Repositioning Proton and Electron Donors in a Family of Proton-Coupled Electron Transfer Mediators Authors: Enric H. Adillon, Alexander Alabugin, Sungchan Kim, Jonas C. Peters Published: 2026 ABSTRACT Reagents for reductive PCET often tether an acid and a reductant with a bridging moiety, lowering the effective molecularity of PCET to substrates. The effect of this connecting group on the thermodynamics and kinetics of PCET, however, is not systematically understood. To interrogate the role of the bridge in a well-defined system, we measured PCET rates from a set of anilinium-appended cobaltocene complexes to a series of aryl ketones. Synthesis of four cobaltocene PCET mediators, including one via late-stage Pd-catalyzed cross-coupling of a cobaltocenium salt, enabled systematic variation of the covalent bridge between the cobaltocene electron donor and anilinium proton donor. Measurements of electron and proton transfer equilibria show that the overall PCET driving force is minimally perturbed. Kinetic analysis establishes that aryl ketone reduction by our PCET mediators proceeds by a concerted mechanism in all cases, despite substantial variations in the thermodynamic coupling. Notably, a complex containing an o-phenylene bridge exhibits the fastest PCET rate despite having the least driving force, a result attributed to stabilizing orbital interactions between the ketone substrate and the cyclopentadienyl ligand of the mediator. Introduction Proton-coupled electron transfer (PCET) is ubiquitous in biological redox processes such as nitrogen fixation, cellular respiration, and photosynthesis, and is increasingly exploited in catalytic methods for chemical synthesis. − In recent years, molecular catalysts for PCET (herein referred to as PCET mediators) have emerged as effective tools for the electrolytic reduction of N2, CO2, and unsaturated organic small molecules. Such mediators, upon protonation and reduction, act as well-defined H atom equivalents which can be more reactive and selective than direct electrolysis in acidic media, as shown in Scheme a. In this way, mediated electrolysis can lower the overpotential and improve the Faradaic efficiency of reductive protonation reactions. (a) Reduction of an Unsaturated Substrate by Use of a PCET Mediator, (b) A Sample of Reported PCET Catalysts and Reagents Containing Sites for Proton and Electron Transfer, (c) Exploration of Bridging Group as a Design Parameter for Cobaltocene- and Anilinium-Derived PCET Electrocatalysts One such PCET mediator is an aniline-appended cobaltocene (Scheme a) devised by our group, which effects substrate reduction via concerted proton–electron transfer (CPET) to generate O–H, N–H, C–H, and Ni–H radical intermediates under mild conditions. − In this system, a reduced and protonated Co­(II) state is the key reactive intermediate, with a weak BDFEeff of 37 kcal/mol. Following PCET, protonation and electrochemical reduction of the resulting Co­(III) species turns over the catalytic cycle. While interest in electrocatalytic PCET mediators is growing, reductive PCET mediators remain particularly few in number. Representative examples of PCET mediators are depicted in Scheme b. Mayer and co-workers have established ruthenium-imidazole complexes as tunable PCET platforms, which have been employed as electrocatalysts by Waymouth and co-workers for the PCET-driven oxidation of propanol to acetone. , Nishibayashi and co-workers are exploring iron and cobalt sandwich compounds as stoichiometric PCET reagents as well as a samarium system equipped with a protic arm that may act as a PCET mediator. , A reduced and protonated Cp*Fe-phenol system shows potency for reducing acetophenone, anthracene, and a molybdenum nitride. Other examples include phenol-appended cobalt and nickel pincer complexes from the Siewert group, and a titanium semiquinone complex from the Baek group; both have been used as stoichiometric PCET donors. − These are just a few relevant instances. − With respect to PCET mediator design, a reasonable first choice of proton and electron donors for a target PCET reaction can be made on thermodynamic grounds using known redox potentials, acidities, and bond energies. These thermodynamic terms set the driving force for PCET (either H atom addition or abstraction), which in general influences PCET rate via linear free energy relationships. Yet it is not obvious how the connection of the electron and proton