Fulgide Derivatives as Photo‐Switchable Coatings for Cathodes of Lithium Ion Batteries – A DFT Study

Abstract

Photo‐switchable coatings for lithium ion batteries (LIB) can offer the possibility to control the diffusion processes from the electrode materials to the electrolyte and thus, for example, reducing the energy loss in the fully charged state. Fulgide derivatives, as known photo‐switches, are investigated concerning their use as coating for vanadium pentoxide, a potential cathode material for LIB. With the help of Density Functional Theory calculations, two fulgide derivatives are characterized with respect to their photophysics, their aggregation behaviour on the cathode material and the ability to form self‐assembled monolayers (SAM). Furthermore, the two states of the photo‐switchable coating are tested with respect to lithium diffusion from the cathode material, passing the SAM and entering the electrolyte. We found a difference for the energy barriers depending on the state of the photo‐switch, preferring its closed form. This behaviour can be used to prevent the loss of charge in batteries of portable devices.

Photo‐switchable coatings for electrodes of lithium‐ion batteries offer a potential method for externally controlling diffusion processes. Fulgide derivatives, in the form of self‐assembled monolayers on vanadium pentoxide, are studied using density functional theory simulations to investigate the influence of the state of the photo‐switch on the diffusion barriers of lithium ions, revealing a significant difference.

1. Introduction

Lithium‐ion batteries are a universally used tool to provide electrical energy in portable devices independent of stationary power sources. ^[1] Their use in vehicles and energy storage facilities can contribute to the structural change towards renewable energies. A lithium‐ion battery (LIB) usually consists of a cathode (e. g. a metal oxide like LiCoO₂ ^[2] ), an anode (e. g. graphite) and the electrolyte (e. g. ethylcarbonate (EC) with lithium salts). Due to electrochemical reactions, the electrolyte can be reduced and forms lithium salts on the surface of the electrodes. This is known as solid electrolyte interphase (SEI), mainly formed on the anode of the LIB. ^[3] In contrast, the cathode electrolyte interphase (CEI) is less pronounced and less investigated, since its abundance depends on the cathode material and the conditions of the electrolyte. ^[4] However, there are possibilities to investigate their properties by induced formation e. g. via brief electric shorting. ^[5] The CEI is reported to consist of the typical SEI components: LiF, Li₂O and Li₂CO₃. ^[6] The formation of the CEI can be controlled or even avoided by coating the surface of the cathode with inert materials to improve the electrochemical performance of LIBs. An elegant way to form a dense and regular coating is the use of a self‐assembled monolayer (SAM) on the surface of the cathode. With that, it is possible to form a homogeneous ultra‐thin yet dense layer on the oxide surface.[ ⁷ , ⁸ ] Some studies show that the SAMs can reduce the activation barrier for interfacial Li ion transfer, ^[8] which can improve the battery performance additionally.

In this work, we take a closer look at the influence of the SAM on the permittivity of Li ions. The main idea is to apply a surface coating, whose properties are controllable by an external trigger. This should be done with a SAM consisting of photo‐switchable molecules. Such smart surfaces are already known in other areas like nematic coatings, ^[9] wettability, ^[10] dynamic bonds, ^[11] protein and cell adhesion. ^[12] As cathode material, we choose vanadium pentoxide (V₂O₅) which shows good properties[ ¹³ , ¹⁴ , ¹⁵ , ¹⁶ ] and has also been investigated in context with surface coatings using SAMs. ^[17] For the application as smart surface coating, the organic molecules have to fulfill the following four criteria: 1) They must have a photo‐switchable moiety, meaning the light induced intramolecular reaction must be reversible. Furthermore, the absorption maxima of the two photo states must be well‐separated. This separation should be more than twice the usual full‐width of half maximum of experimentally found electronic transitions (at least 3000 cm⁻¹). 2) The molecule must contain a polar functional group to be adsorbed on the metal oxide surface. 3) Both photo‐isomers must form a stable SAM. 4) The structural changes upon switching are small enough to conserve the SAM and large enough to influence the permittivity of Li ions.

To investigate whether selected fulgide derivatives fulfill these requirements, we apply density functional theory (DFT) simulations tackling the photophysics of the molecules, the aggregation on surface and the Li ion diffusion through the SAM.

