Comparison between nucleation of pentacene monolayer islands on polymericand inorganic substrates

Sirapat Pratontep,1,* Frank Nüesch,2 Libéro Zuppiroli,2 and Martin Brinkmann1,†

1Institut Charles Sadron, 6 rue Boussingault, 67083 Strasbourg, France2Laboratoire D’Optoélectronique des Matériaux Moléculaires, Institut des Matériaux, Ecole Polytechnique Féderale de Lausanne,

CH-1015 Lausanne, Switzerland�Received 23 February 2005; revised manuscript received 23 May 2005; published 11 August 2005�

We present a comparative study of the nucleation and growth of pentacene monolayer islands in the sub-monolayer regime onto inorganic substrates of SiO2 and sapphire �Al2O3� and organic substrates ofpoly�methyl-metacrylate� �PMMA�. We have determined the scaling laws that govern the saturated nucleationdensity per unit area N as a function of two essential deposition parameters: the deposition rate � and thesubstrate temperature Ts. For all substrates, we found N���, with 0.8���1.2, and the Ts dependence of thenucleation density follows a typical activated Arrhenius law. Whereas a critical nucleus size of i=2 is obtainedfor all investigated substrates, the activation energy for nucleation depends significantly on the dielectric. Thedifference in activation energy of nucleation on PMMA and SiO2 is due to different molecule-substrateinteractions.

DOI: 10.1103/PhysRevB.72.085211 PACS number�s�: 61.43.Hv, 81.15.Ef, 68.37.Ps

I. INTRODUCTION

Molecular semiconductors like oligothiophenes andacenes �pentacene and tetracene� are among key materials inthe fabrication of organic field effect transistors �OFETs�.1,2

Understanding the relationship between the morphology andstructure of the organic layer, and the corresponding chargetransport, is of paramount importance in the perspective ofimproving device performances.3 Previous studies have dem-onstrated that the charge carrier mobility in OFETs based onsexithiophene reaches its highest value as the thickness ap-proaches approximately two monolayers.4 It therefore ap-pears essential to master the nucleation and growth of thefirst few layers of organic semiconductors on various dielec-tric substrates, i.e., to control grain size, polymorphism,smoothness, and layer continuity. In particular, scaling lawsshould be established that govern the nucleation ofmonolayer-thick pentacene islands on different substrates,especially on flexible plastic substrates, as a function of thevarious deposition parameters �substrate temperature Ts, cov-erage �, and deposition rate R�. Despite several publishedstudies on the nucleation and growth of conjugated systemsonto inorganic substrates like SiO2 and H-passivatedsilicon,5–7 no study has been reported to date on the nucle-ation mechanism of organic semiconductors onto smoothamorphous polymer dielectrics, for instance PMMA�poly�methyl-methacrylate��.

We present here a comparative study of the nucleation ofpentacene onto inorganic �SiO2 and Al2O3� and organic�PMMA� substrates. Pentacene �C22H14� is chosen for itshigh charge mobility of the order of 1 cm2/V s, its wide-spread use in OFET fabrication, and its ability to form two-dimensional islands. To study the nucleation stage, we limitour investigations to the early stage of growth, i.e., at lowcoverage ��40%, when neither coalescence nor Ostwaldripening of the islands are observed.7 Central to this work isthe measurement of the saturated nucleation density, repre-

sented by the number of pentacene islands per unit area, as afunction of both substrate temperature and deposition rate.

II. EXPERIMENT

We performed high-vacuum deposition of pentacene at abase pressure of 3�10−7 mbar. Pentacene �Fluka�, purifiedtwice by vacuum sublimation, was evaporated from a fusedquartz crucible heated by a tungsten wire. The effect of depo-sition rate � on the maximum nucleation density N was in-vestigated at a fixed substrate temperature, Ts=338 K�65 °C�, while the variation of N with Ts was measured atfixed deposition rate, �, in the range 0.4−0.5 nm/min. Depo-sition rate and nominal film thickness were measured by aquartz crystal balance kept at 287 K �15 °C�, where the ef-fect of reevaporation is minor.7 This ensures the sameamount �fixed at a nominal thickness of 0.5 nm� of pentacenearriving on all substrates. The inorganic substrates wereAl2O3 �polished sapphire with no preferential orientationfrom Stecher Ceramicparts GmbH, cleaned using the meth-odology described in the literature8� and SiO2 �a boron-doped silicon wafer with a 200 nm thick thermal oxidelayer�. Both types of substrates have a rms roughness below0.2 nm �2�2 �m2�. Once transferred in vacuum, the inor-ganic substrates were further cleaned in situ by using a rfoxygen plasma treatment �0.1 W/cm2 power and −50 Vbias� for 5 min with 0.1 mbar of pure oxygen. For the poly-mer substrate, an atactic PMMA �Aldrich, Mw=120 kg/mol,5% weight solution in toluene� was spin-coated on Si�100�substrates at 1000 rpm for 1 min to obtain a smooth polymerfilm with thickness of 200–300 nm. Pentacene thin filmsdeposited on these substrates were investigated by ex situatomic force microscopy �AFM� on a Nanoscope III intapping mode using Si tips �25–50 N/m and 280–365 kHz�.Data analyses were performed by using both WSxM �Nano-tec Electronica S.L.� and AnalySIS �Soft Imaging System�software.

