Collective molecular switching in hybrid superlattices for light-modulated two-dimensional electronics

Molecular switches enable the fabrication of multifunctional devices in which an electrical output can be modulated by external stimuli. The working mechanism of these devices is often hard to prove, since the molecular switching events are only indirectly confirmed through electrical characterization, without real-space visualization. Here, we show how photochromic molecules self-assembled on graphene and MoS2 generate atomically precise superlattices in which a light-induced structural reorganization enables precise control over local charge carrier density in high-performance devices. By combining different experimental and theoretical approaches, we achieve exquisite control over events taking place from the molecular level to the device scale. Unique device functionalities are demonstrated, including the use of spatially confined light irradiation to define reversible lateral heterojunctions between areas possessing different doping levels. Molecular assembly and light-induced doping are analogous for graphene and MoS2, demonstrating the generality of our approach to optically manipulate the electrical output of multi-responsive hybrid devices.

S-3 measured directly after spin-coating, while that of the MC film was recorded after UV irradiation of the same film. As expected, the spectra in the solid state show less structured absorbance bands as compared to those in solution, due to the π-π interaction among molecules. The spectrum of the same film was also measured after 48 h, showing an almost complete recovery.
(d) Evolution of the film absorbance (at 620 nm) normalized at the value at t = 0 s after UV irradiation at room temperature. As compared with the MCSP relaxation in solution, the thermal recovery in films is much slower, and a full recovery of the SP state is achieved only after approximately 20 h. The relatively slow MCSP recovery in thin films warrants us with sufficient time to explore the electrical properties of the metastable MC/2DM system (see main text). HOPG. The assembly shown in this image is different from that presented in Fig. 1e   The devices were never exposed to air: the spin-coating, the electrical measurements and the UV irradiations were performed in-situ in a nitrogen filled glove-box, while the green irradiation was performed ex-situ, but the samples were transported in a sealed chamber with nitrogen atmosphere. However, the measurement shown in panel (a) shows that the SP-MC film is not completely stable even in this controlled nitrogen atmosphere. We hypothesize that the deterioration measured in the devices might be caused by the protonation of the SP/MC forms. 10 This effect could be caused by the interaction of the spin-coated SP/MC films with volatile chemicals used in the nitrogen glovebox while the sample was stored. Φ is defined as the energy required to extract an electron from the bulk to the vacuum level: where V ∞ is the electrostatic potential in vacuum and E F is the Fermi level of the bulk (here graphene). The contribution of the interface potential to the work function shift can be estimated via the charge density difference at the interface: ∆ (z) = sys − ( assembly + ρ slg ) S-21 where ρ sys , ρ assembly , and ρ slg are the charge density of the whole system (interface), the molecular assembly and the graphene layer, respectively. The electrostatic potential ∆V E associated to the charge density difference at the interface ∆ρ is obtained by a numerical integration of the Poisson equation: The work function shift can be expressed as a combination of two contributions: ∆ = Δ assembly + . . = Δ assembly + Δ E + Δ slg (4) where ∆V assembly is the shift of the electrostatic potential induced by the intrinsic dipole moment of the molecular assembly, and B.D. is the potential change at the interface due to the adsorption of the molecular assembly 7 . The latter can be decomposed in two terms, namely the geometric rearrangement of the substrate and the electronic reorganization resulting from the adsorption of the molecular assembly on graphene. Upon physisorption of the molecular assembly on graphene, it is expected that the graphene geometry is almost unperturbed; i.e., no geometric restructuration of the carbon atoms occurs (∆V slg = 0). Therefore, the work function shift, ∆Φ, can be expressed in terms of the local electrostatic potential associated to the charge density redistribution ∆V E between graphene and the monolayer, and the contribution of the molecular assembly ∆V assembly . By computing the potential profile across the molecules in the direction normal to the graphene plane (z-axis), ∆V assembly can also be calculated from the electric dipole of the molecules. Indeed, in the Helmholtz model, the assembly contribution is directly proportional to the electric dipole of the molecules along the axis normal to the graphene surface, μ: where e is the elementary charge, ε 0 the vacuum permittivity, and S the surface area of the unit cell. Note that by using the effective dipole of the molecule from calculations performed on the assembly (using periodic boundary conditions), depolarization and image charge effects are automatically built in.   Fig. 10). As a result, the molecule adopts a 3D conformation on the graphene surface, in contrast to the MC derivative that is 2D (planar molecule). Hence, spontaneous self-assembly of the SP molecules is likely exclusively driven by have been performed on both enantiomers (R and S) using the fine-tuned DREIDING force-field.
The formation of the SP monolayers has next been investigated, adopting the same approach used for the MC derivative. Namely, the lowest-energy conformations have been used as building blocks to build the assemblies with different organizations, including mixture of lowenergy conformations, interdigitation of alkyl chains, racemic or chiral assemblies, etc.
Unfortunately, the quenched dynamics simulations failed in providing well-organized 2D architectures, likely because of the multiple possible arrangements of the aromatic heads of the chiral molecules. We thus turned to a different approach, whereby SP monolayers were prepared using input from STM measurements. From the experimental data, alkyl chains are physisorbed, forming 2D crystalline structure on the graphene surface (Fig. 1). The conjugated core of the SP moiety leads to bright spots, which are aligned perpendicular to the alkyl chain axes (dark regions). The SP isomers can apparently organize in contiguous lamellae adopting either headto-head or head-to-tail ( Supplementary Fig. 3) configurations.
Usually, racemic mixture of a chiral compound dropped on HOPG leads to the formation of a monolayer with various domains (chiral or racemic) 8,9 . Here, no chiral domains can be distinguished on the STM images. Therefore, we assume the monolayers are formed with a statistical ratio of both enantiomers. Considering the geometrical features extracted by the STM images (Fig. 1e), a non-interdigitated SP monolayer has been simulated. We started with a geometry optimization of a SP dimer (with both enantiomer in head-to-head fashion) and use the relaxed structure as a building block for the preparation of the monolayer. In order to preserve that organization, we then ran MD simulations where the tails of the alkyl chains (ten CH 2 units) are spatially constrained. Note that we used in these calculations the upper estimate within the experimental error bar for the value of parameter a (6.1 nm), so as to leave more space to the head groups. This approach thus enforces a self-organization driven by the alkyl side chains in line with the STM data, while allowing the aromatic cores to fully explore their (complex) configurational space.
MD simulation runs over 1 ns yield for SP broad distributions of the aromatic head group orientation that translate into small (compared to MC) interfacial dipoles, with average and standard deviations of 0.23 D and 0.11 D, respectively (see Supplementary Fig. 6).