Theoretical Methods

Methods of Calculation

Molecular properties like conformation preference and electronic absorption were calculated using density functional theory (DFT) with ORCA 5.0. ^[18] Herefore, we tested different functionals (PBE,[ ¹⁹ , ²⁰ ] PBE0, B3LYP,[ ²¹ , ²² , ²³ ] M06‐X, ^[24] M06‐2X, ω‐B97X ^[25] and CAM−B3LYP ^[26] ), all with a def2‐TZVPP basis set ^[27] and the Grimme D3 correction with Becke‐Johnson damping. ^[28] The process included the optimization in the electronic ground state, whose minimum behaviours were tested by calculating the second derivative and finding no negative eigenvalues. Subsequently, the optimized structures were used for calculating theoretical absorption spectra using time‐dependent DFT (TD‐DFT) with the same functionals as in ground state, considering the first 50 singlet excitations. Solvation effects were considered using the Conductor‐like Polarizable Continuum Model ^[29] in ORCA.

In order to calculate the properties of the absorbed molecules on the surface, we performed DFT simulation with Quantum Espresso 6.3.0. ^[30] The Kohn‐Sham equations were solved using the projector augmented wave (PAW) approach for describing electronic core states and a plane‐wave basis set setting the energy cutoff for the wavefunctions to 40 Ry and the kinetic energy cutoff for charge density and potential to 320 Ry. Electron exchange and correlation energies were calculated within the generalized gradient approximation (GGA) in the Perdew‐Burke‐Ernzerhof (PBE) form.[ ¹⁹ , ²⁰ ] For vanadium, the Hubbard correction of 3.0 eV was applied. To consider van‐der‐Waals interactions, we used the D3 dispersion corrections. ^[31] This combination is known for a very good reproduction of the properties of V₂O₅ ^[32] and was also used in our former investigations.[ ¹⁶ , ¹⁷ , ³³ ] All simulations were performed at the Γ point.

Visualizations of molecules and aggregates were done with Avogadro 1.2 ^[34] and Vesta 3.5. ^[35]

Computational Details

For the simulation of the absorbed molecules on the surface of V₂O₅, a slab representing the (001) surface was created. The crystal structure of α‐V₂O₅ with Pmmn symmetry from the materials‐project web‐page ^[36] was used for generating super‐cells of $3\times 3\times 1$ and $3\times 2\times 1$ unit cells, respectively. As we have already investigated the conformation preference of para‐aminobenzoic acid on the (001) surface of V₂O₅, ^[17] we adapted the structural preference of up‐standing molecules for the aggregation of the fulgide derivatives on top of the slab. Above the molecules, the cell was extended by 15 Å of vacuum.

Once the preferred configurations of the fulgide molecules were obtained by optimization (force convergence threshold=1$\times {10}^{-4}$ a.u.), the space above them was filled with solvent molecules (ethylcarbonate=EC), representing a thickness of about 8 Å. To obtain a suitable starting point for the DFT simulations, we used the PACKMOL ^[37] programme to fill the space and repeated the optimization. An example for such a system containing the V₂O₅, the fulgide molecules and the EC solvent is shown in Figure 1, also indicating the height of the particular components.

Figure 1.

Example for the system containing slab, SAM and electrolyte.

For the calculation of the Li pathways, we decided to perform a step‐wise mechanism instead of commonly used packages like NEB, since we assumed a quite complex diffusion mechanism. The Li was placed inside the V₂O₅ slab, between the two upper layers. After optimization of the position, the Li was moved manually 0.5 Å upwards and the respective height coordinate fixed, giving only the opportunity to move freely in the plane normal to the movement. This procedure was repeated 35 times yielding an approximate energy profile for the diffusion pathway from the V₂O₅ slab through the fulgide SAM to the electrolyte. In order to avoid that the whole system containing slab, SAM and electrolyte moves, the bottom layer of the slab had to be fixed in space.

2. Results and Discussion

2.1. Molecular Properties of the Fulgide Derivatives

The fulgides as photo‐switchable molecules are chosen for the following reasons: 1) they are known for its reversible photo‐switching,[ ³⁸ , ³⁹ ] 2) they show well separated absorption bands for both photo‐isomers, 3) their structural rearrangement upon the photo‐switching is weak, so that it can be expected that the SAM will be preserved. For the interaction with a metal oxide surface, a further requirement is necessary: the molecule must contain (at least) one polar group. Since the introduction of functional groups may affect the electronic properties of the fulgide switch, we want to verify its absorption behaviour with the help of TD‐DFT calculations.