PHYSICAL REVIEW B 72, 085211 �2005�

1098-0121/2005/72�8�/085211�5�/$23.00 ©2005 The American Physical Society085211-1

III. RESULTS AND DISCUSSION

The effects of deposition rate on nucleation density forSiO2, Al2O3, and PMMA substrates are comparatively illus-trated in Fig. 1. Irrespectively of the chemical nature of thesubstrate, we observed the formation of monolayer-thickpentacene islands, corresponding to pentacene moleculesstanding upright but slightly inclined on the substrates. Onthe inorganic substrates, we observe highly ramified struc-tures at low deposition rate �below approximately2 nm/min�, resembling typical structures described by theirreversible “hit and stick” picture of the diffusion-limitedaggregation model.9 We note that on SiO2 and Al2O3 theisland morphology becomes less dendritic at higher deposi-tion rates, in accordance with our previous study,7 whereas,on PMMA, pentacene islands are observed to be more com-pact at all deposition rates.

Regarding the effect of deposition rate on saturated nucle-ation density, a common trend is observed for all three sub-

strates: the number of pentacene islands per unit area in-creases significantly as the deposition rate is increased. InFig. 2, we plot the dependence of the nucleation density N�averaged from a few hundred pentacene islands� on thedeposition rate for the three substrates at a fixed substratetemperature of 65 °C. First, we note that the nucleation den-sity on PMMA clearly exceeds that observed on SiO2 andAl2O3 substrates by almost one order of magnitude. Second,we observe that the rate dependence of the nucleation densityis described in all cases by a scaling law, N���. The expo-nents � depend on the type of substrate, and are listed inTable I. SiO2 and PMMA substrates yield almost the sameexponent, �=1.17±0.10, i.e., ��1, whereas for Al2O3 weobserve �=0.8±0.1. We notice the absence of deviation ofthe experimental results from the scaling law at low rate�below 0.5 nm/min�, indicating that heterogeneous nucle-ation by surface impurities on these substrates is still negli-gible with respect to homogeneous nucleation.10 This is inaccordance with the observation in the case of Al2O3, wherethe presence of atomic steps on the sapphire substrate doesnot induce significant heterogeneous nucleation of islands.On PMMA substrates, we note also the presence of 3D is-lands which have most likely been nucleated by surface im-purities and/or surface defects like holes in the PMMA film�see Fig. 1�.

We further probed the pentacene thin-film growth by in-vestigating the nucleation of pentacene as a function of sub-strate temperature, Ts. Figure 3 presents two sequences ofAFM topographic images showing the temperature depen-dence of the morphology obtained for a fixed deposition rate��= 0.5 nm/min�. In a small range of Ts from 25 to 70 °C,the island density on both substrates decreases rapidly as Ts

TABLE I. Experimental data relative to the nucleation of pentacene monolayer islands on inorganic andpolymer substrates �see the text�.

Substrate SiO2 Al2O3 sapphire PMMA

Condensation regime Initially incompletea Initially incomplete Initially incomplete

Exponent � 1.16±0.1 0.8±0.1 1.18±0.1

ENucl�eV� 0.76±0.05 — 0.36±0.05

EA�eV� 0.40–0.50 — 0.1

aUsing the terminology of Ref.11.

FIG. 1. Effect of deposition rate on the surface morphologymeasured by AFM for three different types of substrate: SiO2,Al2O3, and PMMA. For all the samples, the substrate temperaturewas fixed at 65 °C and the nominal thickness of the pentacene filmswas always 0.5 nm. All the images are 10�10 �m2 in size.

FIG. 2. Dependence of pentacene island density on depositionrate SiO2, Al2O3, and PMMA substrates at a fixed substrate tem-perature, Ts=65 °C.

PRATONTEP et al. PHYSICAL REVIEW B 72, 085211 �2005�

085211-2

is increased, especially in the case of SiO2. In agreementwith the deposition rate study, the overall nucleation densityis higher on PMMA than on SiO2. We also observe that theapparent coverage � decreases as Ts increases on both sub-strates, as depicted in Fig. 4�a�. This is a clear indication thatreevaporation occurs during deposition.7 Only at sufficientlyhigh deposition rates and low Ts is there a correspondencebetween the apparent thickness and the nominal thickness ofthe films measured by the quartz balance. It is worth noticing

that, for Ts=299 K, we still observe a small effect of re-evaporation, since the observed coverage �26% on SiO2� liesslightly below the expected value of 33% �0.5 nm� in thecase of complete condensation. Qualitatively, we can accountfor these observations by the fact that, at higher Ts, penta-cene molecules desorb more readily from the PMMA or SiO2substrates during deposition before being captured by stableislands. The fact that reevaporation is more important onSiO2 than PMMA is ascribed to both the different nucleationdensities and the different molecular interaction �adsorptionenergy and diffusion energy� on both types of substrates�vide infra�. As noticed in our previous study,7 these obser-vations underline the interplay existing between two charac-teristic distances in the system: the average NN distance,NN, and the average diffusion distance before desorption,des:

des = �D�A, �1�

where D is the diffusion constant of pentacene molecules onthe substrate and �A is the mean residence time of pentaceneon the substrate before desorption.

The significant level of reevaporation evidenced in thepresent case, especially for Ts�323 K, is an important cluein identifying the regime of condensation at play. In previousreports,6,7 we have demonstrated that the nucleation andgrowth kinetics of organic molecules vapor deposited on in-organic substrates like SiO2 or H-passivated Si�100� can besatisfactorily explained by the rate-equation formalism intro-duced by Venables et al.11 It predicts that N should vary as afunction of deposition rate R and substrate temperature Tsaccording to the following equation:11

FIG. 3. Effect of substrate temperature on the surface morphol-ogy measured by AFM for two types of substrate: SiO2 and PMMA.For all samples the deposition rates are in the range0.45–0.50 nm/min and the nominal thickness of the pentacenefilms was fixed to 0.5 nm. All the images are 10�10 �m2 in size.

FIG. 4. �a� Variation of apparent coverage � with substrate tem-perature Ts for a fixed nominal thickness of 0.5 nm. �b� Depend-ence of pentacene island density on substrate temperature SiO2 andPMMA substrates at fixed deposition rate �0.4 nm/min−0.5 nm/min�.

COMPARISON BETWEEN NUCLEATION OF PENTACENE… PHYSICAL REVIEW B 72, 085211 �2005�

085211-3

N � �� exp� ENucl� , �2�

where = �kBTs�−1. ENucl is the activation energy for homo-geneous nucleation.

ENucl involves various activation energies related to differ-ent physical processes occurring on the substrate surface,e.g., desorption of pentacene molecules from the substrate�EA�, surface diffusion �Edif f�, and the formation of a clusterof critical size i and binding energy Ei. More precisely, Ei isdefined as the difference in free energy between i adsorbedmolecules on the substrate �noninteracting one with another�and i molecules forming an island.11

The exact dependence of ENucl on the various activationenergies EA, Edif f, and Ei depends on the condensation re-gime as well as the dimensionality of the cluster �2D vs 3D�.From the observation of a moderate level of reevaporation,as illustrated in Fig. 4�a�, we propose that the nucleation andgrowth of pentacene occurs in the regime called “initiallyincomplete condensation.”11 This allows us to extract thecritical nucleus size i of pentacene islands from the measuredvalues of �, since �= i /2 for 2D islands.11 Within the frame-work of Venables’s model, the experimental � values in therange 0.8–1.2 lead us to conclude that the critical cluster sizeis i=2 for all investigated substrates. This result is differentfrom that reported by Ruiz et al., who found i=3.5 Thisdifference is most likely related to the different growth mod-els used in both cases. The derived i value is by essencemodel dependent. The model used in Ref. 5 does not takeinto account reevaporation, whose effect on the growth ofpentacene is clearly demonstrated in this study as well as inRef. 7. At this point, it is also worth mentioning the predic-tions of a more recent theoretical approach by Jensen etal.10,12 This approach is distinct from that of Venables in thatreevaporation of molecules �atoms� impinging directly ontop of existing islands is also taken into account. One of thegrowth regimes defined by Jensen et al. predicts a depen-dence N�� with a critical nucleus size i=1 under the con-dition that the diffusion length * of a molecule on top of anisland before reevaporation is larger than the maximum av-erage radius of the islands Rmax, i.e., *�Rmax. However, itmay be argued that, given the low coverages investigated inthis study, the direct capture of pentacene molecules on topof pentacene islands is likely to be negligible with respect tothe diffusive growth by capture of pentacene molecules dif-fusing on the substrate.

