The correlation between gate dielectric, film growth, and charge transport in organic thin film transistors: The case of vacuum-sublimed tetracene thin films

Journal ofMaterials Chemistry C

PAPER

Dow

nloa

ded

by C

NR

Mila

no o

n 07

Jan

uary

201

3Pu

blis

hed

on 2

2 N

ovem

ber

2012

on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/C

2TC

0033

7F

View Article OnlineView Journal | View Issue

aDepartement de genie physique, Ecole

Succursale Centre-Ville, Montreal, Queb

[email protected]; Tel: +1 514 340-4711bIMEM-CNR, Parco Area delle Scienze 37/A,cInstitute of Crystallography, CNR, Strada S

Science Park, 34149 Basovizza, Trieste, ItalydDipartimento di Ingegneria dei Materiali,

38123 Trento, Italy. E-mail: luca.lutterotti@eDepartement de genie chimique, Ecole

Succursale Centre-Ville, Montreal, Quebec H

† Electronic supplementary informationmicroscopy images, XRD reconstructedcharacteristics of tetracene TFTs. See DOI

Cite this: J. Mater. Chem. C, 2013, 1,967

Received 4th October 2012Accepted 18th November 2012

DOI: 10.1039/c2tc00337f

www.rsc.org/MaterialsC

This journal is ª The Royal Society of

The correlation between gate dielectric, film growth,and charge transport in organic thin film transistors: thecase of vacuum-sublimed tetracene thin films†

Julia Wunsche,a Giuseppe Tarabella,b Simone Bertolazzi,a Maimouna Bocoum,a

Nicola Coppede,b Luisa Barba,c Gianmichele Arrighetti,c Luca Lutterotti,d

Salvatore Iannotta,b Fabio Cicoirae and Clara Santato*a

The complex interplay of dielectric substrate properties, semiconducting film growth, crystal structure,

texture, and charge carrier transport is investigated for the case of tetracene films deposited on

different dielectrics (polystyrene, parylene C, polymethylmethacrylate, hexamethyldisilazane-treated

SiO2, and untreated SiO2). The tetracene hole mobility, measured in the bottom-gate organic thin film

transistor (OTFT) configuration, varies over more than one order of magnitude depending upon the

dielectric layer used. Atomic force microscopy and synchrotron grazing incidence X-ray diffraction

measurements, analyzed with the extended Rietveld method, were used to investigate the influence of

film connectivity, crystalline phase, polymorphism, and texture on charge transport. The role of the

surface polarity and the processing conditions of the gate dielectric layer are also discussed. Based on

our results, we propose guidelines for the selection of a gate dielectric material favorable for charge

transport in tetracene films.

Introduction

Organic electronic devices such as organic light-emitting diodes(OLEDs),1 photovoltaic cells,2 thin lm transistors (OTFTs),3

light-emitting transistors,4,5 and biosensors6 are undergoing animpressive progress towards large area and exible applica-tions. OLEDs have already entered the market as components ofat-panel displays and light sources. However, several chal-lenges still need to be overcome to make other organic elec-tronic devices viable for a wider range of applications.

Organic semiconducting thin lms constitute the basis forthe large majority of organic electronic devices, such as OTFTs.The characteristics of OTFTs are affected by several factors,such as the molecular arrangement in the lm, the lmmorphology, and the properties of the interfaces between the

Polytechnique de Montreal, CP 6079,

ec H3C 3A7, Canada. E-mail: clara.

# 2586

43100 Parma, Italy

tatale 14, Basovizza, Km 163,5 in AREA

Universita di Trento, Via Mesiano, 77,

ensicaen.fr

Polytechnique de Montreal, CP 6079,

3C 3A7, Canada

(ESI) available: AFM and uorescencepole gures, output and transfer

: 10.1039/c2tc00337f

Chemistry 2013

organic lm and other device materials, such as metal elec-trodes, affecting charge injection,7,8 and gate dielectrics.9–12 InOTFTs, where it also acts as a substrate, the gate dielectricmaterial plays a primary role in establishing charge transportproperties. Along this line, we recently demonstrated that anappropriate choice of the gate dielectric material can enhancethe hole mobility in tetracene TFTs by more than one order ofmagnitude.13

For lms of a given molecular organic semiconductor,several growth modes and crystal structures are observed, as aresult of the weak van der Waals intermolecular forces.14,15 As aconsequence, the substrate properties radically affect themorphology, crystal structure, and texture of organic lms.16

The gate dielectric can also inuence the charge transportproperties of OTFTs via the interfacial chemistry with thesemiconductor, which can control the charge carrier trapdensity at the semiconducting lm–dielectric interface.10,11

Moreover, a gate dielectric with high dielectric constant canaffect the charge transport in the organic semiconductor inproximity to the dielectric due to the increase in both static(statistically oriented surface dipoles) and dynamic (enhancedelectron–phonon coupling) disorder.17–19

The objective of this work is to shed light on the role of thegate dielectric material in establishing the lm morphology,crystal structure, texture and charge transport in tetracenelms. Tetracene was selected for its combined charge carriertransport and electroluminescence properties, which have beenexploited in organic (light-emitting) transistors.20–24 We studied

J. Mater. Chem. C, 2013, 1, 967–976 | 967

Fig. 1 Structure of the tetracene thin film transistors and molecular structures of the organic dielectrics.

Journal of Materials Chemistry C Paper

Dow

nloa

ded

by C

NR

Mila

no o

n 07

Jan

uary

201

3Pu

blis

hed

on 2

2 N

ovem

ber

2012

on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/C

2TC

0033

7FView Article Online

lms deposited on three polymer dielectrics, compatible withall-plastic TFTs:25–27 polystyrene (PS), parylene C (PARY C), andpolymethylmethacrylate (PMMA), and on hexamethyldisilazane(HMDS) (Fig. 1).28 Atomic Force Microscopy (AFM) revealed thatthe growth morphology of tetracene lms is strongly affected bythe gate dielectric substrate. Grazing Incidence X-Ray Diffrac-tion (GIXRD) data, analyzed with the extended Rietveldmethod,29 gave unprecedented insights into the crystal struc-ture and texture of tetracene lms. The analysis revealed theeffect of the substrate on the volume fraction of the differentobserved crystallographic phases. Film textures were extractedto determine the orientation of the grains with respect to thecharge transport direction. Inclined and centered textures wereobserved depending upon the dielectric.

