Materials Research Bulletin 47 (2012) 952–956

Catalyst and its diameter dependent growth kinetics of CVD grown GaNnanowires

Chandan Samanta a, D. Sathish Chander a,b, J. Ramkumar b, S. Dhamodaran a,*a Department of Physics, Indian Institute of Technology Kanpur, Indiab Department of Mechanical Engineering, Indian Institute of Technology Kanpur, India

A R T I C L E I N F O

Article history:

Received 20 May 2011

Received in revised form 26 December 2011

Accepted 11 January 2012

Available online 20 January 2012

Keywords:

A. Nitrides

B. Vapor deposition

C. Electron microscopy

D. Lattice dynamics

A B S T R A C T

GaN nanowires were grown using chemical vapor deposition with controlled aspect ratio. The catalyst

and catalyst-diameter dependent growth kinetics is investigated in detail. We first discuss gold catalyst

diameter dependent growth kinetics and subsequently compare with nickel and palladium catalyst. For

different diameters of gold catalyst there was hardly any variation in the length of the nanowires but for

other catalysts with different diameter a strong length variation of the nanowires was observed. We

calculated the critical diameter dependence on adatoms pressure inside the reactor and inside the

catalytic particle. This gives an increasing trend in critical diameter as per the order gold, nickel and

palladium for the current set of experimental conditions. Based on the critical diameter, with gold and

nickel catalyst the nanowire growth was understood to be governed by limited surface diffusion of

adatoms and by Gibbs–Thomson effect for the palladium catalyst.

� 2012 Elsevier Ltd. All rights reserved.

Contents lists available at SciVerse ScienceDirect

Materials Research Bulletin

jo u rn al h om ep age: ww w.els evier .c o m/lo c ate /mat res b u

1. Introduction

Miniaturization of devices based on nanotechnology has led toincreased interest in different nano-scale materials. Inorganicnano-structures exhibit peculiar and unique properties differentfrom their bulk counterparts due to quantum effects and also dueto large surface to volume ratio. In electronic, photonic and energyrelated applications GaN has become an important material for itssuperior properties in comparison to other semiconductors [1,2].Most vital properties are its large bandgap, high melting point,carrier mobility, and electrical breakdown field. Additionally GaNexhibit spontaneous polarization and piezoelectric fields depend-ing on growth orientations [3]. Such properties at the nano-scalewould be of immense help in fabricating nano-devices usingquantum dots and wires [4,5]. Though superior quantum effectsare observed for quantum dots, nanowires have drawn moreattention in the last decade. It is because of the difficulty in makingcontacts with quantum dots whereas, it is easy to make contactswith nanowires. However, diameter and aspect ratio control ofnanowires has been challenging and important for the fabricationof nanowire based electrical and optical devices [6,7]. For catalysisgrowth of nanowires, the diameter of catalytic particle determinesthe diameter of nanowires. On the other hand length growth rate ofnanowires depends on the diameter [8,9]. There have been

* Corresponding author. Tel.: +91 512 2596238; fax: +91 512 2590914.

E-mail address: kdams2003@gmail.com (S. Dhamodaran).

0025-5408/$ – see front matter � 2012 Elsevier Ltd. All rights reserved.

doi:10.1016/j.materresbull.2012.01.012

theoretical models explaining the growth kinetics of nanowiresgrown by molecular beam epitaxy and metal organic chemicalvapor deposition. However, extending these models for theanalysis of nanowire grown by thermal chemical vapor deposition(CVD) has not been attempted. But it is very important tounderstand the growth kinetics for CVD grown nanowires.Johansson et al. proposed a mass-transport-limited diffusionmodel for calculating the diameter dependent length growth rateof nanowires [10],

dL

dt¼ 2VRwlw

rwtanh

L

lw

� �� 2VJsw

rwCoshðL=lwÞþ 2VRtop (1)

where L is the length of nanowire; t is the time; V is the the atomicvolume of adatom in the catalyst particle; Rw is the deposition rate;lw is the diffusion length along the side of the nanowire; rw is theradius of the growing nanowire; Jsw is the adatom flux and Rtop isthe nanowire radius at the tip.