transfer subunits will affect the PCET mechanism or its rate. Treating bridge position as a PCET design variable, and obtaining benchmark kinetic data on its effect, serves the growing interest in PCET mediatorsespecially if kinetic activity can be improved at decreased thermodynamic overpotential. To address this gap, we have synthesized and examined a series of cobaltocene-derived PCET mediators based on our original design (Scheme a), herein differing solely in the covalent bridge (Scheme c). Importantly, this series maintains the same electron and proton-bearing fragments and thereby controls for the anticipated influence of driving force. This allows the intrinsic effect of the bridging group to be isolated. Through determination of the crystal structures, measurements of ET and PT equilibria, PCET kinetics, isotope effects, Hammett parameters, and computational analysis, we quantify the bridging group’s influence on the mechanism and rate of PCET. We find that the PCET rate is relatively insensitive to changes in the bridge but also identify a surprising electronic rationale for how a specific bridge can augment metallocene PCET reactivity: the geometry of the acid can direct donor–acceptor orbital interactions reminiscent of a pericyclic reaction with the substrate, promoting electron transfer. As a result, an ortho-substituted analog, which features the least driving force for the model PCET reaction, reacts most rapidly in contrast to the expected rate-driving force relationship. This finding points to secondary-sphere design (inclusive of the linker) as a means to control activity and selectivity in mediated PCET. Results and Discussion Synthesis Our series of phenylene-bridged mediators (discussed herein as [Co(II/III)bridge, N/NH]0/+/2+) was prepared in two steps by a variation of our previously reported route to the p-phenylene isomer (Scheme a). Aryllithium addition to [Cp2Co]­[PF6] followed by abstraction of the ring endohydride from Cp­(η4-CpH-PhNMe2)­Co­(I) with the tritylium reagent [Ph3C]­[BArF 4] (ArF = 3,5-bis­(trifluoromethyl)­phenyl) afforded [Co(III) o,m,p‑Ph, N]­BArF 4 salts. − As necessary, Co­(II) forms of the mediators were prepared by thermolysis of the corresponding Co­(I) diene complexes or by reduction of Co­(III) with sodium amalgam (detailed in the Supporting Information). Synthetic Routes for the Preparation of (a) [Co o–Ph, N], [Co m–Ph, N], and [Co p–Ph, N] and (b) [Co(III) p−Ph2, N]­BArF 4, (c) X-ray Crystal Structures Listed from Top to Bottom: [Co(II) o−Ph, N], [Co(II) m−Ph, N], [Co(II) p−Ph, N], and [Co(II) p−Ph2, N] A parallel mediator synthesis was devised which places the harshest and least chemoselective stepsCp-aryl bond formation and cobalt oxidationahead of coupling to the Brønsted acid. As shown in the route in Scheme b, replacement of the lithioaniline synthons with p-lithiobromobenzene or the complex magnesiate Mg­(PhBr)3LiBu yielded Cp­(η4-CpH-PhBr)­Co­(I), an air-stable Co­(I) diene with an exo aryl bromide. Hydride abstraction afforded the cobaltocenium salt [CpCo­(III)­CpBr]­BArF 4, a highly activated electrophilic substrate for Pd cross-coupling chemistry under aqueous Suzuki-Miyaura conditions. The use of a BArF 4 counterion makes purification straightforward via silica gel chromatography. A key advantage of this new cross-coupling methodology is that it enables late-stage diversification of cobaltocenes with functional groups intrinsically incompatible with organolithium addition to [Cp2Co]+ and oxidation of Co­(I). In addition to the newly reported o-PhNMe2, m-PhNMe2 and p-PhBr substituted Co­(I) dienes, each metallocene complex has been structurally authenticated by single-crystal X-ray diffraction in the neutral (CoII, shown in Scheme c) and cationic (CoIII, see Supporting Information S4) states. The aryl substituents do not significantly perturb the bond metrics of the cobaltocene fragment. Significant C­(Cp)–C­(Ph) torsional strain is evident in [Co o–Ph, N], while the biphenylene moiety in [Co(II) p−Ph2, N] exhibits slight torsion about the C­(Ph)–C­(Ph) bond. Thermochemistry Relevant thermochemistry of the cobaltocene series was obtained by measurements of the proton and electron transfer equilibria, as illustrated in Figure a. Accordingly, the pK a of each [Co(III), NH]2+ was measured by titration with anilinium acids and quantification of the relevant species’ concentrations by 1H NMR spectroscopy (see Supporting Figure