single-crystal organic charge-transfer interfaces probed using schottky-gated heterostructures

ARTICLESPUBLISHED ONLINE: 22 JULY 2012 | DOI: 10.1038/NMAT3383

Single-crystal organic charge-transfer interfacesprobed using Schottky-gated heterostructuresIgnacio Gutiérrez Lezama1, Masaki Nakano2, Nikolas A. Minder1, Zhihua Chen3,4,Flavia V. Di Girolamo1,5, Antonio Facchetti3,4 and Alberto F. Morpurgo1*

Organic semiconductors based on small conjugated molecules generally behave as insulators when undoped, but theheterointerfaces of two such materials can show electrical conductivity as large as in a metal. Although charge transfer iscommonly invoked to explain the phenomenon, the details of the process and the nature of the interfacial charge carriersremain largely unexplored. Herewe use Schottky-gated heterostructures to probe the conducting layer at the interface betweenrubrene and PDIF-CN2 single crystals. Gate-modulated conductivity measurements demonstrate that interfacial transportis due to electrons, whose mobility exhibits band-like behaviour from room temperature to ∼150K, and remains as highas ∼1 cm2 V−1 s−1 at 30K for the best devices. The electron density decreases linearly with decreasing temperature, anobservation that can be explained quantitatively on the basis of the heterostructure band diagram. These results elucidatethe electronic structure of rubrene/PDIF-CN2 interfaces and show the potential of Schottky-gated organic heterostructures forthe investigation of transport in molecular semiconductors.

Organic semiconductor interfaces determine the operationof devices based on molecular materials1, such as organiclight-emitting diodes2,3, p–n junctions4,5 and solar cells6–8.

In most cases, electronic transport perpendicular to the interfaceis the key process of interest. Transport in the interface plane hasattracted attention only recently9–18 (for early work, see ref. 9),both in the applied9–14, and fundamental domains15. It has beenfound that interfaces between twoorganic semiconductors regularlyexhibit an enhanced electrical conductivity, and that, when singlecrystals of specific molecules are used, this interfacial conductivitycan be as large as in a metal—a remarkable observation, given theinsulating character of the constituentmaterials. There is consensusthat charge transfer from the surface of one organic semiconductorto that of the other determines the interfacial conductivity15,16,19,but even the most basic properties of these organic charge-transferinterfaces could not be accessed experimentally so far, preventinga true understanding of the phenomenon. Indeed, interpretationof experiments has so far relied on plausible physical scenarios andestimates15,16, but no direct measurements could be performed todetermine whether the interfacial charge carriers are electrons orholes (or both), their density, their mobility, how these quantitieschange as a function of temperature or whether they can be tunedin device structures.

Here we address these problems by performing experimentson a new type of device: organic Schottky-gated heterostructuresbased on organic single crystals. The devices that we discuss (seeFig. 1a–e) are based on rubrene20 and PDIF-CN2 single crystals21—the materials that perform best in p- and n-channel organicsingle-crystal field-effect transistors (FETs)—at whose interfacea highly conducting layer forms spontaneously (SupplementaryFig. S1). The heterostructures consist of a Cr film (the Schottkygate) deposited on a polydimethylsiloxane (PDMS) support, ontowhich first a rubrene and then a PDIF-CN2 single crystal are

1DPMC and GAP, University of Geneva, 24 quai Ernest Ansermet, CH1211 Geneva, Switzerland, 2CERG, RIKEN Advanced Science Institute, Wako, Saitama351-0198, Japan, 3Polyera Corporation, 8045 Lamon Avenue, Skokie, Illinois 60077, USA, 4CHEM, Northwestern University, 2145 Sheridan Road, Evanston,Illinois 60208, USA, 5CNR-SPIN and University of Naples, p.le Tecchio 80 80125 Naples, Italy. *e-mail: [email protected].

laminated. Before laminating the PDIF-CN2 crystal, gold electrodesare evaporated onto rubrene through a shadow mask, to contactthe interfacial conducting layer. A schematic diagram and actualimages of devices are shown in Fig. 1 (the details on the fabricationprocess are discussed in the Supplementary Information). Devicesof this type have never been previously realized with organicsemiconductors, but have clear analogies with Schottky-gatedheterostructures based on iii–v inorganic semiconductors22.

The Schottky barrier present at theCr/rubrene interface preventscharge injection for one polarity of applied bias, effectively isolatingthe gate electrode from the rubrene/PDIF-CN2 interface23,24. Owingto the depletion region associated with the Schottky barrier andthe very low density of states in the bandgap of rubrene singlecrystals, virtually all charge induced on application of a gate voltageis accumulated at the rubrene/PDIF-CN2 interface. This eliminatesproblems present in heterostructures with a conventional gate—that is, a gate separated from the semiconductor by an insulatinglayer—in which the induced charge is also accumulated at theinterface between the gate insulator and the semiconductor (see, forexample, ref. 19). As it is important to both ensure correct deviceoperation and characterize the heterostructures—and given thatSchottky gates are not commonly employed in conjunction withorganic semiconductors—we start by discussing measurementsshowing that the gate electrode indeed functions as expected fora Schottky contact.

