Carbon Nanotube Electronics and Photonics

8/2/2019 Carbon Nanotube Electronics and Photonics

http://slidepdf.com/reader/full/carbon-nanotube-electronics-and-photonics 1/8

Carbon nanotubeelectronics and

photonicsPhaedon Avouris

Their geometry, mechanical flexibility, and unique charge-transport properties make carbon nano-

tubes ideally suited to supplant silicon in the next generation of FETs.

Phae don Avo uris is an IBM fellow and manager of nanometer-scale science and technology at the IBM Thomas J. Watson Research

Center in Yorktown Heights, New York.

Since the demonstration of fhe first transistors in 1947,computing electronics has been hased mostly on one mate-rial, silicon. The emerge nce of MOSFETs in the 1960s and thedevelopment of microlifhography and other fabricationtechniques have allowed the continuous miniaturization ofsilicon devices. The resulting increases in density andswitching speed and the reduced cost have made today's in-formation age possible. Unfortunately, for various techno-logical and even fundamental physical reasons, that scalingprocess is reaching its useful limits. In resp onse, researchersare exploring ideas for logic devices based on either differ-ent operating principles or different materials whose prop-erties can sustain the scaling to yet smaller size and greaterperformance.

Among the alternative technologies, carbon-hased elec-

tronics is particularly significant. It is based on the 1991 dis-covery of carbon nanotubes' and the more recent studies ofindividual graphite layers, so-called graphenes.^ A single-walled CNT can be thou ght of as a ribbon of graphene rolledinto a seamless, one-atom-thick cylinder. The ribbon's w idthdetermines the CNT's diameter (typically 1-2 nm), and thedirection in which the graphene is rolled defines the CNTchirality. Moreover, mo st of the CN T's electrical prop ertiescan be deduced from those of graphene after additionalboundary conditions are introduced.^ Both materials displayoutstanding electrical properties such as either ballistictransport or diffusive transport with long mean free paths.And both are mechanically strong, flexible, and thermallyconductive. But unlike graphene, semiconducting nan-

otubes have fairly large bandgaps that allow them to pro-duce large on/off current rafios when fashioned into logicdevices such as an FET,

The structure of traditional MOSFETs is described in thebox on page 35 . In a CNT FET, the silicon chann el is rep lacedby an individual single-walled CNT (see figure 1); thefirst such transistors were independently reported in 1998from groups at Delft University of Technology and IBM'sThomas J. Watson Research Center (see the article by CeesDekker in PHYSICS TODAY, May 1999, page 22).''

For the last several decades optoelectronics has been dom-inated by m-V semiconductors, particularly gallium arsenideand its alloys. In 2002, Rice University researchers observed

photolum inescence from CNTs," and the following year, IBMresearchers reported electroluminescence.'' Those later devel-opm ents open ed th e possibility of basing both com puting elec-

tronics and optoelectronics on the same material; for recentreviews, see reference 3 .

Nanotube transistors

A CNT has an atomic and electronic structu re that gives itunique advantages as an FET channel. Its small, 1- to 2-nmdiameter enhances the gate's ability to control the potentialof the channel, particularly when the gate is configured towrap around the CNT. The strong coup ling makes the CNTthe ultimate thin-body semiconductor system and allows oneto shrink an FET in size yet still avoid the dreaded "short-channel effects," in which the gate field basically loses con-trol of the device.

That all bonds in the CNT are satisfied and the surfaceis atomically smoo th also has impo rtant im plications. Absent

are the scattering of electrons by surface states and the rou gh-ness that plagues conventional FETs, especially at high gatevoltages. TTie electrical ban dg ap of a semicon ducting CNT isinversely proportional to the tube diameter,' which allowssuch tubes to be used in various different applications. Asdiscussed later, the key advantage is the CNT's high chargemobility /J and its low capacitance. For typical performancecharacteristics as an FET channel, see figure 2.

The "simple" substitiition of a silicon channel with aCNT dramatically alters the detailed physics of the device.The CNT's quasi-one-dimensional character, strong electronconfinement, nanometer width, and strong covalent bondingdrastically affect the channel's electrical- and thermal-transport properties. The chemistry of the metal-carbon

contacts, the interaction of insulators with CNTs, and the sen-sitivity of those one-atom-thick structures to environmentalinfluences all raise fascinating challenges and opp ortun ities.