As a first step, we evaluate the performance of seven functionals with respect to a feasible reproduction of the experimental absorption bands. Therefore, we tested PBE, PBE0, B3LYP, M06, M06‐2X, ω‐B97X and CAM−B3LYP for bis(benzilidene)succinic anhydride (fulgide 1), whose structure is shown in Figure 2.

Figure 2.

Scheme for the different fulgide derivatives investigated in this paper.

The simulated UV/Vis spectra of the open and closed form of 1 are shown in Figure S1. The experimental values (${\lambda }_{\max ,\mathrm{open}}=355\mathrm{nm}$ ; ${\lambda }_{\max ,\mathrm{close}}=485\mathrm{nm}$ ^[40] ) are indicated as vertical grey lines. Comparing the results, we can see that none of the functionals shows a perfect fit with both experimental values. In gas phase, the smallest derivations are obtained with the hybrid functionals PBE0 and B3LYP as well as M06. Since the value of the Open isomer is underestimated and the value for the Closed isomer is overestimated, the splitting between both isomers is calculated significantly smaller than in the experiment. Nevertheless, we can demonstrate that the absorption of both isomers is well separated. Considering solvation effects by taking into account the dielectric constant of 2‐methyl tetrahydrofuran ($ϵ=6.97$ ), the experimental solvent, ^[40] we see a bathochromic shift of the entire spectrum maintaining the splitting. With that the functionals PBE0, B3LYP and M06 represent the spectra of the Closed form very well, but not the Open form. It may be associated to the geometry of the HOMO and LUMO, shown in Figure S2, where the orbitals in the case of the Closed isomer are located only on one aromatic moiety, while in the Open isomer the whole molecule is involved. As shown in the simulated absorption spectra, this issue cannot be tackled with range‐separated functionals. CAM−B3LYP shows a good fitting for the Open isomer, but not for the Closed one. Consequently, we can stack with the above mentioned hybrid functionals, because our goal is to have a tool for evaluating the influences of different functional groups on the photophysics of the fulgide backbone.

As a next step, we were looking for a possibility for the interaction of the fulgide with the (001) surface of V₂O₅. This surface is known to be the most stable one.[ ⁴¹ , ⁴² ] The experience from our former project ^[17] generated the idea of carboxylic groups, which can interact via hydrogen bonds with the exposed oxygen atoms on the surface as well as via electrostatic interactions between their oxygen atoms and the vanadium atoms of the surface. Due to the distance of about 9 Å between two exposed oxygen atoms, the surface has a valley for the aggregation of the organic molecules. This distance also allows the interaction via two hydrogen bonds, hence we suppose that a fulgide molecule with two carboxylic groups shall adsorb better on the surface and form a more stable SAM. Hence, our molecules for further investigation are the 4‐((Z)‐((E)‐4‐benzylidene‐2,5‐dioxodihydrofuran‐3(2H)‐ylidene)methyl)phtalic acid (2, o‐dicarboxy fulgide) and its derivative 4‐((Z)‐((E)‐4‐(3’‐methyl)benzylidene‐2,5‐dioxodihydrofuran‐3(2H)‐ylidene)methyl)phtalic acid (3, methyl‐o‐dicarboxy fulgide), see Figure 2.

For the photo‐switchability, the fulgide derivative has to have one double bond in E conformation and one in Z conformation directly bound to the succinic anhydride moiety. Due to the fact that we propose the functional groups only at one of the phenyl rings, both rings are distinguishable, which results in a configurational difference between Z‐E and E‐Z isomers. Furthermore, the 1,3,4 substitution pattern of the phthalic acid ring breaks the local symmetry. Hence, we have to consider four configurational isomers of 2 and 3, respectively, being able to undergo the photo‐switching. These configurational isomers together with their corresponding photo‐isomers are summarized in Figure 3.

Figure 3.

Summary of isomers of 2 and 3 in their open form (top) and closed form (bottom).

Both sets of eight isomers are optimized with ORCA (PBE0‐D3(BJ)/def2‐TZVPP) and QuantumEspresso (PBE based ultrasoft pseudo‐potentials in a 20×20×20 Å box). Their energies are compared in relation to the isomer of lowest energy. These relative energy values are shown in Table 1.

Table 1.

Relative energies E _rel of the photo‐isomers of 2 and 3, respectively.