Figure 2a shows the current flowing through the rubrene crystalon biasing the gate (that is, the current between the gate and both thedrain and source contacts; during device operation this correspondsto the gate leakage current). This current depends on the biaspolarity, as expected for a Schottky gate. For negative gate voltages(Fig. 2b), the gate extracts holes injected at the rubrene/PDIF-CN2interface, giving rise to a space-charge-limited current25 throughthe rubrene crystal. Accordingly, the current–voltage (I–V ) curves

788 NATURE MATERIALS | VOL 11 | SEPTEMBER 2012 | www.nature.com/naturematerials

© 2012 Macmillan Publishers Limited. All rights reserved

http://www.nature.com/doifinder/10.1038/nmat3383

mailto:[email protected]

http://www.nature.com/naturematerials

NATUREMATERIALS DOI: 10.1038/NMAT3383 ARTICLES

AuAu

Rubrene

PDIF-CN2

Cr100 µm

300 m

Rubrene edgeRubrene

Au

Cr

PDMS

Rubrene

PDIF-CN2

a

c

e

b d

200 µm

Figure 1 | Rubrene–PDIF-CN2 Schottky-gated heterostructures. a,b, Chemical structures of the N,N′-bis(n-alkyl)-(1,7 and1,6)-dicyanoperylene-3,4:9,10-bis(dicarboximide) (PDIF-CN2) (a) and the rubrene molecule (b) used for the realization of the Schottky-gated organicheterostructures. c, Schematic representation of the Schottky-gated heterostructures, comprising a PDMS support, the metal gate electrode, a bottomrubrene crystal, the Au source and drain contacts and a top PDIF-CN2 crystal. The interfaces are formed by the high-mobility planes of both the rubreneand the PDIF-CN2 crystals. d,e, Optical micrographs of rubrene–PDIF-CN2 Schottky-gated heterostructures. d zooms in on the core of one device ande shows a larger field image of another (multi-terminal) device that includes the epoxy-bonded contact wires. The white arrows in e point to the edges ofthe rubrene crystal.

(Fig. 2c) show an approximately quadratic bias dependence. Forpositive gate voltages, the current is much smaller (Fig. 2b),because—as mentioned above—holes injected from the chromiumgate into rubrene have to overcome the Schottky barrier (thework function of chromium is ∼4.5 eV (ref. 23), correspondingto a Schottky barrier height (φB) of approximately 0.8 eV (refs 24,26)). As the temperature is reduced, the leakage current decreasesfollowing a thermally activated mechanism. For positive gate biasthis is due to thermal activation over the barrier, whereas fornegative bias (Fig. 2d) the thermally activated behaviour originatesfrom space-charge-limited current through states in the bandgapof rubrene (that is, through traps27,28). The activation energy(Fig. 2d inset), corresponding roughly to the energy differencebetween the Fermi level and the bottom of the valence band ofrubrene, ranges between 250 and 400meV, in agreement withrecent studies of rubrene metal–semiconductor FETs (ref. 24)and previous investigations of space-charge-limited current28. Theobserved variations are indicative of sample-to-sample fluctuationsin the concentration of dopants unintentionally present in rubreneand in the density of in-gap states, which determine the position ofthe Fermi level relative to the top of the valence band.

Having ensured the proper operation of the Schottky gate, weproceed to discuss transport in the interface plane, which wasinvestigated as a function of source–drain bias (VDS), gate voltage(VG) and temperature (T ). Figure 3a shows that, even at VG= 0V,the I–V characteristics of the heterostructures are linear throughoutthe investigated temperature range, and that the resistance isreproducibly close to 1M�/� at room temperature (see the tablein the Supplementary Information for a summary of data from

all working devices). This observation directly indicates that ahighly conducting layer does indeed form at the rubrene/PDIF-CN2interface, because the resistance of individual rubrene and PDIF-CN2 crystals is much larger than 1G�. The resistance (Fig. 3b)exhibits only rather small variations when T is reduced down to∼100–150K, whereas further lowering causes a more pronouncedresistance increase for all devices. Figure 3c,d shows the gate-voltage dependence of the current flowing in these heterostructuredevices. The measured current always increases with applying apositive voltage to the Schottky gate and decreases for the oppositepolarity, exhibiting a nearly linear VG dependence throughout theinvestigated range. This observation indicates an n-type behaviourof the devices, from which we conclude that the current in theinterfacial conducting layer is carried by electrons.

