Zener Tunneling and Photocurrent Generation in Quasi-MetallicCarbon Nanotube pn-DevicesMoh. R. Amer,† Shun-Wen Chang,§ Rohan Dhall,† Jing Qiu,‡ and Stephen B. Cronin*,†,‡,§

†Department of Electrical Engineering, ‡Department of Material Science, and §Department of Physics and Astronomy, University ofSouthern California, Los Angeles, California 90089, United States

*S Supporting Information

ABSTRACT: We investigate the electronic and optoelec-tronic properties of quasi-metallic nanotube pn-devices, whichhave smaller band gaps than most known bulk semi-conductors. These carbon nanotube-based devices deviatefrom conventional bulk semiconductor device behavior due totheir low-dimensional nature. We observe rectifying behaviorbased on Zener tunneling of ballistic carriers instead of idealdiode behavior, as limited by the diffusive transport of carriers.We observe substantial photocurrents at room temperature,suggesting that these quasi-metallic pn-devices may have abroader impact in optoelectronic devices. A new technique based on photocurrent spectroscopy is presented to identify theunique chirality of nanotubes in a functional device. This chirality information is crucial in obtaining a theoretical understandingof the underlying device physics that depends sensitively on nanotube chirality, as is the case for quasi-metallic nanotube devices.A detailed model is developed to fit the observed I−V characteristics, which enables us to verify the band gap from thesemeasurements as well as the dimensions of the insulating tunneling barrier region.

KEYWORDS: Quasi-metallic nanotube, band gap, Zener tunneling, photocurrent, exciton transition energy, chirality assignment

Metallic carbon nanotubes (i.e., (n-m) mod 3 = 0) areknown to exhibit small band gaps on the order of 10−

100 meV.1−3 Early scanning tunneling microscopy (STM)studies attributed these small band gaps to curvature-inducedeffects, which produce band gaps in the range 0−50 meV forchemical vapor deposition (CVD) grown nanotubes.4,5

However, more recent studies show that suspended quasi-metallic nanotubes may exhibit band gaps considerably largerthan those induced by curvature.1 To date, there have been nopublished works reporting the use of individual quasi-metallicnanotubes for optoelectronic devices. The optoelectronicproperties of individual semiconducting carbon nanotube pn-junctions, however, have been studied by several groups.6−8 Leeet al. reported nearly ideal diode behavior in suspended carbonnanotube pn-junctions, as well as photocurrent spectroscopyshowing prominent peaks corresponding to E11 and E22

transitions.9,10 Semiconducting pn-devices also exhibit efficientelectron−hole pair generation due to impact ionization.11

Mueller et al. showed efficient narrow-band light emission forsemiconducting nanotubes with similar device structure.12

Optoelectronic devices consisting of individual metallic nano-tubes, however, have not been investigated. For most quasi-metallic nanotubes on substrate, features associated with themini-band gap are largely washed out by the trapped charges inthe underlying substrate.13 Thus, suspended nanotubes thatminimize environmental effects are required in order to buildfunctional devices that operate in this small energy range.

The ability to identify the precise chirality of carbonnanotubes optically has had a tremendous impact in the fieldby enabling a direct comparison of theory and experiment.14

However, there is currently no comprehensive technique toidentify the chirality of a nanotube in a functional device (i.e.,field-effect-transistor structures). For metallic nanotubes,Rayleigh scattering and tunable Raman spectroscopy are theonly optical methods able to unambiguously establish nanotubechiralities. Tunable Raman spectroscopy is difficult to performon individual nanotubes, requiring lasers and notch filters to betuned for each point in the spectrum.15 This technique is alsolimited by the range of the tunable laser, which typically onlycovers a small portion of the Kataura plot. Rayleigh scatteringrequires that the nanotube be suspended over several tens ofmicrometers and cannot be performed on a nanotube in a field-effect-transistor (FET) configuration.16,17 This has greatlylimited our theoretical understanding of nanotube devicephysics, particularly for devices that depend sensitively onnanotube chirality, as is the case in quasi-metallic nanotubedevices. For example, even the effect of electrostatic gating onthe optical transitions in metallic nanotubes (e.g., E11

M) has notbeen investigated.In the work presented here, we measure the electronic and

optoelectronic properties of quasi-metallic nanotube pn-

Received: June 26, 2013Revised: October 7, 2013

Letter

pubs.acs.org/NanoLett

© XXXX American Chemical Society A dx.doi.org/10.1021/nl402334e | Nano Lett. XXXX, XXX, XXX−XXX

devices, which deviate from conventional bulk semiconductordevice behavior due to their low dimensionality. The I−Vcharacteristics of this pn-region are measured as a function ofelectrostatic doping conditions at room temperature and downto 4 K. The rectifying behavior observed at cryogenictemperatures is fit to a Zener (i.e., band-to-band) tunnelingmodel for several different nanotubes with a variety of bandgaps. The photocurrent response of these devices is alsoexplored over a range of electrostatic doping conditions andlaser wavelengths. Photocurrent spectra taken from thesenanotube devices are used to unambiguously identify theirchirality.Figure 1a shows a schematic diagram of the device geometry

