|Home | About | Journals | Submit | Contact Us | Français|
Coupling spins of molecular magnets to two-dimensional (2D) materials provides a framework to manipulate the magneto-conductance of 2D materials. However, with most molecules, the spin coupling is usually weak and devices fabricated from these require operation at low temperatures, which prevents practical applications. Here, we demonstrate field-effect transistors based on the coupling of a magnetic molecule quinoidal dithienyl perylenequinodimethane (QDTP) to 2D materials. Uniquely, QDTP switches from a spin-singlet state at low temperature to a spin-triplet state above 370K, and the spin transition can be electrically transduced by both graphene and molybdenum disulphide. Graphene-QDTP shows hole-doping and a large positive magnetoresistance (~50%), while molybdenum disulphide-QDTP demonstrates electron-doping and a switch to large negative magnetoresistance (~100%) above the magnetic transition. Our work shows the promise of spin detection at high temperature by coupling 2D materials and molecular magnets.
Hybrid electronic devices containing magnetic molecules are attractive, as data storage units in quantum information processing, as it enables electric-field controlled spin manipulation1, 2. However, for an efficient read-out of the spin state of the magnetic molecule, it is important to optimize the magnetoconductance (MC), which is the change in electrical conductance under an external magnetic field, since this parameter characterizes the coupling of the spin state to the local electronic environment3. The electronic coupling between a non-magnetic conductor and the magnetic molecule is used to electrically detect the spin polarization: ideally, this coupling can give rise to a giant magnetoconductance effect that can be electrically transduced. For example, quantum dots directly connected to two metallic or superconducting contacts (source-drain electrode) and capacitively coupled to a third gate electrode, can be used as spin valves; the electrical currents are influenced by the high spin state of the weakly coupled molecular magnet (SMM) crystals physically coupled to the quantum dots1, 4, 5. However, such a device performs only at extremely low temperatures (below 4K) and hence, these materials have not been applied so far in practical devices4–8. It is important therefore, to develop magnetic nanomaterials with a room temperature spin state that can be coupled with a suitable electrical conductor via electrostatic, dipolar or flux coupling to detect the spin polarization.
In this context, polycyclic aromatic hydrocarbons with an open-shell singlet diradical ground state have been investigated for potential applications in organic spintronics9. Recently, Zeng et al. reported a molecule quinoidal dithienyl perylenequinodimethane (QDTP) (Fig. 1a), that, while existing as a singlet diradical in the ground state, is in equilibrium with a triplet biradical that is thermally excited around 360K10 (for SQUID measurement, Fig. 1b). Thus, QDTP, with a spin transition at room or higher temperature, has a clear advantage over magnetic nanoparticles or SMMs, and therefore, is well adapted for use in spintronic devices. To electrically detect these molecules, it is necessary to introduce a spin detector, for which a two-dimensional (2D) atomic thin layer that has a high sensitivity to spin states arising either from orbital hybridization effects or from charge transfer, is an ideal candidate10.
Recently, electrical transport experiments conducted at room temperature have identified a number of advantages of using 2D atomic layers, such as long spin diffusion length, large spin signal, and relatively long spin lifetime as compared to metals (Cu, Ag, or Al) or semiconductors (heavily doped Si, GaAs, or Ge)11. Moreover, 2D atomic layers have also been exploited as non-magnetic spacers in the conventional giant magnetoresistive or tunneling magnetoresistive multilayer devices and have shown greater potential for enhancing spin polarization as compared with oxide films12, 13.
Until now, experimental and theoretical investigations of the role of disorder in magnetoconductance have focussed mainly on enhancing the spin-orbit coupling strength of pristine graphene by introducing adatoms, such as heavy metal atoms or atomic hydrogen14–17. In contrast, the effective coupling of room-temperature stable molecular spins and graphene’s magnetoconductance allows one to study spin state-dependent electron transport, which can form the basis of a high-temperature magnetoelectronic device18, 19. On the other hand, MoS2, a semiconducting 2D transition metal dichalcogenide, provides a good platform for studying the role of Coulomb interaction in molecular spin detection. A large electron interaction strength (three orders larger than graphene) due to reduced dielectric screening in semiconducting 2D crystals and the relatively heavy electron band masses of Mo d-orbitals in the presence of high carrier densities indicates that MoS2 should be an ideal system for studying carrier–spin interaction20.