According to Eq. �1�, the Ts dependence of the totalnucleation density N is expected to follow the exponentialArrhenius law. In Fig. 4�b�, we verify its validity for penta-cene island nucleation on smooth polymer and inorganic sub-strates by plotting a natural logarithm of the density againstan inverse of Ts. Clearly, we observe in both cases a good fitto the activated character of the nucleation density with Ts.Using Eq. �1�, we can extract the corresponding activationenergies for homogeneous nucleation of pentacene islands,ENucl: 0.78±0.05 eV on SiO2 and 0.34±0.05 eV on PMMA.In the range of 299 K�Ts�353 K, we do not observe achange in the activation energy, ENucl. This observation sug-gests that no change of condensation regime occurs in this Tsrange.11

The derived values for the activation energy for nucle-ation clearly indicate that nucleation is favored on polymericsubstrates with respect to inorganic dielectric substrates likeSiO2. In the regime of “initially incomplete condensation,”the activation energy for the nucleation of 2D islands isgiven by ENucl=1/2�Ei+ iEA�.11 Assuming that the value ofEi depends mainly on the structure of the critical island, i.e.,it does not depend on the type of dielectric substrate, we canestimate the activation energy for desorption of pentacene,EA. The critical cluster binding energy Ei can be estimatedfrom a simple calculation of the van der Waals interactionsbetween two pentacene molecules in a dimer.13 We shall as-sume that the structure of the critical cluster is identical tothat of the dimer present in crystalline pentacene.14,15 Thecalculated van der Waals interactions between two pentacenemolecules in the dimer amount to approximately15.3 kcal/mol �0.67 eV� for the structure by Holmes et al.14

�d001=1.41 nm� and 13.3 kcal/mol �0.58 eV� for the struc-ture by Campbell et al.15 �d001=1.45 nm�. Accordingly, if wetake Ei in the range 0.6–0.7 eV, we can estimate the activa-tion for desorption EA on SiO2 in the range 0.4–0.5 eV andEA0.1 eV on PMMA.

At first glance, from the higher value of EA for SiO2 wewould expect a smaller reevaporation on this substrate withrespect to PMMA, which contradicts our observations �videsupra�. However, the extent of reevaporation on a given sub-strate also has to take into account the difference in molecu-lar diffusion of pentacene on the surface. Indeed, from Eq.�1�, the typical diffusion distance of pentacene before de-sorption des depends on the diffusion constant of a penta-cene molecule on the surface, namely D=Do exp�− Edif f��Edif f is the activation energy for diffusion�. Using, for thetypical residence time of a molecule on the substrate beforedesorption,�A, the expression given by Venables et al.,11 �A=�−1 exp� EA� � =1/kBTs and � is a typical vibration fre-quency of the substrate�, we see that the Ts dependence ofthe diffusion distance before desorption des varies as des�exp� �EA−Edif f� /2�. Hence, des depends on the difference�EA−Edif f� and not solely on EA. It is therefore not contra-dictory to observe both a larger value of EA and a largereffect of reevaporation on SiO2 with respect to PMMA. Infact, this result suggests a larger diffusion constant of penta-cene molecules on a polymer surface like PMMA as com-pared to SiO2 or Al2O3. This result is further supported bythe fact that the nucleation density of pentacene islands islarger on PMMA than on SiO2. Indeed, from Ref. 11, the rateof nucleation of pentacene islands is proportional to the dif-fusion constant D of molecules on the substrate. Hence, thelarger nucleation density observed on PMMA also supportsour assumption of a larger diffusion constant of pentacene onPMMA with respect to SiO2 and Al2O3.