used in this study, where two gate electrodes positionedbeneath an individual, suspended quasi-metallic carbon nano-tube creating p- and n-doped regions. The energy band diagramdrawn above the nanotube in this figure illustrates the Zenertunneling process, where electrons in the p-type valence bandtunnel across the insulating barrier region to the n-region. Thecurrent−voltage characteristics of this four-terminal device areplotted in Figure 1b, showing rectifying behavior characteristicof Zener tunneling, particularly for low gate doping (i.e., Vg1 =−Vg2 = 1 V). Photocurrent generation at room temperature ismeasured as a function of laser energy under different pn-gatepotentials (i.e., Vg1 = −Vg2), as shown in Figure 1c. A dominantpeak is observed in these photocurrent spectra correspondingto the E11

M optical transition.The conductance-gate voltage characteristics taken with Vg1

= Vg2 (i.e., three-terminal device configuration) exhibit aminimum conductance near zero gate voltage at the chargeneutrality point as shown in Figure 2b. This is typical ofsuspended quasi-metallic nanotubes and indicates the presenceof a quasi-metallic band gap. Under pn-gating conditions (i.e.,Vg1 = −Vg2) at room temperature, the nanotube exhibits Ohmicbehavior varying between 450.5.kΩ and 2.5MΩ, as shown inFigure 2a. Figure 2c shows the current−voltage characteristicsof the same quasi-metallic nanotube gated in the pn-configuration at 4 K, which shows rectifying behavior andreverse breakdown at high negative bias voltages. The differentdata sets in the plot were taken under different electrostaticgating conditions ranging from Vg1 = −Vg2 = 1−V. In the I−Vbias curves shown in Figure 2c, the current saturation at highbias arises from the finite contact resistance and optical phononemission by hot electrons, discussed previously.13 The reversebreakdown voltage depends strongly on the applied gatepotentials, as plotted in Figure 2d, with the largest reversebreakdown occurring under the weakest gating conditions.Under heavy gating, the insulating barrier region becomessmaller resulting in lower breakdown voltages. The forward biasbreakdown is also plotted in Figure 2d, which follows the sametrend as the reverse breakdown voltage but is approximately 0.1V lower in magnitude. These reverse breakdown voltages aremuch lower than those reported previously on semiconductingnanotube pn-diodes, which have considerably larger band gaps(Eg = 0.6 eV).18 Comparing the low bias conductance under ppand pn gating conditions, we observe a 5 order of magnitudechange from 6 × 10−6 S to 3 × 10−11 S, respectively.In order to establish the breakdown mechanism, we

measured the I−V characteristics by sweeping the voltage inthe forward and reverse directions, as shown in Figure 3a. Here,no hysteresis is observed, indicating that this effect is caused byZener tunneling in the junction rather than avalanchebreakdown, as observed in semiconducting nanotube pn-

junctions.18 A schematic diagram of the Zener tunnelingprocess is illustrated in the energy band diagram in Figure 1a,where electrons from the p-type valence band tunnel across theinsulating region to the n-type conduction band. Electronstunneling across the i-region see a potential barrier ∼Egap inheight. This mechanism is similar to the tunneling observed intunnel diodes, except that Zener tunneling occurs at higher biasvoltages due to larger barrier widths.19 The electric field in thejunction is calculated according to

Figure 1. Dual gate device geometry. (a) Schematic diagram of thedevice geometry and energy band diagram illustrating Zener tunneling.(b) Waterfall plot of the tunable rectifying I−Vbias characteristics takenat 4 K. (c) Photocurrent spectra taken at room temperature showingthe evolution of the E11

M dominant peak.

Nano Letters Letter

dx.doi.org/10.1021/nl402334e | Nano Lett. XXXX, XXX, XXX−XXXB

ε =+V V

L V V( , )bi reverse

g1 g2 (1)

where Vbi is the built-in voltage of the junction, Vreverse is theapplied reverse bias voltage, and L is the width of the insulatingregion. Here, the width L and built-in voltage Vbi are stronglydependent on the electrostatic doping (Vg1 and Vg2). Thetunneling probability can be estimated using the Wentzel−Kramer−Brillouin (WKB) approximation given by18,20

ν ε=

−ℏ

⎡⎣⎢⎢

⎤⎦⎥⎥T

Eexp

4

3 2gap2

F (2)

where Egap is the band gap energy, νF is the Fermi velocity (8.4× 105 m/s), and ε is the electric field in the junction, given byeq 1. The total tunneling current can be calculated as18,20

