Enhanced Graphene Photodetector With Fractal Metasurface
Letterpubs.acs.org/NanoLettEnhanced Graphene Photodetector with Fractal MetasurfaceJieran Fang,†,§ Di Wang,†,§ Clayton T. DeVault,‡,§ Ting-Fung Chung,‡,§ Yong P. Chen,†,‡,§Alexandra Boltasseva,†,§, Vladimir M. Shalaev,*,†,§ and Alexander V. Kildishev*,†,§†School of Electrical and Computer Engineering, ‡Department of Physics and Astronomy, and §Birck Nanotechnology Center andPurdue Quantum Center, Purdue University, West Lafayette, Indiana 47907, United States DTU Fotonik, Department of Photonics Engineering, Technical University of Denmark, Lyngby, DK-2800, DenmarkS Supporting Information*ABSTRACT: Graphene has been demonstrated to be a promisingphotodetection material because of its ultrabroadband optical absorption,compatibility with CMOS technology, and dynamic tunability in opticaland electrical properties. However, being a single atomic layer thick,graphene has intrinsically small optical absorption, which hinders itsincorporation with modern photodetecting systems. In this work, wepropose a gold snowflake-like fractal metasurface design to realizebroadband and polarization-insensitive plasmonic enhancement ingraphene photodetector. We experimentally obtain an enhanced photovoltage from the fractal metasurface that is an order of magnitude greaterthan that generated at a plain gold graphene edge and such an enhancement in the photovoltage sustains over the entire visiblespectrum. We also observed a relatively constant photoresponse with respect to polarization angles of incident light, as a result ofthe combination of two orthogonally oriented concentric hexagonal fractal geometries in one metasurface.KEYWORDS: Photodetector, graphene, fractal metasurface, plasmonicsGphotogain, are not necessary for such detectors to be operative.Despite all the advantages of MGM-PDs (broadband operation,uniform sensitivity against wavelength, fast response speed, zeropower consumption), the responsivity (sensitivity) in MGMPDs is notoriously low because graphene, which absorbs merely2.3% of incident light,1 is used as the photoabsorption material.On the other hand, plasmon oscillations have been known andwidely utilized to enhance optical absorption and generate hotelectrons in various optical systems. Unsurprisingly, efforts havealso been made to enhance the responsivity in MGM-PDs usingplasmonic nanostructures. However, the previously proposedplasmonic enhancement methods are mostly narrowband andpolarization dependent.28 31 Optical waveguides have also beendemonstrated to enhance the optical absorption by almost100% in graphene photodetectors,10,27 but this method requirescoupling light into waveguides that could result in a bulkyexperimental setup. Indeed, although narrowband and polarization sensitive enhanced photodetectors are desirable inspectrally resolved and polarization specific photodetectionscenarios, there is definitely a strong need for broadband andpolarization insensitive enhanced photodetectors. In this work,we propose a gold fractal metasurface design that has arelatively flat optical absorption in the visible part of thespectrum to realize broadband and polarization insensitiveplasmonic-enhanced graphene photodetector.raphene has been demonstrated as an appealing materialfor photodetection due to its unique properties such aswide optical absorption spectrum, wavelength independentabsorption, high room-temperature electron and hole mobilities, mechanical flexibility, and dynamic tunability in opticaland electrical properties.1 9 So far there are primarily fiveknown physical mechanisms that enable photodetection ingraphene: the photovoltaic effect,10 12 the photothermoelectriceffect,13 16 the bolometric effect,17 the photogating effect,18 20and the surface plasmon-assisted mechanism.21,22 Among these,the photovoltaic (PV) effect makes use of the built-in electricfield that is induced by the differently doped regions ingraphene to separate the optically excited electron hole pairs ingraphene and give rise to photovoltage, whereas the photothermoelectric (PTE) effect is associated with the photovoltageproduced by the optically generated hot electrons in regionswith different thermoelectric powers (Seebeck coefficients) ingraphene: VPTE (S1 S2)ΔT, where S1 and S2 are the Seebeckcoefficients of two regions in graphene with different dopinglevels, and ΔT is the electron temperature difference across thetwo regions. Both effects are thought to contribute inphotovoltage generation in metal graphene metal photodetectors (MGM-PDs),23 and make MGM-PDs the prioritizedcandidates for ultrafast graphene photodetector, owing to thehigh carrier transport velocity12,24 and extremely short carrierheating and cooling times25 27 in graphene. Moreover, MGMPDs are ideal for applications where zero power consumptionand zero dark current are desired, because source drain bias andgate voltage, although useful for dynamically tuning the 2016 American Chemical SocietyReceived: August 1, 2016Revised: December 13, 2016Published: December 14, 201657DOI: 10.1021/acs.nanolett.6b03202Nano Lett. 2017, 17, 57 62