Carrier scattering

A key reason to use CNTs in electronics is their excellenttransport properties. The elastic scattering of charge carriersis weak, and the carrier mean free paths are long—on theorder of a micrometer for pure CNTs.' Inelastic phonon-scattering processes therefore tend to dominate. Thestrength of the scattering depends on the energy of the car-riers. At low temperatures and low bias, only acousticphonons can couple to electrons, which gives rise to an in-verse temperature dependence of the carrier mobility insemiconducting CNTs. The weak temperature depend encein quasi-one dimension is due to phase-space restrictions



The pr inc iple of a M O S F E T

The basic structure ofa MOSFET involves a channel made of asemiconductor, typically silicon, connected to two electrodes: asource and drain. An insulating thin film, usually silicon diox ide,separates the channe l, source, and d rain from a third electrodecalled the gate (G), as pictured schematically. By applying a volt-age Vg^ between thegate and source, the conductance of the

semiconducting channel can be modulated. Charge carriers

(electrons or holes) traveling between source and drainencounter a material- and structure-dependent energy barrierin the bulk of the semiconductor.

The barrier, shown in the energy profile as a fun rtion of thechannel position, lies in the conduction band for an electron(the charge carrier in ann-typesemiconductor) or in the valenceband for a hole (the carrier in a p-type semiconductor). The elec-tric field generated by the biased gate, depending on its direc-tion and strength, lowers or raises the barrier and thus changesthe conductivity. Forexample, a positive gate voltage lowers thebarrier for an electron and a negative gate voltage raises it. Thedrain-source voltage f,,, provides the d riving force that movescarriers across the channel.

The scaling of a MOSFET involves reducing its dimensionsand supply voltage by the same factor a. The speed of the

device thus increases by a and the density of devices, whenpacked together, by a', while the power density remainsconstant.

A number of limitations arise as dimensions reach the

nanometer regime. Two variables are already close to their scal-ing limits. One of them is the thickness of the gate insulator, typ -ically SiO;, which is currently about 1 nm . Were that thickness to

shrink much m ore, the leakage current from quantum tunnelingwould undermine the switching action of the FET. The leakagecurrent also contributes to a major problem—excess powerdissipation.

Another parameter that is difficult to scale dow n is the oper-ating voltage V usually about 1 V, The dynamic switching energyof a device is given by ViiC^^^^ + CJV^, where Q^,^ and Q are the

device and wiring capacitance contributions, respectively. IfV Isnot reduced as devices shrink in size and are more closelypacked, the dissipated power density will increase. High tem-peratures then develop locally and decrease both the perfor-mance and lifetime of devices. Ways to reduce V and the totalcapacitance are, therefore, being vigorously pursued.

To maintain thebasic princ iple of the FET—that Is, the con-

Substrate

POSITION

trol of transport in the channel using gate bias—one also needsto m aintain a certain relationship between the screening lengthX—the scale over which The elearic file is screened out in the

semiconductor—and the length of the gate, /. ,. Specifically, i^should be a few times X, which depends on the channel and

oxide-insulator thicknesses d^ and d^,, respectively, and theirdielectric constants, e and £„.. In a planar FET, \ = Íd,d,,£(-/£j^'\Therefore, thin-d^- devices with thin, high-E,„ dielectrics are

desired. To increase the effective coup ling of the gate and chan-nel, a geometry in w hich the m etal gate wraps around the chan-nel is optimal. In that cylindrical geometry, X is shorter and

depends less on d„,. Consequently, devices can be madeshorter, and the scaling process can continue.

and the absence of low-angle scattering.

As a result of energy- and momentum-conservation re-

quirements and the mismatch between the phonon sound ve-locity and electron band velocity, only a small fraction of

phonons can effectively participate in the scattering in CNTs.

The consequence is a very high low-electric-field mobility

(typically about 10 000 cmVVs), and values as high as

100 000 cmVVs have been reported, even at room tempera-

ture, by the University of Maryland's Michael Fuhrer. Tliat is

unlike other matt-rials such as III-V semiconductors, whose

charge mobilities are high at low temperatures but substan-

tially degraded at room temperature.

