High Quantum Efficiency Hot Electron Electrochemistry€¦ · High Quantum Efficiency Hot Electron Electrochemistry ... 3−Si metal−insulator−semiconductor (MIS) junction. Despite

High Quantum Efficiency Hot Electron ElectrochemistryHyun Uk Chae,† Ragib Ahsan,† Qingfeng Lin, Debarghya Sarkar, Fatemeh Rezaeifar,Stephen B. Cronin, and Rehan Kapadia*

Department of Electrical and Computer Engineering, University of Southern California, Los Angeles, California 90089, UnitedStates

*S Supporting Information

ABSTRACT: Using hot electrons to drive electrochemicalreactions has drawn considerable interest in driving high-barrier reactions and enabling efficient solar to fuel conversion.However, the conversion efficiency from hot electrons toelectrochemical products is typically low due to high hotelectron scattering rates. Here, it is shown that the hydrogenevolution reaction (HER) in an acidic solution can beefficiently modulated by hot electrons injected into a thingold film by an Au−Al2O3−Si metal−insulator−semiconductor(MIS) junction. Despite the large scattering rates in gold, it isshown that the hot electron driven HER can reach quantum efficiencies as high as ∼85% with a shift in the onset of hydrogenevolution by ∼0.6 V. By simultaneously measuring the currents from the solution, gold, and silicon terminals during theexperiments, we find that the HER rate can be decomposed into three components: (i) thermal electron, corresponding to thethermal electron distribution in gold; (ii) hot electron, corresponding to electrons injected from silicon into gold which drivethe HER before fully thermalizing; and (iii) silicon direct injection, corresponding to electrons injected from Si into gold thatdrive the HER before electron−electron scattering occurs. Through a series of control experiments, we eliminate the possibilityof the observed HER rate modulation coming from lateral resistivity of the thin gold film, pinholes in the gold, oxidation of theMIS device, and measurement circuit artifacts. Next, we theoretically evaluate the feasibility of hot electron injection modifyingthe available supply of electrons. Considering electron−electron and electron−phonon scattering, we track how hot electronsinjected at different energies interact with the gold−solution interface as they scatter and thermalize. The simulator is first usedto reproduce other published experimental pump−probe hot electron measurements, and then simulate the experimentalconditions used here. These simulations predict that hot electron injection first increases the supply of electrons to the gold−solution interface at higher energies by several orders of magnitude and causes a peaked electron interaction with the gold−solution interface at the electron injection energy. The first prediction corresponds to the observed hot electron electrochemicalcurrent, while the second prediction corresponds to the observed silicon direct injection current. These results indicate that MISdevices offer a versatile platform for hot electron sources that can efficiently drive electrochemical reactions.KEYWORDS: MIS devices, hot electron, hydrogen evolution reaction, quantum efficiency, Monte Carlo simulation,scattering mechanism

Efficiently using hot electrons before thermalization hasbeen an aim of fields such as hot-electron transistors,1−4

solar cells,5−7 plasmonics,8−14 photoemission,15,16 memo-ry,17,18 and solar-to-fuel19 devices. However, nonequilibriumelectrons exhibit lifetimes of ∼1−100 fs due to electron−electron and electron−phonon interactions.20−22 These ultra-short hot carrier lifetimes drive the thermalization process todominate over most other technologically relevant processes,causing devices to generally have low hot-electron quantumefficiencies. Multiple strategies have been explored to over-come these challenges, including engineering systems with low-electron densities and weak electron−phonon coupling, suchas quantum dots,23 minimizing the transit length of hotelectrons, by creating devices using 2-D materials,1,24,25 and thesearch for new materials with naturally favorable scatteringrates, such as perovskites.26 However, the overall efficiency27,28

of these hot electron devices have mainly precluded their

practical use, with photoemitters29,30 and Flash memory31 asthe two exceptions.In this work, we use a metal−insulator−semiconductor

junction to controllably create a hot electron population viatunneling of electrons from the semiconductor conductionband into the metal and then use that hot electron populationto modulate the electrochemical reaction rate at a metal−electrolyte junction. Figure 1a shows the basic device structurewhich consists of an n-type silicon wafer, an aluminum oxideinsulator layer, and a thin gold layer. The entire device isencapsulated in an epoxy, leaving only the top gold layerexposed, and is immersed in a 0.5 M H2SO4 solution. A

Received: June 5, 2019Revised: August 2, 2019Published: August 21, 2019

Letter

pubs.acs.org/NanoLettCite This: Nano Lett. 2019, 19, 6227−6234

© 2019 American Chemical Society 6227 DOI: 10.1021/acs.nanolett.9b02289Nano Lett. 2019, 19, 6227−6234

Dow

nloa

ded

via

UN

IV O

F SO

UT

HE

RN

CA

LIF

OR

NIA

on

Mar

ch 1

, 202

0 at

22:

56:0

9 (U

TC

).Se

e ht

tps:

//pub

s.ac

s.or

g/sh

arin

ggui

delin

es f

or o

ptio

ns o

n ho

w to

legi

timat

ely

shar

e pu

blis

hed

artic

les.