LetterNano LettersWe begin by introducing a new fractal metasurface design.The metasurface is realized through a fractal tree to mimic thesnowflake geometry. As demonstrated in previous works, thefractal metasurface has been shown to exhibit broadbandabsorption and multiple resonances with increased levels,32 34where the practical examples include plasmonic elements33 thatfollow the Cayley tree topology35 or the nanostructuredaluminum electrodes34 following the classical space-fillingcurves of Hilbert and Peano.36,37 Here we choose the sixpoint asterisk shape as our seed geometry (the first-level fractal,Figure 1a). Construction of the branching is generated byTable 1. Arm Lengths for the Whole Fractal Metasurfacelevel number, N4-level fractal arm lengths (μm, bluesegments in Figure 1b)3-level fractal arm lengths (μm, redsegments in Figure 1b)12343.471.150.380.131.390.460.15NAmodel are adopted from an online database.40 (See SupportingInformation for details.) We show one full-wave simulated inplane electric field distribution just underneath the gold fractalmetasurface when it is illuminated at the wavelength of 530 nm(Figure. 1c). The high intensity regions (hot spots) are tightlylocalized around the branches and edges of the fractalstructures.To further validate the performance of our metasurface, weperformed near-field scanning optical microscopy (NSOM,MultiView 2000, Nanonics Imaging Ltd.) to elucidate the nearfield characteristics of the plasmonic fractal. For near-fieldmeasurements, we fabricated gold fractal patterns of exactly thesame dimensions on top of a bare glass substrate; this wasnecessary due to the strong absorption that would haveotherwise occurred using a silicon substrate. The metasurfacesample is illuminated from the far-field using a weakly focused532 nm diode laser incident on the bottom of the metasurface,that is, the glass substrate side. The near-field signal is obtainedby scanning a metal-coated (chromium and gold) tapered fiberwith a 50 nm aperture above the surface of the sample. Thescan is performed at a fixed distance of 100 nm in order tomitigate damage to the tip and/or sample and to eliminatetopographic artifacts in the signal.41 Figure 2a shows theFigure 1. (a) Construction of the fractal design with “snowflake”geometry from level 1 to level 4. (b) Total structure comprising of afour-level (blue part) and a three-level (red part) fractals utilized in thestudy. (c) FDTD simulated in-plane electric field (of the incidentelectromagnetic wave) distribution just underneath the gold fractalmetasurface on a glass substrate under the excitation wavelength of530 nm. The electric field is linearly polarized along y-direction.recursive iterations that is different from the classicalapproaches to n-flake generation.35 In our case, upon eachiteration six new branches are only spread from each new rootpoint. The angle between each branch is kept at 60 . Thebranch width and thickness are set to be 40 nm for all levels ofiterations, while branch lengths at each level are decreased bythe scaling ratio of one-third as the level increases with a totallevel of 4, denoted by the red color in Figure 1a. For the sake offabrication and measurement conveniences, we designed thefractal metasurface with a diameter of 10 μm to cover thewhole area of illumination by the laser spot in our system.Nonetheless, due to the simple and inward scalability of fractalmetasurfaces, it is convenient to design metasurfaces that arefitted for other spot sizes and for enhancement at other desiredwavelength ranges, while keeping the overall coverage area ofthe metasurface unchanged with increasing fractal levels. Inorder to increase the density of the branches within theillumination spot and to compensate for the intrinsicpolarization anisotropy, we add another three-level fractalstructure concentrically with a 30 mismatch to the four-levelfractal structure, denoted in red color in Figure 1b. All the armlengths for the whole fractal metasurface are listed in Table 1.Next, we investigate the optical characteristics of such goldfractal metasurface numerically through the finite-differencetime-domain (FDTD) method.38 We employ a dispersivemodel for gold, which is defined as the sum of a Drude termand two critical point terms and is implemented through ageneralized dispersion material model.39 The parameters of theFigure 2. (a) Experimental near-field extinction map of the fractalmetasurface, obtained using near-field scanning optical microscopy(NSOM). The measurement is done in collection mode at awavelength of 532 nm. (b) The simulated electric field distributionat the wavelength of 532 nm. A floating window average is applied tomimic the 50 nm diameter aperture that is used in the NSOMexperiment in (a).experimental near-field extinction map of the metasurface. Tocompare with experiments, we simulated, using an FDTDmethod, the electromagnetic fields in a plane 100 nm above theplasmonic metasurface assuming a 532 nm plane wave incidentfrom the substrate side and applied a moving average weightedwith a 50 nm disk to account for the convolution of the tip.42Our results are well matched with experiment as illustrated inFigure 2b and indicate strong plasmonic extinction near thebranches and edges of the fractal structure.When visible light is incident upon the fractal metasurface, itexcites plasmon oscillation in the gold fractal structure, which inturn confines and enhances the electric field of the incidentelectromagnetic wave within nanometers of the structure,contributing to an extensive electron hole pair generation andelevating the electron temperature through electron electron58DOI: 10.1021/acs.nanolett.6b03202Nano Lett. 2017, 17, 57 62
LetterNano Lettersinteractions in graphene.43 The generated carriers are thenspatially separated/driven via the aforementioned built-inelectric field (PV) and thermoelectric power differential(PTE) at gold graphene interface, giving rise to a detectablephotovoltage. Additionally, due to a combination of twoorthogonally oriented concentric hexagonal fractal geometriesin an integrated metasurface design (see Figure 1b, red and bluestructures), the enhancement in photovoltage detection isindependent of the polarization angle of the incidentelectromagnetic wave, providing yet another feature unprecedented by the previously reported plasmonic enhanced MGMPDs.In our experiment, we integrated the fractal metasurface withthe drain contact of the graphene field effect transistor (FET)device, so that the metasurface is at the same electrical potentialas the bulk drain contact to facilitate the electron (hole)collection and also the theoretical analysis. The devicefabrication starts with the transfer of a monolayer graphenesheet44 grown by chemical vapor deposition (CVD) onto ahighly p-doped silicon substrate (0.001 0.005 Ω-cm) with a300 nm thick dry thermal dioxide on top. The fractalmetasurface with gold rod for electrical connection with thedrain contact and a ring encircling the fractal metasurface tomaximize electron (hole) harvesting at the source contact weredefined by electron beam lithography (EBL), Ti (3 nm)/Au(40 nm) metallization and liftoff. The large sheet of graphenewas then etched into smaller rectangles using photolithographyand O2 plasma etch. The bulk source and drain contact pads (3nm Ti, 80 nm Au) were fabricated to directly cover the goldrods and partially the graphene sheet. Finally wire bonding thefabricated chip to printed circuit board was performed forelectrical measurements.The photovoltage response of our device was measured bythe setup illustrated in Figure 3a. A continuous wave laser (Ar Kr) chopped at 1.1 kHz by an optical chopper was coupled to a10 microscope and was then focused on the photodetectorwith a spot diameter of 7 μm. The generated photovoltagewas then measured via the source drain contacts by a lock-inamplifier synchronized with the optical chopper. We firstinvestigated the enhancement in photovoltage generation fromthe ring encircling the fractal metasurface. To do this, onsample 1 we fabricated a tip-and-ring structure without fractalmetasurface and placed it in parallel with the structure withfractal metasurface (see Figure 3b). By measuring thephotovoltage generated on the tip (“spot B” in Figure 3b)and on the plain gold-graphene edge (“spot C” in Figure 3b),we observed an average of 5 times photovoltage enhancementon the tip. We then measured the photovoltage generated whenthe laser spot was incident upon the fractal metasurface (“spotA” in Figure 3b), denoted as Vfractal, and when the laser spot wasincident upon the plain gold graphene edge (“spot C” inFigure 3b), denoted as Vedge, and defined Vfractal/Vedge as theenhancement factor of photovoltage generated on fractalmetasurface