Here we demonstrate a hybrid device comprising of a layer of QDTP molecules, which are charged coupled to 2D materials such as graphene or MoS2. We have chosen graphene as a channel material due to its semimetallic band structure and an intrinsically low-spin-orbit coupling, which allows the study of long spin coherence. The MC at high temperatures (300–390K) of these two hybrid systems, QDTP-graphene, and QDTP-MoS2, are systematically investigated. Spin transition from the singlet to the triplet state (~360K) enhances MC in the hybrid devices. Different charge transfer behavior (p or n-type, respectively) is observed for graphene and MoS2 due to the difference in how QDTP molecules bind to these substrates. On graphene, we observe the creation of inhomogeneous charge impurities that lead to an increase in resistance with magnetic field, whereas, the exchange interaction between the triplet spin state of the molecule and MoS2 results in a sharp transition of magnetoconductance (~100%) at the spin transition temperature.
Monolayers of graphene or MoS2 were micromechanically exfoliated from the respective bulk crystals using standard scotch tape method and mounted on degenerately doped silicon substrates with a 300-nm-thick SiO2 overlayer that served as the gate dielectric. The monolayer films were characterized by photoluminescence (PL) (Supplementary Fig. 1), Raman spectroscopy and atomic force microscopy (AFM) (Supplementary Fig. 2). Electrical contacts of Hall-bar-type configuration were fabricated using electron-beam lithography followed by the evaporation of 3nm of titanium (Ti) and 75nm of gold (Au). The device was annealed at 420K to remove resist residues and to reduce contact resistance (see METHODS). The magnetconductance (σ xx) and Hall conductivity (σ xy) were measured as a function of temperature (T) and magnetic field (B) using Physical Property Measurement System from Quantum Design (PPMS). Applying a gate voltage V g, to the Si substrate controlled the charge density in the active channel (graphene or MoS2) device.
QDTP molecules (Fig. 1a) dispersed in dichloromethane (DCM) were drop casted on the devices followed by annealing under vacuum at 400K to remove the solvent (see METHODS). Atomic force microscopy (AFM) confirmed the successful attachment of the molecule on graphene/MoS2 (Supplementary Fig. 2a, b). In the following two sections, we will discuss the electrical transport at high temperature in graphene and MoS2 based hybrid devices.
Figure 2a shows the typical two-probe room temperature (RT) transfer characteristics (σ–V g plot) of pristine graphene field effect transistor, which reveals an ambipolar behavior with Dirac point at V g=1.0V. A positive shift (ΔV g~17V) of the charge neutrality point (V D) after the deposition of QDTP indicates p-type doping by the molecule (for detailed statistics see Supplementary Figs. 4a and 5a, and Supplementary Table 1). The experimental data σ(V g) can be fitted well by considering effective medium theory (EMT)14 and using fitting parameters such as carrier mobility μ, and fluctuation of carrier density n r.m.s., which are used to calculate the Boltzmann–RPA conductivity due to screened Coulomb scattering. This Boltzmann–RPA conductivity σB[n,μ] enters into zero magnetic field EMT equation , where P[n,n 0,n r.m.s] is a Gaussian distribution centered at an average carrier density n 0 with an r.m.s. fluctuation given by n r.m.s. For QDTP-graphene hybrid, we have used an additional parameter, i.e., the short-range conductivity σ S for Boltzmann–RPA conductivity, which captures the saturation of conductivity at high carrier density. The carrier mobility μ = (23200 ± 1000) cm2 V-1 s-1 and fluctuation of carrier density nr.m.s. = (118 ± 8)1010 cm-2 for pristine graphene and μ = (25000 ± 1000) cm2 V-1 s-1, nr.m.s. = (109 ± 8)1010 cm-2 and σs = (335 ± 23)e2 ∕ h for graphene-QDTP hybrid can be obtained after careful fitting. Experiments carried out on graphene flake with relatively smaller mobility (2800cm2V−1s−1) also shows qualitatively similar behavior (mobility of hybrid device μ~6400cm2V−1s−1) after deposition of QDTP molecules (Supplementary Fig. 5a). Increment of carrier mobility in graphene after the deposition of QDTP molecule can be explained by compensation of charged impurities, as has been reported in the previous studies of nanoparticles coated graphene21, 22. From the difference of V D, we get the concentration of holes introduced by molecule for the two devices as Δp=252×1010cm−2 and 105×1010cm−2, respectively.