We estimate both the electron density and field-effect mo-bility using the same type of analysis commonly adopted forconventional FETs. The interfacial conductivity can be written asσ (VG)= n(VG)eµFET, where n(VG) is the density of electrons atthe rubrene/PDIF-CN2 interface, given by n(VG)= C(VG−VT)/e(C = εε0/d is the capacitance per unit area between the interfaceand the gate, ε is the relative dielectric constant of rubrene and d isthe thickness of the rubrene crystal—close to 2 µm in most devices;VT is the threshold voltage, corresponding to the extrapolated gatevoltage at which the source–drain current vanishes). The mobil-ity can therefore be extracted as µFET = L/W (1/CVDS)dIDS/dVG(where L and W are the interface length and width, respectively)and is found to be nearly gate-voltage independent. The electrondensity as a function of VG is then given by n= σ/eµFET, whereσ is the conductivity.

NATURE MATERIALS | VOL 11 | SEPTEMBER 2012 | www.nature.com/naturematerials 789




ARTICLES NATUREMATERIALS DOI: 10.1038/NMAT3383

205 K

271 K

296 K

Reverse biasForward bias0.2

0.0

–0.2

–0.4

–0.6

–0.8

I G (

μA)

VG (V)

|VG| (V)

|I G| (

A)

Ln (

|I G|)

220 K

295 K

Forward bias

–40 V

Forward bias

VG = –10 V

VG (V)

E A (

meV

)

Cr

3.5 4.0

1000/T (K–1)

4.5 5.0

Rubrene PDIF-CN2

EV

EV

S1

S1

S1

S1

S2

–50 –25 0 25 50 75 100

1 × 10-10

1 × 10-9

1 × 10-8

1 × 10-7

1 × 10-6

1031 50

Forward bias (VG < 0)

Bφ

Reverse bias (VG > 0)

Cr PDIF-CN2Rubrene

0

100

200

300

400

–50 –40 –30 –20 –10 0

a b

c d–14

–15

–16

–17

–18

–19

–20

–21

Bφ

Figure 2 | Characterization of the Schottky gate. A careful characterization of the Schottky gate is essential to ensure correct device operation. a, CurrentIG (measured on device S1) flowing through the rubrene crystal on biasing the gate with respect to the source and drain electrodes. The bias dependence isstrongly asymmetric, as expected for a Schottky gate: space-charge-limited current flows through the device for negative VG (forward bias), whereas thecurrent is injection limited by the Schottky barrier at the Cr/rubrene interface for positive VG (reverse bias; see the schematic energy band diagrams in b).As the temperature is lowered, the current is strongly suppressed also for negative VG, increasing the range of operation where the device is not affectedby leakage current. c, Bias dependence of IG measured for negative VG at different temperatures (device S1). The data show a quadratic dependence,characteristic of space-charge-limited current. With lowering T the current decreases exponentially, as shown in d, because space-charge-limitedtransport occurs through trap states in the bandgap of rubrene crystals (see also ref. 27). The corresponding activation energy, ranging between 250 and400 meV (in agreement with ref. 28), is a measure of the energy difference between the Fermi level and the top of the valance band of rubrene (the insetshows data from devices S1 and S2).

The temperature dependence of the field-effect mobility andelectron density at VG = 0 for three devices is shown in Fig. 4a,b.The room-temperature electron mobility values compare well tothose found for FETs based on PDIF-CN2 single crystals withCytop or PMMA as the gate dielectric29, which have approximatelythe same dielectric constant as rubrene30–32. The consistency inthese mobility values supports the validity of our analysis andconfirms that the electrons are accumulated at the rubrene/PDIF-CN2 interface. It also indicates that the surfaces of the PDIF-CN2 crystals maintain their integrity; that is, the crystal surfaceis not damaged by the proximity of rubrene, as could occurby molecular interdiffusion. Interestingly, as the temperatureis lowered, the electron mobility exhibits a more pronouncedband-like transport20,33 behaviour when compared with PDIF-CN2single-crystal FETswith a suspended channel (that is, with a vacuumas the gate dielectric), peaking at temperatures ∼150–170K (µFETvalues up to∼4–5 cm2 V−1 s−1are observed) and remaining close to1 cm2 V−1 s−1 in the best devices even at temperatures as low as 30K.This observation shows that rubrene–PDIF-CN2 Schottky-gateddevices lead to significantly improved low-temperature electron

transport. The electron density n is found to range between 1×1012and∼5×1012 cm−2 in different devices, and in all cases it decreaseslinearly as the temperature is decreased, with only small (<10%)device-to-device deviations in the slope. The high reproducibility ofthe linear temperature dependence and of its slope are particularlyremarkable, in view of themuch larger sample-to-sample variationsin the total electron density.