‡ It should be noted that the validity of different approaches to determine surfaceand adhesion energies is still a matter of scientic debate (ref. 31–34).

ExperimentalSample preparation

All samples were prepared on SiO2/n-Si wafers, where theheavily doped h100i Si wafer (resistivity 0.001–0.005 U cm)served as the gate electrode and the 200 nm thick thermallygrown SiO2 top-layer served as part of the gate dielectric. All thepolymer dielectrics were deposited on SiO2 (Fig. 1). Thisapproach permits us to study the effect of a specic dielectricsubstrate, while ensuring low gate leakage currents.10 The SiO2

substrates were cleaned by sequential sonication in acetone andisopropanol, followed by 20 min UV-ozone treatment. Thebottom-gate/top-contact OTFT structure employed is shown inFig. 1. The polymer dielectrics and the HMDS SAMs wereprepared as follows. PS (211 000 g mol�1, Sigma Aldrich) andPMMA (100 000 g mol�1, Sigma Aldrich) were spin-coated fromtoluene solutions (5000 rpm, 10 mg ml�1 for PS, 15 mg ml�1 forPMMA) in an ambient atmosphere and dried overnight underlow vacuum at 80 �C.10,25 The solution processing of the polymerdielectrics was optimized until homogeneous and pinhole-freelms were obtained, as veried by AFM images. PARY C wasdeposited by chemical vapor deposition (10 mTorr base pres-sure) using a Parylene Deposition System (PDS).30 The lmthickness t of the polymers was measured by stylus prolometryand ellipsometry (Table 1). HMDS (Gelest, Inc.) was spin-coated(2000 rpm) in a N2 glove box and, subsequently, annealed at120 �C for 1 h. Films of tetracene (Tokyo Chemical Industries,purity > 97%) were deposited by vacuum sublimation on PS,PARY C, PMMA, HMDS, and SiO2 dielectrics at 3.5 A s�1 (base

968 | J. Mater. Chem. C, 2013, 1, 967–976

pressure of 5 � 10�6 Torr).22 The thickness of the tetracenelms used for GIXRD measurements and OTFTs was about 50nm. Au source and drain top contacts (channel length L ¼ 60,80, 100, or 150 mm and channel widths W ¼ 2 or 4 mm) weredeposited by thermal evaporation through a shadow mask(Thin Metal Parts, Colorado Springs, CO).

Measurement of surface energy

To determine the surface energies of the different dielectriclayers, gd, and their adhesion energies with tetracene, Wa, thecontact angles, q, of six liquids (water, ethylene glycol, glycerol,n-dodecane, formamide, tricresyl phosphate) were measured.The surface tension component approach by Owens and Wendtwas used to calculate g andWa.31‡ In this approach, g is the sumof its polar (gP) and dispersive (gD) components and interac-tions are described by geometric means. To obtain gd,glð1þ cos qÞ=ð2 ffiffiffiffiffiffi

gDl

p Þ is plotted versusffiffiffiffiffiffiffiffiffiffiffiffiffigPl =g

Dl

p, where gl, g

Dl ,

gPl are the surface energy and its components for the liquid

used.35 A linear t then gives the square root of the componentsof the dielectric surface energy,

ffiffiffiffiffiffigPd

pand

ffiffiffiffiffiffigDd

p, as the offset

and slope, respectively. The same approach was applied toobtain the surface energy of tetracene lms vacuum-sublimedon untreated and HMDS-treated SiO2. The measured contactangles were in good agreement for the two substrates.Wa can becalculated from the surface energies of the tetracene lm (gtc

and its components gPtc and gD

tc) and of the dielectric layer as:Wa ¼ 2

ffiffiffiffiffiffiffiffiffiffiffigPdg

Ptc

p þ 2ffiffiffiffiffiffiffiffiffiffiffiffigDdg

Dtc

p.35

Investigation of morphology and structure

The morphology of the tetracene lms and of the dielectricsurfaces was investigated by Atomic Force Microscopy (AFM,Digital Instruments Dimension 3100) in tapping mode, underambient conditions, using an aluminum-coated silicon canti-lever (radius < 10 nm, spring constant 42 N m�1). At least fourimages were taken for each sample, in four distinct regions. TheAFM images were analyzed using WSxM soware.36 The islandgrowth was quantitatively analyzed with the Flooding functionof WSxM on two images, 10 mm � 10 mm- and 15 mm � 15 mm-sized, for 2 and 5 nm nominal thickness. At higher nominalthicknesses, the Flooding function was not used since the

This journal is ª The Royal Society of Chemistry 2013

Table 1 Properties of the dielectric layers used in this work: film thickness, t; average RMS surface roughness for 1 mm � 1 mm images, av. Rq; water contact angle, qw;surface energy, g, and its polar, gP, and dispersive component, gD; the adhesion energy with tetracene Wa. The table also includes the dielectric constants, 3r, of thedielectric layers and the total capacitance, Ctot, for the bilayer structures with SiO2

Dielectric t [nm] av. Rq [nm] qw g [mJ m�2] gP [mJ m�2] gD [mJ m�2] Wa [mJ m�2] 3r Ctot [nF cm�2]

PS 33 � 1 0.15 96� � 2� 32.7 � 0.6 0.8 � 0.3 31.8 � 0.3 66.4 � 1.2 2.6 (ref. 10 and 41) 13.9PARY C 93 � 4 1.2 89� � 2� 37.4 � 1.0 2.2 � 0.5 35.2 � 0.5 71.3 � 1.5 3.1 (ref. 42) 10.9PMMA 35 � 1 0.15 73� � 2� 35.0 � 0.8 8.0 � 0.4 27.0 � 0.4 66.8 � 1.4 3.5 (ref. 25 and 41) 14.5HMDS — 0.27 68� � 3� 38.4 � 0.5 13.0 � 0.3 24.5 � 0.6 67.2 � 1.3 — z17.3SiO2 200 � 2 0.11 4� � 3� 67.9 � 1.1 46.0 � 0.6 21.9 � 0.5 72.1 � 2.3 3.9 17.3

Paper Journal of Materials Chemistry C

Dow

nloa

ded

by C

NR

Mila

no o

n 07

Jan

uary

201

3Pu

blis

hed

on 2

2 N

ovem

ber

2012

on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/C

2TC

0033

7FView Article Online

tetracene islands are no longer well separated, yielding unreli-able results.