to plain edge. The study of fractal metasurfacephotovoltage generation was carried out at six experimentallyavailable wavelengths, 476, 488, 514, 530, 568, and 647 nm, toinvestigate the broadband enhancement effect in the visiblespectrum. In this work, all measurements were done with zerogate voltage (VG 0) and source drain bias (VSD 0), unlessotherwise indicated. As an illustration, we show the photovoltage generated as a function of incident power at thewavelength of 568 nm in Figure 3c,d (the measurements atother wavelengths are provided in the Supporting Information),Figure 3. (a) Experimental setup for photovoltage measurement; (b)scanning electron micrograph (SEM) of sample 1 the graphenephotodetector with the fractal metasurface and tip-and-ring structure(the white scale bar is 10 μm); inset image shows the zoomed-in viewof gold fractal metasurface (the white scale bar is 1 μm). (c) Blue lineswith error bars: measured photovoltage when laser is incident on tipwithout fractal metasurface (spot B in (b), solid line) and on plainedge (spot C in (b), dotted line) as a function of incident power, theerror bars are experimentally measured data points and blue lines arelinear fits to the experimental data. Red circles: enhancement factorsfrom tip to edge at individual tested incident powers. (d) Similar to(c), the measured photovoltage generated on fractal metasurface (spotA in (b), solid blue line) and on plain metal/graphene edge (spot C in(b), dashed blue line) as a function of incident power andenhancement factors (red dots, right vertical axis) at each testedpower. The measurements in (c,d) were carried out at the wavelengthof 568 nm.from which a linear relationship between the two can be seen,indicating that we were operating the device before absorptionsaturation.29To show the broadband nature of the photovoltageenhancement, we plot the enhancement factors Vfractal/Vedge atthe six tested wavelengths in Figure 4a. The error bars comefrom the fact that the enhancement factors vary with varyingincident optical powers. As is evident in Figure 4a, enhancement factors ranging from 10 to 16 are achieved at the testedvisible-spectrum wavelengths. We notice that although thesimulated optical absorption of the fractal metasurface is ratherflat in the investigated spectral range (shown by the cyan solidline in Figure 4a), the enhancement factors in photovoltagegeneration exhibit slight wavelength dependence and theenhancement factors increase with increasing wavelength. Webelieve that the reason for such behavior is two-fold. On onehand, it is caused by the greater photovoltage generated on theplain gold graphene edge at shorter wavelengths due tostronger heating of gold pad, accounted for by the PTE effect inphotovoltage generation. On the fractal metasurface, the entiremetasurface area contributes to generating photovoltage(Vfractal), independent of incident light wavelength. On theplain gold graphene edge (Vedge), however, larger/smaller areaextending into the gold pad within the laser spot contributes togenerating photovoltage at respectively shorter/longer wavelengths.23,26 As a result, the enhancement factors decrease atshorter wavelengths simply because the area of graphene thatcontributes to Vedge increases compared to that at longer59DOI: 10.1021/acs.nanolett.6b03202Nano Lett. 2017, 17, 57 62
LetterNano Letters(E-field along y-axis in Figure 1c) polarizations and observeonly negligible difference between the two (see SupportingInformation). The experimental and simulation resultsdemonstrate the robustness of the photovoltage enhancementby the proposed fractal metasurface with respect to polarizationof incident light.In order to compensate for the partial enhancement due tothe reduced source-drain distance for spot A in sample 1, wefabricated sample 2 where we placed a fractal metasurfaceencircled by a 30 μm ring in parallel with a plain source drainstructure with 30 μm separation on the same graphene sheetand studied the enhancement of photovoltage between the twostructures. Similar to sample 1, in sample 2 we study the ratio ofphotovoltage generated when laser is incident on fractalmetasurface (spot A in Figure 5a) to that generated whenFigure 4. (a) Red markers with error bars: measured enhancement ofphotovoltage generation (Vfractal/Vedge, spot A to spot C in Figure 3b)over a wavelength range from 476 to 647 nm. Cyan curve: thesimulated absorption spectrum of the fractal metasurface. (b)Measured photovoltage as a function of source drain bias VSD.Measurement was done at the wavelength of 514 nm with an inputpower of 1 mW. Blue markers are measured data points and orangecurve is linear fit to the data points. (c) Normalized photovoltage as afunction of incident light polarization angles. Blue markers aremeasured data points and orange curve is linear fit to the data points.