	2 (o‐dicarboxy fulgide)
	Open 1	Open 2	Open 3	Open 4
$E_{\mathrm{rel};\mathrm{Orca}}$ /kJ/mol	1.6	5.4	1.9	0.0
$E_{\mathrm{rel};\mathrm{QE}}$ /kJ/mol	0.0	7.5	7.0	2.4

	Closed 1	Closed 2	Closed 3	Closed 4
$E_{\mathrm{rel};\mathrm{Orca}}$ /kJ/mol	28.4	33.1	27.5	39.0
$E_{\mathrm{rel};\mathrm{QE}}$ /kJ/mol	47.3	48.1	40.4	60.0

	3 (methyl‐o‐dicarboxy fulgide)
	Open 1	Open 2	Open 3	Open 4
$E_{\mathrm{rel};\mathrm{Orca}}$ /kJ/mol	2.2	6.0	2.8	0.0
$E_{\mathrm{rel};\mathrm{QE}}$ /kJ/mol	0.0	8.1	1.2	1.6

	Closed 1	Closed 2	Closed 3	Closed 4
$E_{\mathrm{rel};\mathrm{Orca}}$ /kJ/mol	24.2	32.0	28.1	40.0
$E_{\mathrm{rel};\mathrm{QE}}$ /kJ/mol	39.0	44.6	40.6	59.7

It can be seen that the closed form is less stable than the open form, which indicate both methods of calculation. The origin of this difference can be seen in the loss of the extended π system in the closed form as well as a large steric repulsion. The energetic order of the isomers is not influenced by the methyl group, as the values of 2 and 3 are similar. The only significant difference can be found for the motif Open 3, especially when referring to the QE simulations. The PBE0‐D3 simulation performed with Orca claims the motif Open 4 as minimum, while in the simulation with QuantumEspresso, the motif Open 1 is the most stable. Due to the aspect that the methods are fundamentally different (hybrid vs. GGA functional), a resulting energy difference of few kJ/mol is not surprising.

In order to verify the photo‐switchability, we performed the TD‐DFT calculations on the optimized geometries of both sets of the eight isomers. The results are shown in Figure 4.

Figure 4.

Comparison of the simulated absorption spectra for the eight isomers of 2 and 3 together with the respective spectrum of 1, TD‐PBE0‐D3(BJ)/def2‐TZVPP.

The four open isomers of 2 show the first absorption wavenumber between 26188 and 26627 cm⁻¹, while corresponding values of 3 range from 25805 to 26489 cm⁻¹. For the closed isomers, wavenumbers between 22576 and 23506 cm⁻¹ are found for 2 and between 22413 and 23371 cm⁻¹ for 3. With that, separations of the photoisomers between 2593 and 3624 cm⁻¹ are obtained, values which are smaller than one of the reference for 1 (3836 cm⁻¹). For the selection shown in Section 2.2, it is important to mention that the conformers 1 and 3 show the larger values of absorption separation (3624 and 3368 cm⁻¹ for 2 and 3392 cm⁻¹ and 3414 cm⁻¹ for 3, respectively).

The comparison of the simulated absorption spectra can give information of the effect of the functional groups. The maxima of 2 are shifted bathorchromic (−107 to −546 cm⁻¹) in comparison to 1 in the open form, while difference for the closed form exhibits shifts between −322 and $+609$  cm⁻¹. For 3, these shifts are similar, for the open form a bathochromic shift of −245 to −929 cm⁻¹ is observed, while the closed form shows shifts between −485 and $+473$  cm⁻¹. The introduction of the carboxylic groups only shows a systematic influence on one of the photo‐isomers, while the influence on the other one depends on the individual isomer. The result is a reduction of the difference between the absorption of the photoisomers. However, this reduction is not very large, hence we can conclude that the reversible photo‐switch remains. In contrast to the carboxylic group, the effect of the methyl group is systematic. Comparing the individual isomers of 2 and 3, a hypsochromic shift between 136 and 395 cm⁻¹ can be found for both the open and closed forms.

2.2. Aggregation Behaviour of Fulgide Derivatives on the V₂O₅ Surface

As the chosen molecules shall be used as a surface coating, the next step is the calculation of their adsorption behaviour. For this purpose, we simulate the up‐standing arrangements of all eight isomers of 2 and 3 on a slab of V₂O₅. The molecules are placed in the valley in the (001) surface, maximising the interaction with the surface as reported for para‐aminobenzoic acid. ^[17] The resulting geometries of the adsorbed molecules do not differ significantly between 2 and 3, hence we show them only for 2 in Figure 5. In order to differentiate between the pure molecules and their aggregates on the surface, we reduce the nomenclature to sO1 and sC1 refering to Open 1 and Closed 1 on the surface (indicated with s), respectively.

Figure 5.