To confirm the validity of the conclusions extracted from theanalysis of the field-effect response, we have also performed Halleffect measurements29 in the presence of a perpendicular magneticfield. The fabrication of devices in a Hall bar geometry is complex,as it requires a very precise alignment of the PDIF-CN2 crystalto the electrodes measuring the Hall voltage. Nevertheless, wesucceeded inmeasuring the Hall effect as a function of temperature,enabling us to extract both the density of charge carriers and themobility. The results are shown in Fig. 4c,d, respectively (see theSupplementary Information for the rawHall voltage data) and agreewith the analysis of the field-effect data. In particular, the chargecarrier density extracted from the Hall effect decreases linearly asthe temperature is lowered, with the same slope extracted from






160 K

295 K

75 K

55 K

33 K

VG = 0 V

VDS = 10 V VDS = 10 V

2

4

I DS

(μA

)I D

S (μ

A)

I DS

(μA

)R S

(MΩ

)

0

–2

–4

–15 –10 –5 0

S1

S1S1

VDS (V)5 10

100

1.2

1.1

S1

S2

S3

R s (M

Ω)

1.0

0.9

100 150

S1

200T (K)

250 30010

1

15

–1000

1

3

4

5

–50VG (V) VG (V)

0 50 100 150

T (K)

200 250 300

0 50 100 –50–100 0 50 100

0.8

0.6

0.4

0.2

0.0

160 K

7.5 V

295 K

33 K

33 K

5.0 V

2.5 V

1.0 V

a b

c d

Figure 3 | In-plane transport characteristics of Schottky-gated rubrene/PDIF-CN2 interfaces. a, I–V characteristics measured at different temperaturesfor VG=0 V. The linearity signals the presence of a conducting layer throughout the investigated temperature range (and indicates the good contactquality). b, The resistivity depends weakly on temperature between room temperature and 100 K, and increases rapidly on further cooling in all devices.The inset zooms in on the data taken on device S1, showing that in the best samples the resistance decreases with lowering T between room temperatureand approximately 150 K. c, In all devices and at all temperatures, the current at the interface increases when a positive gate voltage is applied, indicatingthat charge carriers are electrons. As the temperature is lowered, the leakage current at negative VG is strongly suppressed, allowing measurementsthroughout a larger VG range (in all of the measurements the source–drain current IDS is much larger than the gate leakage current IG) as shown in Fig. 2.d, VG dependence of the source–drain current at different source–drain voltages, measured at the lowest temperature reached in our experiments (33 K).The labels S1, S2 and S3 indicate on which device the data were measured.

the field-effect analysis (we found that in Hall bar devices theelectron mobility is consistently lower, probably owing to the effectof the mechanical stress in the PDIF-CN2 crystals, induced by thealignment process during fabrication).

As we now proceed to show, the experimental observationscan be rationalized in terms of the semiconductor heterostructureband diagram, accounting for two sources of charge accumulationat the rubrene/PDIF-CN2 interface: the charge transfer acrossthe interface from the surface of the rubrene crystal to thatof the PDIF-CN2, and the charge displaced across the entireheterostructure. We start by discussing the first contribution.Having observed that the interfacial carrier density decreases onlylinearly—and not exponentially—with decreasing T , implies thatno gap (or at most a very small one) is present between thetop of the rubrene valence band and the bottom of the PDIF-CN2 conduction band (as is the case, for instance, for interfacesbased on tetramethyltetraselenafulvalene (TMTSF) and 7,7,8,8-tetracyanoquinodimethane (TCNQ) single crystals16). At the sametime, the density of transferred charge carriers—a few times1012 cm−2, 100× smaller than in tetrathiafulvalene (TTF)–TCNQinterfaces15—and its linear decrease for decreasing T also imply theabsence of any sizable band overlap. This implies that the top of the

rubrene valence band and the bottom of the PDIF-CN2 conductionband are aligned at the same energy before the two materials arebrought into contact (Fig. 5a). As an independent and quantitativeconfirmation of the validity of this conclusion, we have measuredthe value of the work function difference between rubrene andPDIF-CN2 crystals using scanning Kelvin force probe microscopy34(SKFPM). As the energy difference between the Fermi level and theband edge in both rubrene and PDIF-CN2 can be extracted fromtransport measurements on individual crystals, the difference inwork function between the two materials allows us to determinethe energy level alignment directly. We find that the two levels areindeed aligned with a precision of a few tens of milli-electronvolts(see Supplementary Information for details).