cartoon schematic of the band diagram is shown in Figure 1b.MIS junctions enable injection of the highest energy hotelectrons when compared to both metal−insulator−metaltunnel (MIM) junctions, or metal−semiconductor (MS)junctions. In MIM junctions, the large density of states aroundthe Fermi levels of the metals cause large currents to flow withrelatively smaller applied biases, limiting the energy at which

hot electrons can be injected. For MS junctions, the offsetbetween the semiconductor conduction band and metal Fermilevel is pinned at the interface, and this causes the injectedenergy of the hot electrons to be fixed by the Schottky barrierheight. In this device, as the voltage across the MIS junctionincreases, there will be an increase in both current and theenergy at which the hot electrons are injected. This behavioroccurs due to the insulator layer depinning the semiconductorconduction band from the metal Fermi level at the junction.Thus, the MIS structure is expected to generate the hottestelectrons in the metal when compared to MS or MIMstructures.Figure 1c shows the current density plotted as a function of

applied bias for an Au/Al2O3/Si device with a 12 nm thick goldlayer, a 6 nm thick Al2O3 layer, and a moderately doped n-type(5 × 1016 cm−3) Si wafer. From the device measurements, wesee that the current increases exponentially until VAu−Si ≈ 0.4 Vand then increases linearly. To understand this behavior, wesimulate the device using a 2-D Technology computer-aideddesign (TCAD) Sentaurus simulation, which self-consistentlysolves the drift-diffusion and Poisson equations. From thissimulation, we extract the silicon surface electron density,plotted in Figure 1d on both linear and log scales. Importantly,the sheet charge density also shows a clear exponential andlinear regime. These two regimes occur depending on wherethe applied voltage is dropped. In the exponential regime (0 V< VAu−Si < 0.4 V), the majority of the applied voltage isdropped across the semiconductor depletion region, changingthe semiconductor band bending and therefore surface chargedensity, but without causing any significant change in the bandoffset between the metal Fermi level and semiconductor

Figure 1. Device schematics and hot electron injection process. (a)Schematic of the structure of metal insulator semiconductor deviceused here. (b) Band diagram of the device and two different paths forinjected hot electrons into the Au region. (c) Current density vsapplied voltage for the Au/Al2O3/Si diode used here in both linearand log scale. (d) TCAD Sentaurus simulations of surface electronconcentration as a function of applied bias in the Au/Al2O3/Si device.

Figure 2. Linear sweep voltammetry (LSV) curves of 12 nm Au MIS device. (a) Linear scale solution current density vs applied Au-Solutionvoltage for varying Au−Si diode voltages and (b) log scale of (a). (c) Solution current density vs applied Au−Si voltage for varying Au-Solutionvoltages and (d) log scale of (c). (e) Tafel relation for the low VAu−Si and (f) high VAu−Si..

Nano Letters Letter

DOI: 10.1021/acs.nanolett.9b02289Nano Lett. 2019, 19, 6227−6234

6228

conduction band. This occurs due to the oxide capacitancebeing much greater than the semiconductor capacitance. Thus,although more electrons are injected from the semiconductorinto the metal in this regime, the relative energy at which theyare injected does not significantly change. Next, the linearcharge density regime (0.4 V < VAu−Si) occurs when themajority of the applied voltage is dropped across the oxide. Inthis regime there will be a nearly 1:1 ratio between the appliedbias and the change in the semiconductor conduction bandedge position with respect to the metal Fermi level. Thisregime occurs due to the semiconductor capacitance inaccumulation being much greater than the oxide capacitance.From these measurements and simulations, we estimate thatthe initial 0.4 V applied bias does not change the offsetbetween the silicon conduction band and metal, but that, forvoltages above 0.5 V, the voltage is primarily dropped acrossthe oxide until the series resistance limits the current.The electrochemical measurement was carried out using a

potentiostat with two working electrodes to apply independentbias voltages between the gold and solution terminals(VAu‑Solution) and the gold and silicon terminals (VAu−Si). Weuse a platinum wire as a counter electrode in the solution andan Ag/AgCl reference electrode. A schematic diagram of themeasurement setup and configuration is shown in Figure S1.Unlike traditional electrochemical cells, where the workingelectrode (WE) and the counter electrode (CE) are theprimary current components (ignoring current flow into thereference electrode), and are by definition equal in magnitude,in this device there are three currents: the current leaving thesilicon, ISi; the current leaving the gold, IAu; and the currentleaving the counter electrode, ISolution. Supplementary FigureS1a shows the three measured current components (ISi, IAu,ISolution) and the internal current components which will flow inthe device. In our MIS device, the current flow is due primarilyto electrons, and thus the electron flux directions are oppositeto the current arrows. Supplementary section S1 also shows themathematical relationships between current components inthis measurement setup.To study the redox behavior of these devices, we carried out

two types of ISolution measurements. First, we sweep the voltageof the Au-Solution (VAu‑Solution) junction and simultaneouslystep the voltage of the Au-Silicon (VAu−Si) junction. Figure 2a,bplot the linear and log scale results of these measurements,respectively. From the linear scale plot (Figure 2a), as thevoltage between the Au−Silicon junction increases, the turn-onvoltage for hydrogen reduction is reduced, and current densityincreases. For an applied VAu‑Solution = −1.5 V, the currentincreases from ∼13 mA/cm2 to ∼42 mA/cm2 when VAu−Si = 2V. From the log-scale current plots (Figure 2b), we can moreclearly see the reduction curves shift as the VAu−Si voltage isincreased. This shift to lower voltages between the gold andsolution is attributed to the increased flux of electronsimpinging on the gold solution surface, caused by the injectionof hot electrons into the gold from the Si. To understand theeffect of the Au−Si junction voltage, we have swept VAu−Siwhile stepping VAu‑Sol. Figure 2c shows the results on a linearscale. In all cases of current vs VAu−Si, there is a turn-on voltage,which becomes smaller as the VAu‑Solution voltage is increased.This should occur as the higher voltage will change both theconcentration of [H+] at the gold/solution interface andmodify the energy barrier. Figure 2d shows the same graphs ona log scale. At low diode voltages, the solution current is nearlyconstant, limited by the thermal electrons in gold. Once VAu−Si