Resulting structures of the adsorption process of all eight isomers on a slab of V₂O₅, exemplarily shown for 2.

The adsorption behaviour of the different isomers is analysed with respect to the binding energy E _bind:

$${\displaystyle E_{\mathrm{bind}}=E_{\mathrm{sO}/\mathrm{Cn}}-E_{\mathrm{Open}/\mathrm{Closed}\mathrm{n}}-E_{\mathrm{surface}}}$$ (1)

with $n=1-4$ . Furthermore, we analysed the percentage of London dispersion energy in the binding energy:

$${\displaystyle \Delta E_{\mathrm{disp}}=E_{\mathrm{disp},\mathrm{sO}/\mathrm{Cn}}-E_{\mathrm{disp},\mathrm{Open}/\mathrm{Closed}\mathrm{n}}-E_{\mathrm{disp},\mathrm{surface}}}$$ (2)

The results can be found in Table 2.

Table 2.

Relative energies E _rel, binding energy E _bind and dispersion energy upon adsorption $\Delta E_{\mathrm{disp}}$ for the photo‐isomers of 2 and 3 on V₂O₅.

	2 (o‐dicarboxy fulgide)
	sO1	sO2	sO3	sO4
E rel/kJ/mol	19.8	27.2	13.8	0.00
E bind/kJ/mol	−92	−92	−105	−114
$\Delta E_{\mathrm{disp}}$ /kJ/mol	−52	−58	−57	−75

	sC1	sC2	sC3	sC4
E rel/kJ/mol	60.5	51.0	72.4	84.5
E bind/kJ/mol	−99	−109	−80	−88
$\Delta E_{\mathrm{disp}}$ /kJ/mol	−60	−48	−49	−65

	3 (methyl‐o‐dicarboxy fulgide)
	sO1	sO2	sO3	sO4
E rel/kJ/mol	37.0	44.7	33.0	0.00
E bind/kJ/mol	−94	−95	−100	−133
$\Delta E_{\mathrm{disp}}$ /kJ/mol	−53	−58	−49	−88

	sC1	sC2	sC3	sC4
E rel/kJ/mol	46.7	58.7	89.7	96.6
E bind/kJ/mol	−124	−117	−82	−94
$\Delta E_{\mathrm{disp}}$ /kJ/mol	−72	−62	−55	−70

All calculated binding energies are negative that means the molecules tend to adsorb on the surface. Furthermore, the values range between −80 and −130 kJ/mol, which is common value for physisorption ^[43] and also similar to the result of pABA on V₂O₅ ^[17] (−94 kJ/mol). The most stable motif for adsorption is sO4, which can be explained with the fact that outer phenyl ring also interacts with the surface, see Figure 5. This can also be seen in the rise of the dispersion energy in comparison with the other sOn motifs (at least 20 kJ/mol more). With the objective of building a SAM, this behaviour of sO4 is counterproductive since it yields an occupation of space on the surface, which reduces the possibility of intermolecular stabilization. The motifs with next higher relative energies are sO1 and sO3. These motifs are almost planar and with that better candidates for the formation of a stable SAM. The higher relative energies of the sCn motifs result from the difference in the molecular energy (Section 2.1) of the respective photo‐isomers. This molecular origin can be see in the fact, that all sOn and sCn motifs have similar binding energies. Actually, these similar binding energies are not surprising since the main interaction between slab and molecules can be associated to the phthalic acid unit interacting with V₂O₅ and should only be slightly influenced by the backbone. The exception sO4 is discussed above. For the same reason, it is also not surprising that the introduction of the methyl group does not fundamentally change the binding energies. However, there is a tendency that they increase slightly, which is a result of a higher dispersion energy induced by an additional interaction between the methyl group and the surface.

For the next step towards the SAM, we filled the slab with three molecules of each motif. Due to the periodic boundary conditions in our simulations, we can represent the situation that every molecule is sandwiched between two of its kind resulting a regular coverage of the surface. For better comparability, we show the resulting binding and dispersion energies as molecular values (indicated with the index m) that means the purely obtained values are divided by three. The results are shown in Table 3.

Table 3.

Relative energies E _rel, molar binding energy $E_{\mathrm{bind},\mathrm{m}}$ and molar dispersion energy upon adsorption $\Delta E_{\mathrm{disp},\mathrm{m}}$ of the arrangement of three molecules on the slab of V₂O₅ for the photo‐isomers of 2 and 3 on V₂O₅.

