Surface Processes at the Nanoscale: how crystals meet the outside world 1) Motivation: let's start with snowflak es 2) Microscopy and diffraction techniques 3) Nucleation, growth and nanofabricatio n 4) Specific systems: what do we want to John A. Venables Physics Department, Arizona State University and London Centre for Nanotechnology, UCL
62
Embed
Surface Processes at the Nanoscale: how crystals meet the outside world 1) Motivation: let's start with snowflakesMotivation: let's start with snowflakes.
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
Surface Processes at the Nanoscale:how crystals meet the outside world
1) Motivation: let's start with snowflakes
2) Microscopy and diffraction techniques
3) Nucleation, growth and nanofabrication
4) Specific systems: what do we want to know?
Pd/MgO(001); Cu/Cu(111); Ge/Si(001), etc
5) 2D Modeling: recent work in progress
6) Nanostructures : disciplines and technology
John A. VenablesPhysics Department, Arizona State University
and London Centre for Nanotechnology, UCL
Scientific and Technological Motivation
• We understand binding in bulk crystals: what is special and different at the surface?
• We understand thermodynamic equilibrium: but useful structures are grown kinetically...
• It's the science behind the chip business: epitaxial growth of heterostructures, lasers
• And catalytic reactions at small particles: only chemical firms don't share their secrets
• Plus energy, health, environment and art: enjoy nanotubes, tetrapods, snowflakes, etc.
• Wilson Bentley (1865-1931) was a farmer near Jericho, Vermont, who during his lifetime captured some 5000 snow crystal images. More than 2000 were published in 1931 in his famous book, Snow Crystals
• Ukichiro Nakaya (1900-1962) was the first person to perform a true systematic study of snow crystals. Trained as a nuclear physicist, Nakaya was appointed to a professorship in Hokkaido, the North Island of Japan, in 1932, where there were no facilities for nuclear research. Undaunted, Nakaya turned his attention to snow crystals, which were locally very abundant. Book (1954), Snow Crystals: Natural and Artificial.
The morphology diagramthe role of supersaturation
Driving force = kTln(S);
Supersaturation S = (p/pe)Early MC calculations, 1979John Weeks, George Gilmer
Facets, dendrites, pattern formation
• Pt(111) monolayers by high-T STM:
• top row 0.15 ML
• bottom row 1 ML
• Thomas Michely & Joachim Krug, book Islands, Mounds and Atoms, 2004
1.5 mm 310 nm field of view
• Ice Photos: Ken Libbrecht, book Snowflake, Winter's secret beauty, 2003
Early TEM pictures: Au/NaCl(001)
Donohoe and John L. Robins (1972) Journal of Crystal Growth
Field ion microscopy: diffusion of adatoms
Gert Ehrlich group UIUC, 1988-1997; Gary Kellogg 1994-1997
W(211) substrate23.06 kcal/mol = 1 eV/atom
Scanned Probe Microscopy
UHV STM: Pt/Pt(111) T = 424 K: ML
Helium atom scattering at different T (K)Thomas Michely, George Comsa group (1990-1995)
0.35 3.0
12 90
621 K
424 K
275 K
High resolution TEM: CoSi2/Si(111)
left: a) plan view TEM
b) platelets, c) wires:
lattice resolution cross
-section David Smith
(100)
(-111)
(111)
(-511)
[100
]
(100
)300 nm300 nm300 nm
above: platelets & nanowires by AFM Anouk Rougee
Zhian He, David J. Smith, Peter A. Bennett PRL 93 (2004) 256102
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…
Competitive capture
dn1/dt = R – n1/;
an
c…
Venables (1987) Phys. Rev. B
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. (2001) PRL; Venables et al. (2003) Proc. Roy. Soc.