With such an alignment of the valence and conduction bandin rubrene and PDIF-CN2, interfacial charge transfer occurs atany finite temperature on establishing contact, and stops when theelectrostatic potential difference between the surfaces of the twomaterials, generated by the transferred charge itself, is sufficientlylarge. In other words, the electrostatic potential difference (aninterface dipole) generated by the transferred charge shifts apart thetop of the rubrene valence band and the bottom of the PDIF-CN2band, effectively opening a gap proportional to the density of






0

2

FET

S3

S3

S1

S1

S2

S2

4 400

1

2

60

T (K)

VG = 0 V

VG = 0 V

80

6 (

cm2

V–1

s–1

)

(cm

2 V

–1 s

–1)

n (1

012

cm

–2)

0 50 100 150

T (K)

200 250 300

0

1

Hal

l (cm

2 V

–1 s

–1)

0

1

2

n Hal

l (cm

2 V

–1 s

–1)

0S4 S4

0 100 200

T (K)

300 0 100 200

T (K)

300

50 100 150

T (K)

200 250 3000

1

2

3

4

a

b

c d

µµ

µ

Figure 4 | Temperature dependence of electron mobility and density atrubrene/PDIF-CN2 interfaces. a, The black, red and green circles representthe field-effect mobility of electrons for heterostructures S1, S2 and S3,respectively (the mobility value are taken at VG=0 V; at roomtemperature, where the applied VDS shifts the onset of the leakage currentto positive VG values, the data are taken at the smallest possible VG). Thefilled blue circles represent the field-effect mobility measured in aPDIF-CN2 single-crystal FET with vacuum as the gate dielectric, which isthe n-channel transistor with the highest mobility so far. The comparisonshows that the heterostructure devices exhibit a band-liketransport—increasing mobility with decreasing T—in a broadertemperature range. In the heterostructures, the mobility remains as high as∼1 cm2 V−1 s−1 at the lowest temperature values (see inset), orders ofmagnitude larger than in the best organic n-channel FETs at the sametemperatures. b, Electron density at VG=0 V for the same three deviceswhose mobility data are shown in a. All devices exhibit the same (within±10%) linear temperature dependence, whose slope matches theprediction of the model calculations discussed in the main text(represented by the dashed–dotted lines). c,d, The temperature-dependentelectron mobility and electron density (at VG=0 V), extracted from Hallmeasurements performed on device S4, which provide independentconfirmation of the analysis of the field-effect data and of the comparisonwith the model prediction (dashed–dotted line).

transferred charge nS, so that the value of nS at equilibrium needsto be found self-consistently. Within the simplest approximation—identical to the one that successfully captures the order ofmagnitude of the conductivity at TMTSF/TCNQ interfaces16—wecanwrite (for details, see Supplementary Information):

nS=∫∞

nSe2d/2εε0

2NSe−(E/kT )dE = 2NSkTe−nSe2d/2εε0kT (1)

where nSe2d/εε0 is the electrostatic potential difference between therubrene and the PDIF-CN2 surfaces,NS is the density of states at thesurface per unit energy (∼2× 1015 cm−2 eV−1), ε ∼ 3 is the valuefor the relative dielectric constant of the organic materials used,and d ∼ 1 nm is the distance between the electrons accumulatedon the PDIF-CN2 surface and the holes on the rubrene surface. Inequation (1) Boltzmann statistics is used because the occupationprobability of states in the organic materials due to the transferredcharge is much less than 1, owing to the large density of states (thephysical picture used for this analysis is based on the model that wehave developed to reproduce quantitatively band-like transport inorganic single-crystal FETs of differentmolecules29,35). By defining

y ≡nS

2NSkT(2)

equation (1) becomes y = e−e2dNSy/εε0 , which does not depend ontemperature anymore. If we denote its solution y∗, we directly findfrom equation (2) that

nS= y∗2NSkT (3)

predicting a linear temperature dependence of the electron densityon temperature, as observed experimentally. Equation (3) gives aquantitative estimate for the slope of the linear T dependence thatis in excellent agreement with the data. (See the dashed–dotted linesin Fig. 4b.With the values of the parameters that we have indicated,the numerical solution of equation (2) gives y∗=0.029.)

In the absence of other contributions to the interfacial chargedensity, an equal amount of electrons and holes should bepresent at the surface of PDIF-CN2 and rubrene. In the transportmeasurements, however, hole conduction is not observed. This isbecause the total carrier density n at the interface also includesthe—more conventional—contribution due to charge displacedacross the entire structure, which ensures that the electrochemicalpotential is spatially uniform23. A schematic energy band diagramof a rubrene–PDIF-CN2 heterostructure is shown in Fig. 5b (thecharge transferred across the entire heterostructure does not changesignificantly the band discontinuity—gap—at the rubrene/PDIF-CN2 interface, which depends on the amount of charge transferredfrom the surface of rubrene to that of PDIF-CN2). At thechromium/rubrene interface, the top of the rubrene valenceband is lower than the electrochemical potential by an amountcorresponding to the height of the Schottky barrier. Deep inthe PDIF-CN2 crystal, the electrochemical potential is below thebottom of the conduction band by an amount determined by theunintentional dopants present in the crystal (which we measuredto be 100–120meV, estimated by the activation energy of theresidual small conductivity of individual PDIF-CN2 crystals). At aquantitative level, therefore, the details of the band diagram dependon the doping level in the rubrene and PDIF-CN2 crystals, as well ason the distance between the Schottky gate and the rubrene/PDIF-CN2 interface (for∼2 µm-thick crystals, it is comparable to the sizeof the depletion region, see ref. 24). Irrespective of these details,however, the net effect is to accumulate electrons at the interface(see Supplementary Information for more details) that fill thehole states of the rubrene surface and shift the chemical potentialtowards the electron side (thereby explaining the absence of hole






