Pentacene/ TiO 2 anatase hybrid interface study by scanning probe microscopy and ﬁrst principles calculations

The understanding and control of the buried interface between functional materials in optoelectronic devices is key to improving device performance. We combined atomic resolution scanning probe microscopy with ﬁrst principles calculations to characterize the technologically relevant organic/inorganic interface structure between pentacene molecules and the TiO 2 anatase (101) surface. A multipass AFM imaging technique overcomes the technical challenge of imaging simultaneously the corrugated anatase substrate, molecular adsorbates, monolayers and bilayers at the same level of detail. Submolecular resolution images revealed the orientation of the adsorbates with respect to the substrate and allowed direct insight into interface formation. Pentacene molecules were found to physisorb parallel to the anatase substrate in the ﬁrst contact layer, passivating the surface and promoting bulk-like growth in further organic layers. While molecular electronic states were not signiﬁcantly hybridized by the substrate, simulations predicted localized pathways for molecule-surface charge injection. The localized states were associated with the molecular LUMO inside the oxide conduction band, pointing to eﬃcient transfer of photo-induced electron charge carriers across this interface in prospective photovoltaic devices. In uncovering the atomic arrangement and favorable electronic properties of the pentacene/anatase interface, our ﬁndings testify to the maturity and analytic power of our methodology in further studies of organic/inorganic interfaces.


Introduction
Organic photovoltaic (OPV) devices, where thin films of organic molecules coated with metal oxide buffers are layered between metallic electrodes, provide an affordable route to manufacturing large-area, lightweight solar cells for universal use.Device operation is enhanced by efficient electron-hole pair (exciton) separation into carrier charges in the organic region and directional intra-layer charge mobility.Favorable band alignment at the organic/inorganic interface is particularly important for effective charge injection across the layers, whereas poor contact between the two materials leads to carrier scattering and recombination that degrade device performance.All these factors depend on the electronic structure of the buried interface; by controlling it we can enhance the power conversion efficiency of OPV devices. 1 Identifying the atomic arrangement at the organic/inorganic interface is critical to determining the local electronic band alignment.Atomic details of the interface structure are difficult to establish experimentally, so computational techniques are frequently deployed to simulate various molecule-solid contact configurations.Such calculations are made challenging by the large numbers of atoms and degrees of freedom involved; chemical intuition helps to identify relevant configurations, but without experimental verification there is no guarantee that the computed interface structure resembles the one in the actual device.
In this work, we pursue a combined theoretical and experimental approach to interface structure determination.A metal oxide surface was functionalised by organic molecules at varying degrees of coverage.TiO 2 anatase served as the prototype model oxide buffer material on account of its optical activity and high charge carrier mobility. 2,3The (101) facet of an anatase natural crystal was covered with photoactive pentacene molecules; this is an electron acceptor molecule widely employed in organic optoelectronics. 4,5These two materials play significant roles in the recent revival of dye-sensitized solar cells, 6,7 TiO 2 as substrate [8][9][10] and pentacene as a candidate for metal-free dyes, 11,12 which makes their interface a topic of current scientific interest.
Scanning probe microscopy (SPM), atomic force microscopy (AFM) 13 in particular, can now achieve outstanding sub-molecular imaging of molecules on flat surfaces. 14,15Structural detail is delivered by constant-height close-range scanning techniques with chemically inert CO-functionalized probes. 16The considerable substrate corrugation of reconstructed metal oxide surfaces or large adsorbates presents a challenge for this imaging method, making it difficult to functionalize the probe with a CO molecule using vertical manipulation techniques or to scan the surface at constant height.To overcome this obstacle we opt for a 'multipass' imaging method.Developed to address high-resolution AFM imaging of large molecules on corrugated substrates, it allows us to obtain atomic-scale resolution detail of both surface and adsorbate simultaneously. 17perimental characterization was carried out by combining 'multipass' AFM with bias spectroscopy imaging 18 and scanning tunnelling microscopy (STM). 19In multi-channel measurements, probe-surface force and tunneling current signals were recorded simultaneously to obtain real-space local information on the chemical and electrical response of the functionalized surface.We employed first-principles density functional theory methods to interpret the contrast formation mechanisms behind AFM and STM imaging, and analyze the electronic properties at the surface.Such a comprehensive approach allowed us to study local bonding of the molecules to the crystal substrate, monolayer formation and thin film morphology, and thus reconstruct the typical interface structure.A similar approach may be employed to characterize the assembly and growth of complex surface nanostructures.