becomes sufficiently large, we see an exponential increase inthe current until the linear regime shown in Figure 2c. Theinitial exponential increase is attributed to the increase inenergy of the electrons injected into gold. Finally, there is aclear difference in the current levels at which the crossoverfrom exponential to linear occurs. These data show theelectrochemical reaction rate on the gold surface dramaticallyshifting due to the Au−Si junction.To analyze the electrocatalytic activity and to elucidate the

reaction mechanism of hot electron devices, a Tafel analysis isintroduced. In conventional electrochemistry, the Tafelequation is well-defined as

η = +a b Jlog10 solution (1)

where η is the overpotential, which is the difference betweenthe electrode potential and the standard electrode potential,

=α

a Jlog ( )RTF

2.30310 0 , where R = 8.314 J·mol−1·K−1 is the

universal gas constant, F = 96485.3 C mol−1 is the Faradayconstant, J0 is the exchange current density, α is thephenomenological charge transfer coefficient, and =

αb RT

F2.303

is called the Tafel slope. For a single electron transfer process,α is often found to be ∼0.5 which leads the Tafel slope to be∼120 mV/dec.32−34 It is noteworthy that the Tafel equationoriginates from the Butler−Volmer equation:

= · −α η α η− −J J e e( )F RT F RTSolution 0

/ (1 ) /(2)

where the second exponential term becomes negligible at largeoverpotential and reduces to the simplified Tafel equation.32,33

Figure 2e,f shows the Tafel slopes of different regions of thesolution current density at different VAu−Si conditions. Asshown in the Figure 2e, when VAu−Si = 0 and 0.5 V, whichgenerates no or less hot electrons, the Tafel slope is ∼127 mV/dec, which is close to the often observed 120 mV/dec. ThisTafel slope indicates that the hydrogen evolution reactionhappening at the electrode is predominantly limited by thesingle electron transfer step which is popularly known as theVolmer reaction step (H+ + e− = Hads).

32−34 With increasingoverpotential, the Tafel slope starts increasing and the solutioncurrent density starts getting saturated. This saturation can beattributed to a number of factors: (i) as the current increases,the reaction gets limited by the mass transport to and from theelectrode,32−34 (ii) adsorption of reduced hydrogen atoms atthe electrode,32−34 and (iii) deviation from the conventionalTafel equation at higher overpotentials. The charge transfercoefficient, α = 0.5, is generally not applicable at higheroverpotentials when the change in the activation energy of theredox reaction with overpotential starts becoming nonlinear.34

Since our MIS device does not show any Tafel slope that isbelow 100 mV/dec, the reaction mechanism is not beinglimited by either the Heyrovsky step (H+ + Hads + e− = H2,Tafel slope of 40 mV/dec) or the Tafel step (Hads + Hads = H2,Tafel slope of 30 mV/dec).32−34 When there is a large positiveVAu−Si, as shown in the Figure 2f, we can see that there are twodifferent regions with different Tafel slopes. We observe a Tafelslope of (i) ∼175 mV/dec and (ii) ∼190 mV/dec at lower andhigher overpotentials, respectively. While the Tafel equationworks reasonably for the lower VAu−Si case, it deviates from theideal form for hot electron cases. This deviation stems from thefact that the derivation of the Butler−Volmer equationconsiders electron flow from the Fermi level of the electrodeto the redox states.34 Since the hot electrons have considerably

Nano Letters Letter

DOI: 10.1021/acs.nanolett.9b02289Nano Lett. 2019, 19, 6227−6234

6229

higher energy than the Fermi level, the conventional Tafelslopes do not manifest themselves in the higher VAu−Si cases.For high VAu−Si, we attribute the Tafel slope (∼175 mV/dec)at the lower overpotential to the hot electrons beingtransmitted to the redox states with considerably largertransmission probability than the thermal electrons in gold.With the increasing overpotential, the transmission probabilityof the hot electrons does not increase considerably while thesupply of the hot electrons remains constant which leads to asaturation of the current followed by the first exponentialincrease. As the overpotential increases further, the thermalelectrons of gold also acquire a considerably large transmissionprobability and we can see the second exponential increase incurrent with a different Tafel slope (∼190 mV/dec). Whileapparently it seems that the charge transfer efficiency (α∼0.31) has decreased compared to the VAu−Si = 0 case (α∼0.47), it is noteworthy that the magnitude of current densityincreased considerably. At this higher current density, the masstransport limitation, ohmic losses, and adsorption will also behigher which may collectively manifest as a larger Tafel slope.To make sure that the observed currents are not resulting

from any experimental artifact, we have carried out a set ofcontrol experiments. First, we have systematically modified thevoltage sources, to ensure that the result was not an artifactfrom the potentiostat (Supplementary Section S2). The keyresults show that if Au−Si junction bias is carried out by anindependent voltage source, the results are identical to the twoworking electrodes based potentiostat measurement setup(Figures S2, S3, and S4). After eliminating the possibility of ameasurement artifact, we studied whether the effect could beattributed to the lateral resistivity of the gold film or pinholesin the gold film using control samples (Figure S5), with detailsin Supplementary Section S3. First, Au film resistivitymeasurements (Figure S6) were carried out, with the measuredresistivity used as an input to a 3-D TCAD Sentaurussimulation which allowed us to accurately simulate theexpected current density and voltage drop across the goldfilms. The simulation results (Figure S7) show that the lateralresistance of the gold films only causes a maximum voltagedrop of ∼6 mV for a current density of 10 mA/cm2, which isnegligible with respect to the current and voltage shifts here.Once lateral resistivity was eliminated as a potential source ofthe observed current shift, we studied the effect of pinholes inthe gold films on the current. It should be noted that, fromatomic force microscopy (AFM) measurements, the root-mean-square (RMS) roughness of the thin gold films are ∼0.71nm (Figure S8). To controllably test the effect of these holes,we fabricated a thick gold film (100 nm) with lithographicallydefined holes (Figure S5b). By then carrying out the sameVAu−Si and VAu‑Solution sweeps, it can be determined if pinholesin the film could explain the results. However, Figure S9 showsthat even with engineered holes, there is a small change inISolution as a function of bias between the silicon and gold,dramatically smaller than the experimental data. We have alsostudied metal−semiconductor (MS) junctions, discussed inSupplementary Section S4. The key results show that an MSjunction with 12 nm gold and moderately doped Si (5 × 1016

cm−3) give similar overall behavior, but with much lowercurrent and voltage shifts (Figure S10a,d). As the thickness ofthe gold is increased to 100 nm, the effect becomes negligible(Figure S10b,e). Furthermore, if a thin gold layer is used withheavily doped Si, then the Schottky barrier becomes thin andturns into a tunnel barrier, which causes electrons to be

injected into the gold near the Fermi level, which alsoeliminates the observed behavior (Figure S10 c,f). Theseresults also validate the idea that an MIS junction will providehotter electrons than an equivalent MS junction.To determine if the oxide or silicon is degrading, causing

current due to the dissolution of the electrode, we carried outstability tests (Supplementary Section S5). If the sample wasattacked during the electrochemical measurements, we expectthe diode characteristics to change dramatically. Figure S11a,bshow the I−V curve of the MIS diode when it was firstfabricated, before any electrochemical measurements werecarried out, and after all the measurements in this paper werecarried out, with minimal difference. To highlight thismeasurement, the before and after curves shown in FigureS11a,b were separated by about one year, highlighting thestability of the Au/Al2O3/Si devices used here under ourexperimental conditions. Figure S11c shows the current vs timecurve for the MIS device, highlighting the stability. Figures S12and S13 show the three current components for our MISdevice vs time, to highlight the fact that the observed currentsare stable, and not time dependent or due to any kind ofcapacitive effects. These control experiments shed light on themechanism behind the observed behavior and eliminatemeasurement error, lateral resistivity, pinholes, or oxidationof the substrate itself as the possible cause of the observedcurrent behavior. Thus, we conclude that hot electrons injectedinto the gold are responsible for the modulation of the reactionrate at the gold/solution interface.Next, we study how each of the three measured current

components (ISolution, IAu, ISi) change as a function of appliedvoltage. Figure 3a shows the measured current components,and the internal current components which comprise them.The source of hot electrons in the gold is the electron injection

Figure 3. Current flow mechanism and measurement result. (a)Schematic diagram of main current components. Major currents arecomposed with several minor current components. (b−d) Threemajor current measurements of the closed system (i.e., ISi, IAu, andISolution) under different Au−Si voltages.

Nano Letters Letter

DOI: 10.1021/acs.nanolett.9b02289Nano Lett. 2019, 19, 6227−6234

6230

from the silicon conduction band. The electrochemicalreduction current on the gold is separated into twocomponents: reduction due to the thermal electron population,IThermal Electron, and reduction due to the hot-electronpopulation, IHot Electron. Finally, there is a direct electrochemicalreduction component from the silicon to the solution,IDirect Injection. These correspond to the measured componentsfrom the following relationships.

= − + +I I I I( )Solution Thermal Electron Hot Electron Direct Injection

(3)

= + − −I I I IAu Thermal Electron Hot Electron Au Si (4)

= +−I I ISi Au Si Direct Injection (5)

From these relationships, we can see that when VAu−Si = 0 V,IAu−Si = 0 A, this becomes the traditional three-electrodemeasurement where the gold is the working electrode, andISolution = −IAu. Figure 3b shows the three measured currentcomponents during a VAu‑Solution voltage sweep for VAu−Si = 0 V.The observed behavior is as expected, with the IAu = −ISolution.Figure 3c,d show the currents for VAu−Si = 0.5 V and VAu−Si =1.5 V, respectively. Surprisingly, ISi increases with the voltageapplied between the solution and gold. Since the voltagebetween the gold and silicon is fixed, and it was previouslyshown that the silicon does not directly inject current into thesolution (Figure S9), the current injected from the silicon intothe gold should be constant with respect to VAu−Si. However, asseen in Figure 3c,d, an increase in the measured current, ΔISiexists for both applied voltages. This increase in ISi can beexplained by four mechanisms: (i) a change in the goldelectrode voltage due to the lateral currents, (ii) holes in thegold which allow direct reduction of hydrogen due to thepotential of the silicon with respect to the solution, (iii)