Cr

φB

Rubrene

PDIF-CN2

Rubrene

nS =0

E

DOS

EV

EV

EC

EF

EV

EF

EC

EVEC

n (T) = constant n (T) = cTn (T) = e– /kT

e e e

EC

EV EV

EC

PDIF-CN2

a

b

TMTSF TCNQ Rubrene PDIF-CN2TTF TCNQ

Δ

Δ εεnSe2d

Figure 5 | Energy level alignment of organic interfaces and band diagram of rubrene–PDIF-CN2 Schottky-gated heterostructures. a, Schematicrepresentations of the energy-band alignment in organic single-crystal interfaces. TTF–TCNQ (left; see ref. 15) is characterized by a large overlap betweenthe valence band of the electron donor (TTF) and the conduction band of the electron acceptor (TCNQ), resulting in a temperature-independent chargecarrier density at the interface. In TMTSF/TCNQ (middle; see ref. 16), a small energy gap ∆ between the top of the valence and the bottom of theconduction band of the two materials leads to a thermally activated temperature dependence of the interfacial carrier density. In rubrene/PDIF-CN2

interfaces (right; this work) the top of the valence band of rubrene is nearly perfectly aligned with the bottom of the conduction band of PDIF-CN2,resulting in a linear temperature dependence of the carrier density transferred at the interface. b, Band diagram of rubrene–PDIF-CN2 Schottky-gatedheterostructures (see the Supplementary Information for details). The small gap ∆nS at the rubrene/PDIF-CN2 interface originates from the difference inelectrostatic (dipole) potential generated by the transferred interfacial charge. The zoom-in of the interfacial region represents the density of states (DOS)at the surface of rubrene (red) and PDIF-CN2 (blue) as a function of energy. It includes the DOS associated with the shallow traps of the two materials,which decays rapidly away from the valence and conduction band edges (represented by the two horizontal lines labelled EV and EC). As discussed in thetext, the Fermi level EF is pinned very close to the bottom of the PDIF-CN2 conduction band.

conduction). Note how the overall band diagram of the device isanalogous to that of conventional Schottky-gated heterostructuresbased on inorganic materials22.

The linear temperature dependence of the carrier density isa manifestation of the band alignment at the rubrene/PDIF-CN2interface, which is why the effect is highly reproducible in differentsamples. The total electron density, on the other hand, is deter-mined by the (uncontrolled) concentration of dopants uninten-tionally present and by the distance between the interface and theSchottky gate, which explains the sample-to-sample fluctuations.The net result of these two contributions is to fix the electrochemicalpotential at the interface, close to the bottom of the conductionband of PDIF-CN2, in the tail of states inevitably present owingto disorder20,35,36 (that is, the shallow traps of PDIF-CN2). As thetemperature is lowered, the electron density decreases, lowering thechemical potential at the PDIF-CN2 surface. At the same time the

gap between the rubrene valence band and the PDIF-CN2 conduc-tion band—which is proportional to the charge density transferredfrom one surface to the other—also decreases. As a result, the po-sition of the chemical potential relative to the bottom of the PDIF-CN2 conduction band is notmuch affected (that is, it is pinned closeto the bottom of the PDIF-CN2 conduction band). This is the mainmechanism for the high FET mobility values observed at 30–40K,because the proximity of the chemical potential to the bottom ofthe PDIF-CN2 conduction band results in a large concentration ofelectrons in the band (thermally activated out of the shallow traps),even at the lowest temperature of our measurements. Indeed, thedata in Fig. 4a,b show that at a low temperature there is a correlationbetween electron density and field-effect mobility, with higher den-sities corresponding to larger mobility values. The situation is dif-ferent in PDIF-CN2 FETs based on suspended single crystals—thedevices that exhibit the largest room-temperature mobility29—in






which the maximum density of charge carriers that can be gateinduced is ∼1011 cm−2. The resulting larger distance between thechemical potential and the bottom of the PDIF-CN2 conductionband causes an exponential suppression of the number of chargecarriers activated out of the shallow trap states, leading to orders-of-magnitude lower field-effectmobility at 30–40K (Fig. 4a).