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
• Metals on insulators (Au/NaCl, Fe/CaF2, Pd/MgO...): energies, role of defects, metal catalytic properties
• Metals on and in semiconductors (Ag/Si, Ti, Dy/Si...) energies, subsurface growth, nanowires, magnetism
Experiment - kinetic model - quantum calculation
A particular case: Pd/MgO (001)
G. Haas et al. 2000 PRB; Venables and Harding 2000 JCG
Defect nucleation, i = 3 at high T
Pd/MgO (001): parameter sensitivity
Venables and Harding 2000 J. Crystal Growth 211, 27-33
TrappingPair Binding
Rate equations & KMC with DFT parameters
Rate equations Venables, Giordano & Harding, J. Phys. C.M. 2006
KMC: L. Xu, G. Henkelman, 2005-07, G. Barcaro et al, 2005
Pd2
Pd4
Pd3
FS+ center
Conclusions #11) Nucleation & growth models have been developed
where "experimental" energies for adsorption, diffusion, binding & trapping can be extracted.
2) Small 2D and 3D clusters are mobile on the surface, can even be liquid; competing configurations
3) Many theoretical methods are now available to see if such energy values are reasonable. The cases of Pd and Ag/MgO(001) have been investigated in detail, but results have been controversial. Are we now OK?
4) The Chemists seem to be winning! Embedded clusters, spin polarized calculations seem to be needed to get good values, especially for Pd, which has competing singlet and triplet ground states.
REs: integrate to 2 or 20 min. anneal with given V0.
KMC: hexagonal lattice simulations (1000 x 1155) sites with EB = V0.
Conclusions #2: time-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.
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
Sizes and shapes in Ge/Si(001)
TEM, AFM: Chaparro, Zhang, Drucker, Smith J. Appl. Phys (2000)
Size distributions and alloying
T = 450 °C
1.5 x109
1 x109
0
5 x108
3.2 x109
1.6 x109
0
4.8 x109
0 40 12080 160
X 2
(d)
(b)T = 600 °C
5 ML6.5 ML8.0 ML9.5 ML11.0 ML12.5 ML
Diameter (nm)
Num
ber
of is
land
s /
cm2
/ 2.
5 nm
bin Strain relief via
1) interdiffusion 2) change of shape
Hut-dome transitionsreversible via alloying athigh T > 500 oC
S. Chaparro, Jeff Drucker et al. PRL 1999, JAP 2000
Ge/Si(001) STM Movies: watching paint dry at 450 OC
Mike McKay, John Venables and Jeff Drucker, 2007-08
gas-source MBE from Ge2H6
Ge = 5.0ML, 0.1 ML / minT = 450 °C, 26 min/frame62 hrs total elapsed timefirst frame after 33min annealField of view 600nm x 600 nm
Ge = 5.6ML, 0.2 ML / minT = 500 °C, 7 min/frame14 hrs total elapsed timefirst frame after 160min annealField of view 400nm x 400 nm
Ge/Si(001) hut clusters:Annealing at T = 450 oC
1
4
2
103
56
7
89
1,255
30 nm
1
4
2
103
56
7
89
33
1
4
2
103
56
7
89
2,503
1
4
2
103
56
7
89
3,751
100
150
200
250
300
350
400
450
500
10
15
20
25
30
0 1000 2000 3000 4000
9
Anneal Time (min)
Width
Volume
Length
100
150
200
250
300
350
400
450
500
10
15
20
25
30
0 1000 2000 3000 4000
8
Anneal Time (min)
Width
Volume
Length
Most islands static, smallest island grows (8).
Conclusions #3: Long term annealing with barriers
1) Long term meta-stability in the Ge/Si(001) system at intermediate T = 450 oC, ripening at 500 oC, over long times, several days.
2) Some hut clusters to grow via growth of the short side, but other sides do not grow. Individual facets nucleate and grow: volume proportional to length; nucleation barrier smallest on the shorter sides.
3) Large ad-dimer mobility and some coarsening on and in the wetting layer. Finally dislocated dome clusters grow, and coarsening accelerates, with much mass transfer over large distances (many m).
Starting Simplifications: 1) low concentration D = D*; 2) no distributed sinks = 0; 3) annealing G = 0.