oxidation of the silicon/Al2O3 substrate, or (iv) hydrogenreduction by injected electrons at a gold/hydrogen complexwhich directly accepts electrons from the silicon, drivingreduction without the need for a multistep electron transfer tothe “bulk” gold and then to the solution. Our controlexperiments analyzing the lateral gold potential drop (FigureS7), with lithographically defined holes (Figure S9), andstability measurements (Figure S11) eliminate mechanisms (i),(ii), and (iii). From this, we conclude that the increase in ISi isdue to direct injection of electrons into a species that iscomplexed with the gold. This is observable due to theindependent voltage control and current measurement of thegold and silicon terminals.We note that since Figure 3b−d are steady state measure-

ments, for all cases, ISolution + IAu + ISi = 0. Most importantly,from these graphs, we can quantify the three components ofthe electrochemical reduction current, which are thermalelectrons from the gold, hot electrons from the gold, and directinjection from the silicon. Using Figure 3b, we can identify thecontribution of the thermal electrons in gold as a function ofapplied bias. At any given VAu‑Solution, the thermal contributionfrom the gold should stay constant. Thus, we can define thenet change in solution current as

Δ = − =

= Δ +−I I I V

I I

( 0)Solution Solution Solution Au Si

Si Hot Electron (6)

where ISolution(VAu−Si = 0) refers to the current composed onlywith Au thermal electrons at the nonbiased condition, ΔISirepresents the change in ISi due to direct injection of electronsfrom the silicon to the solution, and IHot Electron is composed ofhot electrons in the Au region. So, the net change of solutioncurrent shown in the eq 6 can be explained by analyzing theindividual components which are presented in the eqs 3−5.

Figure 4. Hot electron measurement and characterization. (a) Current component ratio map along the increase of Au−Si voltage. Portion of thehot electrons in the total current keeps increasing as Au-Si voltage increases. (b) Current density from direct injected electrons from Si toelectrolyte at different fixed VAu‑Solution. (c) Quantum efficiency of hot electron device at fixed 2.0 V Au−Si voltage and (d) at fixed −0.8 V Au-Solution voltage.

Nano Letters Letter

DOI: 10.1021/acs.nanolett.9b02289Nano Lett. 2019, 19, 6227−6234

6231

From these curves, we can then extract the thermal, hotelectron, and direct injection components as a function ofapplied bias. Figure 4a shows the solution current densityJSolution vs VAu‑Solution sweeps for five diode bias potentials(VAu−Si = 0, 0.5, 1, 1.5, 2 V), with the total solution currentdensity separated into thermal electron, hot electron, anddirect injection current density components. When VAu−Si = 0V, all the measured current is due to the thermal electrons inthe gold. However, as the applied bias across the diodeincreases, both the direct injection and hot electroncomponents increase. However, while the hot electron currentincreases monotonically, the direct injection current appears tohave a peak. Figure 4b plots the direct injection current densityas a function of Au−Si bias. At higher VAu‑Solution, a clear peak isobserved. We attribute this behavior to the direct injection ofcarriers from the silicon into hydrogen ions on the surface ofthe gold, with a well-defined energy state.Figure 4c shows the quantum efficiency of the hot electron

induced hydrogen reduction, defined as JHot Electron/JSi, whereJHot Electron is the current density of hot electrons as a functionof VAu‑Solution, while JSi is the diode current density. Critically, itis seen that the hot electron efficiency increases to ∼85%before saturating at high solution potentials. Harvesting the hotelectrons in the Platinum based MS device35 and transferringplasmon induced hot electrons from Au nanoparticles36 haveshown ∼5.6%−8.5% and ∼40% efficiency, respectively. Thiswork presents the highest quantum efficiency reported to datefor a hot-electron mediated electrochemical process. Fur-thermore, we also show the current density efficiency as afunction of VAu−Si and demonstrate that, at high diode biases,the current efficiency is >50%. Figure S14 shows the overallquantum efficiency in different biased conditions characterizedin Figure 2. These efficiencies suggest there is a clear pathtoward using MIS structures as efficient sources for hotelectron devices.To gain further insight into the hot electron dynamics in

gold, we have carried out detailed simulations using a modified2D Monte Carlo simulation package for semiconductortransport, Archimedes.37 We have modified the simulator byimplementing the gold density of states, electron−electron (e−e), and electron−phonon (e−p) scattering. We describe thedetails of the implementation in Supplementary section S11.Figure 5a shows the scattering rates for the e−e, e−p, and totalscattering as a function of energy above the Fermi level in gold.To verify the scattering rates used here, we show that theelectron−electron scattering rates used here match thosepublished in literature38 (Supplementary Figure S19).Furthermore, we show that the simulator used here canaccurately reproduce experimental electron temperature vstime profiles as shown in Figure S20. While the e−p scatteringrate is high compared to the e-e scattering rate, due to therelatively small energy of acoustic phonons, at high energy, theenergy loss per fs for hot electrons is dominated by e−escattering events (Figure 5b), due to the relatively largeaverage energy loss per scattering event for high energyelectrons interacting with electrons near the Fermi surface.Next, we carry out a simulation where we inject electrons atvarying energies above the gold Fermi level and track thedecay.From these results, the rate of attempts at the gold surface

due to hot electrons is obtained as a function of energy. Usingthe attempt rate enables us to account for the interactions ofthe electrons in the gold with the surface. Essentially, this

approach is analogous to the attempt frequency approach usedwhen calculating tunneling rates out of quantum wells, wherethe tunneling rate is proportional to the rate at which electronsreflect off the quantum well barriers. Here we use the attemptrate as an analog for the attempt rate for tunneling in quantumwells. This approach simultaneously allows us to capture therate of interactions between the hot electrons and the gold/solution interface and normalize this rate accurately with theinteraction rate between the thermal electrons and the gold/solution interface. The rate of attempts at the gold surface dueto the thermal electrons in gold is also obtained from thesimulation. By then normalizing these two rates of attempts asdescribed in Supplementary section S11, we plot the attemptrate for gold with no hot electron injection and gold with 26mA/cm2 of hot electrons injected 2 eV above the Fermi level.This is the current density of our MIS device at VAu−Si = 2 V.While many injection energies were simulated (Figure S20a),we show the 2 eV injection result as it closely approximates theexpected offset given by considering the initial band offsetbetween our n-doped Si and a gold electrode with aworkfunction of 5.1 eV, which is ∼1 eV, and then adding inthe expected increase in energy due to the applied voltage.While the applied voltage is 2 V, the initial ∼0.5 V can beassumed to drop over the silicon depletion region, and the final∼0.5 V is expected to be limited by series resistances (FigureS15), giving us an approximate injection energy of 2 eV.Figure 5c shows the results with the energy with respect to

the gold Fermi level on the y-axis and attempt rate on the x-axis in log scale, and Figure 5d shows a zoomed in view inlinear scale. We immediately see that, for high energies, there issignificant increase in the hot-electron attempt rate. Note thatthese results are in units of 1/cm2·s·eV; thus we can see thatthe injection of hot electrons at a rate of 26 mA/cm2 creates an

Figure 5. Hot electron simulations. (a) Electron−electron andelectron phonon scattering rates in gold plotted as a function ofenergy above Fermi level. (b) Energy loss rate per fs, obtained bymultiplying scattering rate at a given energy by average energy loss perscattering event. (c) Log scale attempt rate of electrons tunneling intogold/solution interface plotted as a function of energy with andwithout hot electron injection. (d) Linear scale attempt rate plot.

Nano Letters Letter

DOI: 10.1021/acs.nanolett.9b02289Nano Lett. 2019, 19, 6227−6234

6232

attempt rate on the order of ∼1018 /cm2·s·eV. While these mayseem high, they can be understood by considering the velocityof hot electrons in gold to be vHE ≈ 108 cm/s, leading to atransit time across 10 nm thick gold of τAu ≈ 10−14 s.Considering an energy loss rate of PHE ≈ 10 meV/fs, aninjected particle would lose 1 eV in τHE ≈ 100 fs, leading to nrefl≈ 10 attempts/injected particle. A current level of 10 mA/cm2

leads to a hot electron flux of FHE ≈ 1017 /cm2·s. Taking FHE×nrefl we get an interaction rate of ∼1018 /cm2·s·eV. This is asignificant interaction rate which allows us to explain why thesedevices exhibit such high efficiencies with respect to hotelectron driven electrochemistry; despite the high scatteringrates, there is still a significant number of interactions betweenthe hot electron and the gold/solution interface.Finally, we also see a peak in the attempt rate (Figure 5c) at

the hot electron injection energy. This occurs when electronshave been injected into the gold, but have not yet undergonee−e scattering, as e−e scattering for electrons at 2 eV abovethe Fermi level would cause the loss of an average of ∼700meV per scattering event, which is obtained by dividing thevalues in Figure 5a and 5b. Before e−e scattering occurs, theinjected electrons still have significant energy, and if they arebackscattered due to acoustic phonons, they will have anonzero chance of being injected back to the Si. However, ifthe injected electrons are immediately transferred to thesolution before e−e scattering, no current will flow in the golddue to this injected electron and, simultaneously, the net fluxacross the Si/Au interface will increase due to suppression ofbackscattering from the Au into the Si. For our observeddevices, we attribute the observed ΔISi, shown in Figure 3c,d,to this mechanism and call it “direct injection current” here.Furthermore, the peaked shape shown in Figure 4b indicatedthe silicon direct injection current is related to a sharperfeature in the electron distribution, and not the hot electrontail which we see in Figure 5d.In conclusion, we demonstrate that a MIS tunnel diode can

act as a source of hot electrons for efficiently drivingelectrochemical reductions, with efficiencies reaching ∼85%for high biases. This approach is general, and not limited to theSi/Al2O3/Au device with hydrogen reduction shown here.Future experiments could explore hot holes for high-energyoxidation reactions, other redox reactions for carbon-to-fuelreactions, and other materials such as graphene and other 2-Dmaterials that transport electrons more efficiently to see if highefficiencies can be achieved at lower voltages.Methods. Sample Preparation. Moderately phosphorus

doped (Nd = 5 × 1016 cm−3) (100) and heavily phosphorusdoped (Nd = 1 × 1019 cm−3) (100) silicon wafers (MTICorporation) were used as the substrate. Native SiO2 wasremoved with a 1:10 ratio of HF/H2O (Sigma-Aldrich, 49%CMOS grade) etching for 1 min. After oxide etching, 1 nm oftitanium and 100 nm of silver back contact metals wereevaporated in an electron beam evaporator (Temescal,SL1800). To prevent front side damage, a blank Si handlewafer was used after an acetone, IPA, and D.I. water rinse. Themetal insulator semiconductor (MIS) structure was fabricatedby depositing an aluminum oxide insulator layer with AtomicLayer Deposition (Ultratech/Cambridge Savannah ALD)using Trimethyl aluminum (Aldrich, 1001278062) and water(Aldrich, W4502) precursors. Au films (10−100 nm thick)were evaporated under two different conditions. Roomtemperature Au film was deposited with an Electron beamevaporator (Temescal). Cryo (90 K) temperature Au films