	2 (o‐dicarboxy fulgide)
	sO1O1O1	sO2O2O2	sO3O3O3	sO4O4O4
E rel/kJ/mol	0.00	40.1	14.0	661
$E_{\mathrm{bind},\mathrm{m}}$ /kJ/mol	−127	−122	−130	$+91$
$\Delta E_{\mathrm{disp},\mathrm{m}}$ /kJ/mol	−147	−139	−143	−151

	sC1C1C1	sC2C2C2	sC3C3C3	sC4C4C4
E rel/kJ/mol	332	676	335	1480
$E_{\mathrm{bind},\mathrm{m}}$ /kJ/mol	−64	$+49$	−56	$+306$
$\Delta E_{\mathrm{disp},\mathrm{m}}$ /kJ/mol	−150	−143	−150	−134

	3 (methyl‐o‐dicarboxy fulgide)
	sO1O1O1	sO2O2O2	sO3O3O3	sO4O4O4
E rel/kJ/mol	0.00	39.8	6.65	2990
$E_{\mathrm{bind},\mathrm{m}}$ /kJ/mol	−131	−126	−130	$+864$
$\Delta E_{\mathrm{disp},\mathrm{m}}$ /kJ/mol	−160	−146	−152	−175

	sC1C1C1	sC2C2C2	sC3C3C3	sC4C4C4
E rel/kJ/mol	293	1567	320	1206
$E_{\mathrm{bind},\mathrm{m}}$ /kJ/mol	−72	$+347$	−65	$+211$
$\Delta E_{\mathrm{disp},\mathrm{m}}$ /kJ/mol	−164	−139	−162	−166

The resulting binding energies indicate the stability of the resulting layers. Here it can be seen, that for both fulgide derivatives 2 and 3 only the motifs sO1O1O1, sO2O2O2, sO3O3O3, as well as sC1C1C1 and sC3C3C3 remain attached on the surface. For the other cases, the repulsion between the molecules is larger than the attraction on the surface (the optimizations yield partially non‐physical bond cleavage due to overlap of the molecules). Hence, it can be concluded that sO4O4O4, sC2C2C2 and sC4C4C4 cannot form a stable SAM. As both photo‐isomers shall form a stable SAM, we will focus in the following only on the switches corresponding to Open/Closed 1 and Open/Closed 3. The structures of the switches sO1O1O1 – sC1C1C1 and sO3O3O3 – sC1C1C1 are shown in Figure 6.

Figure 6.

Resulting structures of three molecules of 2 adsorbed on V₂O₅. Similar structures obtained for fulgide derivative 3.

As it is possible that a domain (either sOxOxOx or sCxCxCx) is only partly switched by irradiation, we simulated also the intermediate cases. Besides the above showed homo‐trimers sOxOxOx and sCxCxCx, we simulated also the hetero‐trimers sOxCxOx and sOxCxCx (with x either 1 or 3). The molecular binding energy $E_{\mathrm{bind},\mathrm{m}}$ in dependence on the completeness of the switching process from sOxOxOx to sCxCxCx is shown in Figure 7.

Figure 7.

Molecular binding energy $E_{\mathrm{bind},\mathrm{m}}$ in dependence on the completeness of the photo‐switching process. The values for the slabs with one molecule are given as reference.

The figure shows that the molecular binding energy is the highest when all three molecules are in the open form. All four combinations of sOxOxOx have larger negative binding energy (−127–−131 kJ/mol) than the corresponding sOx motifs. (−92–−124 kJ/mol). This can be interpreted as an attraction between the molecules, which is caused by the attractive dispersion interaction between the molecules with large π systems. The pure dispersion energy values change from −49–−57 kJ/mol for sOx to −147–−164 kJ/mol for sOxOxOx. This means that there have to exist further contributions to the intermolecular interaction energy, probably caused by steric repulsion. In contrast, the fully closed structures sCxCxCx show molecular binding energies in the range −56–−72 kJ/mol, which is lower than the corresponding values of the sCx motifs (−80–−105 kJ/mol). The change of the dispersion energies for sCxCxCx (−150–−164 kJ/mol) is comparable to the ones of the sOxOxOx. Hence, the steric repulsion in the case of the sCxCxCx is strong and cannot be (fully) compensated by the attraction of the dispersion. Nevertheless, the repulsion is not strong enough to release the molecules from the surface, since their overall binding energies are negative. For the step‐wise switching, we find an almost linear dependency of the binding energy with the amount of Closed molecules for the conformer 1 and a slight preference of structures with Open molecules for conformer 3. From that we can conclude that in the latter case, a complete change of the domain from sO3O3O3 to sC3C3C3 will be challenging and the intermediate state has to be considered for the upcoming discussion with respect to the diffusion of lithium.