S6). The half-wave potentials were measured by cyclic voltammetry in both the aniline (E 1/2(N)) and anilinium (E 1/2(NH)) protonation states, providing the necessary data to determine the BDFEN–H and the thermodynamic coupling values (ΔE 1/2). (a) Square scheme of [Co(II), NH]+ with coupling highlighted around the outside of the square. (b) pK a (CoIII)-E 1/2(NH) plot of the set of mediators bracketed by diagonal lines of constant BDFE. (c) Electrochemical (blue, left) and spectroscopic (red, right) measurements plotted as a function of BDFEN–H, demonstrating that coupling and driving force in this series are essentially independent. (d) Cyclic voltammograms of [Co(III), N]+ (blue trace) and [Co(III), NH]2+ with 10 mM [4‑CNPhNH3]­[OTf] (black trace) which exemplify how thermochemical coupling is determined electrochemically. (e) UV–vis spectra of [Co(II) o‑Ph, N] and [Co(II) o‑Ph, NH]­BArF 4, exemplifying how spectroscopic coupling is determined. The thermochemical profiles are summarized in Table and depicted graphically in Figure b,c. There are small variations in pK a and E 1/2 across the series, but the spread in the BDFEs is very small (36.5 to 37.2 kcal/mol with an estimated error of ±0.4 kcal/mol). Variation in pK a is compensated for by a proportional change in E 1/2 (and vice versa). The exception is [Co(II) o‑Ph, NH]+, with a BDFEN–H (38.8 kcal/mol) that is modestly higher, likely due to electrostatic and steric interactions unique to ortho substitution. In the absence of a covalent linker, the formal BDFEN–H of a combination of cobaltocene (E 1/2 = −1.33 V) and N,N-dimethylanilinium (pK a = 11.47) is estimated as 38 kcal/mol. , 1Thermochemical and Spectroscopic Parameters of Aniline-appended Cobaltocene Complexeslinker E 1/2 (N) (V vs Fc0/+) E 1/2 (NH) (V vs Fc0/+)pK a (CoIII)pK a (CoII)BDFE (kcal/mol)ΔE 1/2 (N/NH) (mV)Δλmax (N/NH) (meV) o-Ph–1.32–1.118.612.138.821087 m-Ph–1.31–1.229.310.837.290261 p-Ph–1.35–1.218.611.036.5140302 p-Ph2 –1.30–1.2810.210.537.120153aThe relative error in the bond dissociation free energy measurement is estimated to be ± 0.4 kcal/mol. Thermochemical and Spectroscopic Parameters of Aniline-appended Cobaltocene Complexes The relative error in the bond dissociation free energy measurement is estimated to be ± 0.4 kcal/mol. While the substitution pattern of the linker has little influence on BDFEN–H, it has a significant effect on the thermodynamic (acid–reductant) coupling, which has been inferred to bias systems toward concerted PCET. , We measure thermodynamic coupling electrochemically as the shift in E 1/2 upon changing protonation state. For example, the thermodynamic coupling for [Co(III) o−Ph, N]+ is shown in Figure d as the shift in E 1/2 from [Co(III) o−Ph, N]+ (blue trace) to [Co(III) o−Ph, NH]2+ (black trace) upon addition of acid, approximately +210 mV. An alternative reporter of acid–reductant coupling can be gleaned spectroscopically as a shift in the visible metal-to-ligand charge transfer (MLCT) band upon protonation (Δλmax). Herein we refer to Δλmax, a proxy for orbital coupling between the site of the proton and electron, as the spectroscopic coupling. We illustrate spectroscopic coupling using [Co(II) o−Ph, N], whose UV–vis spectrum (blue trace) is depicted in Figure e. Upon addition of acid to form [Co(II) o−Ph,NH]+ (black trace), the feature at 526 nm shifts to 546 nm. For comparison to the thermodynamic coupling, the spectroscopic coupling is reported in Table . It reflects the sensitivity of the (CoII/Cp) → (Ar π*) MLCT to N-protonation, which deactivates the resonance of the aniline N lone pair into the Ar π system. For three of the complexes in the series, the thermodynamic and spectroscopic coupling are linearly correlated, reflected by an overlay of the red and blue data points in Figure c. However, [Co o–Ph, N] is a clear outlier, with the strongest thermodynamic coupling and weakest spectroscopic coupling. The divergent coupling behavior in [Co o–Ph, N] arises from its combination of internal electrostatic contacts and steric strain. Ortho substitution leads to particularly close proximity between the two formal charge centers (Co and N) and thus a large observed thermodynamic coupling. The ortho-disposed groups also induce torsion along the C­(Cp)–C­(Ph) bond axis, which disrupts conjugation between the two π systems, rotates the N lone pair out of alignment with the Ar π system, and destabilizes the phenyl-centered