were evaporated with a thermal evaporator (Denton VacuumInc., DV-502A). For the device reported in the main text, cryoevaporated Au films were used. Image reversal photo-lithography was done for the control device with holes (KarlSuss, 100UV030). For contact wire attachment, copper wirewrapped with aluminum foil at the one end was used. Twowires are connected to the front and back side of the deviceseach with fast drying silver paint (Ted Pella Inc., 16040-30). Aring contact was drawn in the front side of the device in Aufilm region. For device encapsulation, a glass slide (VMRMicro slides) was used as a back holder. Fabricated deviceswith contacts were placed on the glass with epoxy (GorillaEpoxy clear) to encapsulate the device while leaving the Auelectrode surface exposed.

Electrical Measurements. All the electrochemical I−Vmeasurements were done by using a potentiostat (AdmiralInstrument, Squidstat Prime). Two different channels wereused to control separate voltages applied to the system. Eachchannel has its own working, reference, and counter electrode.The first channel was connected to the Au as the workingelectrode, a platinum wire as the counter electrode, and a Ag/AgCl reference electrode; the second channel consists of a Siemitter as the working electrode, a platinum wire as thecounter electrode, and Au as reference electrode to bias theAu−Si junction. 0.5 M H2SO4 was used as the electrolytesolution. Schottky, ohmic, and four-probe I−V measurement ofdevices were characterized by a Semiconductor ParameterAnalyzer (Keysight B1500a). The roughness of the Au filmevaporated at both 300 K and 90 K substrate temperature werecharacterized using an Atomic Force Microscope.

■ ASSOCIATED CONTENT

*S Supporting InformationThe Supporting Information is available free of charge on theACS Publications website at DOI: 10.1021/acs.nano-lett.9b02289.

Additional figures and explanation of simulations (PDF)

■ AUTHOR INFORMATION

Corresponding Author*E-mail: [email protected].

ORCIDHyun Uk Chae: 0000-0002-7655-2907Ragib Ahsan: 0000-0002-3833-7851Debarghya Sarkar: 0000-0002-5411-7066Stephen B. Cronin: 0000-0001-7089-6672Rehan Kapadia: 0000-0002-7611-0551Author ContributionsH.U.C., Q.L., and R.K. designed the experiments. H.U.C. andQ.L. carried out the sample fabrication and measurements.R.A., D.S., F.R., and R.K. carried out the simulations. H.U.C.,R.A., Q.L., D.S., S.C., and R.K. contributed to analyzing thedata. H.U.C., R.A., and R.K. wrote the paper while all authorsprovided feedback.

Author Contributions†H.U.C. and R.A. contributed equally to this work.

NotesThe authors declare no competing financial interest.

Nano Letters Letter

DOI: 10.1021/acs.nanolett.9b02289Nano Lett. 2019, 19, 6227−6234

6233

■ ACKNOWLEDGMENTS

This work was supported by ACS-PRF Grant No. 55993-ND5(R.K.), AFOSR Grant No. FA9550-16-1-0306 (R.K.), NationalScience Foundation Award No. 1610604 (R.K.), ArmyResearch Office ARO Award No. W911NF-17-1-0325 (S.C.),and the Molecular Foundry at Lawrence Berkeley NationalLaboratory, a user facility supported by the Office of Science,Office of Basic Energy Sciences of the U.S. Department ofEnergy (DOE) under Contract No. DEAC02-05CH11231.R.A. acknowledges a USC Provost Graduate Fellowship.