2D Rate-diffusion simulations
J.A. Venables, J. DeGraffenreid, D. Kay & P. Yang, PRB 2006
R. Grima, DeGraffenreid, Venables, PRB 76, 233405 2007
Frame from 2D Movies
Connect to MatLab filesmovie #1; isometricmovie #2;plan viewmovie #3; capture numberwith/ without repulsive fields
Mean-field equations from microscopic dynamics
From Shu, Liu, Gong et al:
For Ge/Si(001): 1 = 1.75 eV; at lattice sites
21 = .75 eV fast diffusion direction
1 2 1
1 1
( , ) exp( ( ) ( , ))
( ) ( )ˆ ˆ( , ) ( , )
D x y D x y
x y D x y x yx y
V
Strain dependent Diffusion D and Drift velocity Vas deduced by Grima, DeGraffenreid, Venables 2007
Ge/Si(001) concentration profiles
R. Grima, J. DeGraffenreid and J.A. Venables, 2007, PRB
2= 1= 1.75 eV
2 1= 0.75 eV
2- 1= 1.50 eV
Conclusions• Three approaches to diffusion in potential fields (Ovesson
2002, Venables & Brune 2002, Grima & Newman 2004) "same for constant D"; but this is not generally the case. V&B thermodynamics correct, G&N advection-diffusion
• Capture numbers are much reduced due to island potential fields; (rectangular) updateable potentials for "strain".
• Grima & Newman's MED algorithm has been solved for 2D problems; Sum rules are exactly satisfied, including general potential fields. Nanowire systems studied with Ge/Si(001) model parameters (Venables et al., 2006).
• MED with "potentials due to strain" are studied (Grima, DeGraffenreid, Venables 2007). Explicit microscopic expressions for D & drift velocity V obtained; D changes are more important than drift for the Ge/Si(001) model.
Y. Li, M.C. Bartelt, J.W. Evans et al.: PRB 56 12539 (1997)
P. Yang & J.A. Venables: MRS 859E JJ3.2 (2005)
Numerical methods, Capture numbers in potential fields
S. Ovesson: PRL 88 116102 (2002)
J.A. Venables & H. Brune: PRB 66 195404 (2002)
R. Grima & T.J. Newman: PRE 70 036703 (2004)
J.A. Venables, J. Degraffenreid et al.: PRB 74 075412 (2006)
R. Grima, DeGraffenreid, Venables, PRB 76 233405 (2007)
Nanostructures : disciplines and technology
• Interdisciplinary environment: Physicists, Chemists, Materials Scientists, Engineers. Interchangeable jobs: what does each discipline bring to the table?
• Electrochemistry, solution chemistry, single molecules: more knobs to turn, but fewer in-situ analysis tools? Do all Inventions lead to Innovation? If not, why not?
• What will we really learn from biology? Is nano-bio-anything the wave of the future, or is it just the latest bubble, and already past its prime? Stick to basics...