In Eq. (1), the first term represented the diffusion of materialdirectly deposited on the nanowire sidewalls, and the second termaccounts for adatom diffusion from the substrate surface up alongthe nanowire. The third term accounts for the material depositeddirectly on the catalytic particle. This equation indicates that as thenanowire diameter increases the length decreases. However forvery small but increasing diameters of nanowires, increase innanowire length has also been reported [11,12]. Through a detailedmodeling by taking into account both increasing and decreasing oflength growth rate with increasing diameter was reported byFroberg et al. [11]. Assuming that the direct impingement into

Fig. 1. SEM images of nanowires grown using gold catalyst with varying diameter

(a) 50 nm, (b) 85 nm, and (c) 120 nm.

C. Samanta et al. / Materials Research Bulletin 47 (2012) 952–956 953

catalytic particle is negligible, there seems to be a critical diameterbelow which length growth rate decreases with decreasingdiameter due to Gibbs–Thompson effect (GTE) and above whichlength growth rate decrease with increasing diameter due tolimited adatoms diffusion. The critical diameter is given by [11],

dc ¼4sV

kTln ðP=P1Þ(2)

where s is the surface energy of the catalyst in its bulk form (J/m2);V is the the atomic volume of adatom in the catalyst particle (m3);k is the Boltzman’s constant (J/K); T is the growth temperature (K);P is the adatoms vapor pressure in reactor (Pascal) and P1 is theadatoms vapor pressure inside the large (infinite radius) catalyst(Pascal). So critical diameter at which the nanowire attainsmaximum length is used to explain the growth kinetics. However,observable critical diameter depends chiefly on the ratio ofadatoms pressure in the reactor to the adatoms pressure inside thecatalyst. By controlling the catalyst particle diameter one can alsocontrol aspect ratio of nanowires. GaN nanowires with differentcatalytic particles have been reported by several groups [13–15].But there are hardly any reports on controlling the nanowirediameter hence the aspect ratio by choosing proper catalyst andwith suitable diameters. In addition to the catalytic particle, thegrowth condition is also very important to choose the growthkinetics. In this paper we demonstrate experimentally the controlof growth kinetics by a suitable choice of catalyst, their diametersand growth conditions. We show that the mass transport limiteddiffusion model can explain the growth kinetics where thediffusion through nanowire sidewall dominates. Adoption of asuitable catalyst, its diameter with known growth kinetics isimportant to have controlled growth of nanowires for nano-scaledevice applications.

2. Experimental

2.1. Synthesis

GaN nanowires were grown on silicon substrates withthermally evaporated gold, nickel and palladium as catalysis withaverage particle diameters 120, 45 and 20 nm, respectively. Forgold catalyst the average particle diameter was also varied to be120, 85 and 50 nm. High purity gallium-metal and ammonia gaswere used as source and the growth temperature and ammoniaflow rates were fixed to be 1173 K and 25 standard cubiccentimeter per minute (sccm), respectively. For all the cases thegrowth time was also fixed to be 3 h and the distance between thegallium source and growth substrate was about 3 mm [6].

2.2. Characterizations

The nanowires were characterized by field emission scanningelectron microscopy (FESEM) (operated at 10 keV), transmissionelectron microscopy (TEM) (operated at 200 keV), energy disper-sive analysis of X-rays (EDAX) and Raman spectroscopy. TheRaman spectra were recorded using 514.5 nm of Ar-ion laser in thebackscattering mode at room temperature. The laser power was10 mW through a 20�-objective and the collection time was 10 s.