■ REFERENCES(1) Yan, J.; Kim, M. H.; Elle, J. A.; Sushkov, A. B.; Jenkins, G. S.;Milchberg, H. M.; Fuhrer, M. S.; Drew, H. Nat. Nanotechnol. 2012, 7,472.(2) Vicarelli, L.; Vitiello, M.; Coquillat, D.; Lombardo, A.; Ferrari,A.; Knap, W.; Polini, M.; Pellegrini, V.; Tredicucci, A. Nat. Mater.2012, 11, 865.(3) Yajima, T.; Hikita, Y.; Hwang, H. Y. Nat. Mater. 2011, 10, 198.(4) Brongersma, M. L.; Halas, N. J.; Nordlander, P. Nat.Nanotechnol. 2015, 10, 25.(5) Koppens, F.; Mueller, T.; Avouris, P.; Ferrari, A.; Vitiello, M.;Polini, M. Nat. Nanotechnol. 2014, 9, 780.(6) Wurfel, P. Sol. Energy Mater. Sol. Cells 1997, 46, 43−52.(7) Reineck, P.; Lee, G. P.; Brick, D.; Karg, M.; Mulvaney, P.; Bach,U. Adv. Mater. 2012, 24, 4750−4755.(8) Wu, K.; Chen, J.; McBride, J. R.; Lian, T. Science 2015, 349,632−635.(9) Mukherjee, S.; Libisch, F.; Large, N.; Neumann, O.; Brown, L.V.; Cheng, J.; Lassiter, J. B.; Carter, E. A.; Nordlander, P.; Halas, N. J.Nano Lett. 2013, 13, 240−247.(10) Brongersma, M. L.; Halas, N. J.; Nordlander, P. Nat.Nanotechnol. 2015, 10 (1), 25.(11) Clavero, C. Nat. Photonics 2014, 8, 95−103.(12) Sobhani, A.; Knight, M. W.; Wang, Y.; Zheng, B.; King, N. S.;Brown, L. V.; Fang, Z.; Nordlander, P.; Halas, N. J. Nat. Commun.2013, 4, 1643.(13) Kale, M. J.; Avanesian, T.; Christopher, P. ACS Catal. 2014, 4,116−128.(14) Mohammadparast, F.; Dadgar, A. P.; Tirumala, R. T. A.;Mohammad, S. A.; Topal, C. O.; Kalkan, A. K.; Andiappan, M. J. Phys.Chem. C 2019, 123, 11539−11545.(15) Plummer, E.; Eberhardt, W. Advances in Chemical Physics 2007,533−656.(16) Knight, M. W.; Sobhani, H.; Nordlander, P.; Halas, N. J. Science2011, 332, 702−704.(17) Tomiye, H.; Terano, T.; Nomoto, K.; Kobayashi, T. In A novel2-bit/cell MONOS memory device with a wrapped-control-gate structurethat applies source-side hot-electron injection, VLSI Technology, 2002.Digest of Technical Papers. 2002 Symposium on, 2002; IEEE: pp206−207.(18) Ohnakado, T.; Mitsunaga, K.; Nunoshita, M.; Onoda, H.;Sakakibara, K.; Tsuji, N.; Ajika, N.; Hatanaka, M.; Miyoshi, H. InNovel electron injection method using band-to-band tunneling induced hotelectrons (BBHE) for flash memory with a P-channel cell, ElectronDevices Meeting, 1995. IEDM’95, International, 1995; IEEE: pp279−282.(19) Linic, S.; Christopher, P.; Ingram, D. B. Nat. Mater. 2011, 10,911.(20) Sun, C.-K.; Vallee, F.; Acioli, L.; Ippen, E.; Fujimoto, J. Phys.Rev. B: Condens. Matter Mater. Phys. 1994, 50, 15337.(21) Guo, C.; Rodriguez, G.; Taylor, A. J. Phys. Rev. Lett. 2001, 86,1638.(22) Brandt, N. C.; Keller, E. L.; Frontiera, R. R. J. Phys. Chem. Lett.2016, 7, 3179−3185.

(23) Zibik, E.; Grange, T.; Carpenter, B.; Porter, N.; Ferreira, R.;Bastard, G.; Stehr, D.; Winnerl, S.; Helm, M.; Liu, H. Nat. Mater.2009, 8, 803.(24) Vaziri, S.; Lupina, G.; Henkel, C.; Smith, A. D.; Ostling, M.;Dabrowski, J.; Lippert, G.; Mehr, W.; Lemme, M. C. Nano Lett. 2013,13, 1435−1439.(25) Zeng, C.; Song, E. B.; Wang, M.; Lee, S.; Torres, C. M., Jr;Tang, J.; Weiller, B. H.; Wang, K. L. Nano Lett. 2013, 13, 2370−2375.(26) Niu, S.; Huyan, H.; Liu, Y.; Yeung, M.; Ye, K.; Blankemeier, L.;Orvis, T.; Sarkar, D.; Singh, D. J.; Kapadia, R. Adv. Mater. 2017, 29,1604733.(27) Hou, B.; Shen, L.; Shi, H.; Kapadia, R.; Cronin, S. B. Phys.Chem. Chem. Phys. 2017, 19, 2877−2881.(28) Key, M.; Cable, M.; Cowan, T.; Estabrook, K.; Hammel, B.;Hatchett, S.; Henry, E.; Hinkel, D.; Kilkenny, J.; Koch, J. Phys.Plasmas 1998, 5, 1966−1972.(29) Mooney, J. M.; Silverman, J. IEEE Trans. Electron Devices 1985,32, 33−39.(30) Schmidt, O.; Bauer, M.; Wiemann, C.; Porath, R.; Scharte, M.;Andreyev, O.; Schonhense, G.; Aeschlimann, M. Appl. Phys. B: LasersOpt. 2002, 74, 223−227.(31) Sarkar, D.; Tao, J.; Wang, W.; Lin, Q.; Yeung, M.; Ren, C.;Kapadia, R. ACS Nano 2018, 12, 1656−1663.(32) Shinagawa, T.; Garcia-Esparza, A. T.; Takanabe, K. Sci. Rep.2015, 5, 13801.(33) Fletcher, S. J. Solid State Electrochem. 2009, 13, 537−549.(34) Vanmaekelbergh, D. Electron Transfer in Chemistry 2001, 126−188.(35) Nedrygailov, I. I.; Moon, S. Y.; Park, J. Y. Sci. Rep. 2019, 9,6208.(36) Furube, A.; Du, L.; Hara, K.; Katoh, R.; Tachiya, M. J. Am.Chem. Soc. 2007, 129, 14852−14853.(37) Sellier, J. M.; Fonseca, J. E.; Klimeck, G. In Archimedes, the freeMonte Carlo simulator: A GNU package for submicron semiconductordevices on nanoHUB, 2012. 15th International Workshop onComputational Electronics, 2012; IEEE: pp 1−4.(38) Ladstadter, F.; Hohenester, U.; Puschnig, P.; Ambrosch-Draxl,C. Phys. Rev. B: Condens. Matter Mater. Phys. 2004, 70, 235125.

Nano Letters Letter

DOI: 10.1021/acs.nanolett.9b02289Nano Lett. 2019, 19, 6227−6234

6234