Abstract
The tip of a scanning tunnelling microscope is an atomicscale source of electrons and holes. As the injected charge spreads out, it can induce adsorbed molecules to react. By comparing largescale ‘before’ and ‘after’ images of an adsorbate covered surface, the spatial extent of the nonlocal manipulation is revealed. Here, we measure the nonlocal manipulation of toluene molecules on the Si(111)7 × 7 surface at room temperature. Both the range and probability of nonlocal manipulation have a voltage dependence. A region within 5–15 nm of the injection site shows a marked reduction in manipulation. We propose that this region marks the extent of the initial coherent (that is, ballistic) timedependent evolution of the injected charge carrier. Using scanning tunnelling spectroscopy, we develop a model of this timedependent expansion of the initially localized hole wavepacket within a particular surface state and deduce a quantum coherence (ballistic) lifetime of ∼10 fs.
Introduction
During imaging, the scanning tunnelling microscope (STM) injects charge into a surface on average every 1 nanosecond. Less than 1 picosecond after injection a charge carrier will have undergone a hierarchy of scattering events leading to eventual thermalization with the surface. In this interval, a charge carrier may transfer energy to individual atoms and molecules on the surface and induce, for example, bond breaking^{1} or atomic desorption^{2}. Thus, atomic manipulation is the signature of the underlying dynamics of the charge carriers. Conventional atomic manipulation is restricted to the tunnel junction and the timescale of the charge dynamics has to be inferred from the complementary calculations^{3}. Conversely, nonlocal manipulation, in which molecules located at some tens of nanometres^{4,5,6,7,8} away from the tunnelling site respond to charge injection, can expose the ∼200 femtosecond (fs) diffusive dynamics of injected electrons^{9}.
Here, we employ the nonlocal manipulation technique at room temperature to probe the ultrafast ballistic dynamics of the injected charge carrier, that is, while it coherently evolves from its original quantum state. We find a coherence length scale of up to 15 nm, and deduce an associated coherence lifetime of ∼10 fs. Recent advances in the ability to construct atomicscale structures by conventional atomic manipulation have opened the way to atomic architectures that exhibit engineered quantum effects on metals^{10,11}, semiconductors^{12,13} and even graphene^{14}. Our results indicate the possibility of harnessing such quantum behaviour well above normal 4 K operating temperatures, possibly reaching the more practical room temperature regime.
Results
Nonlocal manipulation of toluene molecules with holes
Figure 1 shows three pairs of 30 × 30 nm STM images taken before and after the injection of holes (negative sample bias) at the location marked ‘X’ at voltages (A,D) −1.6 V, (B,E) −1.9 V and (C,F) −2.2 V. The bright spots are silicon adatoms of the Si(111)7 × 7 surface and the darkspots indicate chemisorbed toluene molecules^{15}. Before hole injection, there is a random distribution of toluene molecules (see Supplementary Fig. 1 for fullscale 50 × 50 nm version of Fig. 1a,d), while after injection, there is a markedly lower toluene (darkspot) coverage surrounding the injection site. We quantify this nonlocal desorption (manipulation) behaviour by identifying the crystallographic location^{16} relative to the injection site of each molecule before and after the injection. Figure 1g–i show, for each voltage, the radial dependence of the ratio of molecules that are manipulated (that is, leave their initial adsorption site), N(r), to the total number of molecules that were originally at that radial distance, N_{0}(r). Here, we inject into unfaultedmiddle (UM) sites of the Si(111)7 × 7 surface (Fig. 1j,k) and report the behaviour of molecules initially bonded to crystallographically equivalent UM sites (other sites produce similar behaviour—see Supplementary Figs 3–5 and 7–9).
‘Suppression’ of manipulation close to injection site
At radial distances >15 nm, the radial distributions are well described by a onehole manipulation process coupled to a 2 dimensional (2D) diffusive chargetransport model with a single decay channel, see Supplementary Fig. 2 (ref. 9). (The manipulation of the molecule, that is, induced movement away from the original binding site, could in principle be due either to a phonon emission during this decay or a nonadiabatic decay channel^{17}). The diffusive model parameters are β, the probability per generated hole of inducing a manipulation event, and λ, the range of the diffusive transport. The bluedashed line in Fig. 1g–i show fits of this model to the data, and Fig. 2a,b show the bias dependence for these parameters. The principal result of this work is that at distances less than 15 nm there is a strongly voltagedependent ‘suppression’ region R, where the probability of manipulation is far below the predicted value. Furthermore, the probability β of inducing a manipulation event increases monotonically with increasing negative voltage (Fig. 2a), whereas the range R of the suppression region has a sawtooth behaviour and returns to a minimum value at a voltage of −1.5 V. The same voltage marks a transition between near constant values for the diffusion length scale. (Similar behaviour is found for the other molecular binding sites of the Si(111)7 × 7 surface unit cell, Supplementary Figs 3–5 and there is evidence of a near identical voltage discontinuity for STMinduced nonlocal desorption of chlorine atoms from the Si(111)7 × 7 surface by holes^{18}).
Connection between surface states and transport regions
We associate the plateau region of the diffusion length λ with diffusive charge transport through a particular prominent surface state^{9,19}, and identify the threshold voltage corresponding to each plateau region with the onset of particular surface states. This assignment is supported by scanning tunnelling spectroscopy (STS) measurements of UM silicon adatoms (the injection site) shown in Fig. 2d. The nonlocal threshold at −1.2 V corresponds to the onset of the S_{3} backbond surface state^{20}, which peaks at −1.6 V in Fig. 2d, and the second threshold at −1.6 V corresponds to the onset of a state that peaks at −2.2 V^{18}, which for consistency we label S_{4}. Thus, the nonlocal manipulation thresholds are matched with the onsets of particular surface states, not their peakpositions. We therefore label the two chargetransport regimes as between −1.2 and −1.5 V and for manipulation between −1.5 and −2.3 V. We find analogous behaviour for the other adsorption sites (Supplementary Figs 3–5).
Discussion
The discontinuity in the range of the suppression region R with increasing voltage immediately rules out an electricfield effect as the cause of the suppression, since the Efield strength monotonically increases with injection voltage (Supplementary Fig. 6B). It also eliminates Efield induced danglingbond charging as a possible ‘suppression’ mechanism^{21}. Furthermore, to maintain a constant current as the voltage increases the, STM tip withdraws from the surface (Supplementary Fig. 6A), and so any mechanical, shortrange tipmolecule interaction would be reduced (not enhanced) as the voltage increased.
Instead, here we propose that the relative suppression of the molecular manipulation probability close to the STM tip reflects a (counterintuitive) reduction of the number of charge carriers that interact, either directly through a shortlived ionic state or indirectly as the carrier decays via phonon emission, with the adsorbed molecule at distances close to the injection site: after injection, the probability distribution undergoes rapid ballistic expansion, or ‘inflation’, followed by relatively slow 2D diffusion. The diffusion is therefore initiated at a radius that is distant from the injection site. The solid radial curves in Fig. 1g–i correspond to such a twostep (ballisticinflation followed by diffusion) model, using the initial conditions for diffusion an annulus of width 2 nm centred on the injection site at a radius R, the range of the suppression region. The excellent fit to the measured data across all lengthscales is evident.
Charge injected from the tip of an STM undergoes a series of processes eventually leading to thermalization with the bulk crystal. These follow a hierarchy of time scales: elastic momentum scattering (that is, directional scattering); inelastic scattering within an electronic state to form a quasiequilibrated distribution (it is during this phase that the nonlocal diffusive transport occurs); inelastic scattering out of the state towards lower lying states where the charge has insufficient energy to induce manipulation (this is the decay channel of the diffusive model). Since the twostep inflationdiffusion model accounts so well for the experimental observation, we conclude that the inflation region is the result of the charge dynamics before quasiequilibration. Furthermore, since any incoherent scattering, whether elastic or inelastic, would lead to diffusive behaviour, the inflation region must correspond to the dynamics of the injected charge before it undergoes any scattering, while it remains a coherently expanding quantum wavepacket. That is, the inflation region is determined by the positions that the injected charges undergo their first scattering event.
We develop a simple model to connect the range of the inflation region with the bandstructure of the surface and hence access the timescale of the coherentinflation process. In line with their prominence in our STS measurement, the inflation process is modelled on the initial occupation (by tunnelling) of an electronic surface state. The state is modelled as an azimuthally isotropic 2D state with a tightbinding dispersion . Figure 3 shows the model for the S_{3} state, with the dispersion shown in Fig. 3a. At a particular injection voltage, we compute the probability of tunnelling with a particular wavevector in the surface plane k_{} (Fig. 3b) and use these values to construct a timedependent cylindrically symmetric wavepacket. Figure 3c shows the radial probability distribution of the initial wavepacket. Figure 3d shows the time evolution of the wavepacket until the end of the inflation time at 10 fs, and Fig. 3e shows the radial extent of the wavepacket at the end of the inflation period. The spread of the wavepacket from the injection site is evident. This radial distribution is used as the starting point for diffusion (Fig. 3f) with an exponential decay (Fig. 3g).
The timeintegrated radial distribution provides the total number of charges that can interact with a molecule at a particular distance. Figure 4 shows a series of these curves fitted to voltagedependent experimental data. Within a transport region each curve has (i) the same diffusive parameters derived from Fig. 2 ( λ=(9.0±0.2) nm (note all errors are s.e.m.), λ=(18.0±0.3) nm); (ii) the same fitting parameter for the inflation time, and (iii) a voltagedependent amplitude. The excellent global fits of the inflationdiffusion model give inflation times for =(11±1) fs and =(9±1) fs. We also find good global fits for the other three possible molecular binding sites and good correlations with their STS measurements (Supplementary Figs 7–9). We conclude that the nonlocal STM manipulation measurements capture the initial coherent dynamics of the injected charge carriers.
The inflation time represents the coherent or ballistic transport regime. We find no direct comparison in the literature, but the relaxation time of electrons within the dangling bond U_{1} state has been measured to be ∼40 fs^{22}. This is longer than our ∼10 fs timescale and, may be due to the U_{1} state lying within the bulk bandgap unlike our higherlying states. However, recent calculations for holes in bulk silicon giving relaxation times of ∼10 fs^{23} have demonstrated the sensitivity to both the identity of the band that the electronic excitation resides in, and to the energy of the electronic excitation within the band. We also note, again for bulk states, that the mean free paths computed for holes in Si are in the tens of nanometres range, mirroring our surfacesensitive measurements.
The STM injects charges with a range of energies, from the bias voltage level to the Fermi level. Figure 3b shows this relationship in terms of parallel momenta. Our experiments are therefore a weighted average over a range of energies and hence, according to Bernardi et al.^{23}, a range of coherent lifetimes. In our simple scheme, we fit a single inflation (coherence) lifetime to each surface state. The sitetosite variation of the lifetimes we obtain may reflect the spatial distribution of the particular surface states within a Si(111)7 × 7 unit cell giving rise to differing sitetosite coherent propagation. Phonon amplitudes will also vary within the large 7 × 7 unit cell^{24}, so that ephonon relaxation rates can be expected to be sitespecific.
The coherentinflation time itself most likely has two main components, a temperatureindependent coupling to bulk states and a temperaturedependent phonon scattering. Our earlier work with temperaturedependent electron injection found no obvious increase in the inflation region at lower temperatures, suggesting that the measured inflation region is an intrinsic property of the surface state. It therefore offers the possibility (as in the case of C_{60} (ref. 25) and other molecular overlayers^{26}) of designing a system with reduced coupling to bulk states and hence an increased quantum coherentinflation range, thereby extending the range over which a possible quantum device could work.
Methods
Sample preparation and imaging
All experiments were performed at room temperature with an Omicron STM1 in an ultrahigh vacuum chamber at base pressure 1 × 10^{−10} mbar. Tungsten tips were electrochemically etched in a 2 M NaOH solution and outgassed by resistive heating in high vacuum. Precut silicon samples from an ntype phosphorus doped Si(111) wafer (0.001–0.002 Ω cm) were prepared using an automated flashing routine^{19}. Toluene was purified by the freezepumpthaw technique with liquid nitrogen and checked for purity with a quadrupole mass spectrometer. Typically, the surface was saturated with a toluene dose of 4 Langmuirs, corresponding to a coverage of 1,500–2,000 molecules in a 50 × 50 nm image. All images were obtained in constant current mode with passive scanning conditions (+1 V and 100 pA)^{16}.
Nonlocal injections
Experiments were performed with a Nanonis SPM control system programmed to inject into userselected atomic locations. Stability during the injections was provided by inhouse developed drifttracking software, and a featurelocking technique employed to ensure the correct injection position. The individual locations of all molecules adsorbed to the surface before and after an injection experiment with respect to the injection site were determined using an inhouse computer analysis suite^{16}. This resulted in a molecule identification accuracy of more than 99%. The results were corrected for thermally induced desorption using the technique outlined in ref. 16. To ensure good statistical results, the total injected charge (number of holes) was varied so that approximately half the molecules at half the distance to the edge of the full image (50 × 50 nm) were manipulated.
The Si(111)7 × 7 surface presents four distinct adatoms that a toluene molecule can bind to: faulted corner (FC), faulted middle (FM), unfaulted corner (UC) and unfaulted middle (UM) (Fig. 1j,k). Each site has slightly different electronic properties^{27} and binding energies for a chemisorbed toluene molecule^{28}. To ensure a consistent experimental scenario here, we only inject into clean UM adatom sites (that is, with no adsorbate molecule) and report in the main text the manipulation of toluene molecules initially bonded to UM sites. Other injection and adsorption sites produce qualitatively similar results (Supplementary Figs 3–5 and 6–9), but the trends in the data are clearest for UM injection and adsorption sites. To further ensure the robust nature of the results, we use an automated experimental system allowing many repeats at each set of injection parameters.
2D diffusion model
As derived in ref. 9 the mathematical model for 2D diffusion with a single decay channel gives
where n_{i} is the number of generated holes, D is the diffusion constant of the charge carriers and the combined term β gives the probability (a crosssection per unit time) that an injected hole will induce a manipulation event. Finally K_{0} is a modified Bessel function of the second kind with argument that includes the parameter λ, which gives the range of the diffusive transport.
Determination of suppression region R
To accurately determine the size of the suppression region, R, the 2D diffusion model was fitted to the data from the largest radial distance to a smaller distance r_{0}. By comparing the coefficient of determination and the fitted value of λ as a function of r_{0}, we could identify the minimum value of r_{0} that gave a coefficient of determination value >0.9, which was the smallest value of r_{0} that had a near constant λ fitting parameter.
Scanning tunnelling spectroscopy
Variable gap dI/dV spectra of individual Si adatoms were obtained directly with a lockin amplifier in the region from 0 to −2.5 V from an initial stabilizing voltage of +1 V at 100 pA. The tunnelling gap was reduced by 25 pm V^{−1} to amplify the signal at low bias within the bandgap region. A typical spectrum contained 200 data points each acquired over 150 ms with a lockin modulation of 20 mV at 521 Hz. The STS curves presented here are the average of 38 such measurements.
Inflation model: 2D surface state
For each molecular adsorption site we find sitespecific nonlocal manipulation and STS properties, reflecting sitespecific coupling to the specific chargetransport surface state. We therefore determine sitespecific energy onsets and the bandwidths. We model the 2D surface state, which mediates the coherent expansion as a cosine tightbinding like electronic state with dispersion relation E=E_{0}+ΔE/2[cos(ka)−1], where E_{0} is the bandedge, ΔE is the bandwidth and a is the Si(111)7 × 7 unit cell length of 2.68 nm. The energy onset E_{0} of the state is given by the thresholds found for the various transport regions identified in the diffusion lengthscale λ, see for example Fig. 2b, giving for UM injections and UM molecular manipulations for the S_{3} state an onset of −1.15±0.1 V and for the S_{4} state −1.55±0.1 V; for UC molecules, −1.15 and −1.55 eV; for FM molecules, −1.15 and −1.35 eV; and for FC molecules −1.15, −1.65 eV and a third transport region starting at −2.15 eV. The S_{3} onsets match with that reported by photoemission for the S_{3} state and the fact that a section of the ARUPS measured results lie within the projected bulk bandgap is strong evidence for a surface state of the Si(111)7 × 7 unit cell^{29,30}. The S_{3} state has been found to have downward dispersion^{29,31} and be coupled to p_{x} and p_{y} orbitals confirming the downwards dispersion.
To determine a statespecific bandwidth ΔE we use the width of the peaks found on our sitespecific STS measurements averaged across all the sites giving a mean peak full width at half maximum (FWHM) of 0.49 eV for S_{3} and 0.76 eV for S_{4}. In ref. 29 a ∼0.35 eV dispersion between the point and half way to the point is reported, giving a full bandwidth of ∼0.7 eV. Hence, our measured STS FWHM of 0.49 eV corresponds to a bandwidth of 0.7 eV for S_{3} and a bandwidth of 1.1 eV for S_{4}. We introduce a Lorentz broadening with a half width at half maximum σ=ħ/ with the inflation lifetime. The resulting energy and momentum spectral density n(E, k_{}) is given by
and shown in Fig. 3a for the S_{3} state.
Tunnelling probability
Charge carriers tunnel across the STM junction with a probability that depends on their energy and lateral momentum k_{}. The standard transmission coefficient for tunnelling^{32} is
where we take the barrier height E_{b} to be the mean height of the barrier of the tunnel junction and assume a vacuum level E_{v} at 4.6 eV for both sides giving E_{b}=E_{v}+E_{i}/2, where E_{i} is the injection energy.
Wavepacket construction
The solutions to the 2D cylindrical Schrödinger equation are the Bessel functions. We construct a timedependent wavepacket from a weighted sum over kspace,
where A is an overall constant of normalization, is the normalization constant for each k_{} state and is the relative amplitude of the particular k_{} reflecting the probability of tunnelling into it:
where E_{F} is the Fermi level in the sample and E_{i} the injection energy (STM bias voltage).
We use the Bessel functions in equation (4) to describe the surface band eigenfunctions, since detailed knowledge of the actual wave functions is not available. Strictly speaking, the Bessel functions are only applicable to states exhibiting freeelectron like dispersion, which is only the case towards the edges of the surface bands here. To assess the likely impact of this we have also considered inflation of the initial wavepacket within the opposing limit, solving on a hexagonal lattice the timedependent Schrödinger equation iħ∂/∂t=H assuming the tightbinding Hamiltonian,
corresponding to one orbital i〉 per Si(111)7 × 7 unit cell, nearestneighbour hopping, and with , γ appropriate to the known dispersion of the surface band. Supplementary Fig. 10 shows the radial evolution predicted in the two cases, and we see both yield comparable expansion properties, with the state density advancing at similar rates, and being relatively uniformly distributed behind the front. This demonstrates the inflation is relatively insensitive to the precise nature of the assumed surface band wave functions.
Radial diffusion and decay
Numerical calculation use an evenly spaced grid of radii up to 300 nm in Δr=1 nm steps and 400 timesteps Δt to a time 2. We use a reduced time t′=t/, Δt′=Δt/ and a reduced distance =r_{i}/λ and Δr′=Δr/λ to write the standard iterative scheme including an exponential decay term to compute the radial probability distribution P(, ) of the diffusing charge
The shape of the timeintegrated radial probability density, used below to fit to the experimental data, is independent of the choice of the lifetime . The lifetime does affect the overall computed probability per hole of inducing manipulation β. In this work we focus on the shape and range of the nonlocal effect not the absolute probability per hole. Nevertheless, to give consistent and realistic values we use =200 fs in line with the results found for nonlocal electron manipulation^{9}.
Fit to experimental data
Experimentally we measure the number of molecules that leave their original location N(r) due to an injection of n_{i} holes and the original number of molecules N_{0}(r) within an annulus at radius r. Hence,
See Lock et al.^{9} for full derivation. Thus, the only fitting parameters in the twostep inflationdiffusion model are (i) the inflation lifetime and (ii) β, the probability per hole of inducing manipulation. We justify the use of a single inflation time for all voltages within a transport region, since the energy relaxation time for hot holes in bulk silicon is near constant for holes with at least 0.1 eV of excess energy above the band edge^{23}, and for electrons in the U_{1} band of the Si(111)7 × 7 surface with 0.1–0.3 eV of excess energy^{22}.
Data availability
All data supporting this study are openly available from the University of Bath data archive at http://doi.org/10.15125/BATH00230.
Additional information
How to cite this article: Rusimova, K. R. et al. Initiating and imaging the coherent surface dynamics of charge carriers in real space. Nat. Commun. 7:12839 doi: 10.1038/ncomms12839 (2016).
References
 1
Sloan, P. A. & Palmer, R. E. Twoelectron dissociation of single molecules by atomic manipulation at room temperature. Nature 434, 367–371 (2005).
 2
Minato, T. et al. Tunneling desorption of single hydrogen on the surface of titanium dioxide. ACS Nano 9, 6837–6842 (2015).
 3
Morgenstern, K., Lorente, N. & Rieder, K.H. Controlled manipulation of single atoms and small molecules using the scanning tunnelling microscope. Phys. Status Solidi B 250, 1671–1751 (2013).
 4
Maksymovych, P., Dougherty, D. B., Zhu, X.Y. & Yates, J. T. Nonlocal dissociative chemistry of adsorbed molecules induced by localized electron injection into metal surfaces. Phys. Rev. Lett. 99, 016101 (2007).
 5
Gawronski, H., Carrasco, J., Michaelides, A. & Morgenstern, K. Manipulation and control of hydrogen bond dynamics in absorbed ice nanoclusters. Phys. Rev. Lett. 101, 136102 (2008).
 6
Hahn, J. R., Jang, S. H., Kim, K. W. & Son, S. B. Hot carrierselective chemical reactions on Ag(110). J. Chem. Phys. 139, 074707 (2013).
 7
Ladenthin, J. N. et al. Hot carrierinduced tautomerization within a single porphycene molecule on Cu(111). ACS Nano 9, 7287–7295 (2015).
 8
Schendel, V. et al. Remotely controlled isomer selective molecular switching. Nano Lett. 16, 93–97 (2016).
 9
Lock, D., Rusimova, K. R., Pan, T., Palmer, R. E. & Sloan, P. A. Atomically resolved realspace imaging of hot electron dynamics. Nat. Commun. 6, 8365 (2015).
 10
Crommie, M., Lutz, C. & Eigler, D. Confinement of electrons to quantum corrals on a metalsurface. Science 262, 218–220 (1993).
 11
Nilius, N., Wallis, T. & Ho, W. Development of onedimensional band structure in artificial gold chains. Science 297, 1853–1856 (2002).
 12
Naydenov, B., Rungger, I., Mantega, M., Sanvito, S. & Boland, J. J. Singleatom based coherent quantum interference device structure. Nano Lett. 15, 2881–2886 (2015).
 13
Schofield, S. R. et al. Quantum engineering at the silicon surface using dangling bonds. Nat. Commun. 4, 1649 (2013).
 14
Wang, Y. et al. Observing atomic collapse resonances in artificial nuclei on graphene. Science 340, 734–737 (2013).
 15
Tomimoto, H. et al. Study of adsorption structure of benzene and toluene on Si(111)7 × 7 surfaces. Surf. Sci. 566, 664–670 (2004).
 16
Sakulsermsuk, S., Sloan, P. A., Theis, W. & Palmer, R. E. Calibrating thermal and scanning tunnelling microscope induced desorption and diffusion for the chemisorbed chlorobenzene/Si(111)7 × 7 system. J. Phys. Condens. Matter 22, 084002 (2010).
 17
Palmer, R. & Rous, P. Resonances in electronscattering by molecules on surfaces. Rev. Mod. Phys. 64, 383–440 (1992).
 18
Nakamura, Y., Mera, Y. & Maeda, K. Desorption of chlorine atoms on Si(111)(7 × 7) surfaces induced by hole injection from scanning tunneling microscope tips. Surf. Sci. 601, 2189–2193 (2007).
 19
Sloan, P. A., Sakulsermsuk, S. & Palmer, R. E. Nonlocal desorption of chlorobenzene molecules from the Si (111)(7 × 7) surface by charge injection from the tip of a scanning tunneling microscope: remote control of atomic manipulation. Phys. Rev. Lett. 105, 048301 (2010).
 20
Hamers, R., Tromp, R. & Demuth, J. Surface electronicstructure of Si(111)(7 × 7) resolved in real space. Phys. Rev. Lett. 56, 1972–1975 (1986).
 21
Teichmann, K. et al. Controlled charge switching on a single donor with a scanning tunneling microscope. Phys. Rev. Lett. 101, 076103 (2008).
 22
Mauerer, M., Shumay, I. L., Berthold, W. & Hofer, U. Ultrafast carrier dynamics in Si(111)7 × 7 dangling bonds probed by timeresolved secondharmonic generation and twophoton photoemission. Phys. Rev. B 73, 245305 (2006).
 23
Bernardi, M., VigilFowler, D., Lischner, J., Neaton, J. B. & Louie, S. G. Ab initio study of hot carriers in the first picosecond after sunlight absorption in silicon. Phys. Rev. Lett. 112, 257402 (2014).
 24
Liebhaber, M. et al. Surface phonons of the si(111)(7 × 7) reconstruction observed by raman spectroscopy. Phys. Rev. B 89, 045313 (2014).
 25
Nouchi, R., Masunari, K., Ohta, T., Kubozono, Y. & Iwasa, Y. Ring of c60 polymers formed by electron or hole injection from a scanning tunneling microscope tip. Phys. Rev. Lett. 97, 196101 (2006).
 26
Chen, L., Li, H. & Wee, A. T. S. Nonlocal chemical reactivity at organicmetal interfaces. ACS Nano 3, 3684–3690 (2009).
 27
Myslivecek, J. et al. Structure of the adatom electron band of the Si(111)7 × 7 surface. Phys. Rev. B 73, 161302 (2006).
 28
Lock, D., Sakulsermsuk, S., Palmer, R. E. & Sloan, P. A. Mapping the sitespecific potential energy landscape for chemisorbed and physisorbed aromatic molecules on the Si(111)7 × 7 surface by timelapse STM. J. Phys. Condens. Matter 27, 054003 (2015).
 29
Martensson, P., Ni, W.X., Hansson, G. V., Nicholls, J. M. & Reihl, B. Surface electronic structure of Si(111)7 × 7Ge and Si(111)5 × 5Ge studied with photoemission and inverse photoemission. Phys. Rev. B 36, 5974–5981 (1987).
 30
Uhrberg, R. I. G., Hansson, G. V., Nicholls, J. M., Persson, P. E. S. & Flodstrom, S. A. Photoemission study of the surface and bulk electronic structures of Si(111)7 × 7 and Si(111) × :Al. Phys. Rev. B 31, 3805–3810 (1985).
 31
Northrup, J. E. Origin of surface states on Si(111)(7 × 7). Phys. Rev. Lett. 57, 154–154 (1986).
 32
Wiesendanger, R. Scanning Probe Microscopy and Spectroscopy: Methods and Applications Cambridge University Press (1994).
Acknowledgements
P.A.S. gratefully acknowledges support from the EPSRC Grant EP/K00137X/1. K.R.R. was supported by a University of Bath studentship and D.L. by an EPSRC DTA studentship. R.E.P. is grateful to EPSRC for the award of a Fellowship.
Author information
Affiliations
Contributions
All authors have contributed significantly to this work. K.R.R. performed the nonlocal experiments and data analysis. N.B., P.H. and D.L. performed initial nonlocal experiments. N.B., P.H. and K.R.R. performed and analyzed the STS. S.C. provided theoretical support and performed timedependent tightbinding simulations. P.A.S. and R.E.P. conceived, designed and led the research.
Corresponding author
Correspondence to P. A. Sloan.
Ethics declarations
Competing interests
The authors declare no competing financial interests.
Supplementary information
Supplementary Information
Supplementary Figures 110 (PDF 1072 kb)
Rights and permissions
This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/
About this article
Cite this article
Rusimova, K., Bannister, N., Harrison, P. et al. Initiating and imaging the coherent surface dynamics of charge carriers in real space. Nat Commun 7, 12839 (2016). https://doi.org/10.1038/ncomms12839
Received:
Accepted:
Published:
Further reading

Remotely Controlling the Charge State of Oxygen Adatoms on a Rutile TiO2(110) Surface Using Atomic Force Microscopy
The Journal of Physical Chemistry C (2020)

The nanometre limits of ballistic and diffusive hothole mediated nonlocal molecular manipulation
Nanotechnology (2020)

Combining scanning tunneling microscope (STM) imaging and local manipulation to probe the high dose oxidation structure of the Si(111)7×7 surface
Nano Research (2020)

STMInduced Desorption and Lithographic Patterning of Cl–Si(100)(2 × 1)
The Journal of Physical Chemistry A (2019)

Common source of light emission and nonlocal molecular manipulation on the Si(111)−7 × 7 surface
Journal of Physics Communications (2019)
Comments
By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.