2.3. Lithium Diffusion through the Fulgide‐V₂O₅ Slab Models

In order to calculate the Li ion diffusion, we decided to reduce the slab model to be able to calculate the movement of the Li ions in a reasonable amount of time. While the three sheets of V₂O₅ are mandatory, the surface model was reduced in the [010] direction from three to two unit cells, which results also in a reduction to two fulgide molecules in the model system. Due to the periodic boundary conditions, our new model represents the SAM in the same way as the ones presented in Section 2.2. We did not reduce the model system to one unit cell in the [010] direction, since we want to consider the intermediate states of the switching process, as well. Furthermore, this dimension of the model avoids an undesired interaction of Li with its image in the next cell. On the top of the fulgide molecules, we place four molecules of EC. This is necessary, since we want to investigate the diffusion of Li through the SAM, that means starting in the V₂O₅ slab and ending in the electrolyte.

Within the (001) plane of V₂O₅, several degenerate minima exist.[ ³³ , ⁴⁴ ] Since the fulgide molecules occupy only the valleys of the (001) surface, the degeneration vanishes. Hence, two different positions of the Li in the (001) plane have to be considered for diffusion. These two positions are shown in Figure 8, presenting one starting point between the molecules (P1) and one directly below the molecules (P2). The diffusion barrier between these two positions is known to be about 1.2 eV (115 kJ/mol). ^[33] These two position are not at the same height due to the wavy structure of the V₂O₅ sheets. The movement of the Li ions is implied; the actual positions during the diffusion process are shown in several snapshots in the Supporting Information.

Figure 8.

Two different starting positions for the diffusion of the Li in [001] direction through the SAM.

In Figure 9, the relative energy profiles of the pathways are shown. The movement of the Li ions is given relative to the surface, using the position of the vanadyl oxygen atoms as a reference, as shown in Figure 8. The pathways are ordered in the following manner: in the left images, the conformation 1 of the fulgide derivatives is shown, while the conformation 3 is shown in the right images. The color code represents the state of the photo‐switch (red=sOxOx, green=sOxCx, blue=sCxCx), and the starting positions of the Li (as in Figure 8) are differentiated with empty squares and filled dots.

Figure 9.

Comparison of the energy profiles for the movement of Li starting in the V₂O₅ slab, passing the SAM (top 2, bottom 3) and ending in the electrolyte. The energy reference point for the relative energy is starting point of the pathways, the distance reference point is the postion of the vanadyl oxygen atoms.

Since all the calculated pathways start with optimized positions of Li in the (001) plane, we can associate the progress with the three parts of the diffusion process: the diffusion from −5 to 0 Å (Step 0–8/10) represents the movement through the top layer of the V₂O₅ slab, the movement up to 10 Å (Step 9/11–29/31) represents the passe through the SAM of fulgide molecules; and the last steps (29–34) show the movement between the EC molecules of the electrolyte. For all pathways, we show the structures of the starting point (0), the minimum after reaching the surface (usually between 7 and 10), the maximum of the pathway between 3 and 8 Å (steps 15 and 25) and the end of the path (34) in the Figures S3–S8.

In the first part, all pathways show similar behaviour. The energy rises until a distance of 2–3 Å below the surface, with an energy barrier between 106 and 145 kJ/mol. These similarities are not surprising since the diffusion through the V₂O₅ layer should not be influenced by the molecules in the SAM. The obtained values are also similar to the corresponding diffusion barrier in the bulk (115 kJ/mol,[ ¹⁶ , ³³ ]).

After passing the V₂O₅ layer, larger differences between the pathways appear. While the ones starting in P1 show an energy drop at 0 Å to a value of about 50 kJ/mol, the ones starting in P2 show there a relative energy of −80–$-90\mathrm{kJ}/\mathrm{mol}$ , in all cases the global minimum of energy. This large energy drop can be explained by the special position the Li ion is passing through. The Li ion is situated between oxygen atoms of the surface and the oxygen atoms of the carboxlic groups of the fulgide molecules, see Figures S3–S8. The result is a strong stabilization of this structure by the Coulombic interactions between Li and O.

As the idea of this project is to study the structural influence of the SAM on the permeability of Li ions, the most important part of the pathways is what happens between 0 and 10 Å. To get an idea for the energy barrier of the diffusion of Li through the SAM, we look at the highest energetic point in this part of the pathway. The results are shown in Table 4 relative to the energy of the initial step.