GIXRD measurements were performed at the XRD1 beam-line at the ELETTRA synchrotron facility in Trieste, Italy. Amonochromatic beam of 8.05 keV was directed at 1� incidentangle on the samples. The diffraction patterns were recorded inreection mode with the 2D detector MarCCD165 (MarRe-search) positioned perpendicular to the incident beam at adistance of 130 mm. The sample inclination to the beam waschanged from u ¼ 1� to u ¼ 9� in steps of 1� yielding 9diffraction images that were simultaneously tted to increasethe pole gure coverage. The images were analyzed by means ofthe Maud soware,37 which makes use of the extended Rietveldmethod. The Rietveld method is used to rene structuralparameters by non-linear least square tting.38 In thiscombined analysis, other parameters, such as microstrain,grain size, and texture are added in the renement process. Forthe tetracene lms, a triclinic crystal structure was assumed, asretrieved from the Cambridge Structural Database (CSD, Allen2002).

Since only a few 00l reections could be measured withreliable intensity, only the unit cell parameter c was optimizedin the Rietveld renement, while the angles were kept constantand a and b were adjusted by the user in such a way that theagreement between recalculated and experimental 2D plot wasmaximized.§ For the texture analysis, the texture model ofstandard functions was used.39,40

Electrical characterization

The OTFT characteristics were measured in an inert (N2)atmosphere using a micromanipulated probe station (LakeShore CPX6) and a semiconductor parameter analyzer (AgilentB1500A). Hole mobility, mh, threshold voltage, VT, and currenton–off ratio were extracted from the transfer curves in thesaturation regime at a drain–source voltage VDS of�60 V, duringthe forward scan, and averaged over at least 5 devices for eachdielectric. The following equation models the current–voltagebehavior in the saturation regime and permits us to extract mhand VT,

IDS ¼ W

2LmhCtotðVGS � VTÞ2

§ A manual procedure was necessary due to the low intensity of the reections,insufficient for the least-square tting of the in-plane unit cell parameters inMaud.

This journal is ª The Royal Society of Chemistry 2013

where IDS is the drain–source current and VGS is the gate–sourcevoltage. The bilayer capacitance Ctot was calculated fromCtot

�1 ¼ Corganic�1 + CSiO2

�1. The capacitance of each layer isgiven by C ¼ 3r30/t, where t, 3r, and 30 are the lm thickness, therelative, and absolute dielectric permittivity, respectively.

Results and discussionCharacterization of dielectric layers

To gain insight into the factors governing lm nucleation andgrowth, we investigated the surface roughness and the surfaceenergy of the dielectric layers. PS and PMMA showed a rootmeansquare roughness (Rq) as low as that of untreated SiO2, whereasPARY C showed the highest roughness. The HMDS treatmentalso increased Rq with respect to untreated SiO2 (Table 1). Allorganic dielectrics provided a more hydrophobic surface thanSiO2, as indicated by their water contact angles, qw (Table 1). Thepolar component of the surface energy (gP) increased in thefollowing order: PS, PARY C, PMMA, HMDS, SiO2 (Table 1). Thedispersive component (gD) followed the opposite order, exceptfor PARY C, which had a higher gD than PS (Table 1).

Using polar and dispersive components of the surface energyis a macroscopic approach to describe the interaction of thesurface with polar and non-polar molecules. The componentsare used to calculate the adhesion energy of the semiconductinglayer to the dielectric substrates. Apart from its role in thecalculation of the adhesion energy, the polar component of thesurface energy is also a qualitative measure for the presence ofsurface dipoles, which can be detrimental to charge transport,as discussed below.

Despite the differences between the polar and dispersivecomponents of the surface energy of the organic dielectrics, theadhesion energy of tetracene (Wa) on the different dielectricshad similar values, with the exception of PARY C, where theadhesion energy was slightly higher and comparable to that oftetracene on SiO2.‡

Film growth and morphology

The growth and morphology of tetracene lms deposited on PS,PARY C, PMMA, HMDS, and untreated SiO2 were studied byAFM for a nominal thickness between 2 and 35 nm (Fig. 2). Themorphology at the early stages of growth is critical for chargetransport in bottom-gate OTFTs, where carriers move within athin layer of the semiconducting lm at the interface with thedielectric.3

J. Mater. Chem. C, 2013, 1, 967–976 | 969

Fig. 2 10 mm � 10 mm AFM images of tetracene thin films on different dielectric surfaces at different nominal thickness. Z-scale: 50 nm.

Journal of Materials Chemistry C Paper

Dow

nloa

ded

by C

NR

Mila

no o

n 07

Jan

uary

201

3Pu

blis

hed

on 2

2 N

ovem

ber

2012

on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/C

2TC

0033

7FView Article Online

In the early stages of growth, the island height is larger thanthe nominal lm thickness, in all dielectrics (Fig. 3). Thisindicates that tetracene lms grow in a three-dimensional (3D)mode (Volmer–Weber like growth). The size and coalescence ofthe islands were found to depend on the substrate. Underthermodynamic equilibrium conditions, the formation of 3Dislands is energetically favorable if the adhesion energy betweenthe dielectric and tetracene, Wa, is smaller than twice the tet-racene surface energy (gtc).{ Nevertheless, if growth kineticsdominates over thermodynamics, the actual growth behavior

{ Wa < 2gtc follows from the condition that gi + gtc > gs for the formation of 3Dislands and the denition of the adhesion energy Wa ¼ gtc + gs � gi, where gtc,gs, and gi are the surface energy of tetracene, the surface energy of thesubstrate, and the interfacial energy, respectively.

970 | J. Mater. Chem. C, 2013, 1, 967–976

can be different.16 Our data show that the adhesion energy oftetracene on the organic dielectrics is close to 2gtc for alldielectrics, which makes it difficult to predict a 3D versus a two-dimensional (2D) growth mode. As revealed by AFM images, theslightly higher adhesion energy of tetracene on PARY C and SiO2

did not lead to 2D growth under the deposition conditionsemployed in this work (growth rate of 3.5 A s�1).