∫= − −−∞

∞I

eh

T F E eV F E E4

[ ( ) ( )]dtunnel

2

v c (3)

where V is the applied bias voltage, and Fv and Fc are the FermiDirac distributions in the valence and conduction bands,respectively.Figure 3b shows a fit using this Zener model to data taken

under Vg1 = −Vg2 = 3.5 V gating conditions showing excellentagreement with the experimental data. In our fits, the barrierheight (or band gap), L, and Vbi are used as fitting parameters.The tunnel barrier width and built-in potential from these fitsfor each of the I−Vbias characteristics shown in Figure 2c areplotted in Figure 3c using the curvature induced band gap ofEgap = 53.4 meV obtained from the identified chirality, asdiscussed below. Here, the barrier width decreases withincreasing gate voltage while the built-in potential, correspond-

ingly, increases with increasing gating. It should be noted thatthese fits agree well with the experimental data under highelectrostatic doping conditions (i.e., |Vg| >1 V) but not verywell under low doping conditions (i.e., |Vg| = 1 V, see Figure S3in the Supporting Information), most likely due the Schottkybarriers at the electrical contacts, which can significantly affectthe transport at low doping.21,22

Another interesting aspect of these quasi-metallic nanotubepn-devices is their ability to produce a finite photocurrent atroom temperature, even though they do not exhibit rectifyingbehavior in this temperature range. Figure 4a shows thephotocurrent map taken under all possible electrostatic gatingconditions. Maximum photocurrent is observed when thenanotube is doped in either a pn or np configuration (i.e., Vg1 =−Vg2). Here, the photocurrent changes sign with pn/nppolarity and increases linearly with laser power indicating theexistence of a charge separation region. Spatially mappedphotocurrent shows a peak photocurrent in the center of thenanotube, which further demonstrates that the photocurrentoriginates from the pn-junction rather than at the contacts(inset of Figure S4a in the Supporting Information). By tuningthe incident laser energy, we collect photocurrent spectra, asshown in Figures 1c and 4b, under various pn-gating conditions(i.e., Vg1 = −Vg2). A dominant peak is observed at 2.14 eVcorresponding to the E11

M optical transition of the nanotube.This nanotube also exhibited a radial breathing mode (RBM) inits Raman spectrum at 215.7 cm−1, which corresponds to adiameter of 1.05 nm by the relation ωRBM = 227/dt.

23 Together,the nanotube diameter and the E11

M optical transition enableunambiguous identification of this nanotube chirality to be(13,1), as shown in Figure 4c.24 As the pn-gate potentials (Vg1

= −g2) are increased, the E11M peak shifts upward in energy and

Figure 2. Electron transport of quasi-metallic nanotube pn-device. Current-bias voltage characteristics taken at (a) room temperature and (c) 4 Kwith the device gated in a pn configuration (i.e., Vg1 = −Vg2). Here, no rectifying behavior is observed at room temperure because of the small bandgap. Rectifying behavior is observed at 4 K with tunable forward and reverse breakdown voltages. (b) Current-gate voltage characteristics at 300 Kshowing a minimum near 0 V. This dip in the conductance arises from the existence of a band gap. (d) Forward and reverse bias breakdown voltagesobserved under different doping conditions, demonstrating tunability between 1 and 0.2 V.

Nano Letters Letter

dx.doi.org/10.1021/nl402334e | Nano Lett. XXXX, XXX, XXX−XXXC

increases by a factor of 5 in intensity, as shown in Figure 4d.This relatively small shift could either be due to electrostaticforce-induced strain25 or band renormalization.26 Tight bindingpredicts two E11

M transitions in nonarmchair metallic nanotubes(i.e., E11

− and E11+ ), corresponding to different quantized

circumferential momenta in the zone folding scheme of carbonnanotubes. Here, however, we only observe one peak at 2.14 eVin the photocurrent spectrum, corresponding to E11

− . Theabsence of the corresponding E11

+ peak at 2.58 eV from thephotocurrent spectrum is interesting. In previous tunableRaman studies, the E11

+ feature was also missing for smalldiameter nanotubes dt < 1.3 nm24 and was attributed to thenodal nature of exciton−phonon coupling.24

On the basis of the (13,1) nanotube chirality, which gives adiameter of dt = 1.05 nm and a chiral angle of θ = 3.7°, wecalculate the curvature-induced band gap to be 53.4 meV basedon the formula

θ= =Ec

dcos 3 53.4 meVgap

t2

(4)

where c is 60 meV·nm2.5 The results described abovedemonstrate that these mini band gaps can produceelectron−hole pair separation at room temperature, whichcan be utilized in optoelectronic devices. The effective bandgaps of the nanotubes measured in this study can also bedetermined by fitting the I−Vgate of each sample to a Landauertransport model,13 in which the conductance of the nanotube isgiven by

∫ λλ

= =+

∂∂−∞

∞⎛⎝⎜

⎞⎠⎟

⎛⎝⎜

⎞⎠⎟⎛⎝⎜

⎞⎠⎟G

Rqh

E TE T L

fE

E1 4 ( , )

( , )dCNT

CNT

2eff

eff

(5)

Here λeff is the effective mean free path for a scattered electronin the system given by Mathiessen’s rule.27,28 The mean freepath for the acoustic phonon scattering length and the opticalphonon scattering length depend on the nanotube’s diameter(dt), as discussed previously.29−31 For the nanotube in Figure3b, the diameter is 1.05 nm as measured by the RBM peak. L isthe nanotube channel length, which is 2.5 μm, and f is theFermi-Dirac distribution. The integral is taken over the densityof states, which is given by

∑==

−

D EE k

k( )

d ( )

dJ

Nj

1

1

where the derivative is taken over a hyperbolic dispersionrelation with a small band gap. Using this Landauer transportmodel to fit the I−Vgate data in Figure 2b, gives a band gap of180 meV, which is considerably larger than the value predictedfrom curvature-induced effects (i.e., 53.4 meV). Thisdiscrepancy between the transport band gap extracted fromthe conductance-gate voltage curve and the curvature-inducedband gap could be related to recent work indicating theexistence of Mott insulating states.32−34 In this interpretation,the band gap extracted from transport data is caused by atransition to an insulating state, which causes the largemodulation observed in the conductance around the chargeneutrality point, and the actual energy separation between theconduction and valence bands is considerably smaller (i.e., 53.4meV). For the I−Vbias data in Figures 2c and 3b, we wereunable to obtain a reasonable fit using the Zener tunnelingmodel with the large band gap value (i.e., 180 meV), providingfurther evidence that the actual energy separation between theconduction and valence bands is on the order of the curvatureinduced band gap.In summary, the mini band gaps in ultraclean, quasi-metallic

nanotubes enable the creation of pn devices using dualelectrostatic gate structures. While these pn junctions showOhmic behavior at room temperature, rectifying behavior isobserved at cryogenic temperatures with a tunable breakdownvoltage. This breakdown is attributed to band to band (orZener) tunneling inside the junction and can be controlled byvarying the electrostatic doping. These pn-devices produce afinite photocurrent at room temperature enabling the preciseidentification of nanotube chirality in a functional device. Adominant E11

− peak is observed in the photocurrent spectrawhile the E11

+ peak is absent, consistent with previous opticalstudies. The electrostatic doping dependence of these photo-current spectra show an increase in the photocurrent intensityand a slight upshift of the E11

− optical transition energy underheavily doped conditions. These results demonstrate that quasi-

Figure 3. Zener tunneling in quasi-metallic nanotube pn diodes. (a)Forward and reverse sweeps for a quasi-metallic pn diode, showing nohysteresis. (b) A fit of the measured rectifying behavior (black circles)to the Zener model (red curve). (c) The extracted barrier width andthe built-in voltage used to fit the measured rectifying I−Vbias curves.

Nano Letters Letter

dx.doi.org/10.1021/nl402334e | Nano Lett. XXXX, XXX, XXX−XXXD

metallic pn-devices may have a broader impact in optoelec-tronic devices.Methods. Fabrication. Devices are fabricated by first

etching a 500 nm deep trench in pregrown silicon oxide (1μm thick) and silicon nitride (100 nm thick) on silicon, asillustrated in Figure 1a. Pt electrodes are deposited on top ofthe silicon nitride, which act as the source and drain electrodes,and 1 μm wide gate electrodes are deposited in the trench with200 nm separation. Mo catalyst is deposited on top of thesource and drain electrodes to initiate the nanotube growth at875 °C for 10 min, as described previously.13,35−37 Thenanotube growth is the final step in this sample fabricationprocess, which ensures that these nanotubes are notcontaminated by any chemical residues from the lithographicfabrication processes.Characterization. We characterize the devices by measuring

I−Vbias of each device. Devices that show Imax ∼ 10/L, where Lis the trench width in micrometers and Imax is the maximumcurrent attained in the nanotube in microamperes, indicateindividual nanotubes and are selected for further study.37

Raman spectra are collected using 532 nm and 633 nm lasers toidentify the nanotube’s RBM frequency and diameter.38,39

Photocurrent spectra were collected with a Fianium super-continuum white light laser used in conjunction with aPrinceton Instruments double monochromator. The powerspectral density was measured and compared with themeasured photocurrent in order to distinguish the nanotube’speaks from the laser peaks. Low-temperature measurements

were carried out under vacuum using a continuous flow liquidhelium optical cryostat.

■ ASSOCIATED CONTENT*S Supporting InformationAdditional device images, rectifying I−Vbias fits, energy banddiagrams, photocurrent generation from different quasi-metallicnanotube pn devices, and electrostatic switching betweenohmic and rectifying behavior. This material is available free ofcharge via the Internet at http://pubs.acs.org.

■ AUTHOR INFORMATIONCorresponding Author*E-mail: scronin@usc.edu.NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTSThis work was supported in part by Department of Energy(DOE) Award No. DE-FG02-07ER46376. A portion of thiswork was carried out in the University of California SantaBarbara (UCSB) nanofabrication facility, part of the NSFfunded National Nanotechnology Infrastructure Network(NNIN) network.

■ REFERENCES(1) Deshpande, V. V.; Chandra, B.; Caldwell, R.; Novikov, D. S.;Hone, J.; Bockrath, M. Mott insulating state in ultraclean carbonnanotubes. Science 2009, 323, 106−110.