Our results represent the first, detailed characterization ofthe electronic properties of single-crystal organic charge-transferinterfaces, enabling the determination of the nature of theinterfacial charge carriers, as well as their temperature-dependentdensity and mobility. Their theoretical analysis successfully linksthe observed transport properties to the interfacial electronicstructure, in a way that can be cross-checked experimentally (forinstance, through Hall effect measurements and SKFPM). It isnot yet common, in the field of organic semiconductors, to beable to establish such a detailed and consistent description of thelow-energy electronic properties of artificially realized structures.The possibility to do so in this case is largely due to the use, forthe experimental investigations, of Schottky-gated heterostructuresbased on the highest quality organic single-crystals available atpresent. As these same devices allow measurements of electrontransport down to temperatures ∼30K—much lower than whatis normally accessed with conventional FETs—while preservinga high charge carrier mobility, organic single-crystal Schottky-gated interfaces can be expected to play an important role inthe exploration of the intrinsic transport properties of molecularsemiconductors33,37,38.

Received 21 November 2011; accepted 19 June 2012;published online 22 July 2012

References1. Forrest, S. R. The path to ubiquitous and low-cost organic electronic appliances

on plastic. Nature 428, 911–918 (2004).2. Tang, C. W. & VanSlyke, S. A. Organic electroluminescent diodes. Appl. Phys.

Lett. 51, 913–915 (1987).3. Koch, N. Organic electronic devices and their functional interfaces.

ChemPhysChem 8, 1438–1455 (2007).4. Harada, K. et al. Organic homojunction diodes with a high built-in potential:

Interpretation of the current–voltage characteristics by a generalized Einsteinrelation. Phys. Rev. Lett. 94, 036601 (2005).

5. Hamwi, S., Riedl, T. & Kowalsky, W. An organic p–i–n homojunction as ultraviolet light emitting diode and visible-blind photodiode in one. Appl. Phys. Lett.99, 053301 (2011).

6. Tang, C. W. Two-layer organic photovoltaic cell. Appl. Phys. Lett. 48,183–185 (1986).

7. Peumans, P. & Forrest, S. R. Very-high-efficiency double-heterostructurecopper phthalocyanine /C60 photovoltaic cells. Appl. Phys. Lett. 79,126–128 (2001).

8. Armstrong, N. R. et al. Organic/organic’ heterojunctions: Organic lightemitting diodes and organic photovoltaic devices.Macromol. Rapid Commun.30, 717–731 (2009).

9. Dodabalapur, A., Katz, H. E., Torsi, L. &Haddon, R. C.Organic heterostructurefield-effect transistors. Science 269, 1560–1562 (1995).

10. Rost, C., Gundlach, D. J., Karg, S. & Rieß, W. Ambipolar organicfield-effect transistor based on an organic heterostructure. J. Appl. Phys.95, 5782–5787 (2004).

11. Wang, J., Wang, H., Yan, X., Huang, H. & Yan, D. Organic heterojunction andits application for double channel field-effect transistors. Appl. Phys. Lett. 87,093507 (2005).

12. Wang, H. et al. Ambipolar organic field-effect transistors with air stability, highmobility, and balanced transport. Appl. Phys. Lett. 88, 133508 (2006).

13. Dinelli, F. et al. High-mobility ambipolar transport in organic light-emittingtransistors. Adv. Mater. 18, 1416–1420 (2006).

14. Muccini, M. A bright future for organic field-effect transistors. Nature Mater.5, 605–613 (2006).

15. Alves, H., Molinari, A. S., Xie, H. & Morpurgo, A. F. Metallic conduction atorganic charge-transfer interfaces. Nature Mater. 7, 574–580 (2008).

16. Nakano, M. et al. Small gap semiconducting organic charge-transfer interfaces.Appl. Phys. Lett. 96, 232102 (2010).

17. Kronik, L. & Koch, N. Electronic properties of organic-based interfaces.Mater. Res. Soc. Bull. 35, 417–421 (2010).

18. Soeda, J. et al. Solution-crystallized organic field-effect transistors withcharge-acceptor layers: High-mobility and low-threshold voltage operation inair. Adv. Mater. 23, 3309–3314 (2011).

19. Wang, H. & Yan, D. Organic heterostructures in organic field-effect transistors.NPG Asia Mater. 2, 69–78 (2010).

20. Podzorov, V. et al. Intrinsic charge transport on the surface of organicsemiconductors. Phys. Rev. Lett. 93, 086602 (2004).