Unlike acoustic-phonon scattering, óptica I-phonon scat-

tering is very strong in CNTs. The phonons contract and elon-

gate tht' carbon-carbon bonds, thus strongly modulating the

electronic structure and coupling the electrons and phonons.

At ambient temperatures, however, the phonon population

is negligible, and electrons must have energies larger than

about 160 meV to emit such a [ihonon. That can only be

achieved imder high voltage-bias conditions.

Unusual, nonequilibrium phonon distributions can beproduced by current flow in nantitubes at (hose high biases

due to differences in electron-phonon coupling among the

various phonon modes, the large disparity in phonon fre-

quencies between acoustic and optical modes, and the poor

thermal interface between the CNT and the insulating sub-

strate. As the carrier energy increases further, other inelastic

processes take place that involve CNT excited states, which

are discussed later.

Current injection

The semiconducting CNTs forming FET channels are con-

nected to metal electrodes, typically palladium, gold, tita-

nium, or aluminum. The differences in the metal and CNTwork functions lead to charge transfer at their interface.

TIic resulting interface dipoles produce a potential-energy



Figure 1 . A single-walled car-bon nanotube, shown in this

scanning tunneling microscopeimage (a), can form the basis ofan FET. (b) In a top-gated con-

figuration, a single nanotube issandwiched between tw o insu-

lating layers, hafnium oxide and

silicon dioxide in this case. Palla-dium metal electrodes serve as

source and drain. (Adapted from

A. Javey e t al., Nar)o Lett. 4,1319,2004.) (c) !n an alternate config-

uration, an insulating layer and metal gate are wrapped aroundeach nanotube to op timize the gate-channel c oupling. Thenanotube segments away from the gate are doped. (Adapted

from Ph. Avouris et al., ref. 3.)

barr ier , the so-called Schottky barr ier . The a lignment of theFermi levels of the metal and the CNT, and therefore the SBheight, depend on their respective work functions, the CNTbandgap , and chemica l -bonding de ta i l s a t the in te r f ace /

An SB exists a t each end of an FET chann el. W hen oneof the two barr iers is much higher than the other—their summu st equa l the ban dga p — the FET operates as a unip olar de-vice , mean ing i t t ranspo r ts only on e type of carr ier , e lec tronsor holes. For example , a metal such as Pd with a high workfunction could be used to form a near ly barr ier less contact

-0.5 0.0 0.5 1.0

GATE VOLTAGE Vg^ (V)

1.5 2.0

-1.2 -0.8 -0.4

DRA IN-SOURC E VOLTAGE Vj (V)

Doped insulator

for holes. But e lectron injection would then exper ience themaximum barr ier with a value c lose to the bandgap. Ali Javeyand colleagues a t Stanford Universi ty used Pd as a metal incontac t wi th 2-nm-diame te r CNTs and obse rved quas i -ballistic tra ns po rt in short, p -ty pe C NT FETs.**

In genera! , the current s trongly depends not only on thetype of metal but a lso on the diameter of the CNT. Moreover ,because carr ier transpor t through the metal-CNT interface is

domina ted by quantum mechanica l tunne l ing th rough theSB, the barr ier 's thickness becomes cr i t ica l . But thanks to theenhanced screening provided by a metal gate near the thinoxide layer , both SBs can be made thin enough to a l low theinjection of e i ther e lectrons, holes, or both simultaneously."That type of transistor is called ambipolar. The insets of fig-ure 2 show how the applica tion of dif ferent gate voltag es tr ig-gers the injection of electrons or holes. The SB thickness canbe modif ied by doping the CNT near the contacts or by usingCN T FETs wi th m ul t ip le ga te s .

The ID character and single-a tomic- layer structure ofCNTs mak e them suscep t ib le to envi ron men ta l pe r tu rba -t ions. Intr insic screening is weak, and trapped charges in agate insulator shift electrical characteristics such as the FET

threshold voltage . Specif ic interactions such as charge trans-

Figure 2, Charge-transport characteristics, (a) The curren t /^versus gate voltage V ^measured from an ambipolar carbonnanotube FET as the drain-source voltage V ^decreases (bot-tom to top) from -0.1 V to -1.1 V in steps of -0,2 V. The left insetshows a schematic band structure of a CNT FET under negativegate bias and negative drain-source bias. Holes are injectedinto the CNT from the source (S). The right inset shows the re-verse: Under positive gate bias, electrons are injected from thedrain (D). (b) Output characteristics of a CNT FET plotted as 1^

V^ . From bottom to top, V ^ varies from +1V, the off state

FET, to -2.5 V, the on state, in steps of -0.5 V. (Adapted fromPh. Avouris et al., ref. 3.)



Figure 3, The five-stage ring oscillator pic-tured in these scannintj electron microscopeimages is composed of 10 FETs built entirelyon one 18-^m-long single-walled carbonnanotube. An image of a human hair in the

background provides the relative scale of theleast-magnified image (circled). Higher-magnification images reveal more detail, withthe nanotube visible in the upper right. Toobtain both p- and n-type FETs on the sameCNT, metals with different work functions arechosen as the gates—f)alladium iPd) for thep-type FETs and aluminum (AI) for the n-type

FETs. The gate oxide ts Alp,, and the ring fre-quency is 70 MHz at 1 V. (Adapted fro mZ. Chen et al., 5ae/ice 311,1735, 2006.)

fer to or from the substra te or the absorption of species on theCNT, the metal electrodes, or the substrate can alter the char-acter of the transport, the doping, and the resistance.^ Indeed,such effects on the electrical properties of CNTs provide thebasis for various sensor applications.

Implementation and performance

Early CN T FKTs were simple structures composed of a longCNT channel that is van der W aals-bonded to metal contacts

and back-gated using theSi

substrateitself.''''

Although func-tional, they had rather poor electrical characteristics. ShalomWind and colleagues improved on those characteristics byembedding the CNTs in the metal electrodes and using topmetal gates and thin gate insulators. Javey, Hongjie Dai, andcoworkers further enhanced the channel-gate coupling andperformance by using high-dielectric insulators such as zir-citniimi oxide.'* A number of different device structures havesince evolved, among them m ultiple-gate devices'' and wrap -around-gate configurations fabricated jointly by IBM andhlarvard University.

Tlie on/off curren t ratio in a CNT FET is generally high :10^-10", ideal for logic devices. Another important parameteris its transconductance ,^^—the change in drain-s ource cur-

rent / j , divided by tlie change in the gate voltage V at a con-stant drain-source voltage. For example, a top-gated FETwith a high-dielectric insulator and a CNT diameter of about1.7 nm can reach a ;^^^ of about 30 microsiemens." The theo-retical limit is close to 155 ¡.iS, which leaves a lot of room forimprovement.

Another important characteristic of FET operation iswhat's known as subthreshold swing, S = dV^/dlog/j^, whichdescribes the voltage needed to produce an order of magni-tude change in the current. Conventional FETs require achange in the channel potential of at least 60 mV at 300 K toproduce that change, and the minimum subthreshold swingimposes a lower limit on the operating voltage and hence onthe power dissipation in conventional FETs. Researchers

have achieved that thermodynamic limit in CNT FETs bysuppressing SBs in double-gate devices'* or by doping the

contacts. More important, work at the Watson Ke.search Cen-ter has .shown that the limit can be exceeded. By taking ad-vantage of band-to-band tiinneling, Joerg Appenzeller andcolleagues showed that the electron distribution can be fil-tered to remove high-energy electrons and reduce the sub-threshold swing.^

High-speed devices, of course, are a key goal of thewhole effort. Consider a typical measure of the frequency re-sponse of conventional transisto rs, the cut-off frequency

gJlnC^^. Using 5 aF as a typical gate-source Cijpacitance Cand s„ ~ 10-20 fiS for a lOO-nm-U-ngth CNT FE7; we find acut-off frequency in the range of 1 THz.' Direct measurementof such high frequencies is difficult because of the mismatchin impedances of CNT FETs and the measurement systems.Indiret t measu rements on single ('NT FETs have shown thattheir performance is currently limited by the parasitic capac-itance of the circuits.

