illuminated, electrically biased device [4,71].

Modeling enables estimation of water adsorption energies, diffusion barriers at grain boundaries, and ionic-migration activation energies.

Such parameters can inform, but not by themselves predict, long-term device stability.

Environmental simulations should distinguish bulk diffusion, surface diffusion, grain-boundary transport, adsorption, and chemical reaction. Calculated barriers are model-specific and should not be presented as universal material constants. First-principles calculations on a Σ5-(210) grain boundary in MAPbI₃ illustrate this distinction: water migrates along the boundary with a barrier of only about 0.07 eV and from the boundary into the bulk with about 0.24 eV, whereas diffusion through bulk MAPbI₃ requires about 0.70 eV [76] (Figure 8). These values support grain-boundary passivation and improved crystallinity as moisture-mitigation strategies, but they refer to a single idealized boundary and should not be generalized to all microstructures.

6.2.2. Linking Modeling to Stability Reporting

Theophylline–BF₄⁻ studies illustrate how molecular modeling and device experiments can be combined, but stability duration must be reported with humidity, illumination, encapsulation, temperature, and electrical condition. A duration alone is not a transferable stability metric [19].

Modeling can guide the selection of functional groups and possible degradation barriers, but it cannot replace formulation screening or standardized aging. Predictions should be tested using controlled compositions and matched reference devices.

6.3. Multiscale Modeling and Device-Level Correlations

Device behavior emerges from coupled electronic, ionic, chemical, and morphological processes. Multiscale modeling can connect atomistic descriptors with drift–diffusion or kinetic models, but quantitative device prediction requires experimentally constrained parameters and uncertainty analysis.

A practical hierarchy is: molecular or precursor interaction; surface and defect electronic structure; ion or molecular transport; film morphology; and device-scale recombination and transport. Each scale should pass validated information to the next rather than infer PCE directly from adsorption energy.

A multiscale modeling framework integrates electronic-structure calculations for bonding and defect states; atomistic simulations for solvent, ion, and water transport; mesoscale descriptions of nucleation and morphology; and device models constrained by experimentally measured transport and recombination parameters.

The main output of a multiscale workflow should be a ranked, testable set of candidates and process conditions, together with predicted failure modes and discriminating experiments.

6.4. Computational Libraries of Natural Metabolites

Computational libraries of secondary metabolites can support high-throughput screening when structures, protonation states, tautomers, conformers, and calculation settings are standardized. Candidate ranking should combine interaction, electronic, environmental, and process-compatibility descriptors.

Screening parameters may include adsorption energy, donor accessibility, dipole moment, frontier-orbital alignment, charge transfer, protonation state, hydrogen-bond capacity, polarizability, hydrophobicity, solubility, and predicted chemical stability. Rankings should retain the model definition and uncertainty for every descriptor.

Machine-learning models may accelerate screening only when trained on curated, comparable data. Small heterogeneous datasets of champion devices are unsuitable targets without architecture, process, and stability metadata.

6.5. Limitations of Current Modeling Approaches

The major limitations are idealized surfaces, uncertain defect charge states, incomplete solvent representation, restricted time and length scales, and weak connection to manufacturing variables. Iterative theory–experiment validation is therefore essential.

Future modeling efforts should incorporate explicit-solvent simulations, ab initio molecular dynamics at elevated temperatures, coupling between electronic-structure and transport models, and integration with high-throughput experimental screening.

6.6. Implications for Scalable and Sustainable Additive Engineering

For scalable additive engineering, computation should prioritize experimentally measurable decisions: solubility window, precursor compatibility, likely adsorption motif, risk of deep states, hydrophobicity, and degradation susceptibility.

The most effective workflow integrates molecular screening, formulation testing, in situ crystallization analysis, film and interface characterization, device validation, standardized stability testing, and life-cycle assessment [6,14,19].

7. Conclusions and Perspectives

This review has examined alkaloids, terpenoids, and polyphenols as multifunctional, process-active additives for halide perovskite solar cells. The central framework connects molecular descriptors and precursor or surface interactions with nucleation, crystallization, film morphology, defect and interface chemistry, charge transport, device performance, and stability. Bio-inspired synthetic derivatives and non-metabolite molecules were included only as explicitly identified mechanistic comparators.

The evidence supports a molecule-specific rather than class-universal interpretation. Xanthines provide the clearest alkaloid-derived examples, with carbonyl coordination playing a central role. Selected polyphenols and phenolic acids have demonstrated coordination, crystallization, strain, interfacial, or oxidation-control effects. Direct photovoltaic evidence for α-terpineol, camphor/camphorquinone, artemisinin, betulin, and the carotenoids lycopene, β-carotene, and astaxanthin establishes terpenoids and carotenoids as more than a purely prospective design space [37,38,39,40,41,43,45,46]; however, the distinct mechanisms and incorporation routes still preclude class-wide generalization.

Reported device improvements cannot yet be combined into universal performance or lifetime ranges because architectures, controls, concentrations, aging conditions, and reporting practices differ. Future quantitative synthesis requires a predefined extraction protocol and study-level table containing composition, architecture, route, control and treated metrics, uncertainty, stability protocol, and reference.

Most demonstrations remain at an early laboratory proof-of-concept stage, broadly consistent with TRL 3–4 rather than a formal technology-wide assignment. The most advanced case, camphorquinone, combines an independently certified efficiency, a 1.0 cm² device, and 1000 h of MPPT operation under ISOS-L-1 [38]. Progress toward TRL 5–6 requires compatibility with blade, slot-die, or inkjet coating; control of ink rheology and drying kinetics; purified or specification-controlled additives; and reproducible processing windows. Long-term stability should be tested under clearly reported ISOS-D, ISOS-L, and ISOS-T conditions, including combined stress where possible.

Priority research should focus on structure-controlled bio-inspired derivatives, chemically defined additive standards, perovskite-specific screening of underexplored terpenoids, explicit-solvent and finite-temperature modeling, in situ crystallization measurements, and validation under scalable coating and ISOS-aligned aging. Machine-learning models should be trained only on curated study-level data that preserve composition, architecture, concentration, processing, uncertainty, and stability metadata.

A life-cycle perspective is essential for establishing whether a bio-derived additive provides a genuine sustainability benefit. Extraction yield, solvent consumption, purification energy, land or biomass demand, formulation concentration, reproducibility, and end-of-life fate must be evaluated alongside photovoltaic performance. Secondary metabolites should therefore be regarded as chemically diverse design scaffolds—not intrinsically green solutions—and advanced through the same evidence standards applied to efficiency, reliability, and manufacturing.