Tetracene lms on PS show rather large islands and lowisland density at both 2 nm and 5 nm (Fig. 3). The islands on PSare regularly shaped and coalesce well, so that full substratecoverage is already achieved at 10 nm nominal thickness. Themorphology changes only slightly at higher lm thicknesses(between 17 nm and 35 nm). The island surface is at, as can bededuced from a plot of the surface inclination of 35 nm thicklms (ESI, Fig. S1†).

This journal is ª The Royal Society of Chemistry 2013

Fig. 3 Average island surface area, density, and height for vacuum sublimedtetracene films of 2 nm (left bar) and 5 nm (right bar) nominal thicknesses,deposited on the substrates indicated on the x axis.

Paper Journal of Materials Chemistry C

Dow

nloa

ded

by C

NR

Mila

no o

n 07

Jan

uary

201

3Pu

blis

hed

on 2

2 N

ovem

ber

2012

on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/C

2TC

0033

7FView Article Online

On PARY C, tetracene islands are slightly smaller than on PSat 2 nm and 5 nm lm thickness. The increase of island densityfrom 2 nm to 5 nm and the more dendritic shape of the islandsat later stages of growthmight originate from the slightly higheradhesion energy of tetracene on PARY C.43 Full substratecoverage is achieved at 17 nm nominal thickness, as conrmedby uorescence microscopy images (Fig. S2†). At 35 nm, tetra-cene lms on PARY C provide a strongly structured surface. Thesurface inclination reveals that the islands are not at but havea curved surface (Fig. S1†). We suggest that the higher surfaceroughness of PARY C lms contributes to the formation of thesesmaller inclined islands.44 Despite their more dendritic shape,tetracene islands on PARY C coalesce well and form a contin-uous layer.

Tetracene lms on PMMA appear similar to those on PS.However, throughout all stages of growth, the tetracene islandsare smaller and the island density is higher. The island surfaceis slightly more curved than on PS (Fig. S1†).

Fig. 4 Integration of the diffraction image within slices (a) to obtain 2D plots of thethe Maud software.

This journal is ª The Royal Society of Chemistry 2013

On HMDS, tetracene grows in very large and at islands.However, due to the irregular shape of the islands, some voidsremain aer island coalescence. Even at 35 nm, the substrate isnot completely covered. An incomplete coverage of SiO2 withHMDS or defects in the HMDS monolayer may play a role in theformation of irregularly shaped islands.10,45

The tetracene growth on untreated SiO2 from 2 nm to 5 nmnominal thickness shows a strong increase in island density,while the island surface area remains constant. This is anindication of limited surface diffusion of the tetracene mole-cules, possibly due to a strong interaction with the SiO2

surface.46 The increase in island height at the early stages issomewhat surprising given the high surface energy. It could bethe result of the specic interplay between deposition kineticsand high diffusion step edge barrier.47 The irregular shape oftetracene islands on SiO2 results in a discontinuous lm withvoids. Between 17 nm and 35 nm, a morphological transitionoccurs resulting in a top layer of smaller, more regularly shapedtetracene islands. A similar change in morphology was observedpreviously for tetracene on SiO2.22 It was attributed to thedifferent interaction of tetracene molecules with the pre-deposited tetracene layer, as compared to the SiO2 layer.

In summary, tetracene lms on PS, PARY C, and PMMA arecharacterized by regularly shaped, well interconnected islands,and a full substrate coverage at a thickness between 10 and17 nm. The size of tetracene islands on PMMA is particularlysmall, while it is large on HMDS. The irregular shape of islandson SiO2 and HMDS creates deep voids in the lm, present evenat high nominal thickness. The fact that tetracene showsdifferent morphologies on surfaces with similar adhesionenergy proves that the growth behavior of tetracene thin lmscannot be predicted based exclusively on the adhesionenergy.19,43,48

Crystal structure and texture

To investigate the structure of the tetracene lms grown on PS,PARY C, PMMA, HMDS, and untreated SiO2, X-ray diffractionwas performed in reection mode at grazing incidence angleson 50 nm thick lms (the thickness used in OTFTs). 1D patternswere generated from the 2D diffraction images by dividing the180� upper reection part into slices of 1� and integrating thediffracted beam intensity along the azimuthal angle within

diffracted beam intensity as a function of 2q and (b) for the Rietveld analysis with

J. Mater. Chem. C, 2013, 1, 967–976 | 971

Journal of Materials Chemistry C Paper

Dow

nloa

ded

by C

NR

Mila

no o

n 07

Jan

uary

201

3Pu

blis

hed

on 2

2 N

ovem

ber

2012

on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/C

2TC

0033

7FView Article Online

these slices (Fig. 4). The patterns obtained in this way wereplotted along the y axis. The data were analyzed with Maudsoware.49

Representative 2D plots of our samples are shown in Fig. 5.The experimental data are plotted in the bottom half of eachplot, while the top half shows the 2D plot recalculated by theMaud soware, based on the structural and textural parametersobtained during the extended Rietveld analysis. The recalcu-lated plot is generally in good agreement with the experimentaldata. The presence of a few well-dened peaks in all the plots isindicative of a polycrystalline structure with strong texture. Thebrightest reections correspond to (001) and (002) planes andcharacterize the c cell parameter. Other weaker reectionsobservable at different angles were used to adjust the other cellparameters.

The double peak along the 2q direction in the 2D plots(Fig. 5) reveals the presence of two crystal phases, a and b.Polymorphism in tetracene lms has been reported previ-ously.50,51 In both phases, the triclinic unit cell (parametersgiven in Table 2) contains two tetracene molecules with theirlong axes aligned almost parallel to the substrate normal, asdeduced from the Rietveld analysis. The two molecules are til-ted with respect to each other, in a herringbone structure. Themolecular packing is illustrated in Fig. 6. From the unit cellparameters, interplanar spacings of d00l ¼ 1.21 nm and d00l ¼1.27 nm can be calculated for a and b phases, respectively. Theinterplanar spacing of the a phase is close to the value reportedfor tetracene powder samples (1.21–1.23 nm).50,52 Therefore, thea phase should be energetically the most stable and it is here-aer referred to as the bulk phase. The b phase is characterizedby a longer c and a shorter b axis, as compared to the a phase(Table 2). This results in a closer packing along the a–b plane, alarger (00l) interplanar spacing, and, consequently, a moreupright position of the tetracene molecules in the b phase. Theformation of the less stable b phase is induced by the substrateand is referred to as the thin lm phase. A strong variation ofthe volume fraction of the two phases upon the dielectric layerwas found. Tetracene lms on PS show the highest fraction of bphase, whereas the use of PMMA results in the lowest fraction.