CNTs can also help alleviate the pow er dissipation prob-lem that limits the scaling of MOSFETs, To dynamicallyswitch a device takes an energy of '/:(Cj,,^ + C,JV', where C,,,,and C^ are the device and w iring capacitance contributions,respectively. A primary adv antag e of CNT FETs over SiMOSFETs is their much lower capacitance values, roughly

0.05 aF/nm. Careful design, though, has to ensure that para-sitic capacitance and C^, typically 0.2 aF/nm, do not over-whelm Cj^^,, as they might, for example, when the wires arelong. The power dissipation can also be lowered by reducingthe supply voltage V,,,, (power ^fV¿i, where/is the operat-ing frequency). However, V,j cannot be arbitrarily reduced.Tlu' performance, T - 1//, degra des with decreasing V^ .Aresult, there is an optimum Vj^ for a given/

Integrated circuits

riif fabrication of CNT-based devices has advanced beyondsingle transistors to include logic gates and more complexstnictures such as ring oscillators.' The oscillators are com-posed of an odd num ber of pairs of p- and n-type FETs madeby appropriate doping of the CNTs. Controlled doping innarioscale devices is difficult, however, and fluctuations in



the number and position of the dopa nts can have a profoundeffect on device performance.

Fortunately, the ambipolar behavior of an undoped CNTcan be used to im plement CMOS architecture. A CMOS logicdevice is formed using a pair of p- and n-type transistors thatoperate u nder con ditions in which one FET is turned on andthe other is off. The ambipolar characteristics of a CNT FETprovide a pair of p- and n-type branches. But the leakagebranch of one type has a current level comparable to the mainbranch of the other. Controlling the threshold voltage in anambipolar device is therefore key to achieving the p- and n-type characteristics suitable for CMOS logic. One can controlthe threshold voltage by timing the work fimction of themetal gate in each FET; the gate work function can be viewedas an extra voltage source added to V .When the m etals areproperly selected, their different work functions shift thecharacteristics of p- and n-type branches to generate a dis-tinguishable on state in one CNT FET and an off state in the

other.That approach was employed by IBM's Zhihong Chen

and c oworkers to fabricate the most complex electronic struc-ture so far based on a single CNT. Figure 3 shows a scanningelectron microscope image of a five-stage ring oscillator inwhich Pd was chosen as the metal gate for the p-type FETand Al for the n-type FET.

Despite the outstanding performance of single-CNTFETs, their practical implementation faces a number of hur-dles. The principal one is the heterogeneity of as-preparedCNT m ixtures. Synthetic techniques have em erged in recentyears for producing ever-smaller CNT diameter and chiral-ity distributions.

One electronic technology that bypasses the hetero-geneity problem to some extent is that of CNT thin-film tran-sistors (TFTs), introduced by Eric Snow and coworkers in2003.'" A film of randomly oriented CNTs is deposited on aninsulator and electrodes are patterned on top of it. Ensem-ble averaging is expected to correct for the heterogeneity ofthe film. The presence of metallic CNTs in the mixture posesa problem, however, since they tend to short out the FETchannel.

To avoid the problem, the channel is made much longerthan the average length of the CNTs so that no single tubecan bridge the source-drain gap. Charge transport occursalong "percolation paths ," and switching involves numero ustube- tube junctions. The technique achieves reasonably high

on/off curren t ratios, but at the cost of increasing the channelresistance and consequently producing low drive currents.The carrier mobility is also reduced drastically, although it

Figure 4. Carbon nanotube thin-film

transistors in this optica l image of anintegrated circuit are bo nded to a flex-ible plastic surface. (Adapted fromQ.Cao et al., ref. 10.)

remains higher than that of typi-cal organic TFTs. Furthermore, thenanotubes can be deposited at lowtemperatures on a variety of sub-strates for various applications,one of them pictured in figure 4.

One approach to improvingthe performance of TFTs is to use

purely semiconducting CNTs in a film. Another is to usehighly oriented CNTs, which reduces the number of high-resistance tube-tube junctions that are so prevalent in ran-domly oriented CNTs. Researchers at the Watson ResearchCenter and Northwestern University have recently imple-

mented those improvements in their films, and now bothhigh driving currents (up to milliamps) and fairly high on/offcurrent ratios (roughly 10'') can be achieved simultaneously.The mobilities are high for TFTs, but they are still signifi-cantly lower than those of singlc-CNT devices.

Analysis of those results suggests that at least two pro b-lems remain. A distribution of tube diameters, and thusbandgaps, still exists in the films. That leads to a spread inthe threshold voltage for switching the current in the differ-ent tubes. Moreover, high currents require a high CNT den-sity, but closely spaced nanotubes tend to interact and screenthe gate field.

