Crystal growth modeling and nanotechnology: Research & educational opportunities for MatLab programming 1) Growth modes and surface processes 2) Rate equations and algebraic solutions 3) Extensions to defect nucleation 4) 1D and 2D rate-diffusion problems 5) 2D diffusion-growth John A.Venables Arizona State University and University of Sussex
37
Embed
Crystal growth modeling and nanotechnology: Research & educational opportunities for MatLab programming 1) Growth modes and surface processes 2) Rate.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Crystal growth modeling and nanotechnology:Research & educational opportunities for MatLab
programming
1) Growth modes and surface processes
2) Rate equations and algebraic solutions
3) Extensions to defect nucleation
4) 1D and 2D rate-diffusion problems
5) 2D diffusion-growth problems: movies
6) Nanotechnology, modeling & education
John A.VenablesArizona State University and University of Sussex
Explanation for NAN 546: April 09
This talk was given at several departmental seminars in the period 2003-05. These seminars typically require about 40-50 minutes and this one has 36 slides (#1, 3-37 here), with none in reserve this time. The audience is usually a Materials Department, with the level aimed at the Graduate Students
Slide #3 is an Agenda slide to guide one through the talk. The hyperlinks on slide #1 connect to Custom Shows, available under the Slide Show dropdown menu. This particular talk has a tutorial character, which explains why we have been able to use it for sessions on growth modes, diffusion mechanisms, rate equations, stress effects, visualization, etc
This material is copyright of John A. Venables. Author and Journal references are typically given in green, and I do not talk about material that has not been published in some form.
Crystal growth modeling & nanotechnology: Research & educational opportunities for MatLab programming
• Growth modes and surface processes• Rate equations and algebraic solutions• Extensions to defect nucleation• 1D and 2D rate-diffusion problems• 2D diffusion-growth problems: movies• Nanotechnology, modeling & education
John A.VenablesArizona State University and LCN-UCL
Growth modes
Island Layer + Island LayerVolmer-Weber Stranski-Krastanov Frank-VdM
Atomic-level processes
Variables: R (or F), T, time sequences (t)
Parameters: Ea, Ed, Eb, mobility, defects…
Early TEM pictures: Au/NaCl(001)
Donohoe and Robins (1972) JCG
Alternative approaches to modeling
1) Rate and diffusion equations
2) Kinetic Monte Carlo simulations
3) Level-set and related methods
plus
4) Correlation with ab-initio calculations
Issues: Length and time scales, multi-scale; Parameter sets, lumped parameters; Ratsch and Venables, JVST A S96-109 (2003)
Rate Equations (experimental variables T, F or R, t)
dn1/dt = F(or R) –n1/ n1(t), single adatoms....dnj/dt = Uj-1 - Uj = 0 nj(t), via local equilibrium
....dnx/dt = dnj/dt = Ui - ... nx(t),
(j > i +1) stable cluster density
Ui = iDn1ni , with i , x as capture numbers
–1 = iD(i+1)ni + xDnx nucleation, growth
Competitive capture
dn1/dt = R (or F) – n1/; an
c…
Venables PRB 36 4153-62 (1987)
Differential equations versus Algebra
Using cluster shape, assumed or measured, express
nx(Z) (Z). f1(Rpexp(E/kT))
t(Z) (Z). f2(Rpexp(E/kT));
where p and E are functions of i, critical nucleus size
similarly for mean size ax(Z) and condensation coefficient (Z), not much used.
Choice of 1) integrating differential equations, or
2) evaluating near the maximum of nx(Z).
Steady state conditions (dnx/dt, etc = 0) converts a set of ODE’s into a (nonlinear) algebraic solution.
Nucleation density predictions
• Matlab Programs (R, T-1 and cluster size, j)
• Input Energies
• Simultaneous output: Densities and critical cluster size, i.
McDaniels et al. PRL 87 (2001) 176105, data, and work in progress on rate equation and 2D modeling
Nucleation on point and line defects
(a) Point defects (vacancies) (b) Line defects (steps)
Extension to Defect Nucleation (parameters nt, Et)
dn1/dt = R –n1/ n1(t), single terrace adatoms
dn1t/dt = 1tDn1nte - n1tdexp(-(Et+Ed)/kT) n1t(t)
.... empty traps trapped adatomsdnj/dt = Uj-1 - Uj = 0 nj(t), via local equilibrium
REs: integrate to 2 or 20 min. anneal with given V0.