Figure 4. Photocurrent generation and exciton dynamics in quasi-metallic nanotubes. (a) Measured photocurrent map for different Vg1 and Vg2configurations. The maximum photocurrent is observed when the device is gated in either a pn and np configuration. (b) Photocurrent spectra takenunder different pn gating conditions showing the evolution of a prominent excitonic peak corresponding to the E11

M metallic transition. (c) Katauraplot showing unique chiral identification based on the nanotube’s RBM at 215.7 cm−1 and the E11

M energy at 2.14 eV. The intercept identifies ananotube chirality of (13,1). (d) The E11

M optical transition energy (red) and photocurrent magnitude (blue) plotted at various electrostatic dopingconditions.

Nano Letters Letter

dx.doi.org/10.1021/nl402334e | Nano Lett. XXXX, XXX, XXX−XXXE

(2) Han, M. Y.; Ozyilmaz, B.; Zhang, Y.; Kim, P. Energy band-gapengineering of graphene nanoribbons. Phys. Rev. Lett. 2007, 98,206805.(3) Zhang, Y.; Tang, T.-T.; Girit, C.; Hao, Z.; Martin, M. C.; Zettl,A.; Crommie, M. F.; Shen, Y. R.; Wang, F. Direct observation of awidely tunable bandgap in bilayer graphene. Nature 2009, 459, 820−823.(4) Ouyang, M.; Huang, J. L.; Cheung, C. L.; Lieber, C. M. Energygaps in “metallic” single-walled carbon nanotubes. Science 2001, 292,702−705.(5) Sasaki, K.; Farhat, H.; Saito, R.; Dresselhaus, M. S. Kohn anomalyin Raman spectroscopy of single wall carbon nanotubes. Physica E2010, 42, 2005−2015.(6) Barkelid, M.; Steele, G. A.; Zwiller, V. Probing OpticalTransitions in Individual Carbon Nanotubes Using Polarized Photo-current Spectroscopy. Nano Lett. 2012, 12, 5649−5653.(7) Lee, J. U. Band-gap renormalization in carbon nanotubes: Originof the ideal diode behavior in carbon nanotube pn structures. Phys.Rev. B 2007, 75, 075409.(8) Liu, C.-H.; Wu, C.-C.; Zhong, Z. A Fully Tunable Single-WalledCarbon Nanotube Diode. Nano Lett. 2011, 11, 1782−1785.(9) Lee, J. U.; Codella, P. J.; Pietrzykowski, M. Direct probe ofexcitonic and continuum transitions in the photocurrent spectroscopyof individual carbon nanotube pn diodes. Appl. Phys. Lett. 2007, 90,053103−053103−3.(10) Lee, J. U.; Gipp, P.; Heller, C. Carbon nanotube pn junctiondiodes. Appl. Phys. Lett. 2004, 85, 145−147.(11) Gabor, N. M.; Zhong, Z.; Bosnick, K.; Park, J.; McEuen, P. L.Extremely efficient multiple electron-hole pair generation in carbonnanotube photodiodes. Science 2009, 325, 1367−1371.(12) Mueller, T.; Kinoshita, M.; Steiner, M.; Perebeinos, V.; Bol, A.A.; Farmer, D. B.; Avouris, P. Efficient narrow-band light emissionfrom a single carbon nanotube p−n diode. Nat. Nanotechnol. 2009, 5,27−31.(13) Amer, M. R.; Bushmaker, A.; Cronin, S. B. The Influence ofSubstrate in Determining the Band Gap of Metallic CarbonNanotubes. Nano Lett. 2012, 12, 4843−4847.(14) Bachilo, S. M.; Strano, M. S.; Kittrell, C.; Hauge, R. H.; Smalley,R. E.; Weisman, R. B. Structure-assigned optical spectra of single-walled carbon nanotubes. Science 2002, 298, 2361−2366.(15) Cronin, S. B.; Yin, Y.; Walsh, A.; Capaz, R. B.; Stolyarov, A.;Tangney, P.; Cohen, M. L.; Louie, S. G.; Swan, A. K.; Unlu, M. S.;Goldberg, B. B.; Tinkham, M. Temperature dependence of the opticaltransition energies of carbon nanotubes: The role of electron-phononcoupling and thermal expansion. Phys. Rev. Lett. 2006, 96, 127403.(16) Sfeir, M. Y.; Wang, F.; Huang, L. M.; Chuang, C. C.; Hone, J.;O’Brien, S. P.; Heinz, T. F.; Brus, L. E. Probing electronic transitionsin individual carbon nanotubes by Rayleigh scattering. Science 2004,306, 1540−1543.(17) Joh, D. Y.; Herman, L. H.; Ju, S. Y.; Kinder, J.; Segal, M. A.;Johnson, J. N.; Chan, G. K. L.; Park, J. On-Chip Rayleigh Imaging andSpectroscopy of Carbon Nanotubes. Nano Lett. 2011, 11, 1−7.(18) Bosnick, K.; Gabor, N.; McEuen, P. Transport in carbonnanotube p-i-n diodes. Appl. Phys. Lett. 2006, 89, 163121−163121−3.(19) Esaki, L. Discovery of the tunnel diode. IEEE Trans. ElectronDevices 1976, 23, 644−647.(20) Jena, D.; Fang, T.; Zhang, Q.; Xing, H. Zener tunneling insemiconducting nanotube and graphene nanoribbon pn junctions.Appl. Phys. Lett. 2008, 93, 112106−112106−3.(21) Javey, A.; Guo, J.; Wang, Q.; Lundstrom, M.; Dai, H. Ballisticcarbon nanotube field-effect transistors. Nature 2003, 424, 654−657.(22) Martel, R.; Derycke, V.; Lavoie, C.; Appenzeller, J.; Chan, K.;Tersoff, J.; Avouris, P. Ambipolar electrical transport in semi-conducting single-wall carbon nanotubes. Phys. Rev. Lett. 2001, 87,256805.(23) Araujo, P. T.; Maciel, I. O.; Pesce, P. B. C.; Pimenta, M. A.;Doorn, S. K.; Qian, H.; Hartschuh, A.; Steiner, M.; Grigorian, L.; Hata,K.; Jorio, A. Nature of the constant factor in the relation between