Results and Discussion
The wide area AFM scan of pentacene-covered TiO 2 anatase surface shown in Fig. 1(a) features a trapezoidal island typical of the (101) facet morphology. 20Pentacene molecules adsorbed onto the island's terrace are distinguishable as bright, oblong objects.Sub-molecular   direction by half unit cell with respect to the adjacent one, a Ti 5c surface site on one groove is aligned with a mid-point between Ti 5c atoms on the other.Consequently, molecules adsorbed in a straight line stack correspond to a mixed A-B registry, as illustrated in Fig. 2

(e).
A stack of molecules in the same registry can only be obtained if the adjacent pentacenes adsorb with an offset of a half-unit cell, like in Fig. 2(f).We optimized candidate stack structures of mixed A-B and single B-B registries, formed by assuming a maximum overlap of molecules in the stack.Computed adsorption energies per molecule of E AB ADS =-1.96eV and 19eV confirm that molecular stacking with an offset is energetically favorable on this surface.This finding supports the experimental observation of diagonal stack growth, and the presence of kinks in the stacks, in numerous AFM images (Fig. 1).
Stronger binding energies for stacked pentacene molecules compared to isolated ones arise partly from the inter-molecular attraction within the stack.To quantify the strength of this interaction, we computed the reference energy for the molecular stack without the sub- Consequently, each molecule in a stack adopts a local geometry that is almost identical to the isolated molecule, with minimal changes in the tilt angle (see Fig. 2(d)).
The alignment of electronic levels at the molecule-oxide interface is critical to effective flow of charge between the materials.The decomposed electronic states of the surface and the molecule adsorbed at a B registry are shown in Fig. 3(d).The highest occupied electronic states of the molecule HOMO, HOMO-1 and HOMO-2 appear as narrow peaks in the band gap of the oxide.Frontier molecular states of the physisorbed pentacene remain as discrete and sharp as those of the gas phase molecule (within the molecular level broadening 0.05eV), apart from LUMO which exhibits a double peak.5][26] Bader charge analysis 27,28 revealed that pentacene transfers only 0.1 e − to the TiO 2 substrate.
The Fermi level at the interface is located at the highest occupied molecular orbital (HOMO) 0.7eV below the TiO 2 conduction band.We analyzed the spatial distribution of this state, illustrated in Fig.   S2).The appearance of a double force minima is clear when probing the sites that would correspond to the CH groups at the upper long edge of the molecules (Fig. 4(a) and Fig. S2).The development of a second minimum in the probe-surface interaction force over both the hollow and the CH site may correspond to a slight bending of the molecule as the probe exerts force on it, or to the atomic-scale bending of the probe termination.Multiple minima in the interaction force are usually associated with transitions between different solution of the probe-surface potential energy upon reducing the probe-sample separation. 30e absence of energy dissipation 31 (Fig. S3) during the acquisition of the force volume points towards an elastic bending due to the interaction with the probe.
Force spectroscopy experiments were simulated by approaching a probe model to key pentacene sites -an uppermost C atom site and the central ring of the molecule, Fig. 4(c)to clarify in detail probe-molecule interactions and associated atomic rearrangements in the presence of the substrate.We deployed first a TiO 2 cluster probe model with a chemically-passive OH group termination; such a probe was shown to produce the dominant AFM contrast in atomic resolution imaging of the bare (101) anatase surface. 20Computed AFM force spectroscopy curves presented in Fig. 4(c) testify to a stronger attractive interaction above pentacene rings as opposed to the carbon backbone, in agreement with the experimental force curves, displayed in Fig. 4(a).The maximum attraction over the ring site is reached by lowering the probe by a further 50pm, reflecting the topographic 50pm height difference between the carbon and the ring site in the atomistic model.Upon inspection of the calculated forces, the pentacene AFM contrast seems to be in line with the standard AFM imaging modes of pentacene and carbon nanostructures composed of aromatic rings 16,32,33 recorded with different probe terminations, where additional repulsive force contributions above C atom sites locally reduce interaction strength.AFM simulations were repeated over the same sites in a pentacene B-B stack to explore how probe-surface interactions change with monolayer formation.Theoretical force spectroscopy curves for isolated and stacked pentacene are compared in Fig. 4(c).In the stack, the onset of attraction is shifted to lower probe heights but the repulsive interactions are slightly reduced, indicating that the intermolecular bonding within the stack weakens the pentacene-probe interactions.