(d) Blue lines: measured photovoltage generated on fractal metasurface (spot A in Figure 3b, solid lines) and on tip (spot B in Figure 3b,dashed lines) with x-polarized light (upper panel) and y-polarized light(lower panel). See Figure 1c for x- and y-directions.Figure 5. (a) Scanning electron micrograph (SEM) of the graphenephotodetector sample 2 with the fractal metasurface (the white scalebar is 10 μm) encircled by 30 μm ring and source-drain contactsseparated by 30 μm; inset image shows the zoomed-in view of goldfractal metasurface. Similar to sample 1, the ratio of the photovoltagegenerated on spot A to that generated on spot B is defined as theenhancement factor. (b) Similar to Figure 4a, the measuredenhancement of photovoltage generation (red markers with errorbars, left vertical axis) from 476 to 647 nm and the simulatedabsorption spectrum of the fractal metasurface (cyan curve, rightvertical axis) on sample 2.wavelength, while it remains identical across the entire testedwavelength range for Vfractal. On the other hand, the Ti dopingand intrinsic p-doping in bulk graphene induces p p junctionat the metal/graphene interface, where PV and PTE effectscounteract with each other.13,15 The competition between thetwo effects could also be a reason for the
four-level (blue part) and a three-level (red part) fractals utilized in the study. (c) FDTD simulated in-plane electric field (of the incident electromagnetic wave) distribution just underneath the gold fractal metasurface on a glass substrate under the excitation wavelength of 530 nm. The electric field is linearly polarized along y-direction.
edge contours play a dominant role in defining perception of fractals [44]. The importance of edge contours is sup-first such experiments, performed in 1994, I used a chaotic pendulum [34] to generate fractal and non-fractal pat-terns. In perception experiments based on these images, 95% of participants preferred the fractal to the non-fractal
All rights reserv ed. www.arpnjournals.com 8549 dimension of a fractal curve is an indication of achieving better space-filling of that fractal. On the other hand, not all fractal curves can be used in the filter design applications. Some of the fractal geometries have been successfully employed to the design different antennas for
discussed the applications of fractal geometry (in his words, the "geometry of nature"), especially for their use in studying the natural world. [4] In his book Fractal Geometry, Kenneth Falconer avoids giving a precise de nition of a fractal, writing "it seems best to regard a fractal as a set that has properties as those listed below, rather
of geometrical complexity that can model many natural phenomena. Almost all natural objects can be observed as fractals (coastlines, trees, mountains, and clouds). Their fractal dimension strictly exceeds topological dimension [2]. Fractal analysis means the determination of the fractal dimensions of an image. The fractal dimensions
Material Safety Data Sheet (MSDS) 1. Identification of the substances/mixture and of Product information . Synonyms Graphene, graphene sheets, graphene flakes, graphene powder, few layers graphene, graphite powder Use of Substance As an additive in coatings, composites, polymers Biotechnology related applications are restricted to research .
FRACTALS AND FRACTAL DIMENSION A precise physical definition of fractal has not yet appeared, nor is it essential for applications in image processing. More important is the general concept of a Johns Hopkins APL Technical Digest, Volume 12, Number 4 (1991) fractal. Falconer1 defines fractals as objects with some or
Jan 16, 2001 · Fractal antenna theory is built, as is the case with conventional antenna theory, on classic electromagnetic theory. Fractal antenna theory uses a modern (fractal) geometry that is a natural extension of Euclidian geometry. In this report, attention is called to this developing, but already quite large, field of study. In the brief study ofFile Size: 215KB
Abrasive Water Jet Machining (AWJM) is the non-traditional material removal process. It is an effective machining process for processing a variety of Hard and Brittle Material. And has various unique advantages over the other non-traditional cutting process like high machining versatility, minimum stresses on the work piece, high flexibility no thermal distortion, and small cutting forces .