radial breathing mode frequency and tube diameter for single-wallcarbon nanotubes. Phys. Rev. B 2008, 77, 241403.(24) Doorn, S. K.; Araujo, P. T.; Hata, K.; Jorio, A. Excitons andexciton-phonon coupling in metallic single-walled carbon nanotubes:Resonance Raman spectroscopy. Phys. Rev. B 2008, 78, 165408.(25) Cronin, S. B.; Swan, A. K.; Unlu, M. S.; Goldberg, B. B.;Dresselhaus, M. S.; Tinkham, M. Resonant Raman spectroscopy ofindividual metallic and semiconducting single-wall carbon nanotubesunder uniaxial strain. Phys. Rev. B 2005, 72, 035425.(26) Spataru, C. D.; Leonard, F. Quasiparticle and excitonrenormalization effects in electrostatically doped semiconductingcarbon nanotubes. Chem. Phys. 2013, 413, 81−88.(27) Pop, E. Energy dissipation and transport in nanoscale devices.Nano Res. 2010, 3, 147−169.(28) Bushmaker, A. W.; Deshpande, V. V.; Hsieh, S.; Bockrath, M.W.; Cronin, S. B. Gate Voltage Controllable Non-Equilibrium andNon-Ohmic Behavior in Suspended Carbon Nanotubes. Nano Lett.2009, 9, 2862−2866.(29) Liao, A.; Zhao, Y.; Pop, E. Avalanche-induced currentenhancement in semiconducting carbon nanotubes. Phys. Rev. Lett.2008, 101, 256804.(30) Zhao, Y.; Liao, A.; Pop, E. Multiband mobility in semi-conducting carbon nanotubes. IEEE Electron Device Lett. 2009, 30,1078−1080.(31) Amer, M.; Bushmaker, A.; Cronin, S. Anomalous kink behaviorin the current-voltage characteristics of suspended carbon nanotubes.Nano Res. 2012, 1−9.(32) Mott, N. Metal-insulator transition. Rev. Mod. Phys. 1968, 40,677−683.(33) Bushmaker, A. W.; Deshpande, V. V.; Hsieh, S.; Bockrath, M.W.; Cronin, S. B. Large modulations in the intensity of raman-scattered light from pristine carbon nanotubes. Phys. Rev. Lett. 2009,103, 067401.(34) Dhall, R.; Chang, S.-W.; Liu, Z.; Cronin, S. B. Pronouncedelectron-phonon interactions in ultraclean suspended carbon nano-tubes. Phys. Rev. B 2012, 86, 045427.(35) Bushmaker, A. W.; Deshpande, V. V.; Bockrath, M. W.; Cronin,S. B. Direct Observation of Mode Selective Electron−PhononCoupling in Suspended Carbon Nanotubes. Nano Lett. 2007, 7,3618−3622.(36) Deshpande, V.; Hsieh, S.; Bushmaker, A.; Bockrath, M.; Cronin,S. Spatially resolved temperature measurements of electrically heatedcarbon nanotubes. Phys. Rev. Lett. 2009, 102, 105501.(37) Pop, E.; Mann, D.; Cao, J.; Wang, Q.; Goodson, K.; Dai, H.Negative differential conductance and hot phonons in suspendednanotube molecular wires. Phys. Rev. Lett. 2005, 95, 155505.(38) Saito, R.; Hofmann, M.; Dresselhaus, G.; Jorio, A.; Dresselhaus,M. Raman spectroscopy of graphene and carbon nanotubes. Adv. Phys.2011, 60, 413−550.(39) Saito, R.; Dresselhaus, G.; Dresselhaus, M. S. Physical Propertiesof Carbon Nanotubes; : World Scientific: New York, 1998; Vol. 35.