Table 4.

Highest relative energy of the pathways between 0 and 10 Å above the V₂O₅ surface (≈energy barrier), as shown in Figure 9. Values are given in kJ/mol.

2	sO1O1	sC1C1	sO1C1	sO3O3	sC3C3	sO3C3
Pos 1	235	195	202	227	198	280
Pos 2	210	182	244	248	245	233

3	sO1O1	sC1C1	sO1C1	sO3O3	sC3C3	sO3C3
Pos 1	224	228	201	243	164	271
Pos 2	206	205	251	249	206	237

With the help of the barrier heights, some trends can be identified. First, we see that the pathways for sC1C1 and sC3C3 are lower than their Open counter‐parts sO1O1 and sO3O3, respectively, both for the fulgide derivative 2 and 3. This means, there is an influence of the state of the photo‐switch on the diffusion barrier. The energy difference between sOxOx and sCxCx is in average 28 kJ/mol and with that approximately 13 % of the average barrier height. Applying the Arrhenius equation $\frac{k_1}{k_2}={\mathrm{e}}^{-\frac{\Delta E}{RT}}$ , we obtain at room temperature a factor for the rate constants of $1.3·{10}^5$ . It can therefore be regarded as a significant obstacle to the lithium diffusion pathway. With respect to the starting position, it can be seen that the preference depends on the conformation of the fulgide derivative. While the diffusion barriers are smaller for position 2, when the SAM consists of fulgide derivatives in conformer 1, the order is inverted for SAMs of conformer 3. The intermediate states sOxCx do not give consistent results. While some barrier values are between the corresponding sOxOx and sCxCx values, others exhibit barrier higher than the sOxOx cases. Some of the pathways show energetic discontinuities (e. g. 2, sC3C3, P1), which can be associated with rearrangement of the SAM molecules or a reorganization of the electrolyte molecules. These discontinuities appear more frequent for fulgide derivative 3 than 2, which can be associated with the increased steric hindrance induced by the additional methyl group. However, a clear influence on the barrier height is not observable.

Finally, the last steps of all pathways show comparable behaviour. These steps are associated with the movement of the Li ion in the solvent. Interestingly, the deviation of the relative energies at the final position of the pathway is small. This indicates that, regardless of the particular pathway, the Li ion ends in a similar energetic situation. It can also be observed that the energy of the last step is similar to that of the initial step, which suggests that the number of steps was well chosen. In the related Figures S3–S8, it can be seen that the Li ion is coordinated by the oxygen atoms of EC and the succinic anhydride moiety of the fulgide derivatives.

3. Summary

We investigated the formation of a smart surface for lithium ion batteries studying the behaviour of photo‐switchable fulgide derivatives on the cathodic material vanadium pentoxide. With the help of DFT simulations, we confirmed that the chosen fulgide derivatives have separated absorption maxima for their open and closed form, validating their use as photo‐switches. By introducing polar groups, these fulgide derivatives maintain their photo‐switchable capabilities while being able to interact with the surface of V₂O₅. The calculated binding energies show that the molecules are well adsorbed. In certain conformations, it is possible to form stable self‐assembled monolayers on the metal oxide surface for both the open and closed form. Extensive simulations demonstrate differences in the energy barriers for lithium diffusion depending on the state of the photo‐switch. This behaviour suggests the potential use of fulgide derivatives as smart surface for lithium ion batteries. We see this as a starting point for experimental investigations and/or designing other photo‐switches for other substrates. As technological application, we envision their use in portable devices without continued use avoiding a charge loss by switching the LIB into a stand‐by state. A further application could be carsharing with electric cars being available on demand with fully charged batteries.

Author Contributions

F.D.: Conceptualization, Methodology, Software, Validation, Formal analysis, Investigation, Data curation, Writing – original draft, Writing – Review & Editing, Visualization, Supervising, Project administration, Funding acquisition. E.C.: Writing – original draft, Writing – Review & Editing, Funding acquisition.

Conflict of Interests

All authors declare that they have no conflicts of interest.

Supporting information

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supporting Information

Dietrich F., Cisternas E., ChemPlusChem 2024, 89, e202400486. 10.1002/cplu.202400486

Data Availability Statement

Data for this article, including atomic coordinates, input and output files are available at ioChem‐BD repository ^[45] at https://doi.org/10.19061/iochem‐bd‐6‐368.