The gate-dependent conductivity σ(V g) of pristine MoS2 device at room temperature (Fig. 2b) are similar to that reported in the literature23; initially, σ(V g) rises slowly with increasing back gate voltage, followed by a rapid rise in current as the Fermi level approaches the bottom of the conduction band. For the hybrid device, doping with QDTP leads to an enhancement of σ(V g) by a factor of three from its pristine value (σ Doped~0.10e 2/h vs. σPristine~0.03e 2/h at V g=0). Qualitatively similar effect occurs (σ Doped~0.06e 2/h vs. σ Pristine~0.02e 2/h at V g=0) for other device which has lower mobility. Inspite of the fact that the conductivity of both pristine MoS2 and MoS2-QDTP hybrid is less than , it follows a power-law gate-dependence in the metallic regime: σ = A(Vg-V0)β, where A is the global parameter, β is the critical exponent, and V 0 is local parameter. From the difference of V0(V0∣MoS2 = - 28.5 V) and V0∣MoS2+QDTP = - 64.8 V, electron doping by QDTP molecule can be extracted as n=(267±1)1010cm−2 (for detailed statistics see Supplementary Figs. 4b and 5b, and Supplementary Table 2).
The Hall measurement allows us to study the gate modulation of transverse resistance (|R xy|=f (V g)) at different temperatures (T) at a constant perpendicular magnetic field (B=2T). Increasing the temperature to 390K, the Dirac point gradually shifts to 28V, making graphene more p-doped (see Supplementary Fig. 6). The longitudinal conductivity (σ xx) as a function of magnetic field (B) for the temperature window 300–390K is shown in Fig. 2c. The background magneto-conductance due to the presence of intrinsic defects in pristine graphene is weak (Δσ xx~−0.1μS) and temperature-independent24 (Supplemetary Fig. 7a). Compared with graphene device, the hybrid devices (gate fixed at V D) show two distinct features: First, the zero-field conductivity decreases (Δσ xx|B=0=−25μS) when the temperature is increased to 390K, and second, as a function of the B-field, σ xx decreases as temperature increases, as shown in Fig. 3c. At 390K, we recorded a 55% reduction of σ xx from its RT value, measured at B=2T. In addition, Fig. 3a displays the B-field dependent longidinal conductivity ratio (σ xx/σ xx(B=0)) of the hybrid device, averaged over 10 samples, measured at the Dirac point at T=390K (σ xx/σ xx(B=0) at room temperature in Supplementary Fig. 8a). The observed magneto-conductance, which is consistent in all hybrid devices we have measured, points to the influence of external magnetic moments coupled to the graphene device.
