*Corresponding author.Fax:#19085822521;e-mail:ral @
<
Journal of Crystal Growth 187(1998)449—454
Physical vapor growth of organic miconductors
R.A.Laudi *,Ch.Kloc,P.G.Simpkins,T.Siegrist
Bell Laboratories,Lucent Technologies,Room IA-264,600Mountain A v enue,Murray Hill,NJ 07974,USA
Received 16December 1997
Abstract
Physical vapor growth in horizontal and vertical systems has been ud to grow crystals of -hexathiophene ( -6T), -octithiophene ( -8T), -quaterthiophene ( -4T),pentacene,anthracene and copper phthalocyanine.Using 10—30mg of starting material,mm —cm sized crystals,suitable for characterization measurements,have been grown.New polymorphs of -quaterthiophene (high temperature,HT and low temperature,LT)and pentacene were discovered.A horizontal geometry is shown to be advantageous becau the fragile crystals are extracted more easily. 1998Elvier Science B.V.All rights rerved.PACS:81.10.Bk;72.80.L
Keywords:Vapor growth;Organic miconductor; -6T; -4T; -8T;Pentacene;Anthracene;Copper phthalocyanine
1.Introduction
Becau of the very promising mobilities and other electronic properties,oligomers such as -hexathiophene ( -6T)[1,2], -octithiophene ( -8T)[3], -quaterthiophene ( -4T)[4]and related com-pounds such as pentacene [4,5],are under world-wide investigation for possible thin-film transistors (TFT)and other applications.Single-crystal thio-phene oligomers were grown in 1995by Horowitz et al.[6]and by Laudi et al.[7].However,usually the properties of vacuum deposited thin
films have been investigated becau bulk crystals until recently have been elusive since reproducible growth methods have not been published.Single crystals are needed for unambiguously asssing physical properties.
We recently completed a study of the growth of -6T in both the high-temperature (HT)and low-temperature (LT)modifications [8].This work re-ports that physical vapor deposition growth,under conditions where transport was dominated by buoyancy driven convection and where flowing in-ert gas was ud to limit diffusion from the volati-lazing source,could reproducibly yield crystals with a si
ze of veral millimeters while consuming only 10—30mg of starting material.Under appro-priate conditions cm sized platelets could be
0022-0248/98/$19.00 1998Elvier Science B.V.All rights rerved.PII S 0022-0248(98)00034-7
Fig.1.Crystal growth apparatus.The temperature is adjusted for each substance but the steepest temperature gradient is generally at the position shown.
prepared.An approach for modeling transport and growth was developed which we feel should be applicable to the growth of other organic TFT crystals.In the following,we show that using this approa
ch we have been able to choo conditions for the growth of -6T HT, -6T LT, -4T HT, -4T LT, -8T,pentacene,copper phthalocyanine and anthracene.Further,we ud the previously re-ported approach to choo conditions for a more convenient and easily controllable horizontal growth arrangement.
2.Experimental procedure
The source materials, -4T and -8T were syn-thesized by H.Katz in Bell Laboratories.Pen-tacene and copper phthalocyanine were purchad from Aldrich (Milwaukee.;WI), -6T was purchad from Organix (Woburn,MA).Fig.1.shows our current apparatus.It should be noted that in the prent work we ud a horizontal tube for trans-port and growth rather than the vertical tube ud in our earlier -6T crystal growth [8].The reason for this was experimental convenience.As can be en both the source and the deposition regions are in parate tubes within the outer reactor tube.The advantage of this geometry is that starting material may be inrted easily and both residual starting material and crystals may be removed easily after growth.Heating was applied by a resistance wire
wound transparent furnace designed to provide various gradients.All of the experiments were con-ducted with a flowing inert (He,Ar)or forming (85%N ,15%H
)
gas.Temperature control ud
standard controllers with an accuracy of $0.5°C.The source tube is 8cm from the inlet tube,a de-position of growing crystals occurs along the crys-tal growth tube.