acceptor orbital in the MLCT transition (Figure S103). Kinetics To compare PCET reactivity across this cobaltocene series, we chose the reduction of acetophenone as a convenient model reaction that proceeds via a ketyl radical intermediate to a pinacol coupling product (Figure a). We electrolyzed the cobaltocenium complexes at −1.30 V vs Fc0/+ in the presence of excess acetophenone and tosic acid, confirming that each yield electrocatalytic quantities of pinacol (Table S8). For kinetic analysis using UV–vis spectroscopy, the reactive form of the cobaltocene complexes, [Co(II), NH]­BArF 4, was generated in situ by protonation of [Co(II), N] with one equivalent of [H(Et2O)2]­BArF 4 (Figure b). Acetophenone addition (Figure b, black trace) led to formation of [Co(III), N]+, and the resulting kinetic trace was fit to an exponential decay to obtain a first-order rate constant (k obs, Figure c). Plotting k obs as a function of substrate concentration (Figure d) then allowed us to extract k PCET, the second-order rate constant (Figure e). For each complex we measured k PCET using [D(Et2O)2]­BArF 4 to determine the H/D kinetic isotope effect (KIE). We also measured k PCET for a series of p-substituted acetophenones to determine the Hammett slope (ρ), a measure of charge-transfer character in the PCET transition state (Figure e). (a) Strategy for determination of the PCET rate constant. (b) UV–vis spectra of [Co(II) m‑Ph, N] (red trace) and [Co(II) m‑Ph, NH]­BArF 4 (blue trace) which decays upon addition of acetophenone (black trace) to yield [Co(III), N]­BArF 4 (green trace) after 30 min. (c) Kinetic trace (absorbance vs time) fit to an exponential decay, characteristic of first-order kinetics in which the slope of the fit line yields the observed first-order rate constant. (d) Plot of the first-order rate constants (k obs) against substrate concentration, yielding the second-order rate constant (k PCET) as the slope of the fit line. (e) Measured kinetic parameters grouped by mediator. With this thermodynamic and kinetic data set, we can probe the mechanism for acetophenone reduction by each complex. First, we consider the ET-PT pathway, but we find it is inconsistent with the observed primary KIEs (1.5–2.1) and can be further ruled out by a thermodynamic exclusionary argument, as the free energy for ET from cobaltocene to acetophenone (ΔG ET ° = 19–23 kcal/mol) is greater than the kinetically inferred barrier heights (ΔG ‡ = 16–17 kcal/mol). Next, we consider the PT-ET pathway, which cannot be ruled out on thermodynamic grounds (e.g., ΔG PT,calc = 14–16 kcal/mol). However, we observe negative slopes for rate–driving force relationships based on PT to the p-substituted acetophenones (i.e., ln­(k PCET) vs ΔG PT,calc), which is inconsistent with a rate-limiting PT step. We conclude CPET is the mechanism consistent with the observed data, as was previously determined for the [Co(II) p‑Ph, NH]+ mediator. These data demonstrate that the concerted mechanism is conserved across all the bridges. This implies that despite substantial variations in the thermodynamic coupling and the individual PT and ET driving forces available at [Co(II), NH]+, the mode of PCET reactivity is here invariant to the position of the electron and proton donors. In contrast, we have previously shown that modifying the redox subunit from cobaltocene to o-carborane switches the mechanism of the PCET reaction from CPET to ET-PT, pre-empting the carborane’s viability as an electrochemical PCET mediator. Interestingly, Figure e shows a positive and moderately steep Hammett slope (ρ) for [Co(II) o‑Ph, NH]+, [Co(II) m‑Ph, NH]+, and [Co(II) p‑Ph2, NH]+ (ρ = 1.4, 1.3, 1.1), whereas the slope is negative and shallow for [Co(II) p‑Ph, NH]+ (ρ = −0.3). The variation in ρ shows that the bridge modulates the charge-transfer nature of the PCET transition state. ρ does not correlate with PCET rate in this system, suggesting that ET–PT asynchronicity here is not a strong predictor of reactivity. This finding contrasts examples of oxidative PCET where basic asynchronous (PT-led) mechanisms predominate. − The distinctive ρ value for PCET from [Co(II) p‑Ph, NH]+ can be rationalized via Bernasconi’s principle of nonperfect synchronization (PNS). Here, the para-phenylene bridge is particularly suited to stabilize the [Co(III), N]+ product state by delocalizing the N lone pair into the π-accepting CpCo(III)­Cp-Ar moiety. This is evident in the