Excited state5Studies of the excited sta tes of CNTs have contributed greatlyto our understa nding of the electronic structure and interac-tions of CNTs and opened the door to new applications."Early on, the electronic excitations of CNTs were discussed,in the context of the single-particle model, as transitions be-tween corresponding pairs of so-called van Hove singulari-ties in the ID density of states that produce free electron-hole pairs. However, as pointed out by Tsuneya A ndo whileat the University of Tokyo in 1997, electron-electron interac-tions should be strong in a confined ID system and the in-teractions should form bound states, known as excitons,below the free-particle continuum. First-principle calcula-tions by Steven Louie and coworkers at the University of Cal-ifornia, Berkeley, showed that exciton binding energies innanotubes are unusually large (on the order of an eV). Exci-tons carry most of the oscillator strength and should, there-fore, dominate the absorption spectra (see figure 5).'^

The predictions were confirmed by two-pht)ton excita-tion spectroscopy by Tony Heinz's group at Columbia Uni-versity and Chrisfian Thomsen's group at the Technical Uni-versity of Berlin.'^ The exciton binding energies ofsemiconducting CNTs depend inversely on their diameterand, unlike the case of bulk sem iconductors, also depend onthe screening provided by the medium in which the CNTsare embedded."

Experimental studies of the excited states of nanotubes

did not start in earnest until 2002 when Bruce Weisman andcowo rkers at Rice observed photo luminescence from CNTswrapped by surfactants and dispersed individually in solu-



."* Previous attempts were hampered by aggregation ef-

fects and the coexistence of metallic and semiconductingCNTs, with the former acting as luminescence quenchers.Once the conditions for luminescence were established, the

tield grew rapidly, A key application has been the identifica-tion of the structure of emitting CNTs. By monitoring the lu-

minescence intensity as a function of laser frequency, re-

searchers can determine the excitation spectra of individualtubes and deduce their structure.^

Following the observation of photoluminescence fromCNTs, the decay of their excited states became an active re-

search field. M easured room -temp e rature fluore scence life-times from dispersions of CNTs are in the range of 10-100 ps,

although theory predicts CNT radiative lifetimes of 1-10 ns.

Reported fluorescence quantum yields, Q, are also low, but

there are significant uncertainties in their values." However,despite the lack of precise Q values, it is clear that fast non-

radiative decay processes operate inCNTs,Another result of the strong electron-electron interac-

tions in CNTs is that the degeneracy present in the single-particle states is partially removed, which results in new ex-

citonic states. The transition to the lowest energy state was

shown by theory ajid experiment to be "dark" —that is, di-

pole forbidden. Trapping in the dark exciton state might beresponsible for the observed low yield. However, the energyseparation between dark and bright excitons determined by

Junichiro Kono and coworkers at Rice is too small to accountfor the low Q. Recent theoretical work at the Watson ReseardiCenter predicts two fast CNT decay channels: multiphonondecay of trapped excitons and a new, phonon-assisted elec-tronic decay pathway.'^ The latter is made possible by dop-

ing the CNT through charge-transfer interactions with its

environment or by external fields.The photoluminescence statistics of individual CNTs are

also interesting. A recent study by Atac Imamoglu and

coworkers at ETH Zürich found evidence of photon anti-

bunching, a discovery that may lead to the use of CNTs assingle-photon sources and in quantum cryptography.

Electroluminescence

From a photonics-applications perspective, a key finding w asthat light emission from CNTs can be obtained by passing acurrent through them." As noted previously, CNT FETs tendtt) be ambipolar, so electron and hole currents can flow si-multaneously. When those electrons and holes recombine, afraction of the recombination events emit a photon and thusgive rise to electroluminescence. EL is widely used to pro-duce solid-state light sources such as LEDs.

Conventionally, EL is generated near the interface be-tween a hole-doped and an electron-doped material. But in

an ambipolar CNT, doping is unnecessary, which simplifiesLED fabrication. Even more interesting, EL from a long CNTdoes not originate from a fixed point but can actually betranslated at will byapplying an appropriate bias at the gate;"the gate bias determines theposition where electron-hole re-combination takes place in the intrinsic m aterial. The energyE of the EL is the same as that of photoluminescence of theCNT and can thus be tuned by using tubes of differentdiameters d since E,, is proportional to l/d.