In conclusion, this study has uncovered new results rela-tive to the thermodynamics of nucleation of pentacene oninorganic �SiO2 and Al2O3� and polymeric �PMMA� sub-strates. In all cases, nucleation of pentacene 2D islands isfound to be homogeneous. The critical nucleus size i=2 wasextracted from the scaling law N��� and is found to beindependent of the type of dielectric considered in this study.The Ts dependence of the nucleation density yields the acti-

PRATONTEP et al. PHYSICAL REVIEW B 72, 085211 �2005�

085211-4

vation energy for nucleation, which involves two main con-tributions: the binding energy in the critical island and theactivation energy for desorption of pentacene from the sub-strate. The observed differences between PMMA and SiO2substrates arise from the difference in the adsorption energyand surface diffusion of pentacene. Further insight into thenucleation mechanism of pentacene on PMMA and SiO2 willbe obtained from additional measurements involving thermaldesorption spectroscopy in order to evaluate accurately theactivation energy for desorption, EA. These results on themorphology in the first monolayer on dielectric substrates

should have important consequences on the transport prop-erties across pentacene islands, i.e., the nucleation densityshould influence the amount of grain boundaries, which mayact as charge carrier traps and degrade OFET characteristics.

ACKNOWLEDGMENTS

We acknowledge support by EEC Contract HPRN-CT-2002-0327 and the Swiss National Science Foundation underContract 20-67929.02.

*Present address: National Nanotechnology Center, 111 ThailandScience Park, Klong 1, Klong Luang, Pathumtani 12120, Thai-land.

†Corresponding author. Electronic address: brinkman@ics.u-strasbg.fr

1 D. E. Hooks, T. Fritz, and M. D. Ward, Adv. Mater. �Weinheim,Ger.� 13, 227 �2001�; C. D. Dimitrakopoulos and P. L. Malen-fant, ibid. 14, 99 �2002�.

2 J. F. Meyer zu Heringdorf, M. C. Reuter, and R. M. Tromp,Nature �London� 412, 517 �2001�.

3 A. Dodabalapur, L. Torsi, and H. E. Katz, Science 268, 270�1995�; E. L. Granström and C. D. Frisbie, J. Phys. Chem. B103, 8842 �1999�.

4 F. Dinelli, M. Murgia, P.Levy, M. Cavalini, F. Biscarini, andD.Leeuw, Phys. Rev. Lett. 92, 116802 �2004�.

5 F. Biscarini, R. Zamboni, P. Samori, P. Ostoja, and C. Taliani,Phys. Rev. B 52, 14868 �1995�; R. Ruiz, B. Nickel, N. Koch, L.C. Feldman, R. F. Haglund, A. Kahn, and G. Scoles, ibid. 67,125406 �2003�; R. Ruiz, B. Nickel, N. Koch, L. C. Feldman, R.F. Haglund, A. Kahn, F. Family, and G. Scoles, Phys. Rev. Lett.91, 136102 �2003�.

6 M. Brinkmann, F. Biscarini, C. Taliani, I. Aiello, and M. Ghedini,Phys. Rev. B 61, R16339 �2000�; M. Brinkmann, S. Graff, and

F. Biscarini, ibid. 66, 165430 �2002�.7 S. Pratontep, M. Brinkmann, F. Nüesch, and L. Zuppiroli, Phys.

Rev. B 69, 165201 2004.8 M. Carrara, Thesis, Ecole Polytechnique Fédérale de Lausanne

�2002�.9 T. A. Witten and L. M. Sander, Phys. Rev. Lett. 47, 1400 �1981�.

10 P. Jensen, Rev. Mod. Phys. 71, 1695 �1999�, and referencestherein.

11 J. A. Venables, G. D. T. Spiller, and M. Hanbücken, Rep. Prog.Phys. 47, 517 �2001�.

12 P. Jensen, H. Larralde, and A. Pimpinelli, Phys. Rev. B 55, 2556�1997�.

13 Calculations of binding energies between pentacene molecules ina cluster were obtained by using the CERIUS package software�Accelrys, Waltham, MA� with the use of the Universal ForceField �UFF�; A. K. Rappe, C. J. Casewit, K. S. Colwell, W. A.Goddard, and W. M. Skiff J. Am. Chem. Soc. 114, 10024�1992�.

14 D. Holmes, S. Kumaraswamy, A. J. Matzger, and P. C. Vollhardt,Chem.-Eur. J. 5, 3399 �1999�.

15 D. J. Campbell, J. M. Robertson, and J. Trotter, Acta Crystallogr.14, 705 �1961�; 15, 289 �1962�.

COMPARISON BETWEEN NUCLEATION OF PENTACENE… PHYSICAL REVIEW B 72, 085211 �2005�

085211-5