Effects of supersaturation on the crystal structure of gold seeded III–V nanowires

1Jonas Johansson, 2Lisa S. Karlsson, 1Kimberly A. Dick, 1Jessica Eriksson, 1Brent A. Wacaser, 1Knut Deppert, and 1Lars Samuelson

1Solid State Physics and the Nanometer Structure Consortium, Lund University, Sweden

2National Centre for High Resolution Electron Microscopy (nCHREM) / Polymer & Materials Chemistry, Lund University, Sweden

jonas.johansson@ftf.lth.se

Outline

• Introduction

• Experimental

• Nanowire growth results

• Nucleation model

• Conclusions

Introduction• High density of planar defects in III-V

nanowires with zinc blende structure. This can be tuned by growth conditions.* **

• Can we radically change the crystalline properties of gold seeded III–V nanowires by growth at different conditions:– Pulsed growth– Altered seed particle

• Can we find growth conditions that let us grow pure zinc blende or wurtzite wires?

• Can we also find a sharper condition for wurtzite formation in zinc blende nanowires than what has recently been proposed?***

*J Johansson et al, Nat Mater, 5 574 (2006): GaP at 440, 470, and 500°C

**H Joyce et al, Nano Lett, 7 921 (2007): GaAs at 450-390°C

***F Glas et al, Phys Rev Lett, 99 146101 (2007): ΔGWZ < ΔGZB

Stacking sequences

Zinc blende: ABC…

Wurtzite: AB…

fault plane

(twinning)

fault plane

fault plane

(wurtziteformation)

zinc blende

Experimental

• Growth of GaP(111)B nanowires on GaP(111)B substrates in MOVPE at 470°C.

• Growth seeded by aerosol fabricated gold particles, 40 nm diameter.

• Precursors: TMG and PH3.

• In contaminated or clean (In free) susceptor.• Pulsed growth or continuous growth.

• Pulsed growth:– TMG: 10 s ON, 1 min OFF– 20 pulses

– PH3 always ON

• HR TEM for structural characterization

Nanowire growth resultsPulsed growthIn background

Continuous growthIn background

Continuous growthIn free background

Extended zinc blende (ZB) segmentsAverage: 20 MLMaximum: 200 ML

Short ZB segments (lamellar twinning)Average: 3.5 ML

Wurtzite (WZ) segmentsAverage: 3.5 MLMaximum: 21 ML

Composition of the seed alloy particle• EDS measurements of the Ga and In concentrations in the gold alloy particle

• Two different growth terminations• Cooling down in

– H2: Ga concentration during growth, C– PH3 + H2: Ga concentration at eqiulibrium, Ceq

• Conclusion: higher supersaturation during In free growth– In background: C/Ceq ≈ 1–3 – In free: C/Ceq ≈ ~10

In background In free background

H2 PH3 H2 PH3

In (at-%) 20 – 30 20 – 30 0 – 5 0 – 5

Ga (at-%) 0 0 20 – 26 0Detection limit ≈ 3 at-% (0 means less than 3 at-%)

Schematics of pulsing

Growth at different supersaturationsPulsed growthIn background

Continuous growthIn background

Continuous growthIn free background

Extended zinc blende (ZB) segments

Short ZB segments (lamellar twinning)

Extended wurtzite (WZ) segments

low Supersaturation (, C/Ceq) high

Nucleation description• Two assumptions for the nucleation model:

– (i) Layer-by-layer growth, monocenter nucleation at wire edge

– (ii) Poissonian nucleation

• Justifications– (i) Perfect atomic planes in

the wires (no grain boundaries),very low P solubility in Au

– (ii)

2

exp1 s

sf

ms = 4.2 ML

2D nucleation model

rrrG 22

22

0 for ordinary planet for fault plane

Step energy eih 2

eqc

B

C

C

s

Tkln

0 rG

Finding the nucleation barrier for ordinary and fault plane nucleation

2

2*G f

ffG

2

2* f

Nucleation barrier

r

ΔG(r)

ΔG*

r*

rrrG 22

22

Possibility for wurtzite formation?• Fault plane nucleation is favoured:*

• Can happen if Γf < Γ– Inner step energies same: γi

f = γi

– Edge step energies differ: γef < γe

• Rewrite ΔG-inequality:

– Not much data… If step energies follow surface energies, for ZnSf ≈ 0.6–0.7 (WZ σav = 0.57 J/m2, ZB σav = 0.86 J/m2)**

• Is this relation sufficient for the wurtzite structure to form?

** GG f

**H Zhang et al, J Phys Chem B, 107 13051 (2003) *F Glas et al, PRL, 99 146101 (2007)

fft

114

2 2

1 efef

Nucleation probabilities

• Nucleation rates:

• Fault plane nucleation probability

TkGJ B*exp

TkGJ Bff*exp

f

ff JJ

Jp

pf versus

efef

(Edge step energyratio)

Poissonian nucleation

• Can we relate the fault plane nucleation probability, pf, to segment thicknesses?

• Poissonian nucleation: the (geometric) probability distribution of nucleating exactly k fault planes, that is a k ML thick WZ segment, before an ordinary nucleus forms

• The average WZ segment thickness (the average value of k)

,2,1,0,1 kpp fkf

f

fWZ p

ps

1

Interpretation of growth results

swz = 3.5, pf = 0.78

sZB = 3.5, pf = 0.22

sZB = 20, pf = 0.05

Pulsing

Continuous

In-free

Condition for WZ formation revisited

• Fault plane nucleation is favoured when

• Does this mean that WZ formation will occur?

• This is a matter of definition…

• By combining the equations for pf and swz, we can pose a stronger condition for WZ formation:

** GG f

** ln GsTkG WZBf

Conclusions

• Pulsed growth – low supersaturation– zinc blende favoured

• Growth at high supersaturation– wurtzite favoured

• Qualitative nucleation model– planar defect density along wire– zinc blende / wurtzite occurence

MOVPE of gold seeded GaP <111>B oriented nanowires: