Abstract
Tunneling is the most fundamental quantum mechanical phenomenon with wideranging applications. Matter waves such as electrons in solids can tunnel through a onedimensional potential barrier, e.g. an insulating layer sandwiched between conductors. A general approach to control tunneling currents is to apply voltage across the barrier. Here, we form closed loops of tunneling barriers exposed to external optical control to manipulate ultrafast tunneling electrons. Eddy currents induced by incoming electromagnetic pulses project upon the ring, spatiotemporally changing the local potential. The total tunneling current which is determined by the sum of contributions from all the parts along the perimeter is critically dependent upon the symmetry of the loop and the polarization of the incident fields, enabling fullwave rectification of terahertz pulses. By introducing global geometry and local operation to currentdriven circuitry, our work provides a novel platform for ultrafast optoelectronics, macroscopic quantum phenomena, energy harvesting, and multifunctional quantum devices.
Introduction
Rectification of light, transformation of oscillating electric and magnetic fields to experimentally observable direct currents, is a key process for ultrafast science^{1,2,3,4} and energy harvesting^{5,6,7}. Owing to their intrinsic nonlinearity and ultrafast response, tunneling junctions have been widely used in rectification of electromagnetic waves^{4,6,7,8,9,10}. In general, a simple picture of onedimensional potential barrier modulated by light has been successful in describing rectification in tunnel junctions. To achieve efficient rectification from the oscillating light waves using the onedimensional junction, one needs some suitable asymmetry across the potential barrier. Several efforts have been made such as tailoring time traces of incident pulses^{11,12,13}, using electrodes with different workfunctions/geometries^{5,14}, or applying an additional DC bias across the barriers. Recent studies demonstrated precise control of rectified tunneling electrons in sharp nanotips or in metallic nanostructures by illuminating light waves^{15,16,17,18,19}. Subwavelength gap structures are exploited to confine incident waves for field enhancement across the onedimensional barrier^{8,20,21,22,23,24,25}, prerequisite for an efficient tunneling process which has a highly nonlinear character^{26}.
Here, we present a new concept of realizing ultrafast control of tunneling currents by adding a new dimension to the traditional picture of onedimensional tunneling junctions. We use lateral, ringshaped tunneling barriers, encasing a metallic island which is surrounded by a metallic plane (Fig. 1a). The total rectified current emerges as a consequence of the contour integration of the tunneling current flux along the perimeter of the loop. By combining nanometerscale tunneling with twodimensional macroscopic geometry, we achieve ultrafast fullwave rectification of electromagnetic waves in subpicosecond time scale that is visualized by femtosecond optical pulses.
Results
Terahertz (THz) tunneling in a ringshaped barrier
Suppose that an electromagnetic wave impinges on a perfect electric conductor film placed on the x–y plane; the incident magnetic field H_{inc} induces an eddy current per length K_{PEC} = \(\widehat {\mathbf{z}}\) × (2H_{inc}) on the film, which reflects back the incident light and blocks field smearing into the perfect conductor. For a realistic metal film, the amount of induced current is similar to the case of a perfect conductor but now the current per area, J(z), flows inside the film, whose behavior is characterized by the skin depth^{27}. We can simplify the expression of J(z) by introducing an effective surface current \({\mathbf{K}} = {\int}_0^h {\mathbf{J}} \left( z \right){\mathrm{d}}z\) where h is the metal thickness^{28}. Similar to the perfect conductor case, it can be expressed by K ~ \(\widehat {\mathbf{z}}\) × H_{0} where H_{0} (≈2H_{inc}) denotes the magnetic field outside the top metal surface (i.e., illuminating side). In the presence of subwavelength gaps perforated in the metal film, the induced surface current K charges the gap and applies a potential difference^{20,22,29}
across the tunnel junction positioned at an arc length l and n denotes the unit vector perpendicular to the contour, directed outward from the loop (Fig. 2a). As the gap width decreases down to the nanometre scale, the induced electric field in the gap is further enhanced by the induced charges of the opposite sides of the gap pulling each other, making enough potential gradient to drive nonnegligible tunneling current across the point junctions. Temporal response of the resulting current is straightforwardly determined by the time profile of the incoming electromagnetic field. However, if the tunnel junctions are adjoined together forming a closedloop, the resulting total current flowing through the loop, I(t), would be a sum over all the point junctions, expressed by the following contour integration,
where J_{t}(l, t) is the tunneling current flux determined by the electric potential V(l, t) and dA = \(\widehat {\mathbf{n}}\)hdl = hdl × \(\widehat {\mathbf{z}}\) and h are the area element vector and height of the wall surrounding the contour, respectively. Depending on the incident polarization, the vector relation (integrand of Eq. 2) determines the amplitude and direction of the local current. The contour integration is naturally affected by the loop symmetry. Therefore, the response of the integrated junctions under the rapidly oscillating field is fundamentally different from the response on each junction element, which introduces an entirely new degree of freedom for manipulating the tunneling current.
To realize the concept of the closedloop barrier and to measure directly the ultrafast currents under electromagnetic pulse excitations, we used two external light sources, a picosecond THz pulse and a femtosecond optical pulse, to trigger currents across the ringshaped nanogaps fabricated in a 100nmthick Au film on a 500 μm thick silicon or quartz substrate^{30}. To avoid the unwanted optical absorption by the silicon substrate, we used quartz substrate for optical experiments. The central metallic island is completely isolated from the surrounding metal film by a vertically aligned 2 or 4nmthick insulating layer (Fig. 1b, c). By intentionally breaking the inversion symmetry in the new loop geometry, we generate finite total currents via the illumination of THz pulses. The timeintegrated total tunneling currents through the barriers are measured directly by attaching electrical probes on the sample surface (Fig. 1d, see Methods for details). A THz polarizer is placed to control the direction of the surface current K.
Figure 2b describes the polarizationdependent behaviors of the instantaneous total tunneling current of a triangular loop under a THz field illumination. Electric potential along the contour shows an asymmetric distribution as a consequence of the lack of inversion symmetry inherent in the triangle geometry. It is interesting to note that the contour integration of the barrier potential affected by the external surface current sources always vanishes independent of the loop shape and loop orientation, automatically eliminating the Ohmic component (see Supplementary Note 1). However, the nonvanishing total current through the barrier naturally emerges for the triangle shape because of the nonlinearity in tunneling current vs. applied potential relation (see Eq. 3 in Methods) together with the triangle’s lack of inversion symmetry. Figure 2c shows the measured current responses from triangular and square loops as a function of the THz polarizer angle. The results show strikingly different behaviors depending on the loop geometry. A much higher current flows across the triangular barrier than the square one since the asymmetric potential distribution along the equilateral triangle results in a net tunneling current through the contour while the potential distribution at any point of a square is mostly counterbalanced by its corresponding point across the center, independent of the polarization of the incident pulse. Figure 2d displays polar plots of the tunneling currents for the triangular and square geometries, where the current amplitudes are reconstructed by assuming that the incident field maintains its amplitude for different polarization angles (see Methods for details). The total current vs. polarization angle shows the threefold rotational symmetry of an equilateral triangle.
Optical control of THz tunneling currents
Figure 3a depicts an ultrafast modulation of the THz tunneling current by optical pulses. In this scheme, electric fields of the femtosecond optical pulse and picosecond THz pulse are summed at a specific position in the contour. The added optical pulse rapidly distorts the local potential barrier under the quasiconstant THz field, generating an additional, local tunneling current. The resulting current amplitude and direction, driven by the sinusoidal optical field, are critically dependent on the background THz voltage across the gap. The optical current is sensitively affected by the THz field strength at the specific position on the barrier and by the time delay between the THz and optical pulses, thus providing a way to visualize the spatiotemporal dynamics of the THz gap voltage. An interesting aspect of the optically modulated quantum barriers is the positiondependent ultrafast optical gating, as illustrated in Fig. 3b. After the polarization of the THz pulse is intentionally set to generate zero current (φ_{pol} = 90°, see Fig. 2a for the notation), local modulation of the barrier by an optical pulse breaks the potential balance of the contour and generates ultrafast switching signals of opposite polarities depending on the side of the triangle shone by the optical pulse (see Supplementary Fig. 5 for the φ_{pol} = 60° case).
THz tunneling dynamics revealed by optical pulses
Under the THz and optical field illumination, the barrier potential is affected by both fields simultaneously. Tunneling current through the barrier is driven by the sum of the light fields, sensitively affected by their temporal field profiles. Figure 4a describes the situation when an optical pulse and a THz pulse are illuminated at the gap together, where they charge the gap and subsequently apply a potential difference across the barrier. The THz pulse profile illustrated in Fig. 4a is the timedependent potential difference across the gap, which is acquired by using the incident field H_{inc} (shown in Fig. 1d and Fig. 3a) and Eq. (1). Specifically, the timedependent voltage is a result of the surface charge accumulated at the gap by the surface current K induced from the incident field. By integrating the current pulse (K ~ \(\widehat {\mathbf{z}}\) × H_{0}, where H_{0} is the magnetic field just above the metal surface) over time, the applied voltage curve across the gap thus can be described by \(V\left( t \right) \propto {\int}_{  \infty }^t {H_{{\mathrm{inc}}}} \left( {t\prime } \right){\mathrm{d}}t\prime\) where H_{inc} is the incident magnetic field strength which is proportional to H_{0} ≈ 2H_{inc} (see Supplementary Fig. 1b, c).
If we compare the measured current under the simultaneous optical and THz excitation with the case of a THzonly excitation, the femtosecond optical field additionally applies a potential difference at a specific time together with the quasiconstant voltage applied by the picosecond THz field (Fig. 4b, see Supplementary Fig. 3 for details). Due to the strong tunneling nonlinearity (Eq. 3 in Methods), the almost sinusoidal oscillation of the optical field nevertheless drives a nonzero current when riding the quasistatic THz field. Figure 5a shows the measured tunneling current under both the THz and optical field illumination by varying the incident optical power. As one can see, the measured current time traces do not strictly follow the THz voltage across the gap (denoted as green dashed line) owing to the tunneling nonlinearity.
Most of the THz tunneling current (without optical illumination) flows only near the intensity maximum of the THz voltage across the gap for a single THz pulse. By analyzing the half widths of the zerodelay peaks of Fig. 5a, we can estimate the timescale of the THz tunneling current pulse. The inset of Fig. 5a shows the extracted half widths as a function of the incident optical power. The half widths become narrower almost linearly as we decrease the optical power; hence, we conclude that the extrapolated half width near zerooptical power (~0.2 ps) is the THz tunneling current pulse width for the given THz voltage pulse profile.
The optical method enables a quantitative analysis of the THz tunneling current vs. voltage relation. By dividing the rectified charge measured from the powerdependent THz tunneling current (Fig. 5b) by the tunneling current pulse width measured from the optical method (inset of Fig. 5a), the THz tunneling current can be quantified with the applied THz voltage across the gap (Fig. 5c, see Methods for details). The resulting THz tunneling current amounts to ~0.3 A, which corresponds to a net flow of ~4 × 10^{5} electrons within the 0.2 ps timescale given by a THz gate field that reaches up to ~3.3 V nm^{−1} or a voltage of ~6.5 V across these quantum barriers. Such ampere levels of tunneling currents driven by a THz pulse (~3 V nm^{−1}) and additional currents by an optical pulse (~8 V nm^{−1}) can be achieved without damaging the barrier during the ultrafast gating with THz and optical pulses^{11,12,31}. We confirmed a quantitative agreement between experiment and calculation neglecting thermal effects (see Fig. 5c or Supplementary Fig. 4).
THz rectification using ring barriers
A DC bias is another control parameter for manipulating the tunneling current of the ring barriers, enabling ultrafast rectification of electromagnetic waves. Figure 6a shows the total current as a function of the time delay between the two pulses under a DC bias. Here we generated a quasimonochromatic, multicycle THz pulse using spectral filtering, and the polarization is set to φ_{pol} = 180° to maximize the response from the left side of the triangle. By sending an optical pulse to the side, we observed the halfwave rectification of an incoming THz pulse as reflected in the local current. We note that the different noise levels shown in Fig. 6b are due to the unstable current flow at strong DC bias conditions. We observed that the current signals become noisy when the field strength applied by the DC bias under THz illumination reaches ~0.5–1 V nm^{−1} (2–4 V potential difference across the 4 nm gap used in obtaining the results shown in Fig. 6b). This threshold DC field strength also depends on the quality and thickness of the Al_{2}O_{3} film. Near this threshold field, the DC current starts to fluctuate and affect the THz current measurement, which makes the noise shown in Fig. 6b.
The integration of these halfwaverectifying barrier elements along the whole contour results in a fullwave rectification, as illustrated in Fig. 6c. The instantaneous total THz current can be directly visualized by increasing the optical spot size to cover the entire loop. If we set φ_{pol} = 90°, the potential differences across the upperright and lowerright sides of the triangle are equal in magnitude and opposite in sign. Now, if an additional DC bias is applied across the loop barrier, this symmetry is broken, resulting in a finite total THz current across the loop barrier independent of the polarity of the THz voltage pulse (Fig. 6d). An optical pulse illuminating both sides provides the instantaneous information on this total THz current (Fig. 6e). Thanks to the unidirectional (i.e., into or out of the loop) current response from all sides of the contour, determined by the DC bias, the THz wave generates a fullyrectified THz tunneling current. We note that if the loop is an ideal equilateral triangle and if the intensity of the optical pulse is constant over the entire contour, the total current across the barrier is zero if the DC bias is zero. But the existence of nonperfect barrier profiles and (or) the slight misspositioning of the optical pulse (different contribution between upperright and lowerright sides of the triangle in this case) would result in a nonzero current.
Discussion
By adjoining the onedimensional tunneling junctions in two dimensions, the tunneling current flowing across the quantum barriers is determined not only by the external electromagnetic pulse profile (e.g., carrierenvelopephase), but also by the geometry of the barrier (i.e., lateral symmetry of the ring). Using the proposed method, we can now control the extremely phase sensitive ultrafast tunneling current by modifying the lateral shape of the twodimensional barrier and by simply changing the polarization of incoming pulses. This concept profoundly widens our modulation technique of ultrafast nonlinear currents, and naturally leads to such unforeseen phenomena as ultrafast fullwave rectifications of THz pulses (see also Supplementary Fig. 6).
Together with the lateral shape of the ring, the optical technique presented in this work can directly reveal the spatiotemporal dynamics of the ultrafast tunneling phenomena. A previous study on THz control of optical photoemission from a nanotip^{17} showed that photoelectrons follow the time trace of the THz field applied at the tip, demonstrating a temporal control of photoemission by THz pulses. Our method exploits the small beam size of the optical pulses and allows the spatiotemporal control of THz tunneling currents by optical fields. This new combination method enables visualization of the spatiotemporal dynamics of the THz tunneling current in the ring barrier, positionsensitive optical gating of THz pulses, and quantification of THz tunneling timescale across the barrier. These newly developed techniques will have a deep impact on the research community working on ultrafast phenomena.
The peak current density driven by the THz pulse in our data of ~4.3 MA cm^{−2} at a field strength of ~3.3 V nm^{−1} is similar to the current densities and field strengths across onedimensional nanogap junctions used in previous studies. However, the amount of measured current in our experiment is much larger since we utilized the whole loop (the total loop area of ~7 × 10^{6} nm^{2} for a triangle barrier whose side length is 70 μm and height is 100 nm) compared with a single point tunnel junction (the total junction area of ~80 nm^{2} or less, such as STM tips or bowtieshaped nanogaps) used in most other studies.
In conclusion, we demonstrated a highly nonlinear lightmatter interaction, taking advantage of the twodimensional, lateral geometry of closedloop quantum barriers, whose lack of macroscopic inversion symmetry plays a vital role in the ultrafast optoelectronics. The ringshaped quantum barriers introduce a new control method for tunneling currents, further enriched by the femtosecond optical excitation at designated areas and time delays. By implementing the contourintegral concept and lateral symmetry into the conventional onedimensional tunneling, we realized a multifunctional quantum device, providing a platform for optical transistors, ultrahigh bandwidth communications and wireless energy conversion.
Methods
Sample preparation
The ringshaped nanogaps are prepared by atomic layer lithography technique^{30}. On a 500μmthick quartz (or low conductivity silicon) substrate, AZ5214 image reversal photoresist was spincoated at 4000 rpm for 60 s, then prebaked at 90 °C for 60 s. The resistcoated sample was exposed to UV light (350–450 nm wavelength, beam intensity about 20 mW cm^{−2}, MIDAS mask aligner) under photomask for 6 s. Then postbaking was performed at 120 °C for 120 s. After the second exposure to the same UV light for 40 s, samples were immersed in MIF500 developer solution for 30–60 s depending on the pattern size. After the deposition of 100nmthick Au layer by an ebeam evaporator and a subsequent liftoff process using acetone with a 1 min sonication, a 2 or 4nmthick alumina layer was coated by atomic layer deposition, which determines the nanogap size. After this, another Au layer of 100 nm was deposited directly onto the previous pattern. Finally, an adhesive tape was applied to the sample to planarize the surface. The side lengths of triangular and square patterns were varied from 10 to 500 μm depending on the experiments.
Generation and detection of THz waves
In this work, a broadband high power THz source is used, whose spectrum ranges from 0.1 to 3 THz (Supplementary Fig. 1). Singlecycle THz pulse is generated by a prismcut lithium niobate (LiNbO_{3}) crystal via pulsefronttilted optical rectification^{32}. Optical pulse (amplified 1 kHz Ti:sapphire laser: wavelength of 800 nm, pulse energy of 5.3 mJ, pulse width of 35 fs, Spitfire, SpectraPhysics) was divided by a 99:1 beam splitter for THz pulse generation (99%) and timeresolved tunneling measurements (1%). Generated THz beam was guided by a series of offaxis parabolic mirrors. The incident field strength was controlled by a pair of wire grid polarizers. For the polarization resolved measurements, we used a single polarizer. Transmitted THz field through the sample is detected in timedomain through electrooptic (EO) sampling^{33}; A 200μmthick ZnTe crystal and subsequent quarter wave plate with a pair of photodiodes probe the THz field. Supplementary Figure 1b, c show the EO sampling signals of the generated THz waveforms and their timeintegrated pulses (voltage across the gap), showing ~1 ps of singlecycle THz pulse (used in Figs. 1–5) and multicycle THz pulse (used in Fig. 6). Supplementary Figure 1d, e show the measured THz transmitted amplitude of triangle nanogaps, normalized by the substrate transmitted amplitude.
Calculation of tunneling current
We used a full integral expression of the Simmons formula^{34,35} to calculate tunneling currents and model the experimental data in this work. Tunneling current density J_{t}(V) across the onedimensional barrier can be expressed by
where m is the electron mass, e the electron charge, h the Planck constant, η the Fermi level of the metal, V the applied voltage across the barrier, E the tunneling electron energy and D is the electron tunneling probability factor based on the WKB approximation
and the mean value of the effective barrier height
where φ_{0} is the rectangular barrier potential height, s the thickness of the insulating layer, λ = e^{2}ln2/16πϵs (ϵ = dielectric constant of the barrier), and Δs = s_{2} − s_{1} is the effective barrier width where the limits s_{1} and s_{2} are given by the real roots of the cubic equation φ(x) = φ_{0} − eVx/s – 1.15λs/x(s −x) = 0.
Tunneling current measurements
Tunneling currents are measured by attaching electrical probes directly on the sample surface. To safely attach a metallic probe inside the pattern without damaging the sample surface, an electrochemically etched tungsten wire was used as the probe tip, with the tip radius of curvature of ~1 μm. One probe is connected to a Keithley 2450 sourcemeter to apply a DC bias and the other probe is connected to a current preamplifier followed by a lockin amplifier synchronized to the laser repetition rate (1 kHz). Both tips are positioned on the opposite side to the illuminating direction to avoid blocking the incident THz beam. All electrical apparatuses were on the same ground.