21. Molinari, A. S., Alves, H., Chen, Z., Facchetti, A. & Morpurgo, A. F. Highelectron mobility in vacuum and ambient for PDIF-CN2 single-crystaltransistors. J. Am. Chem. Soc. 131, 2462–2463 (2009).

22. Davies, J. H. The Physics of Low-Dimensional Semiconductors: An Introduction(Cambridge Univ. Press, 1998).

23. Sze, S. M. Physics of Semiconductor Devices 2nd edn (Wiley, 1981).24. Kaji, T., Takenobu, T., Morpurgo, A. F. & Iwasa, Y. Organic single-crystal

Schottky gate transistors. Adv. Mater. 21, 3689–3693 (2009).25. Lampart, M. A. & Mark, P. Current Injection in Solids (Academic, 1970).26. Matsushima, T., Goushi, K. & Adachi, C. Charge-carrier injection

characteristics at organic/organic heterojunction interfaces in organiclight-emitting diodes. Chem. Phys. Lett. 435, 327–330 (2007).

27. De Boer, R. W. I., Gershenson, M. E., Morpurgo, A. F. & Podzorov, V.Organic single-crystal field-effect transistors. Phys. Status Solidi A 201,1302–1331 (2004).

28. Krellner, C. et al. Density of bulk trap states in organic semiconductorcrystals: Discrete levels induced by oxygen in rubrene. Phys. Rev. B 75,245115 (2007).

29. Minder, N., Ono, S., Facchetti, A. & Morpurgo, A. F. Band-like electrontransport in organic transistors and implication of the molecular structure forperformance optimization. Adv. Mater. 24, 503–508 (2012).

30. Veres, J., Ogier, S. D., Leeming, S. W., Cupertino, D. C. & Khaffaf Mohialdin, S.Low-k insulators as the choice of dielectrics in organic field-effect transistors.Adv. Funct. Mater. 13, 199–204 (2003).

31. Stassen, A. F., de Boer, R.W. I., Iosad, N. N. &Morpurgo, A. F. Influence of thegate dielectric on the mobility of rubrene single-crystal field-effect transistors.Appl. Phys. Lett. 85, 3899–3901 (2004).

32. Hulea, I. N. et al. Tunable Fröhlich polarons in organic single-crystal transistors.Nature Mater. 5, 982–986 (2006).

33. Fratini, S. & Ciuchi, S. Bandlike motion and mobility saturation in organicmolecular semiconductors. Phys. Rev. Lett. 103, 266601 (2009).

34. Ellison, D. J., Lee, B., Podzorov, V. & Frisbie, C. D. Surface potential mappingof SAM-functionalized organic semiconductors by Kelvin probe forcemicroscopy. Adv. Mater. 23, 502–507 (2011).

35. Xie, H., Alves, H. &Morpurgo, A. F. Quantitative analysis of density-dependenttransport in tetramethyltetraselenafulvalene single-crystal transistors: Intrinsicproperties and trapping. Phys. Rev. B 80, 245305 (2009).

36. Merlo, J. A. et al. P-channel organic semiconductors based on hybridacene-thiophene molecules for thin-film transistor applications. J. Am. Chem.Soc. 127, 3997–4009 (2005).

37. Troisi, A. & Orlandi, G. Charge-transport regime of crystalline organicsemiconductors: Diffusion limited by thermal off-diagonal electronic disorder.Phys. Rev. Lett. 96, 086601 (2006).

38. Coropceanu, V. et al. Charge transport in organic semiconductors. Chem. Rev.107, 926–952 (2007).

AcknowledgementsThe authors would like to thank C. Caillier for his assistance during the SKFPMmeasurements and S. Ono, A. Ferreira and I. Crassee for assistance. This study wasfinancially supported by MaNEP, the Swiss National Science Foundation, NEDO and theAFOSR (FA9550-08-01-0331).

Author contributionsI.G.L. developed and fabricated the devices; performed electrical, Hall and SKFPMmeasurements; analysed the data and interpreted the results. M.N. fabricated andcharacterized the first un-gated rubrene/PDIF-CN2 charge-transfer interfaces. N.A.M.designed the Hall set-up and contributed to the electrical characterization. F.V.D.G.contributed to the device fabrication and to the electrical measurements. Z.C. and A.F.synthesized and provided the sample material from which the PDIF-CN2 single crystalswere grown. A.F. also contributed to the writing of the manuscript. A.F.M. planned andsupervised the work, interpreted the results and wrote the manuscript. All authorscontributed to the scientific discussion of the results.

Additional informationSupplementary information is available in the online version of the paper. Reprints andpermissions information is available online at www.nature.com/reprints.Correspondence and requests for materials should be addressed to A.F.M.

Competing financial interestsThe authors declare no competing financial interests.





http://www.nature.com/reprints


single-crystal organic charge-transfer interfaces probed using schottky-gated heterostructures

Documents