Another mechanism by which a current can excite EL in-volves the presence of hot carriers in CNTs, In a unipolarCNT at high bias, the carriers can achieve high enoug h ener-gies and, because of the strong Coulomb interactions amongthem, can induce impact excitation and ionization.'^ In bulksolids, ionization is the principal process induced, but in thequasi-lD CNTs with high exciton binding energies, neutral

Without e-h interactitin

With e-h interaction

2.0 2.5 3.0

P HOT ON E NE RGY (eV)

Figure 5. (a) Ab initio calculations of the absorption spec-tra ofa single-walled carbon nanotube. The electron-hole(e-h) interactions are included inone case (red) but not in

the other (blue). The energy and intensity differences be-

tween the three lowest interband transitions (A. B, C) and

the corresponding exciton states tA', B', C) indicate the pro-

nounced effect of strongly boun d e-h pairs ina nanotube.f,, and £;; signify the transition energies experimentally ob-served using light polarized along the nanotube axis. Theenergy difference between the first red and blue peaks

(shown by the black horizontal arrow) gives thebinding en-ergy of the first exciton, (b) A real space representation of

the É'-h pair probability distribution, |H^(r,, r^)\\ as a functionof the electron position r when the hole position r ,, the

green do t, is fixed. (Adapted from ref. 12.)

excitations dominate. Typically, the emission is localized at

defect sites, where voltage drops provide enough energy to

the carriers to exceed the excitation threshold. More impor-tant, one can choose the location of the emission by intro-ducing along the CNT channel a voltage drop—for example,by locally modifying the coupling between the gate and

the CNT'^

Electroluminescence from CNTs isbroad and nondiri-c-tional. But most optoelectronics applications, such as optiialinterconnections, require a narrow, directional light beam.IBM's Fengnian Xia and coworkers recently achieved that rf-

feet b) confining a single CNT in a planar photonic half-

wavelength microcavity In that configuration, thf tact thatdifterent diameter nanotubes in a sample have slightly dif-

fen;nt emission energies becomes less importan t, because the

resonant wavelength is determined by the structure of the

cavity itself.

Photoconductivity, essentially the reverse of EL. occurswhen optical radiation produces Iree electrons and holes in

a materia l. The process was first observ ed in CNTs at the Wat-

son Research Center in 2003, but since then many such stud-ies have appeared." The same CNT FET device can be usedas a transistor, pho toswitch, light emitter, or photodetector.



In an open-circuit configuration, internal electric fieldscan separate electrons and holes and generate a photovolt-age. Photovoltage imaging microscopy can provide imagesof electric fields along CNTs such as those at SBs or chargeddefects.

On the horizonWith their unique prop erties, semiconducting nanotub es canform the basis for extremely small, fast, and flexible transis-

tor circuits. CNT devices are molecular devices; therefore,simple and inexpensive self-assembly-based fabrication maybe possible. Transistors with chemically modified CNTs arealready be ing used as sensitive and selective chemical and bi-ological sensors. CNT-based light sources and detectors maygive rise to intrachip optical communication and single-molecule spectroscopy. And the excellent electrical conduc-tivity of metallic CNTs may eventually allow the develop-ment of electronic systems in which both active devices andpassive interconnects are based on CNTs. Continued inte-gration of optics and solid-state devices could set the stagefor a unified, CNT-based electronic-optoelectronic technol-ogy, and the long spin coherence lengths in CNTs could alsolead to quantum interference and spintronic devices.

How soon can we expect to see such developments? Themain hur dle lies in the production of large amounts of iden-tical nanostrucfures. Nanotubes come inmany sizes andstructures, and there is no reliable way at present to direcflyproduce any single type of CNT needed in a large integratedsystem. Promising signs suggest that may change. Already,separation of a single type of CNT from mixtures has beensuccessful. And researchers who investigate self-assembly ofCNT devices are reporting encouraging results.