crystal structure of [Co(III) p‑Ph, N]­OTf, which features a trigonal planar N-center (∑ ∠CNC = 360°). By comparison, the N-centers are more pyramidal in the para-biphenylene, meta- and ortho-phenylene bridged [Co(III), N]+ complexes (∑ ∠CNC = 357°, 356°, and 341°, respectively). Maximizing the quinoidal resonance contribution requires N to rehybridize from a pyramidal to a fully planar geometry, which lags the fast proton motion. The result is an imbalanced transition state characterized by advanced proton transfer and negative ρ (positive charge buildup on the ketone). − A quinoidal resonance form is unavailable to the meta bridge, and extended π-orbital overlap in the para-biphenylene is frustrated by the biphenyl torsion. The ortho analog is sterically blocked from aligning the N lone pair with the Ar π orbitals. As it remains substantially pyramidal in the [Co(III), N]+ product, the ortho compound undergoes the least rehybridization of the 4 mediators. We plotted k PCET (to acetophenone) against the available thermodynamic, structural, and kinetic parameters to determine which factors most affect PCET reactivity. Surprisingly, we found poor correlations between rate and component driving forces (pK a and E 1/2), thermodynamic coupling (ΔE 1/2), and Co–N distance (Figures S86–S90). Plotting k PCET against the spectroscopic coupling (Δλmax), we find a modest correlation (R 2 = 0.61) in which stronger coupling tracks with slower rate. A potential explanation for this behavior is that the degree of aniline N­(p)-(Ar π*) orbital overlap encoded in Δλmax correlates with part of the inner-sphere PCET reorganization energy, e.g., the previously discussed rehybridization of N on deprotonation. However, equilibrium exchange measurements of the PCET reorganization energy in such reactive systems are complicated by spontaneous hydrogen evolution, so establishing a direct rather than inferred relationship between reorganization energy and Δλmax is presently challenging. Plotting the transition-state barrier against PCET driving force (the difference in experimental BDFEN–H and calculated ketyl BDFEO–H) does not show the relationship expected of CPET reactions with structurally similar reactants (Figure a). In the low driving force regime, we expect the slope of the fit line (α) from the plot of ΔG PCET ‡ vs ΔG PCET ° to be approximately 0.5. Rather, three complexes with the same BDFEN–H (within error) measurably differ in their PCET rates, while [Co(II) o‑Ph, NH]+ has the least driving force and yet the fastest rate. Consequently, nondriving force parameters such as vibrational overlap, electronic coupling and reorganization energy are likely the dominant rate-controlling factors for PCET in the series, which is uncommon. Examples of structural studies that intentionally isolate these variables are instructive, but very few in number. , (a) Transition-state barrier–driving force relationship for acetophenone reduction by [Co­(II), NH]+. (b) Computed transition state geometries for PCET from [Co­(II) o‑Ph, NH]+ to acetophenone in which the acid directs the substrate toward the Ph group (left) or the Cp group (right). (c) Transition state SOMO, highlighting the orbital interaction of the Cp group with the carbonyl. Evidently, the rate-contributing factors for PCET across this series of bridges are not captured by ground state (thermochemical and structural) descriptors of the mediator in the absence of substrate. We employed theoretical models of the transition state (TS) for PCET to acetophenone, using density functional theory (DFT) calculations in ORCA 6 with a def2-TZVPP basis, the TPSSh hybrid functional, and implicit SMD­(THF) solvation. , The computed barrier heights (ΔG calc ‡ = 13–18 kcal/mol) varied more than the experimental barriers (ΔG ‡ = 16–17 kcal/mol), meaning that even at a reasonable level of theory, the precision of the DFT energies does not quantitatively resolve the fine differences in rates. Nevertheless, the transition state geometries offer qualitative insights. For each mediator, we identified transition states in which hydrogen bonding directs acetophenone over the phenylene linker (Figures S98–S101). With the meta-, para- and biphenylene bridges, this is the lowest TS barrier. For [Co(II) o‑Ph, NH]+, however, we identified a preferred TS geometry where hydrogen bonding directs acetophenone over the Cp-face of cobaltocene (Figure b). The o-phenylene bridge is uniquely