KMC: hexagonal lattice simulations (1000 x 1155) sites with EB = V0.
Extension to Ge/Si(001)stress-limited capture numbers
• Low dimer formation energy (Ef2 ~ 0.35 eV) gives large i,
even though condensation is complete • Stress grows with island size, x decreases
• Lengthened transient regime results, > 1 ML, source of very mobile ad-dimers (Ed2 ~ 1 eV) for rapid growth
eventually of dislocated islands• Interdiffusion, and diffusion away from high stress regions
around islands, reduces stress at higher T and lower F (e.g. at 600, not 450 oC for F ~1-3 ML/min.)
Chaparro et al. JAP 2000, Venables et al. Roy. Soc. A361 (2003) 311
Conclusions: t-dependent capture numbers
1) Explicit t-dependence involves the transient regime and a finite number of adatoms. Barriers or repulsive potential fields reduce capture numbers, lengthen transients and involve more adatoms.
2) Barrier capture numbers and diffusion capture numbers add inversely. An interpolation scheme is needed to describe t-dependence in the transient.
3) Large critical nucleus size lengthens transient. Annealing a low T deposit with potential fields is a very sensitive test of t-dependent capture numbers, as small capture numbers result in little annealing.
And finally,
Area 2D (x,y,t or r,,t) problems: Shapes, edge diffusion, instabilities, lithography,
quantum dots, anisotropic stress effects, etc.
Geometries to consider are square or rectangular, using an (x,y) mesh: e.g. (2x1) and (1x2) Si(001);
hexagonal, which can be approximated by a 1D cylindrical (r) domain;
triangular lattice, applicable to reconstructed f.c.c. metal and semiconductor (111) surfaces.
Expts: STM of Co/Si(111); and Ag/Ag layers/Pt(111) (Bennett et al.) (Brune et al.)
Questions for 1D and 2D modelling
1) How far can one realistically go without becoming over-dependent on too many unknown parameters??
2) How many types of different experiments can one actually perform?
3) Large-scale (commercial) packages can solve PDEs. But, is the science unique, or do multiple inputs give the same results?
Importance of lumped parameters
FFT Method for time-dependent x-y Diffusion Fields
The general solution to
2 2
X Y2 2D + D
C C C
t x y
on a rectangular grid (X,Y) (kx, ky points) is C = Field = ifft2(fft2(Field).*Pmat)where the propagator Pmat for time t is:
unitmat.*exp(-2t(DX(1-cos(2(X-1)/kx))/a2 +
DY(1-cos(2(Y-1)/ky)))/b2)
Program Structure (MatLab 6.5)
• Initialize, set up Field and Island Masks
• Calculate Propagator, Pmat, for t
• Loop over time steps: ktimes = [1:900]
• Update Field, calculate fluxes and reset boundary conditions on Island and Field
• Plot Field data at plottime = [1 2.. 10.. 900]
• Save calculated capture number data & other calculations for subsequent plotting
height = 5time = 90 t = 0.164*64 grid
(5*11) grows to(19*33)
Dx = 5Dy = 10
Venables & Yang 2004
Annealing: rectangular islands
Capture Numbers during annealing
1D and 2D results and conclusions Attachment-limited solutions EB, V(r) for STM
Cu/Cu(111) Brune (EPFL), Phys. Rev. B 2002; AFM Ge/Si(001) Drucker (ASU), in progress
Anisotropic attachment and growth: AFM/STM/ LEEM/HREM (Co,Pd) silicide nanowires Bennett
Anisotropic 2D nucleation and growth: AFM/ LEEM/HREM Ag/Si nanowires Li & Zuo (UIUC)
Conclusion: It is worth exploring a few models with a
few defined parameters in 1 and 2 spatial dimensions
when a strong connection to experiment is available.
Nanotechnology, modeling & education
Interest in crystal growth, atomistic models and experiments in collaboration
Interest in graduate education: web-based, web-enhanced courses, book
See http://venables.asu.edu/ for detailsOpportunities for undergraduate (REU), and