3.Results and discussion
3.1.Model for crystal growth in a horizontal tube In our typical horizontal reactor experiments the flow is directed over the source at a flow rate in the range of 50ml/min.Thus,the mean velocity of the jet exiting the 3mm diameter inlet tube (Fig.1)is approximately 12cm/s,and the maximum value of the Poiuille pipe flow [9]within the inlet tube is 24cm/s,becau velocity clo to the wall is negli-gible.The structure of the circular jet before it enters the source and growth tubes has been de-scribed by Schlichting [9].The maximum velocity in the jet,which occurs along its axis is given by u
"3K /8 x ,(1)where is the kinematic viscosity of the gas,x is the axial coordinate (distance from the inlet tube)and K "4Q /3 a
(2)
450R.A.Laudi et al./Journal of Crystal Growth 187(1998)449–454
Table 1
Experimental crystal growth conditions for mm or larger crystals Material
Arrangement
Cross-ctional area for gas flow (cm )Source
temperature (°C)Deposition temperature Gas and flow rate (ml/min)
Comments
-6T LT Vertical 2.3280—300
Ar,He,
N #H ,40Mm sized plate
like crystals -6T HT and LT Vertical 4.44320HT 290°C LT 250°C Ar,150
Cm sized plate like crystals -6T HT Horizontal 2.5320290°C Ar,He,
N #H ,40'5mm plates
-4T LT Vertical 0.66140 Ar,40Ca.1mm pale
yellow thin plates -4T HT Vertical 0.66180 Ar,40Up to 3mm yellow plates -4T HT Horizontal 2.5185140°C N #H ,602—3mm plates -8T
Vertical
0.66
350
Ar,20
1—2mm brown plates
Anthracene Horizontal 2.5160150°C He,50
Above 1cm plates Pentacene
Horizontal
2.5
285
Between source and 220°C N #H
,30
10;2mm lath like crystals Copper phthalo-cyanine
Vertical 1.26470
450°C
He,75
Up to 5mm long needles
In vertical systems the area of the annulus between the inside tube delivering the flowing,externally introduced gas and the outside reactor tube (e Fig.1,Ref.[8]);in horizontal systems the area of the reactor cross ction.
The highest temperature at which crystal nuclei were obrved,when the source was at the indicated temperature. Crystals are deposited in a steep temperature gradient between source and room temperature.
is the kinematic momentum.The values for phys-ical properties of gas are taken from Ref.[8]Table 1.In Eq.(2)Q is the volume flow rate and a is the radius of the inlet tube.Eq.(1)shows that the maximum jet velocity decays inverly with dis-tance from the exit plane of the inlet tube.Note that experimental conditions define K and that the car-rier gas properties influence u
.Since the source
tube is approximately 8cm from the inlet tube,the maximum jet velocity at that location (for Q "50ml/min)is about 1.45cm/s for argon and 0.17cm/s for helium.The jet velocity for forming gas approximately equals the velocity for Ar,since the N
kinematic viscosity +Ar.
The velocity distribution across the circular free jet is known in term of a dimensionless parameter [9],
"0.244(K ) r /( x ),
(3)
where r is the radial coordinate measured from the jet axis.Eq.(3)has been ud to estimate,for a par-ticular gas,how the jet broadens as it moves down-stream.Considering Fig.1,the diameters and wall thickness of the three concentric tubes are such that esntially all flow enters the source tube and exits into the crystal growth tube.After traveling 8cm down the reactor tube the jet is substantially broadened to approximately fill the source tube as
R.A.Laudi et al./Journal of Crystal Growth 187(1998)449–454451
Fig.2.Schematic of two circulation patterns for gas at &1atm:(a)buoyance driven convection in a clod ampoule,(b)forced convection in an open system.
it enters.This can be en from Eq.(3)where as x increas is smaller and the jet is broader.Buoyancy driven convection occurs in the growth and source tubes as a result of the applied horizontal temperature gradient.An estimate of the magnitude of this motion can be obtained from a two-dimensional Handley cell model [10]which we apply along the axis of the tubes using a typical measured temperature gradient d t /d x of 10°cm \ (e Fig.1).The velocity distribution in a Handley cell (in our ca modeled as a two-dimensional tube with a horizontal temperature gradient)is a cubic function of the depth.So that if the hot zone is on the left,velocity will be from left to right on the top and reverd at the bottom (e Fig.2a).Using this cubic velocity distribution described by Simpkins and Chen [10],the maximum velocity u*(made dimensionless by dividing by the thermal diffusivity in Ref.[10]and multiplying by the tube diameter)is given by u * 3x 10\ R /¸C
(4)
and occurs at approximately half-way between the tube axis and the top or bottom surfaces.In Eq.(4)R is the Rayleigh number defined as R "g ¹d
,
(5)
where g is the gravitational constant, is the ther-mal expansion coefficient and is the thermal dif-fusivity.Values for the physical constants were tabulated in Ref.[8].Also,in Eq.(5)d is the tube diameter and ¹is the temperature difference across the region of steepest gradient.Returning to Eq.(4),¸is the aspect ratio,length for gradi-ent/tube diameter (&6)and C is a constant (ap-proximately unity [10]).For the carrier gas of interest,argon,helium and nitrogen,R "51800,1120and 64660,respectively.The corresponding maximum velocities are therefore about 14,1.6and 12cm/s,which are substantially greater than the forced convection jet flow.