To check the validity of the tunneling process in our ring barrier structure, we performed currentvoltage (IV) measurement under for two different cases: application of a DC bias voltage and of THz pulses. Blackdots in Fig. 5c shows a DC IV measurement, demonstrating a general exponential behavior of a onedimensional tunneling junction. From the curve fitting process based on the Simmons formula, we extracted the barrier potential of the Al_{2}O_{3} layer used in our experiments (~2.2 eV) and layer thickness (~2 nm), the latter of which is also estimated from the TEM picture analysis (Fig. 1a). To compare the measured DC IV data and the THz measurements on an equal footing, we considered the DC current flowing through one side of the triangle. Thus we divided the DC current by three; assuming an equivalently distributed DC current through all three sides.
Next we measured tunneling current as a function of the incident THz field strength, as shown in Fig. 5b. A pair of wiregrid polarizers were used to vary the THz field strength with the incident polarization set to φ_{pol} = 180°. The resultant THz electric field applied across the gap (xaxis, inset of Fig. 5b) is estimated by the Kirchhoff integral formalism^{20,36}: The measured THz transmitted amplitude (t ~ 3.5 × 10^{−3}, Supplementary Fig. 1d, e) and the coverage ratio of our sample (β ~ 1.07 × 10^{−5}) gives the field enhancement factor (t/β) of ~348. With the measured field enhancement factor, the resulting gap field was estimated by E_{gap} = E_{inc} × (field enhancement), where E_{inc} is the incident field strength. And we converted the measured current data to the total rectified charges (yaxis, inset of Fig. 5b). The bandwidth of our current preamplifier is 2 kHz (with sensitivity of 10 nA V^{−1}) which is much lower than that of the tunneling current pulse (expected to be higher than 1 THz from the tunneling nonlinearity). In this case, the measured signal in the lockin amplifier is proportional to q_{THz}/τ_{rep}, where q_{THz} is the total rectified charges under a single THz pulse,
where τ_{rep} is the pulse to pulse separation time and I(t) the instantaneous total tunneling current in Eq. (2). To find a quantitative relation between the output of the lockin amplifier and the total rectified charges, we modeled the time trace of the current preamplifier response I_{amp}(t) (inset of Supplementary Fig. 1a) by Fourier expansion
where p is an integer and C_{p} is the corresponding Fourier coefficient. Thus C_{1} is directly related to the lockin amplifier output which is synchronized to the pulse repetition rate. Our aim is to find a relation between q_{THz} = I_{0}τ (integration of I_{amp}(t) over time, assuming I_{amp}(t) as a square pulse) and C_{1}, where I_{0} is the maximum of I_{amp}(t) and τ is the current pulse width of I_{amp}(t). By multiplying cos(2πt/τ_{rep}) and integrating in [−τ_{rep}/2, τ_{rep}/2] for both sides, C_{1} can be expressed by
This can be further reduced to C_{1} = 2I_{0}τ/τ_{rep} by assuming τ ≪ τ_{rep} (while this assumption is not rigorously valid under the current preamplifier bandwidth of 2 kHz presented here, we confirmed the same signal level for a larger bandwidth of 20 kHz by decreasing the sensitivity of the preamplifier under a large signal condition, thus justifying our assumption). Considering the rootmeansquare output of the lockin amplifier (C_{1}/\(\sqrt 2\) = measured current), we can write the total rectified charges for the single THz pulse by
We finally converted the measured current to the instantaneous tunneling peak current via dividing q_{THz} by the measured tunneling current pulse width and taking the current through one side of the triangle (Fig. 5c) by considering the current distribution (Fig. 2b).
Modeling polarizationdependent response of tunneling currents in ringshaped barriers
Under the THz field illumination, tunneling current through the triangle barrier sensitively depends on the incident polarization. We modeled the polarizationdependent total instantaneous current I_{total} by the following equation (Supplementary Fig. 2a),
where a_{i} is the asymmetry factor for each side of our quantum barrier figure, J_{t}(E) the tunneling current density as a function of the field strength E, E_{inc} the incident field strength, θ_{i} the angle between the THz induced surface current and the figure side, and i runs through the number of sides of the figure. Here, the asymmetry factor reflects the nonuniformity of a realistic barrier in our sample. The black line of Supplementary Fig. 2b shows the raw data presented in Fig. 2c for a triangle (Malus’ law). Due to the slight nonuniformity of each side, a_{i} values of 1, 0,75 and 0.67 were used for the fitting process. Using these parameters and taking J_{t}(E) = J_{t}(E_{inc}), we can describe the full 360 degrees polarization responses as shown in Supplementary Fig. 2b. Together with the fittings for the squareshapedbarrier sample, the reconstructed polar plots of Fig. 2d are obtained.