• A great field for "emergent phenomena": simple rules lead to complex results (P.W. Anderson, 1972)
Nanotechnology, modeling & education
Interest in crystal growth, atomistic models and collaborative experiments, as illustrated in this talk
Interest in graduate education: web-based, and web-enhanced courses and resources, book
See http://venables.asu.edu/index.html for current projects, reference list, links to courses, resources
New Professional Science Masters (PSM) in Nanoscience degree at ASU, now in second year http://phy.asu.edu/graduate/psm/overview
Charge redistribution, and hence Et in F-centre too large? Pd: Et = 2.72 eV > Hf (PdO) = 0.9 eV; Ag: Et = 1.27 eV > Hf (Ag2O) = 0.34 eV; Hf from Reuter & Scheffler (2004)
Fuks, Kotomin et al. (2002-03)HF+Correl, Ea, Ed too small?: Ag: Ea ~ 0.20, Ed ~ 0.05 eV
Embedded DFT cluster + classical shell
Expt: Pd/ MgO(001) Et > 1.5, Ea = 1.2, Eb = 1.2 and Ed < 0.3 eV? Calculated Ed = 0.34, not 0.86 eV (B&J)
Many details, several XC functionals explored, etc; can only give a flavor here. Giordano..& Pacchioni (2004-05)
Terrace Step F F DiVac
Pd Ea 1.36 1.85 3.99 2.70 3.00
trap Et 0.49 2.63 1.20 1.64
Pd2 Eb 0.50 0.66 0.57 0.91 1.71
trap E2t 1.14 1.34 cluster 1.49
Nanoclusters: Pd2, Pd3 and Pd4 ...Expt: Pd/ MgO(001) Et > 1.5, Ea = 1.2, Eb = 1.2 and Ed < 0.3 eV? Indications of i=3 and desorption at high T
Extension of same Pd2 approaches to Pd3 and Pd4:
Pd2 has minimum binding (0.57 eV) at F-center, Pd3 by a further 0.75 eV, and Pd4 by a another 1.38 eV (all relative to Pdn-1 on the defect and Pd1 on the terrace)
Spin polarized cluster configuration (spin singlets d10 on surface, versus triplet d9s1 in gas phase, E ~ 0.19 eV)
Giordano..& Pacchioni (2005)
Is i=3 likely at high T at defects? Looks good: obvious next question in context of "believing" these energies...
Issues: Length and time scales, multi-scale; Parameter sets, lumped parameters;
Christian Ratsch and John Venables, JVST A S96-109 (2003)
Level Set Function Surface Morphology
t
=0
=0
=0
=0=1
• Continuous level set function is resolved on a discrete numerical grid
• Method is continuous in plane, but has discrete height resolution
The Level Set Method: Schematic
• Governing Equation:
• Obtain velocity of island boundaries by solving diffusion equation:
• Boundary condition
0||
nvt
dt
dNDF
t22
0
=0
• Velocity:
• Nucleation Rate:
• Seeding position chosen stochastically (weighted with local value of 2)
2),( tDdt
dNx
)()( aveedge DDvn nn
Level Set Slides: Christian Ratsch, UCLA Applied Math Department
The Level Set Method: Formalism
KMCLevel SetData: Fe/Fe(001)
J.A. Stroscio and D.T. Pierce, Phys. Rev. B 49 8522 (1994)
Petersen, Ratsch, Caflisch, Zangwill, Phys. Rev. E 64, 061602 (2001)
Reversibility: Sharpening of the Island Size Distribution
Microscopy and Diffraction Techniques • Early TEM: Au/NaCl(001) island growth example• FIM: the first to "see atoms" and diffusion paths • Scanned probe microscopy: STM, AFM and MFM• High resolution TEM, EDAX, EELS, holography • UHV analytical SEM and STEM, AES, EELS• LEEM and PEEM, SPLEEM, etc • coupled with scattering and diffraction techniques:
LEED, THEED, RHEED, X-rays and neutrons, Helium atom scattering (HAS), RBS, ICISS....
Acronym heaven: techniques in red available at ASU
UHV SEM and STEM: AES & SAM• ASU development, mid 80's- present: John Cowley, Peter
Crozier, Jeff Drucker, Gary Hembree, Mohan Krishnamurthy, Jingyue Liu, Mike Scheinfein, John Spence, John Venables et al.
• UHV Applications to electronic materials and catalysis: see my web page at http://venables.asu.edu/research/index.html
• Cowley memorial volume: J. Electron Microscopy 54 (2005) 151
LEEM and PEEM: SPLEEM & XMCDPEEM• ASU and Trieste development, early 90's- present: Ernst Bauer,
Peter Bennett, Assia Pavlovska, Ig Tsong and co-workers
• UHV applications to surface morphology and reconstructions, electronic and magnetic materials and catalysis: see Bauer/ Pavlovska web page at http://physics2.asu.edu/homepages/bauer/
• Review article: Reports on Progress in Physics 57 (1994) 895