In our previous study of growth in a vertical system we identified volatilization and diffusion at the source,transport between source and growing crystals and diffusion to the growing crystal as critical steps in growth [8].We have previously modeled the transport of -6T in a vertical tube in the prence of flowing inert gas at one atmosphere and found that even at substantial flow rate of gas (typically 130cm /min)buoyant convection dom-inated over transport induced by the flowing gas (forced convection).Using procedures described in Ref.[8]where the buoyant velocity is &17 ¹(where ¹is the typical temperature difference between the gas delivery tube wall and reactor wall
452R.A.Laudi et al./Journal of Crystal Growth 187(1998)449–454
of a vertical apparatus),we estimate that the vel-ocities due to buoyant convection in typical vertical apparatus always were more that20cm/s,while that due to forced convection was typically0.47cm/s. Thus,buoyant convection dominates in our verti-cal tube experiments.
Understanding the probable circulation pattern is aided by consideration of Fig.2.Fig.2a shows the idealized circulation pattern for a clod am-poule where the left end is hotter than the right end and transport is by buoyancy driven convection. This circulation pattern can be ud to understand the buoyancy driven component of circulation in an apparatus where the tube is open at the ends(e Fig.1).In the clod ampoule of Fig.2a the buoy-ancy cell revers direction at the end of the am-poule.In the open apparatus,which we ud,the cell will rever where the temperature gradient decays significantly,probably only a little to the right of the crystal deposition region in Fig.1. Fig.2b shows a circulation pattern for forced con-vection where the inlet is a small diameter tube producing a gas jet along the center of the reactor tube.As shown above,becau the velocity profile broadens as it proceeds down stream,the forced convection velocity is not substantial in an appa-ratus with our dimensions.For our apparatus the approximate overall velocity in a particular region of the apparatus is the sum of the buoyancy and forced convection velocities.Thus,even when gasflows thr
ough the system and the gas exit is on the right(Fig.1)counterflow due to the buoy-ancy effect appears to be overriding.It must be emphasized that the prent model is a simplifi-ed one.Local temperature gradients,the three-di-mensional aspect of our real system and the interaction of buoyancy and forced convection fluxes would complicate a detailed analysis(e Ref.
[11]).The prence of the source tube and the crystal deposition tube inside the reactor tube(e Fig.1),which are esntial to remove the unud source material and the fragile platelet crystals from reactor tube,does not substantially alter the convection pattern.Considering the estimated vel-ocities mentioned above,under all our conditions buoyancy convection tends to dominate and net flow patterns similar to Fig.2a are believed to predominate.
As discusd in Ref.[8],however,the small for-ced convection and the associatedflow of inert gass is advantageous in carrying away impurities and decomposition products.It is interesting to point out that even at the lowest calculated buoy-ancy driven velocities(1.6cm/s for helium)more than enough transport occurs to account for the mass of crystals deposited,since large area thin plates are not very heavy.
3.2.Crystal growth of particular materials
Table1lists preliminary results in a arch for the best conditions for growth of the organic mi-conductors studied.The conditions reproducibly yield crystals large enough for physical measure-ments.In early experiments we ud a vertical ap-paratus;more recently,for experimental con-venience we ud a horizontal apparatus.As described above,analysis shows that both the hori-zontal and vertical arrangement lead to transport dominated by buoyancy convection.
Bad on our experience we believe that either a vertical or horizontal growth geometry could be ud for the growth of materials of Table1,al-though the horizontal geometry is experimentally more convenient.
-6T and -4T werefirst grown in a vertical apparatus as discusd in Refs.[8,12].We ud similar source temperatures andflow rates in the horizontal reactor with excellent results in so far as crystal size and quality were concerned.A variety of gas were ud in the horizontal reactor and the grown crystals were always easy to recover.During this work we discovered two -4T polymorphs ana-logous to tho en in -6T[12].The HT(high-temperature form)crystallized when the source was at a high temperature probably becau deposition temperature was high.
-8T has been grown in a small vertical system but so far has not been studied in a horizontal system.
Anthracene crystals have been grown from a melt and from vapour in the past[13].We studied this material purpoly becau a large quantity of starting material was readily available,so that it could be ud to“calibrate”the apparatus and con-ditions.It was also ud to make comparisons with
R.A.Laudi et al./Journal of Crystal Growth187(1998)449–454453