3. Results and discussion

Fig. 1(a)–(c) shows the SEM images of GaN nanowires varyinggold catalyst particle diameters as, 50, 85 and 120 nm, respective-ly. The average nanowire diameters are 52, 85 and 115 nm which isclose to the catalyst particle diameters. However, the strikingfeature is the same length (�3 mm) of the nanowires irrespective of

their diameters and the reason for the same is discussed later.Fig. 2(a)–(c) shows the SEM images of GaN nanowires with Au, Niand Pd catalyst and Fig. 2(d) is a high magnification image ofFig. 2(c). A clear aspect ratio variation is observed for differentcatalytic particles. For palladium catalyst only nanowires (and notother nanostructures) are analyzed and included for furtherdiscussions. In Fig. 2, the diameter of the catalyst particles were120, 45 and 20 nm for gold, nickel and palladium, respectively andthe nanowire diameters are almost equal to the respective catalystparticle sizes. The corresponding lengths measured for nanowires(almost horizontal to the substrate) are approximately 3, 7 and1 mm, respectively. The composition of the nanowires wereanalyzed using EDAX measurements and were found to be almost1:1 (of Ga:N) for gold catalyst but for nickel and palladium it wasrelatively nitrogen rich (please see Fig. S1 of the Supportingdocument for details). Figs. 1 and 2 are the experimentaldemonstration of aspect ratio controlled growth of GaN nanowireswith suitable choice of catalysts. The diameter control is possibleby varying the catalyst particle diameter. The average diameterreported here ranges between 20 nm and 120 nm with aspect ratio

Fig. 2. SEM images of GaN nanowires grown using different catalysts (a) 120 nm gold, (b) 45 nm nickel, (c and d) 20 nm palladium and (d) is high magnification of (c).

Table 1Input parameters for the calculation of critical diameter.

Catalyst s (J/m2) V (m3) kT (J) P1 (Pascal)

Gold 1.2 [10] 1.959E�29 1.618E�20 1.15E�3

Nickel 2.34 [20] 1.959E�29 1.618E�20 2.76E�3

Palladium 1.92 [21] 1.959E�29 1.618E�20 5.75E�3

C. Samanta et al. / Materials Research Bulletin 47 (2012) 952–956954

variation between 35 and 150. As observed there is also a trade-offbetween the nanowire diameter and its length depending on thecatalyst particle for same growth conditions. Fig. 3(a) and (b)shows TEM images of nanowires grown using Au and Ni catalystsand Fig. 3(c) and (d) shows the TEM images of nanowires grownusing Pd catalyst. Fig. 4 shows the Raman spectra for nanowiresgrown using 50 and 120 nm diameter gold catalyst and 45 nmdiameter nickel catalyst. Raman active modes are observed at 566,650, 690 and 724 cm�1 for all the three samples. The peak at566 cm�1 is designated as E2(high) mode, 650 and 690 cm�1 peaksare surface optical phonon modes and 724 cm�1 peak as A1(LO)mode of GaN. Smaller gold catalysts show intense surface phononmodes. But the main difference is the peak intensities of E2(high)and A1(LO) modes for these nanowires. With gold catalyst thenanowires were oriented randomly and with nickel almosthorizontal to the substrate as seen in Figs. 1 and 2. In such acase, E2(high) mode is intense than A1(LO) mode for gold catalystand A1(LO) mode is more intense than E2(high) mode for nickelcatalyst. The Raman measurements were carried out in thebackscattering geometry which would indicate more scatteringfrom the side-facets of nanowires in the nickel case. The exposurenanowire sidewalls would be useful in explaining the growthkinetics discussed hereafter. To explain the growth kinetics of thenanowires with different catalyst and different diameters we canmake use of the critical diameter proposed by Froberg et al. [11].The critical diameters can be calculated using Eq. (2). For this, P isassumed to be the vapor pressure of liquid gallium at 1173 K andP1 is calculated for each catalyst as reported by Johansson et al.[10]. The vapor pressure (P* = P) of liquid Gallium at 1173 K isestimated using the Clausius–Clapeyron equation withDHvap = 258.7 kJ/mol and the reference value is P0 = 9.31E�36 Paat T0 = 303 K. This gives P* = 1.15E�2 Pa at 1173 K. Using theatomic fraction of gallium in each catalytic particle x = 0.1 [10],0.24 [16] and 0.5 [17] for Au, Ni and Pd, respectively and usingRaoult’s law we obtained P1 values (given in Table 1). Additionallywith an approximation that the activity coefficient is unity asassumed by Johnasson et al. [10]. The table also gives the values ofrequired parameters used in calculating critical diameter for all the