able to preorganize the substrate over the Cp ring due to its proximity to Co, while Cp-stacking is geometrically inaccessible in the other complexes. The resulting O–CCp distance in the transition state is 2.8 Å, shorter than the computed O–CCp distance in a tight CoIII-succinyl ion-pair implicated as an intermediate in the two-electron PCET reduction of fumarate by [Co(II) p‑Ph, NH]+. Inspecting the computed transition state SOMO reveals that the carbonyl lies across the face of the Cp ring such that the empty C–O π* orbital overlaps with the half-filled Co–Cp π* orbital. Such an orbital interaction is isolobal with the Diels–Alder transition state (Figure c), contributing to the unexpected transition state stabilization by the o-phenylene bridge. More generally, this could help explain cobaltocene’s apparent ability as an ET mediator, lowering the overpotential for reduction of substrates that are recognizable dienophiles. , Dienophile activation by Brønsted acids is well-known; hydrogen bonding from [Co(II), NH]+ both directs and electronically activates the key orbital interaction. While standard DFT calculations yield an adiabatic description of the ground state surface for PCET, we can anticipate that favorable orbital overlap in the TS maps to a substantially enhanced electronic coupling term (Vel) in a more general nonadiabatic theory of PCET rate. , This is because Vel is proportional to the size of the adiabatic energy gap found at the avoided crossing associated with a PCET reaction. The imbalance between component ET and PT driving forces has been a recent focal point for PCET, as such polar effects can alter rate without changing the overall driving force. Here, we offer an orthogonal consideration to accelerate PCET, independent of intrinsic driving forces and reagent polarity. Since the distance of PT is relatively short, the acid can be considered as a directing group that favorably orients the electron donor and acceptor. We suggest that the geometry enforced by the bridge on this directing group constitutes a design opportunity for PCET mediators. The chemically intuitive picture of a direct π*-π* orbital interaction is reinforced quantitatively when changes in electron density at the ortho TS are dissected with the extended transition state-natural orbitals for chemical valence (ETS-NOCV) method (Figure ). With the reactants acetophenone and [Co(II), NH]+ as the charge and spin reference state, the dominant NOCV contributors to the difference density at the TS are the Co → CO charge transfer interaction and the O–H–N hydrogen bond, i.e., the partial ET and PT component interactions of PCET. Comparing the two ortho TS conformations reveals that the strength of H-bonding is conserved while the Cp-stacked charge transfer NOCV is preferentially stabilized by π*-π* overlap and exhibits a greater fraction of formal ET (v k = 0.71 vs 0.61). Comparison of NOCV acceptor orbitals with corresponding stabilization energies (ΔE k in kcal/mol) and fraction of electron transferred (NOCV pair eigenvalue, ν k). Conclusion To close, we have interrogated how the connection between a Brønsted acid and reductant influences PCET reactivity in a cobaltocene-derived mediator system. We synthesized a series of anilinium-appended cobaltocene complexes that vary only in the identity of the covalent bridging group. Within this set of PCET reductants, thermodynamic measurements establish that the bridge significantly affects acid–reductant coupling while preserving similar H atom transfer driving force. Kinetic measurements establish that the anilinium-appended cobaltocenes all reduce acetophenone via a rate-limiting CPET step, the rate of which is modestly perturbed and contradicts the anticipated rate–driving force relationship. The bridging group exerts its unexpected influence on rate by preorganizing substrates in geometries that differ in their intrinsic barriers, demonstrating the modulation of PCET kinetics at nearly constant driving force. Experimental Section Detailed procedures for the synthesis of all newly reported compounds, measurements to obtain thermodynamic parameters, kinetic measurements, bulk electrolysis data, and data analyses are found in the Supporting Information. Descriptions of quantum chemical calculations and orbital analyses may be found in the SI, and all pertinent ORCA input files and xyz coordinates are provided in a dedicated TXT document. Supplementary Material