Data availability
The data that support the findings of this study are available from the authors on reasonable request.
References
 1.
Paulus, G. G. et al. Absolutephase phenomena in photoionization with fewcycle laser pulses. Nature 414, 182–184 (2001).
 2.
Baltuska, A. et al. Attosecond control of electronic processes by intense light fields. Nature 421, 611–615 (2003).
 3.
Eckle, P. et al. Attosecond ionization and tunneling delay time measurements in helium. Science 322, 1525–1529 (2008).
 4.
Cocker, T. L. et al. An ultrafast terahertz scanning tunnelling microscope. Nat. Photonics 7, 620–625 (2013).
 5.
Zixu, Z., Saumil, J., Sachit, G. & Garret, M. Graphene geometric diodes for terahertz rectennas. J. Phys. D 46, 185101 (2013).
 6.
Sharma, A., Singh, V., Bougher, T. L. & Cola, B. A. A carbon nanotube optical rectenna. Nat. Nanotechnol. 10, 1027–1032 (2015).
 7.
Davids, P. S. et al. Infrared rectification in a nanoantennacoupled metaloxidesemiconductor tunnel diode. Nat. Nanotechnol. 10, 1033–1038 (2015).
 8.
Ward, D. R., Huser, F., Pauly, F., Cuevas, J. C. & Natelson, D. Optical rectification and field enhancement in a plasmonic nanogap. Nat. Nanotechnol. 5, 732–736 (2010).
 9.
Sanchez, A. Jr., Davis, C. F., Liu, K. C. & Javan, A. The MOM tunneling diode: theoretical estimate of its performance at microwave and infrared frequencies. J. Appl. Phys. 49, 5270–5277 (1978).
 10.
Nguyen, H. Q. et al. Mechanisms of current rectification in an STM tunnel junction and the measurement of an operational tunneling time. IEEE Trans. Electron Devices 36, 2671–2678 (1989).
 11.
Rybka, T. et al. Subcycle optical phase control of nanotunnelling in the singleelectron regime. Nat. Photonics 10, 667–670 (2016).
 12.
Yoshioka, K. et al. Realspace coherent manipulation of electrons in a single tunnel junction by singlecycle terahertz electric fields. Nat. Photonics 10, 762–765 (2016).
 13.
Schiffrin, A. et al. Opticalfieldinduced current in dielectrics. Nature 493, 70–74 (2013).
 14.
Cowell, E. W. et al. Advancing MIM electronics: amorphous metal electrodes. Adv. Mater. 23, 74–78 (2011).
 15.
Cocker, T. L., Peller, D., Yu, P., Repp, J. & Huber, R. Tracking the ultrafast motion of a single molecule by femtosecond orbital imaging. Nature 539, 263–267 (2016).
 16.
Krüger, M., Schenk, M. & Hommelhoff, P. Attosecond control of electrons emitted from a nanoscale metal tip. Nature 475, 78–81 (2011).
 17.
Wimmer, L. et al. Terahertz control of nanotip photoemission. Nat. Phys. 10, 432–436 (2014).
 18.
Zherebtsov, S. et al. Controlled nearfield enhanced electron acceleration from dielectric nanospheres with intense fewcycle laser fields. Nat. Phys. 7, 656–662 (2011).
 19.
Park, D. J. et al. Strong field acceleration and steering of ultrafast electron pulses from a sharp metallic nanotip. Phys. Rev. Lett. 109, 244803 (2012).
 20.
Seo, M. A. et al. Terahertz field enhancement by a metallic nano slit operating beyond the skindepth limit. Nat. Photonics 3, 152–156 (2009).
 21.
Bahk, Y.M. et al. Electromagnetic saturation of angstromsized quantum barriers at terahertz frequencies. Phys. Rev. Lett. 115, 125501 (2015).