The other samples were found to contain a and b phases insimilar volume fraction. The dependence of the volume fractionof a and b phases on the substrate, despite similar adhesionenergies (Table 1), points to the presence of specic molecule–substrate interactions. Our results suggest that the non-polarsurface of PS is a favorable substrate for the growth of the tet-racene thin lm phase.

Film texture is of paramount importance since it determinesthe orientation of the grains with respect to the charge transportdirection. Information about the lm texture can be deducedfrom the intensity variation along the y axis for a given 2q in the2D-plot (Fig. 5).k The symmetry of the peaks with respect to thecenter of the plot along the vertical direction, together with theinvariance of the diffraction images to in-plane rotations of thesample, are characteristic of a ber texture, which is expected for

k This corresponds to the intensity variation along the so-called h angle for a given2q in the diffraction images (i.e. along the diffraction rings).

972 | J. Mater. Chem. C, 2013, 1, 967–976

thin lms on amorphous substrates. Indeed, two types of bertextures were found: centered and inclined. The presence of onlyone peak centered along the vertical direction for the tetracenelms on PS, HMDS, and SiO2 indicates that the (00l) planes arealigned parallel to the substrate, i.e. they have a centered texture.In contrast, the split of the peaks for the tetracene lms on PARYC and PMMA results from an inclined texture, i.e., the a–b planeis inclined with respect to the substrate.

Using the Rietveld renement, we determined the orienta-tion distribution function, expressed by the orientation of theber axis (the symmetry axis of the texture) and the angularspread of the orientation of the crystallites around the ber axis(full width at half maximum, FWHM). For all tetracene lms,the ber axis is oriented perpendicular to the sample surface.The direction of the ber axis is further described by the polarangle, qH, and the azimuthal angle, 4H, with respect to the c anda axes of the crystallographic coordinate system (Table 3). Inagreement with their centered texture, tetracene lms on PS,HMDS, and SiO2 yield very similar angles. These angles result inan a–b plane approximately parallel to the substrate, asdeducible from the reconstructed pole gure for the (001)crystallographic plane (Fig. S3†). The ber texture angles for thelms on PARY C and PMMA differ from the angles of thecentered texture and therefore indicate an inclined texture. TheFWHM of the orientation distribution function reveals that thealignment of the crystallites along the texture direction is strongon PS, HDMS, and SiO2, while it is considerably weaker on PARYC and PMMA. This is also reected in the pole gures (Fig. S3†).The origin of these different textures is still under investigation.In the case of PARY C, the observation of a tilted texture withweak alignment could be related to the higher surface rough-ness of the PARY C layer. This explanation does not hold for thePMMA layer, which shows low surface roughness.

The extended Rietveld renement allowed only for a relativecomparison of the out-of-plane grain size in tetracene lmsdeposited on different dielectrics.** No absolute value of theout-of-plane grain size could be obtained, due to the inuenceof the micro-strain and the instrumental broadening. Theheights of tetracene grains on PS, PARY C, HMDS and SiO2 aresimilar. In contrast, the height of tetracene grains on PMMA isconsiderably lower.

OTFT performance

OTFT transfer (drain–source current, IDS, vs. gate–sourcevoltage, VGS, for a xed drain–source voltage, VDS) and output(IDS vs. VDS for a xed VGS) characteristics for OTFTs with PMMA,HMDS (Fig. S4 and S5†), PS, PARY C, and SiO2 (ref. 13) indicatethat all the devices show p-type behavior. The current on–offratio ranges between 105 and 106 for all our devices, whichindicates good switching performance. Fig. 7 shows the averagehole mobility, mh, and threshold voltage, VT, extracted from thetransfer curves in the saturation regime, for the differentdielectrics. The use of organic dielectrics led to an increase of

** The GIXRD measurements in the geometry employed in this work do notprovide any information on the in-plane grain size.

This journal is ª The Royal Society of Chemistry 2013

Fig. 5 2D plots generated by the Maud software from the GIXRD images collected on 50 nm thick tetracene films deposited on different gate dielectrics. For eachdielectric, the lower plot shows the experimental data and the upper plot shows the simulated spectra after the extended Rietveld refinement. Inset: magnified 2D plotfor the (001) reflections.

Table 2 Unit cell parameters for the a and b phases. For both phases, a tricliniccrystal structure was assumed, as reported in the Cambridge Structural Database(CSD, Allen 2002). Since only a few 00l reflections could bemeasured with reliableintensity, only c was optimized in the Rietveld refinement, while the angles werekept constant and a and b were adjusted by the user

a b c a b g

a phase 5.95 7.90 13.00 77.13 72.12 85.79b phase 6.01 7.40 13.72 77.13 72.12 85.79

This journal is ª The Royal Society of Chemistry 2013

Paper Journal of Materials Chemistry C

Dow

nloa

ded

by C

NR

Mila

no o

n 07

Jan

uary

201

3Pu

blis

hed

on 2

2 N

ovem

ber

2012

on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/C

2TC

0033

7FView Article Online

the hole mobility, with respect to untreated SiO2. PARY C and PSyielded the highest values of hole mobility,13 while PMMA yiel-ded the lowest mobility amongst the organic dielectrics. Thevalues of themobility were 0.168� 0.039, 0.064� 0.005, 0.018�0.003, 0.027 � 0.008, and 0.010 � 0.002 cm2 V�1 s�1 for OTFTswith PS, PARY C, PMMA, HMDS, and SiO2, respectively.

VT is higher for devices making use of organic dielectrics,compared to untreated SiO2. This result can be partly explainedby the lower capacitance of the bilayer structures (for PS, PMMA

J. Mater. Chem. C, 2013, 1, 967–976 | 973

Fig. 6 Tetracene unit cell for the b phase.