In the case of MoS2, gate voltage-dependent Hall conductivity (σ xy) at B=2T is plotted in Supplementary Fig. 9 in the same temperature window as for graphene. While σxy of pristine MoS2 is temperature independent (Supplemetary Fig. 9), the hybrid devices exhibit a steep increase in σ xy with temperature; notably, σ xy is enhanced by a factor of 7 as the temperature is increased from 300 to 390K. The similar trend shown in all the measured hybrid devices suggests that first, there is electron doping of MoS2 by the QDTP molecules, and second, the charge transfer from high temperature triplets in the QDTP molecule is greater than that from room temperature singlets. The longitudinal conductivity (σ xx) of MoS2–QDTP hybrid device as a function of magnetic field (B) for the same temperature window as graphene hybrid is shown in Fig. 2d. For pristine MoS2, no change in σ xx (B) is observed as the temperature is increased up to 400K (Supplementary Fig. 7b). However, for the hybrid device at 390K, σ xx increases with applied magnetic field, as shown in Fig. 2d. The differential magneto-conductivity, which is defined by [Δ σ xx=σ xx(B)−σ xx(0)], is slightly negative (−2µS) at 330K, but increases to 7µS at 400K. The observed increase in zero-field σ xx with temperature is consistent with the Hall conductivity (σ xy) measurement. The magneto-conductivity (σ xx/σ xx (B=0)), averaged over 10 samples, is plotted as a function of B-field sweep (0 to 200mT) at different temperatures from 300 to 400K in Fig. 3a. Similar to the graphene hybrid, influence of the spin state of magnetic molecule on the electron transport of MoS2 is evident in all the devices, except for the fact that in the case of MoS2 hybrid, we observe positive magneto-conductance.
QDTP-doped graphene can be considered as an impurity-doped system having an inhomogeneous distribution of electron and holes near the Dirac point25, the magnetoresistivity (or magnetoconductivity) of such system has been analyzed previously using a semi-empirical equation25. Herein, the experimental magnetoconductance at Dirac point for each temperature of our graphene-QDTP hybrid is fitted to a modified semi-empirical equation , as shown in Fig. 2c. The fitting parameter used are high-field conductivity σxx,0, low-field conductivity σ xx,1, and mobility μ. The extra conductivity term (σ xx,1) in the equation accounts for the imbalance in the electron and hole concentrations, unequal electron and hole mobilities, and deviation of sample geometry from the ideal Hall bar26. Above the transition temperature (370K), σ xx,0 (3.36e 2/h) is much larger than σ xx, 1 (1.53e 2/h) indicating that magnetoconductivity is probably better described by the unusual [1+(µB)2]−1/2 dependence at higher temperatures (Supplementary Table 3). The good fitting of the experimental data confirms the model of temperature-dependent macroscopic inhomogeneity. The mean carrier density, given by , decreases in the given temperature window, as shown in Fig. 3b (shown as normalized carrier density n exp/n 300K, averaged over 10 samples). The temperature dependence of carrier density due to thermal smearing predicted by theory is very small, and should not play an important role in the observed temperature dependence extracted by aforementioned method. This suggests a non-trivial role played by the molecule in carrier activation.
In the case of MoS2, the normalized carrier density (n exp/n 300K), averaged over 10 samples, given by two channel model , increases by an order from the RT (n 400K=1.2×1010cm−2), with a sharp change around 360K (Fig. 3b). The temperature dependence of σxx shows a similar trend as the carrier density increasing 8 times from its RT value.
To investigate the origin of high temperature magnetoconductivity in the case of graphene, the gate–voltage dependence of the magnetoconductance (MC=) is calculated at different temperatures (inset of Fig. 3b). At RT, the MC values are relatively constant at different gate voltages. However, a negative MC of 45% is observed at 390K, while fixing the gate voltage at Dirac point (VD). In slightly electron/hole doped regions, the change in MC is much lower even though a similar trend is observed (Supplementary Fig. 10). A highly electron doped region is not expected to show a change in MC with temperature, since the fluctuations in carrier density are negligible far away from the Dirac point when compared with the total number of carriers25. The enhancement of negative magnetoconductance with temperature is due to the temperature-dependent charge transfer between the QDTP and graphene; at higher temperatures (beyond 360K), the density of Coulomb impurities increases due to higher charge transfer with the triplet state of the molecule27. The MC measured in this hybrid device is relatively large (~50%) in comparison with other disordered-graphene devices (~20%)28–30, and graphene based vertical spin valves (~15%)31, 32.