three catalyst. The P value given in the table is the vapor pressure ofliquid gallium at 1173 K which would actually be less at thesubstrate placed at 3 mm distance from the gallium source and itwould depend on the flow rate of ammonia in the reactor. Forexample, the gallium vapor pressure has a profile within thehorizontal reactor dictated by the flow rate of ammonia and themaximum vapor pressure is always down the stream similar towhat Nam et al., reported [18]. In the present case the flow rate was25 sccm and the substrate was placed at about 3 mm from thegallium source which is expected to have lesser pressure than the P

value since the maximum will be far down the stream. Themaximum vapor pressure of adatoms anywhere within the reactorcannot be larger than P itself and at the growth surface cannot besmaller than P1 to have a physical critical diameter value. Hencewe investigated the variation of P and its effect on the criticaldiameter in the range P1 < P < 1E�2. Fig. 5 shows the variation ofcritical diameter with P (calculated using Eq. (2)) for all the threecatalyst with input parameters as given in Table 1. Fig. 5 clearlyindicates that the critical diameter increases in the order of gold,nickel and palladium catalyst. For 0.007 Pa vapor pressure ofgallium the critical diameter for gold catalyst was about 3 nm, fornickel catalyst was about 15 nm and for palladium catalyst it wasabout 40 nm. As P/P1 approaches 1 the critical diameter increasesrapidly for palladium in comparison to gold and nickel. In thepresent case, for same growth conditions gold and nickel will havesmaller critical diameter in comparison to the catalyst particle size.This indicates that the growth is dominated by diffusion ofadatoms and GTE will hardly play any role. Palladium hassignificant critical diameter and also catalyst size is smaller thanthe critical diameter hence the growth is dominated by GTE. The

Fig. 3. TEM images of GaN nanowires grown using different catalysts (a) 120 nm gold, (b) 45 nm nickel, (c and d) 20 nm palladium.

C. Samanta et al. / Materials Research Bulletin 47 (2012) 952–956 955

smaller length growth rate is observed than gold and nickel alsosupports this result. Another striking difference is the length ofnanowires with gold and nickel catalyst. For almost samediameters of gold (about 50 nm) and nickel catalyst (about45 nm) as shown in Figs. 1(a) and 2(b) the average length of the

Fig. 4. Raman spectra of GaN nanowires grown using different catalysts as indicated

in the figure legend.

nanowires are about 3 and 7 mm, respectively. This can beexplained based on the diffusion of adatoms through the sidewallsof the nanowires. As evidenced from the SEM images and Ramandata, nanowires grown with nickel catalyst are mostly parallel tothe substrate exposing more sidewalls. More diffusion fromsidewalls for nickel than gold case possibly results in increasedlength of the nanowires. Since the density of nanowires for goldand nickel cases are very high, diffusion from the substrate can benegligible [19]. The limited surface diffusion is responsible for thesaturated length of the nanowires for the three different gold

Fig. 5. Variation of critical diameter with respect to adatoms vapor pressure inside

the reactor for three different catalysts.

C. Samanta et al. / Materials Research Bulletin 47 (2012) 952–956956

catalyst diameters. Much beyond the critical diameter apparentlythe length growth rate saturates with increase in diameter [19].From Eq. (1) we can clearly see that catalyst with (P/P1) ratio muchlarger than 1 will lead to nanowire growth governed by adatomsdiffusion through sidewalls. In such a case, direct impingement ofadatoms on the catalyst particle and diffusion of adatoms from thesubstrate contributes very little. The ratio (P/P1) �2 for palladium,�4 for nickel and �10 for gold also indicates the growth kinetics. P/P1 close to but slightly larger than 1 indicates the growth kineticsdominated by GTE for higher P/P1 ratio the critical diameterbecome negligible and hence limited diffusion of adatomsdominate the growth kinetics. The ratio (P/P1) �2 for InAsnanowire growth is also governed by GTE as reported by Froberget al., which is consistent with our results [11].