Table 3 Structure and texture parameters extracted from the Rietveld analysisfor tetracene films deposited on different dielectric layers: the volume fraction ofthe b thin film phase; the type of texture (C-centered, I-inclined); the orientationof the fiber axis with respect to the crystallographic cell given by qH and 4H; theangular spread around this orientation given by the FWHM

PS PARY C PMMA HMDS SiO2

b phase [vol%] 85 53 10 52 43Texture C I I C CqH [�] 21.5 28.1 (a), 31.3 (b)a 16.6 21.5 21.54H [�] 215.1 219.7 242.0 214.8 214.9FWHM [�] 1.9 6.0 5.7 1.6 1.4

a Two different qH could be extracted for the a and b phases in the caseof PARY C. The FWHM is approximately the same for both phases. Forlms with centered textures, these angles coincide. In the case ofPMMA, qH could be obtained only for the a phase due to the smallportion of b phase.

Fig. 7 Average hole mobility, mh, and threshold voltage, VT, obtained for tetra-cene OTFTs with different dielectric layers. The error bars represent the standarddeviation within 5 or more tetracene TFTs.

Journal of Materials Chemistry C Paper

Dow

nloa

ded

by C

NR

Mila

no o

n 07

Jan

uary

201

3Pu

blis

hed

on 2

2 N

ovem

ber

2012

on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/C

2TC

0033

7FView Article Online

and PARY C on SiO2, Table 1). However, other effects might alsocome into play, such as contamination during solution pro-cessing (for PS, PMMA and HMDS) and morphological featuresof the tetracene lms.

974 | J. Mater. Chem. C, 2013, 1, 967–976

We note that all the devices, except those using PARY C,show IDS hysteresis, which can be explained by charge carriertrapping at the semiconductor–dielectric interface.53 In the caseof SiO2, charge trapping is related to the presence of silanolgroups.11 The hysteresis in the I–V characteristics of OTFT withPS, PMMA, and HMDS is likely due to impurities or solventtraces resulting from the solution processing of the dielectric,and, for HMDS, to an incomplete SiO2 coverage or defects in theHMDS monolayer.10,45 The possible detrimental effect of solu-tion processing is also supported by the absence of hysteresisand the good reproducibility of the electrical measurements ofOTFTs using a PARY C dielectric. Since the hysteresis behaviorand the change in VT do not correlate with the hole mobility, weconclude that charge trapping can be only one of several factorsaffecting the tetracene hole mobility.

The high hole mobility in OTFT with PS and PARY C can beexplained by the tetracene lm morphology and structure (aswill be discussed below). At the same time, the relatively lowsurface polarity of PS and PARY C might also come into play.Although the difference between the dielectric constants of thedielectric materials employed in this work is small, the polarcomponents of the surface energy of the dielectrics (Table 1)suggest that the surface polarity of PMMA, HMDS-treated SiO2,and untreated SiO2 is considerably higher than that of PS andPARY C. The broadening of the density of states associated withthe statistically oriented surface dipoles of the dielectric layermight, thus, cause charge carrier localization in tetracene lmsdeposited on PMMA, HMDS, and SiO2.17

Correlation between morphology, structure, and chargetransport

One of the requisites for efficient charge transport in OTFTs isgood lm connectivity. AFM measurements show that theregularly shaped islands on the polymer dielectrics (PS, PARY C,PMMA) lead to a complete substrate coverage at low nominalthickness (between 10 nm and 17 nm), with well interconnectedislands. PS is the polymer dielectric, upon which completesubstrate coverage is achieved at the lowest nominal thickness.PS also yields the highest hole mobility.

In general, charge carrier mobility in OTFTs is expected toincrease with increasing lm grain size.54,55 However, it has alsobeen shown that lms with smaller grains can result in highermobility.22,56 Although larger grains reduce the density of grainboundaries that charge carriers have to overcome duringtransport, they might, at the same time, negatively affect thequality of the boundaries.57 To shed light on the effect of thegrain size on charge transport, we employed the island densityand island size, as measured by AFM, to estimate the grain size.Among the dielectrics investigated, PS leads to the growth oflarge islands and to the highest mobility (Fig. 3). Large islands,at all stages of growth, are also formed on HMDS, which insteadleads to modest OTFT performance. As PS leads also tocomplete substrate coverage at low nominal thickness, theseresults indicate that the substrate coverage dominates over thegrain size in determining the tetracene TFT performance, as hasbeen suggested previously.22

This journal is ª The Royal Society of Chemistry 2013

Paper Journal of Materials Chemistry C

Dow

nloa

ded

by C

NR

Mila

no o

n 07

Jan

uary

201

3Pu

blis

hed

on 2

2 N

ovem

ber

2012

on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/C

2TC

0033

7FView Article Online

The presence of the bulk (a) versus thin lm (b) phase has astrong impact on charge transport. It has been shown for tet-racene and pentacene lms that the bulk phase, besides beingpresent at higher lm thickness, can also nucleate on thesubstrate and grow in competition with the thin lmphase.51,58–60 Polymorphism, thus, directly affects the chargetransport at the interface with the dielectric. The thin lmphase should be advantageous for charge transport comparedto the bulk phase, due to the more upright position of tetra-cene molecules, which is expected to favor charge transport inthe direction parallel to the substrate.50,61 For OTFTs withpolymer dielectrics, the volume fraction of the thin lm phaseis on pair with the increase of hole mobility. Moreover, webelieve that the coexistence of thin lm and bulk phases isdisadvantageous, since it leads to a disruption of molecularorder. For pentacene, which has been widely investigated, therelative amount of the bulk with respect to the thin lm phaseincreases with lm thickness.56,60,62 It has been suggested thatthis should also hold for other organic semiconductors.60 Inour case, this would mean that the volume fraction of the thinlm phase at the interface with the dielectric is actually largerthan the value computed with the Rietveld analysis (given inTable 3). Based on these data, the rst tetracene monolayers onPS might be composed almost exclusively of the thin lmphase, favorable to charge transport. In agreement with that,the lower performance of tetracene TFTs with PMMA could beexplained by the mixture of the thin lm phase with aconsiderable amount of bulk phase, at the semiconductor–dielectric interface.