Nano Letters Letter

dx.doi.org/10.1021/nl402334e | Nano Lett. XXXX, XXX, XXX−XXXF

Zener Tunneling and Photocurrent Generation in

Suspended Quasi-Metallic Carbon Nanotube pn-

Devices: Supplementary Information

Moh. R. Amer1, Shun-Wen Chang

3, Rohan Dhall

1, Jing Qiu

2, and Stephen B. Cronin*

1,2,3

Departments of and Electrical Engineering1, Material Science

2, Physics and Astronomy

3,

University of Southern California, Los Angeles, CA, 90089

Table of Contents:

1- Additional information on the device structure:

2- Energy band diagram of each region in the measured rectifying I-Vbias

3- Additional Zener model fits for different electrostatic doping conditions.

4- Photocurrent spectra of additional quasi-metallic samples with different mini-band gaps.

5- Switching between ohmic behavior and rectifying behavior, and tunable rectifying

behavior of additional quasi-metallic samples with narrower mini-band gaps.

1- Additional Information on the Device Structure:

Drain Source

VG2

VG1

CNT

3 µm

Figure S1. Quasi-Metallic nanotube device structure. (a) Schematic diagram showing the final

device structure. (b) SEM image of the suspended nanotube device. (c) Vg1 sweeps at different Vg2

sweeps. The charge neutrality points shift with different electrostatic doping. This effect will cause

the insulating region position to slightly vary along the length of the nanotube. The inset shows the

G-band Raman active mode exhibiting Kohn anomaly, indicative of a metallic nanotube.

(a) (b)

(c)

-6 -4 -2 0 2 4 60.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

Curr

ent (µ

A)

Vg1

(V)

Vg2

=from -7V to +7V, step = 1V

1300 1400 1500 1600 1700

0

1000

2000

3000

Raman Shift (cm-1)

Inte

nsity

2- Energy band diagram of each region in the measured rectifying I-Vbias:

The band diagram of the rectifying I-Vbias when the quasi-metallic nanotube is doped in a pn

configuration is shown in figure S2 for each different highlighted regime.

-0.3 -0.2 -0.1 0.0 0.1 0.2-0.3

-0.2

-0.1

0.0

0.1

0.2

0.3

Cu

rre

nt

(µA

)

Bias Voltage (V)

Vg1

=-Vg2

=3.5V

(b)

(c)

(d)

(a)

(a) (b)

(c) (d)

Figure S2. Energy band diagram of forward and reverse biases for a pn gated metallic

nanotube. The band diagram of each highlighted region on the IVbias curve drawn for illustrative

purposes for the case of (a) positive Vbias, (b) Vbias, = 0 V, (c) negative Vbias, and (d) high negative

Vbias.

Ec

Ev

Ef

Ec

Ev

Efp

Ec

Ev

Efp

Efn

Ec

Ev

Efp Efn

Efn

Egap

eVbias

eVbias

eVbias

Egap

Egap Egap

3- Additional Zener model fits for different electrostatic doping conditions:

-1.0 -0.5 0.0 0.5 1.0-0.4

-0.2

0.0

0.2

0.4

Measured

Fit

Vg1

= -Vg2

= 1.5V

Curr

ent (µ

A)

Bias Voltage (V)

-0.6 -0.4 -0.2 0.0 0.2 0.4 0.6

-0.4

-0.2

0.0

0.2

0.4

Measured

fitsCurr

ent (µ

A)

Bias Voltage (V)

Vg1

= -Vg2

= 2V

-0.4 -0.2 0.0 0.2 0.4

-0.4

-0.2

0.0

0.2

0.4V

g1=-V

g2= 4V

Measured

FitCu

rre

nt

(µA

)

Bias Voltage (V)

-0.4 -0.2 0.0 0.2 0.4-0.4

-0.2

0.0

0.2

0.4

Measured

Fit

Cu

rre

nt

(µA

)

Bias Voltage (V)

Vg1

= -Vg2

= 5V

Figure S3. Zener model and measured rectifying I-Vbias. Zener model fits for different

electrostatic doping conditions (Vg1=-Vg2). (a) 1.5V, (b) 2V, (c) 4V, and (d)5V. The fitting parameters

used to generate these fits are plotted in figure 3c. As discussed in the main paper, the fits agree well

for high electrostatic doping conditions as compared to lower doping conditions.