Simulated AFM spectroscopy curves over the pentacene stack can be directly compared to the short-range interaction data extracted from a force volume, and depicted in Fig. 4(b).
Although the general behavior of the curves is fairly reproduced, we found the computed maximum attractive interactions of 0.20-0.25nNconsiderably larger than the 0.06-0.10nNshort-range force values recorded in the experiments.Such a result points to a different chemical termination of the AFM probe than the initial model considered for these computations.This presents a conundrum, since AFM probes were in situ conditioned to be coated by TiO 2 from the substrate, and no other apex termination has been identified to produce weaker probe-surface interactions than the OH group. 20In the absence of other contaminant material that might alter the chemical composition at the probe apex, we hypothesized that the probe may have picked up a molecule during the scan. 34To verify the possibility of AFM imaging by a probe terminated in pentacene, force spectroscopy simulations over an isolated molecule were repeated with a naphthalene molecule as model probe.Such a probe allowed us to include contrast-defining interactions between the low-lying aromatic rings of the pentacene molecule, while presenting the weakly-interacting C-H bond towards the substrate in an analogy to the OH group terminated probes that produce atomic resolution AFM images in the anatase surface. 20Computed spectroscopy curves with a naphthalene probe are compared to those resulting from the OH-terminated TiO 2 cluster model in Fig. 4(d).The ordering and relative position of the curve minima between the two key sites remains the same, pointing to the same AFM contrast mode in the imaging of pentacene for both probe models.Notwithstanding, the naphthalene-terminated probe exhibits maximum attractive interactions of 0.07-0.11nNthat are 2-3 times smaller than for the TiO 2 cluster case; values that show a reasonable agreement with the experimental results.Small discrepancies, as well as the lack of elastic effects associated with double minimum force spectroscopy signatures, can be explained by the randomly chosen orientation of the molecular probe model and the approximation of a pentacene probe with a naphthalene molecule.
To gain insight into pentacene thin-film formation on the anatase (101) surface, we conducted an experimental AFM characterization of the pentacene growth beyond the first molecular layer, and found that the second pentacene layer presents a completely different morphology.the repeating unit much smaller than the pentacene molecule length, e.g., as illustrated in Fig. 5(d).The 1.45 nm height difference between the domains does not relate to the morphology of the anatase surface (with an atomic step height of approximately 0.34 nm), but rather points towards the second layer of pentacene molecules growing in a slightly tilted, vertically-oriented configuration over a pentacene-passivated anatase surface, e.g. as illustrated in Fig. 5(e).While the reactive anatase (101) surface promotes horizontal monolayer growth, it is sufficiently passivated by the first molecular layer so that further layers grow vertically.High resolution molecular AFM images acquired with the multi-pass method 17 over the region with higher topographic contrast reveal an internal and complex structure within the rectangular unit cell (Fig. 5(f)), denoting a high-density packing of the vertically oriented molecules.Although the packing, orientation and tilting of the molecules in the second layer is difficult to discern from our AFM images, our results suggests that subsequent pentacene monolayers may resemble the bulk configuration of pentacene organic solids.
With prospective applications of pentacene as thin-film component in organic transistors 35 and light-emitting diodes 36 among other organic devices, we extend our discussion to AFM studies addressing pentacene thin-film growth on different buffer materials.On this substrate, not only do we observe horizontal monolayer growth, but the molecules aggregate parallel to each other in the departure from the herringbone arrangement in bulk 44 and vertical monolayers.These features can be attributed to the strong molecule-substrate dispersion interactions and the topological constraint of the sloping grooves of the surface.
A recent study supports our finding that pentacene adsorbs horizontally on the sloping terrace of anatase 11 (albeit with a greater adsorption energy of -2.48eV, the structural details were not disclosed).Similar adsorption behavior was reported for pentacene on the (110) surface of the rutile polymorph. 45We conclude that TiO 2 presents a more reactive substrate for dispersion-dominated pentacene thin films than other metal oxides, which is consistent with the benzene adsorption trend on metal oxide substrates. 11The stronger bonds and orbital overlap with TiO 2 should enhance charge injection across the hybrid organic/inorganic interface.