In contrast to graphene, MoS2-QDTP hybrid device shows a positive MC (inset Fig. 3b), which reaches a maximum positive MC value of 80% at 400K, this is one order of magnitude higher than its RT value of—9%. These results suggests that the carrier density in MoS2 can be tuned by changing the temperature or by external doping and this tunability is the key to efficiently detect the singlet–triplet spin transition in QDTP molecules at high temperatures.
Our electrical transport measurements reveal a strong change in the MC of the hybrid devices at RT vs. 370K, which may have its origin in the temperature-dependent charge transfer interactions of the respective 2d materials with QDTP molecules. Interestingly, graphene and MoS2 show contrasting evolution in the MC (different signs), which should originate from the intrinsic electrical characteristics of the two materials.
To understand the origin of the differences in the transport properties of hybrid devices of graphene or MoS2 with QDTP, first-principles calculations were performed to understand the charge transfer behavior in these hybrid systems. QDTP has three conformational isomers or conformers (such as CF1, CF2, and CF3), as shown in Fig. 1a. Figure 4 shows the charge transfer between graphene/MoS2 and the adsorbed QDTP molecule. For graphene, the differential charge density analysis shows that the CF3 QDTP accepts a larger number of electrons in the triplet state (0.44e per molecule) than in the singlet state (0.29e per molecule) (Fig. 4c). A similar behavior is observed for the CF1 conformer, albeit with a slightly larger charge transfer than the CF3 conformer (Supplementary Figs. 11–14). The results obtained from these calculations agree with the experimental results, both on the p-type doping of QDTP on pristine graphene, and the gradual positive shift of the Dirac point with temperature due to the enhanced doping in the triplet state. In the case of the hybrid QDTP-MoS2 system, we attribute the enhancement of the carrier density to the interaction between QDTP molecule and the sulfur vacancies (V S) in MoS2. It is well known that sulphur vacancies are ubiquitous in MoS2 and can reach concentrations as high as 1010cm−2 33. These defects have unpaired spins and thus have a stronger interaction with spin molecules. V s defects also act as charge traps in MoS2, and can be passivated by adsorbates33, one signature of this is the enhanced photoluminescence (Supplementary Fig. 1) from MoS2 following the adsorption of QTDP molecules. Indeed, based on our spin polarized DFT calculations, the charge transfer between the QDTP and MoS2 plotted in Fig. 4h shows that the triplet-state (0.21e per molecule) adsorption has a much larger charge transfer than that of the singlet-state (0.16e per molecule) adsorption.
Our theoretical calculations show that the different charge transfer behavior observed between graphene and MoS2 has its origin in the different interactions of these 2D layers with the functional groups in the QDTP. The isosurface plot of the differential charge density reveals that the dicyano groups in QDTP act as electron acceptors with respect to graphene (Fig. 4a, b), whereas, the N-methyl (NCH3) group located in the in the mid-section of the molecule acts as an electron donor when present on the surface of MoS2 (Fig. 4f, g).
The local density of states of the spin triplet state for both hybrid systems are shown in Fig. 4d, i, respectively. A tiny amount of spin-polarization is found in the states of graphene and MoS2 due to a weak van der Waals interaction with QTDP. In Fig. 4e, j, our analysis of the spin density shows that significant amount of the magnetic moment is concentrated at the dicyano groups and at the conjugated backbone. As shown by the plot of spin density in Fig. 4j, there is indeed appreciable weight of spin density in the MoS2 sheet. The enhanced spin density in MoS2 is due to the asymmetric coupling of the unpaired electrons in V S localized states with that of the unpaired spins of the triplet state molecule. Such spin density and asymmetric coupling are absent in the singlet state molecule (Fig. 4i).
The observation of large MC in graphene-QDTP at 400K rules out any quantum effects, such as weak (anti)localization28. In our work, the temperature dependent resistivity near Dirac point is a signature of charged-disorder-dominated transport. The conductivity of an inhomogeneously doped graphene film near the Dirac point is different from the homogenous case, due to correlations in impurity positions. The net effect of charge doping with QDTP is the creation of macroscopic regions of charge puddles; these may have a charge excess or charge deficit with respect to the average value. In the inhomogeneous regime, the conductivity near the Dirac point is dominated by the charge-disorder27, 34, 35.