Film texture determines the orientation of grains withrespect to the charge transport direction. A centered textureallows for a direct path of charge carriers parallel to thesubstrate, while an inclined texture leads to a longer path ofcharge carriers. Moreover, the texture can also affect the qualityof the grain boundaries. In our tetracene lms, charge transportdoes not depend signicantly on the type of texture (centeredversus inclined). In contrast, the degree of alignment of thegrains along the texture direction (expressed as FWHM) is likelyto have an impact on charge transport. Along this line, for tet-racene lms grown on PARY C and PMMA, the weak alignmentof grains hinders charge transport.

Conclusion

We have investigated the correlation between dielectricsubstrate properties (surface energy, roughness), lmmorphology/structure, and charge transport in tetracene thinlms deposited on the organic dielectrics PS, PARY C, PMMAand HMDS. The strong texture and the predominance of thethin lm phase in tetracene lms deposited on PS explain thehigher TFT hole mobility, as compared to the other dielectricsemployed. Moreover, charge transport in tetracene lms on PSis favored by the complete substrate coverage at low nominallm thickness and well-interconnected islands. Good OTFTcharacteristics were also obtained with the PARY C dielectric,despite the high PARY C surface roughness, which increases thenucleation density and leads to less ordered lms. We believe

This journal is ª The Royal Society of Chemistry 2013

that the low charge trap density at the tetracene–PARY Cinterface, supported by the absence of hysteresis in the OTFT I–V characteristics, compensates for the lower lm order. The useof PMMA as a gate dielectric leads to a lower hole mobility,compared to PS and PARY C, as a consequence of the weakermolecular alignment in the tetracene lms, the mixture of tet-racene thin lm and bulk phases in proximity to the gatedielectric, and, possibly, charge carrier localization due to thehigher polarity of the PMMA surface.

In conclusion, our results point to some general criteria forthe selection of the gate dielectric for TFTs with vacuum-sublimed organic semiconductors. The ideal dielectric surfaceshould be non-polar, smooth, and free of solvent traces andimpurities. The molecule–substrate interaction should favorcomplete substrate coverage at low lm thickness, a strong lmtexture and the preferential growth of one crystal phase at thesemiconductor–dielectric interface. We are currently investi-gating the validity of the aforementioned criteria for otheracene-based semiconductors.

Acknowledgements

The authors are grateful to P. Moraille, J. Sanchez, K. Laaziri,and J. Bouchard for technical support. Part of this work wascarried out at the Central Facilities of Ecole Polytechnique/Universite de Montreal. CS acknowledges nancial support byFQRNT (grant Nouveau Chercheur). FC acknowledges the MarieCurie Program of the European Commission (grant MC-OIF-CT-2006-040864 TOPOS) and Ecole Polytechnique de Montreal fornancial support. JW is grateful to NSERC for nancial supportthrough Vanier Canada Graduate Scholarship and to C.Schunemann for fruitful discussions.

Notes and references

1 S. Reineke, F. Lindner, G. Schwartz, N. Seidler, K. Walzer,B. Lussem and K. Leo, Nature, 2009, 459, 234.

2 Y. Sun, G. C. Welch, W. L. Leong, C. J. Takacs, G. C. Bazanand A. J. Heeger, Nat. Mater., 2011, 11, 44.

3 H. Klauk, Chem. Soc. Rev., 2010, 39, 2643.4 M. A. McCarthy, B. Liu, E. P. Donoghue, I. Kravchenko,D. Y. Kim, F. So and A. G. Rinzler, Science, 2011, 332, 570.

5 J. Zaumseil and H. Sirringhaus, Chem. Rev., 2007, 107, 1296.6 R. M. Owens and G. G. Malliaras, MRS Bull., 2010, 35, 449.7 F. Cicoira, C. M. Aguirre and R. Martel, ACS Nano, 2011, 5,283.

8 N. Koch, ChemPhysChem, 2007, 8, 1438.9 S. Braun, W. R. Salaneck and M. Fahlman, Adv. Mater., 2009,21, 1450.

10 M. H. Yoon, C. Kim, A. Facchetti and T. J. Marks, J. Am.Chem. Soc., 2006, 128, 12851.

11 L.-L. Chua, J. Zaumseil, J.-F. Chang, E. C.-W. Ou, P. K.-H. Ho,H. Sirringhaus and R. H. Friend, Nature, 2005, 434, 194.

12 C. Effertz, S. Lahme, P. Schulz, I. Segger, M. Wuttig,A. Classen and C. Bolm, Adv. Funct. Mater., 2012, 22, 415.

13 S. Bertolazzi, J. Wunsche, F. Cicoira and C. Santato, Appl.Phys. Lett., 2011, 99, 013301.

J. Mater. Chem. C, 2013, 1, 967–976 | 975

Journal of Materials Chemistry C Paper

Dow

nloa

ded

by C

NR

Mila

no o

n 07

Jan

uary

201

3Pu

blis

hed

on 2

2 N

ovem

ber

2012

on

http

://pu

bs.r

sc.o

rg |

doi:1

0.10

39/C

2TC

0033

7FView Article Online

14 G. Witte and C. Woell, J. Mater. Res., 2004, 19, 1889.15 J. Fraxedas, Adv. Mater., 2002, 14, 1603.16 R. Ruiz, D. Choudhary, B. Nickel, T. Toccoli, K.-C. Chang,

A. C. Mayer, P. Clancy, J. M. Blakely, R. L. Headrick,S. Iannotta and G. G. Malliaras, Chem. Mater., 2004, 16, 4497.

17 J. Veres, S. D. Ogier, S. W. Leeming, D. C. Cupertino andS. Mohialdin Khaffaf, Adv. Funct. Mater., 2003, 13, 199.

18 I. N. Hulea, S. Fratini, H. Xie, C. L. Mulder, N. N. Iossad,G. Rastelli, S. Ciuchi and A. F. Morpurgo, Nat. Mater.,2006, 5, 982.

19 H. Sirringhaus, M. Bird and N. Zhao, Adv. Mater., 2010, 22,3893.

20 D. J. Gundlach, J. A. Nichols, L. Zhou and T. N. Jackson, Appl.Phys. Lett., 2002, 80, 2925.

21 A. Hepp, H. Heil, W. Weise, M. Ahles, R. Schmechel andH. von Seggern, Phys. Rev. Lett., 2003, 91, 157406.

22 F. Cicoira, C. Santato, F. Dinelli, M. Murgia, M. A. Loi,F. Biscarini, R. Zamboni, P. Heremans and M. Muccini,Adv. Funct. Mater., 2005, 15, 375.