The evolution cannot be characterized as an incrementalimprovem ent of existing technology as embodied in the seal-

F A S T I N T R O D U C E S T H E N E X T G E N E R A T I O N

6 4 B I T 5 / ( 6 ) I N P U T 1 0 0 P S M U L T IS T O P T O C ,M U L T IC H A N N E L S C A L E R J O F

M o d e l M C S 6 M u l t i c h a n n e l S e a le r fe a t u r e s :

• No d e a d t i m e b e t w e e n t im e b i n s

• H ig h c o u n t r a t e w i t h 100 ps t im e r e s o l u t io n

• T i m e r a n g e u p to 20 d a y s ( 5 4 - b i t d y n a m i c r a n g e )

• H i g b d a t a t r a n s f e r r a t e to P C by d u a l U S B bus

• F u l l y d i g i t a l d e s i g n , no c o r r e c t i o n s r e q u i r e d

• D r iv e r s fo r L a b V ie w a n d L IN U X a v a i la b l e

IIIIFUSTCornTec: QmtjM

For mare I I I IQS vis i t aur Wf b j i le un der. Mul t ichannel

Eruenwalder Weg 28a. D-82D4I Oberhaching/Gennany

Phon«: 49-10 )89 EES IBO-0. FAX: 49-(0)89 EESIEO 40USA cal l {BI D 429-5727. e-mait : inlo ^ta t tcDm leccain

ing of silicon devices. It involves fundam ental chang es in ma-terials, processing, assembly tecliniques, and physical prin-ciples. The development of CNT technology dep ends, as doeseverything else in industry, oncost-benefit analysis andunanticipated breakth rough s. Already, however, the study ofCNTs has provided us with unique insights into the physicsand chemistry of nanoscale materials.

References

1. S. tljima. Nature 354, 36 (1991).2. K. S. Novose lov et a l . , Nature 438, 197 (200ÍS).3. A. Jor io , M. S . Dresse lhaus , G . Dresse lhaus , eds , . Carbon Nano-

tubes: Admmced Tapies intite Si/nthesis. Struetiire, Properties, ami

Applicntions, Springer, New York (2008); Ph. Avouris , Z. Chen, V.Perebeinos, Nat. Nm ioteehnol. 2, 603 (2 007).

4. S. J . Tans, A. R. M. V erschueren , C. D ekker, Nature 393, 49 (1998);R. Martel et al., Appl. PIn/s. Lett. 73, 2447 (1998).

3. M. J . O'Connel! et a l . . Science 297, 593 (2002); S . M. B achilo et a l . ,Science 298 , 2361 (2002) .

6. J . A. Misewich et al., Snt'í;a' 300, 783 (2003); M. Freitag et al., Plu/s.

Rev. Lett. 93, 0768Ü3 (2004).7. R. Martel et al-, Phys. Rer. Lett. 87, 256805 (2001); S . Heinze et a l . ,

Ptiys. Rev. Lett. 89, 106801 (2002).8- A.'javey et al. . Nature 42 4, 654 (2003); A. Javey et al., Nat. Mater. 1 ,

241 (2002) .9, Y.-M. Lin et al., IEEE Trans. Nanotechnoi. 4, 481 (2005).10. J . Novak et al., Appl. Phys. Lett. 83, 4026 (2003); Q. Cao et al. .

Nature 454, 493 (2008).n . P h . A v o u r i s , M . F re i t a g , V.Perebe inos , Nat. Ptwtonics 2, 34 1

(2008).12 . C. D. Spataru et a l . , Pljys. Rev. Lett. 92, 077402 (2 004).13. F. Wan g et al . . Science 308, 838 (2005); J . Mau ltzch et a l . , Phys. Rei>

8 7 2 , 2 4 1 4 0 2 ( 2 0 0 3 ) .14. V. Perebeinos, Ph. Avouris , Phys. Rei>. Lett. 1 01 , 057401 (2008).13. J. Chen et al. . Science 310 , n7 l" (2 005); V . Perebe inos , Ph . Avouris ,

Pin/s. Rev. B 74, 121410 (2006). •

UHV SURFACESCIENCE

Advanced quadrupole mass spectrometers with UHVsp ecif ic p u lse ion cou nt ing detector f o r : —

• photon and elec tstimulated desorptio

• molecular beam studies

• temperature programmeddesorption

9 . EPIC sys temsfast 100 nanosecond time

resolution

high sensitivity positive

and negative ions

integration of process

data

iow profiie and X-beam

ioniser options

UIDENI

for lurtl ief details ot Hldsn Analyllcal

p r o d u c t s c o n t a ct :

[email protected]

Carbon Nanotube Electronics and Photonics

Documents

Transcript of Carbon Nanotube Electronics and Photonics