Conclusion
The atomistic details of the hybrid organic/inorganic boundary inside heterostructure devices critically define the electronic properties of this interface, and through it, those of the entire device.We have employed several AFM and STM experimental techniques in combination with first-principles simulations to characterize the interface between pentacene thin films and TiO 2 anatase; two materials with superior electrical properties frequently used in optoelectronic devices.A multipass AFM imaging technique allowed us to overcome the technical challenge of simultaneously imaging the substrate, steps, molecular adsorbates and layers with atomic scale resolution.An inert AFM probe termination, here an adsorbed pentacene molecule, met the other necessary condition for submolecular imaging.Consequently we were able to detect the relative orientation of molecular adsorbates with respect to each other and the substrate, which greatly facilitated the computational study of adsorption.
Sub-molecular resolution AFM images reveal that the first layer of pentacene molecules physisorb horizontally on the anatase surface, with the conjugated backbone of the molecule parallel to the typical sloping terraces of anatase.Individual pentacene molecules are also found decorating the step-edges of the anatase (101) surface, with an adsorption geometry that varies with the particular step structure.At the surface terraces, pentacene molecules self-organize into planar molecular stacks for coverages below one monolayer.The anatase surface is passivated by these molecular stacks that compose the first pentacene layer, so that the second pentacene layer adopts a high-density packing of vertically oriented molecules, promoting a thin film growth that is compatible with the bulk structure of the pentacene organic crystal.
Among the metal oxides typically used as buffer layers in electronic heterostructure devices, TiO 2 stands out as the only substrate known to date that binds pentacene in a horizontal geometry at the organic/inorganic interface.Substantial molecule-surface dispersion interactions stabilize the physisorbed adsorption configuration further above the surface, where electronic structure coupling between the pentacene and anatase is limited.Adsorbed pentacene introduced sharp molecular states into the band gap of the substrate, but it is the molecular LUMO that supplied the direct orbital overlap with the substrate necessary for charge transfer.The position of the pentacene LUMO in the conduction band of anatase implies barrier-free injection of electron charge carriers from the organic molecules into the substrate.Higher levels of theory are needed to accurately compute the alignment of electronic levels between the pentacene thin films and the TiO 2 substrate.Nevertheless, the localized charge injection pathways and the favorable level alignment at this interface are indicators that these materials would contribute efficient thin film devices for photovoltaic applications.