To understand the positive MC in case of the MoS2 hybrid, we can consider the magnetic scattering mechanism as reported in diluted magnetic semiconductors (DMS)36–38, together with spin-imbalance caused by the initialization process of randomly oriented spins39, 40. The exchange coupling between the spin triplet state in QDTP and the 2D channel electrons scales linearly with Wigner Seitz radius r s 41. However, for 2D materials, the value of r s depends on both intrinsic factors such as the Fermi velocity, and extrinsic factors like the dielectric screening of the substrate14. For the carrier densities relevant to our experiment, we estimate that for graphene, r s~1, while for MoS2, r s~1000. We attribute the negative magnetoresistance (positive MC) of MoS2 beyond the triplet transition to Ruderman–Kittel–Kasuya–Yosida (RKKY) coupling of the spin triplet state of QDTP to electron spin in MoS2. This is analogous to the coupling between Mn spin states and Ga electrons in the dilute magnetic semiconductor (Ga,Mn)As37, 42. Accordingly, the exchange interaction (J) between the high temperature spin triplet state of the QDTP molecule (S i at site i) and the spin of the carrier in MoS2 (s) is given by the Hamiltonian, H = - J∑iSi.s. This exchange interaction may give rise to the spin-disorder scattering resistivity36–38, 43, which can be reduced by B-field induced alignment of spins, and can be expressed in a simple form as37, 43: (see Supplementary Fig. 15 for fitting details). Supplementary Figs. 15 and 16 show the result of the fit using A(T) as temperature dependent fitting parameter which has the same qualitative behavior as temperature dependent magnetic signal data (χT) from the SQUID measurement (or square of the magnetic moment) of the QDTP molecule10 (Fig. 1b). The coupling constant J can be extracted from DFT calculations using the energy difference between the spin triplet and singlet states (ΔE~0.36eV). The theoretical values of J based on the current structural model is estimated to be around 131eVÅ−2, which is close to the value (150eVÅ3) reported in (Ga, Mn)As37. We have also estimated J to be 200eVÅ−2 from the experiment following Supplementary Eq. 1, as discussed in Supplementary Note 1. The observed magnetoconductance is due to the quadratic dependence of exchange coupling constant (J) and the spin imbalance prevalent in the QDTP molecule, which is enhanced at high temperature due to the spin flip of the molecule (Supplementary Fig. 20). The sudden increase in MC of MoS2 at the transition temperature confirms the effect of the spin transition of the molecule on the electrical transport of MoS2. In comparison with the earlier reports on Mn-substituted ZnO (n-type 3d DMS)36 or Gd-doped Eu-chalcogenides38, the positive MC effect in QDTP-doped MoS2 at high temperatures is much larger (~100%).
Overall, our results show that magnetic molecules on graphene surface can be treated as charge disorder. The transition from singlet to triplet state at high temperature causes a more resistive magneto-transport, which can be enhanced by applying a magnetic field. In the case of MoS2, spin triplets of the adsorbed molecule at high temperature are coupled to charge carriers in MoS2, and an alignment of the non-interacting individual spins with the magnetic field enhances the electron transport. Since the interaction parameter r s in graphene is much weaker than in MoS2, the effect of exchange interaction with the QDTP molecule is much larger in MoS2. The different magnetoconductance behavior of graphene and MoS2 at the spin transition point of the molecule, which can be exploited to detect the spin flip of the magnetic molecules clearly highlights the importance of using the appropriate conduction channel for different targeted application: semimetallic graphene with a higher carrier density is better tailored for impurity-dominated transport, whereas semiconducting 2d materials such as MoS2 are ideal for detecting spin dependent phenomena.
Our work also highlights the importance of understanding the charge transfer and magnetic phenomena taking place at these molecule-2D hybrid interfaces. Clearly, QDTP molecule enhances the sensitivity of the MoS2 FET to small magnetic field (<20mT) by two orders of magnitude. This suggests that modifying 2D materials with magnetic molecules can enhance the MC by orders of magnitude. The advantage of QTDP is that the enhancement is operational at elevated temperatures, thus allowing the sensor to be operated at cryogen-free conditions. Finally, our work suggests that biradical molecules with active spin centers may be good candidates for room temperature hybrid spintronic devices with multi-functional properties, enabling the sensing of magnetic, electrical, optical, and chemical stimuli.
Graphene or MoS2 flake is prepared by mechanical peeling and transferred to heavily doped silicon with a layer of 300nm thick SiO2 (gate capacitance C i=11.6nF/cm2) on top. Optical microscope-based contrast is used to get initial information about the number of layers. To determine the quality and thickness of graphene/MoS2, Raman spectroscopy measurements were carried out. Electrodes were patterned by e-beam lithography using a bilayer resist combination of PMMA and subsequently Ti/Au (3nm/75nm) was deposited by an e-beam evaporator. Standard lift-off procedures using warm acetone were followed after the metal deposition. Typical dimension of the channel is as follows: L=3µm is the channel length, W=1.5µm is the channel width. QDTP molecules dispersed in DCM were drop casted on the devices followed by annealing under vacuum at 400K to remove the solvent.
First-principles calculations based on spin-polarized density functional theory are performed by using Vienna ab initio simulation package44. Becke88 optimization (optB88) is adopted for the description of the van der Waals interaction between the physisorbed QDTP and graphene/MoS2 by using a kinetic energy cutoff of 400eV. Here we consider a single QDTP molecule adsorbed in supercells of graphene and MoS2 containing total atoms of 217 and 183, respectively, with a vacuum layer with a thickness of 15Å. The first Brillouin zone is sampled with a 2×3×1 Monkhorst–Pack grid. All the structures are fully optimized by minimizing the forces on each atom less than 0.005eVÅ−1. The interfacial charge transfer is obtained via calculating the differential charge density Δρ(r). The amount of transferred electron up to z point in the normal direction is obtained using , where Δρ(z) is the plane-average differential charge density through integrating the Δρ(r) in the x–y plane.
All relevant data are available from the corresponding authors.
K.P.L. thanks National Research Foundation, Singapore for NRF Investigator Award: Graphene oxide a new class of catalytic, ionic and molecular sieving materials, award number: NRF-NRF12015. J.W. acknowledges financial support from the MOE Tier 3 programme (MOE2014-T3-1-004) and MOE Tier 2 grant (MOE2014-T2-1-080). S.A. acknowledges financial support from the Singapore National Research Foundation [NRF-NRFF2012-01], Singapore Ministry of Education and Yale-NUS College [Grant No. R-607-265-013121], and the National University of Singapore Young Investigator Award [R-607-000-094-133]. Discussions with Vitor Pereira and Ding Z.J. are gratefully acknowledged. This project is supported by Shenzhen Peacock Plan (Grant No. KQTD2016053112042971).
K.P.L. supervised the project. The hybrid device was made by S.D. Synthesis of QDTP molecule is done by Z.Z. and J.W. S.D. carried out the measurements. The data analysis was made by S.D. with the help of K.P.L. The theory for graphene hybrid was made by I.Y., S.D. and S.A. The theory for MoS2 was done by S.D. and S.A. DFT simulation was done by Y.C. and Y.-W.Z. S.D. and K.P.L. wrote the manuscript with inputs from H.Z., S.A. and Y.C. All authors discussed the data.
The authors declare no competing financial interests.
Subhadeep Datta and Yongqing Cai contributed equally to this work.
Electronic supplementary material
Supplementary Information accompanies this paper at doi:10.1038/s41467-017-00727-w.
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Jishan Wu, Email: gs.ude.sun@juwmhc.
Kian Ping Loh, Email: gs.ude.sun@pkholmhc.