 22.
Kim, J.Y. et al. Terahertz quantum plasmonics of nanoslot antennas in nonlinear regime. Nano Lett. 15, 6683–6688 (2015).
 23.
Ahn, J. S. et al. Optical field enhancement of nanometersized gaps at nearinfrared frequencies. Opt. Express 23, 4897–4907 (2015).
 24.
Han, S. et al. Colossal terahertz nonlinearity in Angstrom and nanometersized gaps. ACS Photonics 3, 1440–1445 (2016).
 25.
Kim, J.Y. et al. Tunnelling currentvoltage characteristics of Angstrom gaps measured with terahertz timedomain spectroscopy. Sci. Rep. 6, 29103 (2016).
 26.
Fowler, R. H. & Nordheim, L. Electron emission in intense electric fields. Proc. R. Soc. Lond. A 119, 173–181 (1928).
 27.
Kang, J.H., Kim, D.S. & Seo, M. Terahertz wave interaction with metallic nanostructures. Nanophotonics 7, 763–793 (2018).
 28.
Jackson, J. D. Classical Electrodynamics. 3rd edn (John Wiley & Sons Inc, Hoboken, NJ, 1999).
 29.
Lee, D. & Kim, D.S. Light scattering of rectangular slot antennas: parallel magnetic vector vs perpendicular electric vector. Sci. Rep. 6, 18935 (2016).
 30.
Chen, X. et al. Atomic layer lithography of waferscale nanogap arrays for extreme confinement of electromagnetic waves. Nat. Commun. 4, 2361 (2013).
 31.
Jelic, V. et al. Ultrafast terahertz control of extreme tunnel currents through single atoms on a silicon surface. Nat. Phys. 13, 591–598 (2017).
 32.
Hirori, H., Doi, A., Blanchard, F. & Tanaka, K. Singlecycle terahertz pulses with amplitudes exceeding 1 MV/cm generated by optical rectification in LiNbO3. Appl. Phys. Lett. 98, 091106 (2011).
 33.
Planken, P. C. M., Nienhuys, H.K., Bakker, H. J. & Wenckebach, T. Measurement and calculation of the orientation dependence of terahertz pulse detection in ZnTe. J. Opt. Soc. Am. B 18, 313–317 (2001).
 34.
Simmons, J. G. Generalized formula for the electric tunnel effect between similar electrodes separated by a thin insulating film. J. Appl. Phys. 34, 2581–2590 (1963).
 35.
Simmons, J. G. Potential barriers and emission‐limited current flow between closely spaced parallel metal electrodes. J. Appl. Phys. 35, 2472–2481 (1964).
 36.
Kyoung, J. S., Seo, M. A., Park, H. R., Ahn, K. J. & Kim, D. S. Far field detection of terahertz near field enhancement of subwavelength slits using Kirchhoff integral formalism. Opt. Commun. 283, 4907–4910 (2010).
Acknowledgements
We thank Jisoon Ihm for helpful discussion. This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT: NRF2015R1A3A2031768) (MOE: BK21 Plus Program21A20131111123).
Author information
Affiliations
Contributions
R.H.J.K. and D.S.K. conceived the study. T.K., R.H.J.K. and G.C. performed experiments and analyzed the data. T.K., R.H.J.K. and J.L. fabricated samples. H.P. and H.J. performed ALD for Al_{2}O_{3} deposition on the samples. T.K. and D.S.K. wrote the manuscript. R.H.J.K., G.C., C.H.P. and D.S.K. commented on the manuscript.
Corresponding author
Correspondence to DaiSik Kim.
Ethics declarations
Competing interests
The authors declare no competing interests.
Additional information
Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Electronic supplementary material
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Kang, T., Kim, R.H.J., Choi, G. et al. Terahertz rectification in ringshaped quantum barriers. Nat Commun 9, 4914 (2018). https://doi.org/10.1038/s4146701807365w
Received:
Accepted:
Published:
Further reading

Terahertz quantum plasmonics at nanoscales and angstrom scales
Nanophotonics (2020)

Terahertz field confinement and enhancement in various subwavelength structures
Journal of Applied Physics (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.