Methods
A UNISOKU Ltd. ultrahigh vacuum (UHV) cryogenic dynamic AFM equipped with a home-built optical interferometer for the detection of the cantilever dynamics was used for the experiments.Measurements were accomplished at a 77 K probe-sample tempera-ture using the frequency modulation detection scheme. 46Platinum iridium coated silicon cantilevers (PPP-NCLPt-20, Nanosensors, Switzerland) were excited to their instantaneous first mechanical resonant frequency keeping the oscillation amplitude (A) constant.The tunnelling current flowing between probe and surface averaged over multiple cantilever oscillation cycles 47 (<I t >) was simultaneously detected during topographic AFM imaging using a commercial scanning probe microscopy controller (Nanonis SPM Control System, SPECS, Germany).The shift of the first mechanical resonant frequency (∆f) from the free-oscillation value upon forces acting on the cantilever probe was used to regulate the probe-surface separation.Sub-molecular resolution AFM imaging was performed using a multi-pass method. 17 Force spectroscopy 48,49 was performed by recording ∆f as a function of the probe-sample relative vertical displacement (Z) over a rectangular matrix of pixels superimposed to a previously imaged area of the surface, closing the feedback loop upon displacement of the probe between adjacent pixels. 17The total probe-surface interaction force was obtained by applying an inversion procedure 50 to the recorded ∆f (Z) curves.Topographic effects due to the acquisition procedure were properly compensated. 17,51Further details about protocols for in situ probe conditioning and anatase sample preparation can be found elsewhere. 20Molecular deposition was carried out by exposing a clean anatase surface to pentacene molecules sublimated from a homebuilt evaporation cell upon reaching 160°C in UHV.The substrate was kept at room temperate during the pentacene deposition, and coverage was regulated by adjusting the exposure time.
All DFT simulations were carried out with the VASP code, 52 the PBE exchange-correlation functional 53 and semi-empirical DFT-D2 dispersion corrections. 54We employed PBE PAW potentials and a planewave cutoff of 500eV.An onsite U=4eV was applied to the substrate Ti atoms to adjust the position of gap states with respect to the conduction band minimum of TiO 2 , following other studies of anatase surfaces. 55,56e employed the TiO 2 anatase (101) surface model previously used for adsorption studies on this surface. 20The 10.57 Å × 23.12 Å × 40 Å model surface area accommodates two sloped surface grooves for pentacene attachment.Single pentacene images were laterally separated by dx=6 Å and dy=9 Å between different simulation cells, while adjacent pentacene stack images were separated by dy=7.4Å.The tall unit cell allowed 24 Å between surface slabs in adsorption simulations, and 13.8 Å between them in AFM simulations.The same unit cells was used for isolated pentacene and surface reference calculations.Geometry optimizations were performed with a gamma-point k-point grid until atomic forces were reduced below 0.01eV/ Å .Such k-space sampling was selected by downscaling the converged (6×6×2) kpoint setting for bulk unit cell calculations of TiO 2 anatase when upscaling the unit cell towards the large (1x6) unit area surface geometry.Adsorption energies and electronic properties were obtained using a 4×2×1 k-point mesh.

Figure 1 :
Figure 1: (a) Topographic AFM image of an area of (101) TiO 2 anatase partially covered with pentacene molecules(20 × 20 nm 2 ).Crystallographic orientation is the same in all images.The trapezoid highlights the contour of an anatase island.(b) Intramolecular resolution AFM image simultaneously acquired with the topography displayed in (a) showing details of the adsorption of pentacene molecules at the terraces and steps of the anatase surface (20 × 20 nm 2 ).The arrows point to pentacene molecules adsorbed at the steps in a sideon configuration.(c) Details of the adsorption of pentacene molecules (blue arrows) on the energetically most stable step of this anatase surface, connecting both upper and lower terrace (8.0 × 8.0 nm 2 ).The line indicate the position of the step edge as characterized from the topography.(d) Details of a stack of pentacene molecules simultaneously resolving atomic features of the (101) anatase surface (4.7 × 4.7 nm 2 ).(e) and (f) Frequency shift images of single pentacene molecules obtained with two different probe terminations (3.0 × 3.0 nm 2 ).Acquisition parameters are: free-oscillation first longitudinal resonant frequency (fo) of the cantilever was 152471 Hz for (a) to (d), and 159202 Hz for (e) and (f); cantilever oscillation amplitude (A) was 95.9 Å for (a) to (d), 169.1 Å for (e), and 135.3 Å for (f); cantilever stiffness (K) was 23.7 N/m for (a) to (d), and 26.9 N/m for (e) and (f); quality factor of the cantilever (Q) was 44163 for (a) to (d), 49188 for (e) and 44246 for (f).Topographic set point (∆f) was -3.0 Hz for (a) to (c), -4.0 Hz for (d), -4.5 Hz for (e) and -14.2 Hz for (f); approach distance towards the surface for sub-molecular imaging (d) was 300 pm for (b), (c), and (e), 250 pm for (d), and 340 pm for (f).Sample bias was -200 mV for (a) to (d), 100mV for (e), and 200 mV for (f).

Figure 2 :
Figure 2: (a) Pentacene adsorption registry A, with the center above a Ti 5c atom.(b) Pentacene adsorption registry B, with the center of the molecule above the mid-point between Ti 5c atoms.(c) Side view of a physisorbed pentacene molecule, indicating adsorption height d and angle Θ.(d) Side view comparison between the geometries of adsorbed pentacene stack (purple) and isolated molecule (yellow).(e) Pentacene mixed registry A-B stacking (straight stack).(f) Pentacene single registry B-B stacking (offset stack).
strate and compared it to energy of gas phase pentacene molecules.This also facilitated the decomposition of the adsorption energy E BB ADS =-2.19eV into intra-stack and stack-surface contributions.We established that molecule-molecule interactions contribute 16% (-0.36eV) and the molecule-surface interactions account for 84% (-1.83eV) of the adsorption energy per molecule in a stack configuration.While the presence of other molecules in the stack appears to reduce the strength of pentacene binding to TiO 2 from -1.91eV to -1.83eV per molecule, the monolayers are overwhelmingly stabilized by molecule-surface interactions (in line with recent experimental observations for conjugated adsorbates on (101) anatase 23 ).