4. Conclusions

In conclusion, we have experimentally demonstrated thecontrol of GaN nanowire diameter and hence the aspect ratio.The growth control is chiefly through the catalyst particlediameter. Nanowires of average diameters ranging from 20 to120 nm and aspect ratios ranging from 35 to 150 have beenrealized. Based on previous reports, critical diameter dependentgrowth kinetics has been elaborated for CVD growth. Nanowiregrowth using gold and nickel catalyst are governed by diffusionthrough sidewalls but for palladium catalyst by GTE. The vaporpressure of adatoms in the reactor and inside the catalyst plays acrucial role in determining the growth kinetics.

Acknowledgments

We are thankful to IITK and DST for funding.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, in

the online version, at doi:10.1016/j.materresbull.2012.01.012.

References

[1] S.J. Pearton, J.C. Zolper, R.J. Shul, F. Ren, J. Appl. Phys. 86 (1999) 1–78.[2] S.J. Pearton, C.R. Abernathy, G.T. Thaler, R.M. Frazier, D.P. Norton, F. Ren, Y.D. Park,

J.M. Zavada, I.A. Buyanova, W.M. Chen, A.F. Hebard, J. Phys.: Condens. Matter 16(2004) R209–R245.

[3] B. Daudin, J. Phys.: Condens. Matter 20 (2008), 473201-1–473201-15.[4] C.M. Lieber, Solid State Commun. 107 (1998) 607–616.[5] M. Meyyappan, Prog. Cryst. Growth Charact. Mater. 55 (2009) 1–21.[6] A. Dabhal, D.S. Chander, J. Ramkumar, S. Dhamodaran, Micro Nano Lett. 6 (2011)

280–283.[7] Justin C. Johnson, H.-J. Choi, K.P. Knutsen, R.D. Schaller, P. Yang, R.J. Saykally, Nat.

Mater. 1 (2002) 106–110.[8] M.T. Borgstrom, G. Immink, B. Ketelaars, R. Algra, E.P.A.M. Bakkers, Nat. Nanotech.

2 (2007) 541–544.[9] L.E. Jensen, M.T. Bjork, S. Jeppesen, A.I. Persson, B.J. Ohlsson, L. Samuelson, Nano

Lett. 4 (2004) 1961–1964.[10] J. Johansson, C.P.T. Svensson, T. Martensson, L. Samuelson, W. Seifert, J. Phys.

Chem. B 109 (2005) 13567–13571.[11] L.E. Froberg, W. Seifert, J. Johansson, Phys. Rev. B 76 (2007), 153401-1–153401-4.[12] S.A. Dayeh, S.T. Picraux, Nano Lett. 10 (2010) 4032–4039.[13] C.-C. Chen, C.-C. Yeh, C.-H. Chen, M.-Y. Yu, H.-L. Liu, J.-J. Wu, K.-H. Chen, L.-C. Chen,

J.-Y. Peng, Y.-F. Chen, J. Am. Chem. Soc. 123 (2001) 2791–2798.[14] T.Y. Kim, S.H. Lee, Y.H. Mo, H.W. Shim, K.S. Nahm, E.-K. Suh, G.S. Park, Korean J.

Chem. Eng. 21 (2004) 257–261.[15] Y.F. Guo, C.S. Xue, W.J. Liu, H.B. Sun, Y.P. Cao, Mater. Charact. 61 (2010) 381–385.[16] X. Weng, R.A. Burke, J.M. Redwing, Nanotechnology 20 (2009), 85610-1–85610-5.[17] From the EDAX spectrum of the catalytic particle attached with the nanowire tip

during TEM measurements (data attached as Fig. S2 in supporting documents).[18] C.Y. Nam, D. Tham, J.E. Fischer, Appl. Phys. 76 (2004) 5676–5678.[19] D.S. Kim, U. Gosele, M. Zacharias, J. Cryst. Growth 311 (2009) 3216–3219.[20] E.A. Clark, R. Yeske, H.K. Birnbaum, Metall. Trans. A 11 (1980) 1903–1908.[21] K.H. Hansen, T. Worren, S. Stempel, E. Lægsgaard, M. Baumer, H.-J. Freund, F.

Besenbacher, I. Stensgaard, Phys. Rev. Lett. 83 (1999), 4120–4123.