(a) (b)

(c) (d)

4- Photocurrent Power Dependence and Photocurrent Spectra of Additional Quasi-

Metallic Samples with Different Mini-Band Gaps:

0 2 4 6 8-150

-100

-50

0

50

100

150

PN (Vg1

= -Vg2

= -7V)

NP (Vg1

= -Vg2

= 7V)Photo

curr

ent (n

A)

Output Laser Power (mW)

1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8 3.0

-4

-2

0

2

4800

800

400

0

200

600

Vg1 =

-V

g2 (

V)

Energy (eV)

800

600

400

200

0

Ph

oto

cu

rren

t (pA

)

1.6 1.8 2.0 2.2 2.4 2.6 2.8 3.0

0

100

200

300

400

Ph

oto

cu

rre

nt

(pA

)

Energy (eV)

Vg1

=-Vg2

=1V

Vg1

=-Vg2

=3V

Vg1

=-Vg2

=5V

Vg1

=-Vg2

=7V

(a) (b)

Figure S4. Photocurrent power dependence and photocurrent spectra at room temperature. (a) pn and np power dependence of the nanotube showing a linear dependence which rolls out the

effect of the thermally generated current. A 633nm laser is used with the laser spot centered in the

middle of the nanotube as illustrated in the graphical inset. The laser output power was measured

using a silicon photodiode while neutral density filters are used to vary the output power. The inset

figure shows the spatial photocurrent scan along the length of a 4µm long nanotube when biased in

a pn configuration. The photocurrent peaks in the middle section of the nanotube indicating of a

charge separation region. (b) Photocurrent spectra map of the sample in figure 1c with additional

doping conditions showing a peak occurring at 2.14eV (c,d) Photocurrent spectra of 2 different

quasi-metallic samples taken at different gating conditions. The spectra show the evolution of

prominent peaks which are attributed to Eii in accordance with the photocurrent peak observed in

figure 1c and S1b.

1.6 1.8 2.0 2.2 2.4

0.0

0.4

0.8

1.2

1.6

Ph

oto

cu

rre

nt

(nA

)

Energy (eV)

Vg1

=-Vg2

= 1

Vg1

=-Vg2

= 3

Vg1

=-Vg2

= 5

Vg1

=-Vg2

= 7

(c)

800

400

0

200

600

(d)

0 1 2 3 4-5

0

5

10

15

20C

urr

en

t (n

A)

Position (µm)

5- Switching Between Ohmic Behavior and Rectifying Behavior, and Tunable

Rectifying Behavior of Additional Quasi-Metallic Samples with narrower mini-band

gaps:

-0.3 -0.2 -0.1 0.0 0.1 0.2 0.3

-4

-2

0

2

4V

g1= -V

g2= -8

Vg1

= Vg2

= -8

Cu

rre

nt

(µA

)

Bias Voltage (V)

T=4K

-10 -5 0 5 100.0

0.2

0.4

0.6

Vg1

=Vg2

(V)

Cu

rre

nt

(µA

)

-1.0 -0.5 0.0 0.5 1.0

-1

0

1

2

Curr

ent (µ

A)

Bias Voltage (V)

Vg1

=Vg2

= -1V

Vg1

=-Vg2

= -1V

T=4K

160 180 200 220 240 260Inte

nsity (

a.u

.)

Raman Shift (cm-1)

Figure S5. Current-Voltage characteristics at 4K of additional samples. (a) I-Vbias at 4K of the

sample in figure 2 showing a pronounced rectifying behavior when gated in a pn configuration

while ohmic behavior is observed when gated in pp configuration. The inset shows the Raman

RBM peak which gives us the advantage of identifying the chirality of the nanotube along with

the photocurrent spectra in figure 1c (b) another sample showing the tunability between ohmic

behavior and rectifying behavior at 4K. The inset shows the Current- Gate Voltage characteristics

at room temperature. (c,d) tunable rectifying behavior with doping conditions of 2 different quasi-

metallic samples at 4K. These samples exhibit narrower band gaps than the samples in figures

S3(a,b) (see insets for I-Vgate taken at room temperature) further cooperating the strong

dependence of Zener tunneling with the quasi-metallic nanotube’s mini-band gap.

(c)

(b) (a)

(d)

-0.2 -0.1 0.0 0.1 0.2-0.4

-0.2

0.0

0.2

0.4

T=4K

Cu

rre

nt

(µA

)

Bias Voltage (V)

Vg1

=-Vg2

=0V

Vg1

=-Vg2

=-1V

Vg1

=-Vg2

=-2V

Vg1

=-Vg2

=-3V

Vg1

=-Vg2

=-4V

Vg1

=-Vg2

=-5V

-6 -4 -2 0 2 4 60.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

Cu

rre

nt

(µA

)

Vg1

= Vg2

(V)

-0.10 -0.05 0.00 0.05 0.10

-0.4

-0.2

0.0

0.2

0.4

Cu

rre

nt

(µA

)

Bias Voltage (V)

Vg1=-Vg2= 0 V Vg1=-Vg2= 0.5 V

Vg1=-Vg2= 1 V Vg1=-Vg2= 1.5 V

Vg1=-Vg2= 2 V Vg1=-Vg2= 3 V

Vg1=-Vg2= 4 V Vg1=-Vg2= 5 V

Vg1=-Vg2= 6 V Vg1=-Vg2= 7 V

Vg1=-Vg2= 8 V Vg1=-Vg2= 9 V

T=4K

-4 -2 0 2 40

2

4

6

8

10

12

Curr

ent (

A)

Vg1

=Vg2

(V)