Figure 3 :
Figure 3: (a) Topography of an isolated pentacene molecule on the surface, over which a bias spectroscopy imaging measurement based on the acquisition of a ∆f(V) curve over each pixel of the image was obtained (image size 3 × 3 nm 2 ).(b) ∆f signal corresponding to a slide of all the measured ∆f(V) curves at a sample bias of V=500mV.(c) <I t > signal associated with the image shown in (b).For additional experimental details see Fig. S1.(d) Partial DOS contributions of the TiO 2 (101) surface and adsorbed pentacene (P5), compared with DOS of pentacene in the gas phase.Gas phase DOS was aligned with the total system DOS at the Fermi level.(e) System HOMO spatial distribution (10 −4 e-/ Å3 isosurface), obtained by integrating total DOS in the energy interval [0, -600] meV.(f) System LUMO spatial distribution, (10 −4 e-/ Å3 isosurface), obtained by integrating total DOS in the energy interval [650, 1000] meV (note that image illustrates both the molecular LUMO and substrate conduction band states).

Figure 4 :
Figure 4: (a) Force curves extracted from a force volume composed by (24×24) ∆f(Z) curves measured over an area of (3x3) nm 2 on a stack of pentacene molecules deposited on the TiO 2 (101) anatase surface.Acquisition parameters are: f o = 152471 Hz, A = 95.9Å, K = 23.7 N/m, Q = 44163.Sample bias was -200 mV.The inset image is a slide of the total force values at the probe-sample separation indicated by the doted vertical line in (a).(b) Details of the short-range forces of two curves shown in (a) obtained close to a CH group site and a hollow site.The dots denote the force points after conversion from the ∆f(Z) curves, and the lines correspond to a spline interpolation of the experimental points.(c) Simulated force curves over a carbon atom and a hollow site -see positions in the model depicted in (b)-of an individual molecule (c-i, h-i; dashed lines) and a molecule within a B-B stack configuration (h-s, c-s), respectively.These forces were calculated using the OH group terminated TiO 2 cluster depicted in (c) as probe model.(d) Comparison of the forces simulated using the TiO 2 cluster and a naphthalene molecule as probe model.A molecular probe produces short-range forces of a magnitude comparable to the experimental counterparts.

Fig. 5 (Figure 5 :
Fig.5(a) displays a surface region where these two layers coexist.A distinct boundary -an abrupt step approximately 1.45 nm high (Fig.5(b)) crossing the image from the upper-left to the lower-right side-separates these two well-differentiated regions.AFM images reveal that the domain with lower topographic contrast corresponds to a phase of pentacene adsorbed into planar molecular stacks that covers the anatase surface almost in full, as depicted in Fig.5(c).Similar AFM images taken on the region showing higher topographic contrast present well ordered periodic features showing a rectangular-like pattern, with a

Table 1 :
Comparison of adsorption structures and energies for single pentacene in configurations A and B.Here d and Θ are the molecule adsorption height and tilt angle, as illustrated in Fig.2(c).We computed the total adsorption energy E ADS and its dispersion contribution E VDW ADS (fractional proportion shown in italic brackets).
Config.d [nm] Θ [ • ] E ADS [eV] E VDW ADS [eV] (%) 38Unlike metals, metal oxides present relatively inert substrates for pentacene attachment.Studies on SiO,38ITO, 39 LaSrMnO 3 , 40 Al 2 O 3 41 and HfO 2 42 surfaces all found that molecules adsorb in standing or tilted configurations, at odds with our findings. Dedritic aggregation and monolayer growth has been well characterised on SiO 43 and Al 2 O 3 , in contrast to the planar stacking growth mechanism observed for the first layer of pentacene on TiO 2 (101) anatase.