23 T. Takahashi, T. Takenobu, J. Takeya and Y. Iwasa, Adv.Funct. Mater., 2007, 17, 1623.

24 Y.-S. Jang, H.-S. Seo, Y. Zhang and J.-H. Choi, Org. Electron.,2009, 10, 222.

25 Y. Yun, C. Pearson and M. C. Petty, J. Appl. Phys., 2009, 105,034508.

26 C. D. Dimitrakopoulos, B. K. Furman, T. Graham, S. Hegdeand S. Purushothaman, Synth. Met., 1998, 92, 47.

27 J. Zaumseil, C. L. Donley, J.-S. Kim, R. H. Friend andH. Sirringhaus, Adv. Mater., 2006, 18, 2708.

28 H. Yang, T. J. Joo, M.-M. Ling, K. Cho, C. Y. Ryu and Z. Bao, J.Am. Chem. Soc., 2005, 127, 11542.

29 L. Lutterotti, D. Chateigner, S. Ferrari and J. Ricote, ThinSolid Films, 2004, 450, 34.

30 J. B. Fortin and T. M. Lu, Chemical Vapor DepositionPolymerization: The Growth and Properties of Parylene ThinFilms, Kluwer Academic Publishers, Boston, 2004.

31 D. K. Owens and R. C. Wendt, J. Appl. Polym. Sci., 1969, 13,1741.

32 M. Michalski, J. Hardy and B. Saramago, J. Colloid InterfaceSci., 1998, 208, 319.

33 Q. Zhao, Y. Liu and E. W. Abel, Adv. Colloid Interface Sci.,2004, 280, 174.

34 D. Y. Kwok and A. W. Neumann, Adv. Colloid Interface Sci.,1999, 81, 167.

35 D.H.Kaelble, P. J.Dynes andE.H.Circlin, J. Adhes., 1974,6, 23.36 I. Horcas, R. Fernandez, J. M. Gomez-Rodriguez, J. Colchero,

J. Gomez-Herrero and A. M. Baro, Rev. Sci. Instrum., 2007, 78,13705.

37 L. Lutterotti, Nucl. Instrum. Methods Phys. Res., Sect. B, 2010,268, 334.

976 | J. Mater. Chem. C, 2013, 1, 967–976

38 D. Chateigner, Combined Analysis: Structure-Texture-Microstructure-Phase-Stresses-Reectivity Analysis by X-rayand Neutron Scattering, Wiley-ISTE, London, 2010.

39 S. Matthies and G. W. Vinel, Phys. Status Solidi B, 1982, 112,K111.

40 I. Lonardelli, H. R. Wenk, L. Lutterotti and M. Goodwin, J.Synchrotron Radiat., 2005, 12, 354.

41 J. Brandrup and E. H. Immergut, Polymer Handbook, Wiley,New York, 1989.

42 S. S. Sabri, P. L. Levesque, C. M. Aguirre, J. Guillemette,R. Martel and T. Szkopek, Appl. Phys. Lett., 2009, 95, 242104.

43 A. A. Virkar, S. Mannsfeld, Z. Bao and N. Stingelin, Adv.Mater., 2010, 22, 3857.

44 S. E. Fritz, T. W. Kelley and C. D. Frisbie, J. Phys. Chem. B,2005, 109, 10574.

45 A. Virkar, S. Mannsfeld, J. H. Oh, M. F. Toney, Y. H. Tan,G.-y. Liu, J. C. Scott, R. Miller and Z. Bao, Adv. Funct.Mater., 2009, 19, 1962.

46 A. Raudino, J. Chem. Phys., 1998, 109, 10430.47 X. Zhang, E. Barrena, D. Goswami, D. de Oteyza, C. Weis and

H. Dosch, Phys. Rev. Lett., 2009, 103, 136101.48 S. Steudel, D. Janssen, S. Verlaak, J. Genoe and P. Heremans,

Appl. Phys. Lett., 2004, 85, 5550.49 L. Lutterotti, M. Bortolotti, G. Ischia, I. Lonardelli and

H.-R. Wenk, Z. Kristallogr., Suppl., 2007, 26, 125.50 S. Milita, M. Servidori, F. Cicoira, C. Santato and A. Pifferi,

Nucl. Instrum. Methods Phys. Res., Sect. B, 2006, 246, 101.51 B. Gompf, D. Faltermeier, C. Redling, M. Dressel and

J. Paum, Eur. Phys. J. E, 2008, 424, 421.52 S. Milita, C. Santato and F. Cicoira, Appl. Surf. Sci., 2006, 252,

8022.53 T. N. Ng, J. H. Daniel, S. Sambandan, A.-C. Arias,

M. L. Chabinyc and R. A. Street, J. Appl. Phys., 2008, 103,044506.

54 G. Horowitz and M. E. Hajlaoui, Synth. Met., 2001, 122, 185.55 J. Chen, C. K. Tee, M. Shtein, J. Anthony and D. C. Martin, J.

Appl. Phys., 2008, 103, 114513.56 D. Knipp, R. A. Street, A. Volkel and J. Ho, J. Appl. Phys., 2003,

93, 347.57 M. M. Ling and Z. Bao, Chem. Mater., 2004, 16, 4824.58 D. H. Kim, H. S. Lee, H. Yang, L. Yang and K. Cho, Adv. Funct.

Mater., 2008, 18, 1363.59 H. Yang, S. H. Kim, L. Yang, S. Y. Yang and C. E. Park, Adv.

Mater., 2007, 19, 2868.60 A. Mayer, A. Kazimirov and G. Malliaras, Phys. Rev. Lett.,

2006, 97, 1.61 J. G. Laquindanum, H. E. Katz, A. J. Lovinger and

A. Dodabalapur, Chem. Mater., 1996, 8, 2542.62 H.-L. Cheng, Y.-S. Mai, W.-Y. Chou, L.-R. Chang and

X.-W. Liang, Adv. Funct. Mater., 2007, 17, 3639.

This journal is ª The Royal Society of Chemistry 2013