Vladan Mlinar 2009 Materials Research Society Spring Meeting

Post on 21-Dec-2014

875 views 0 download

description

For more information about the Spectral Barcoding and establishing structure-spectra relationship in quantum dots, see the following publications: - Vladan Mlinar and Alex Zunger, Phys. Rev. B 80, 035328 (2009). - Vladan Mlinar et al. Phys. Rev. B 80, 165425 (2009). -------------------------- My full publications list can be found at: www.vladanmlinar.com/publications.html

Transcript of Vladan Mlinar 2009 Materials Research Society Spring Meeting

Deciphering Structural Information from the Multiexcitonic Spectra of a Quantum Dot

Vladan Mlinar & Alex Zunger

National Renewable Energy Laboratory

Golden, Colorado USA

Vladan.Mlinar@nrel.gov

QDs: Structure - Spectra relationship

Methods for structural characterization

• TEM based methods

• X-ray diffraction

• X-STM

Single-dot spectroscopy

QDs: Structure - Spectra relationship

(M. Bozkurt, J. M. Ulloa, & P. M. Koenraad)

Methods for structural characterization

• TEM based methods

• X-ray diffraction

• X-STM

Single-dot spectroscopy

• No atomic resolution

• All of the methods require assumption

about composition profile and/or shape!

QDs: Structure - Spectra relationship

(M. Bozkurt, J. M. Ulloa, & P. M. Koenraad)

Methods for structural characterization

• TEM based methods

• X-ray diffraction

• X-STM

Single-dot spectroscopy

• No atomic resolution

• All of the methods require assumption

about composition profile and/or shape!

(M. Ediger &

R. J. Warburton)

QDs: Structure - Spectra relationship

(M. Bozkurt, J. M. Ulloa, & P. M. Koenraad)

Methods for structural characterization

• TEM based methods

• X-ray diffraction

• X-STM

Single-dot spectroscopy

• No atomic resolution

• All of the methods require assumption

about composition profile and/or shape!

• Controllable number of electrons and holes

• μeV resolution

(M. Ediger &

R. J. Warburton)

Typically, Structure is used to predict Spectra

• Since for quantum dots we do not know the structure:

Assume

or

measure

structure

Measured

emission

spectra

Calculate

resulting

spectra

Structure

Typically, Structure is used to predict Spectra

• Since for quantum dots we do not know the structure:

Is this possible?

Assume

or

measure

structure

Measured

emission

spectra

Calculate

resulting

spectra

Structure

Question: What is the structural information encoded in the multiexcitonic spectra of a QD?

?

Spectral Barcoding vs. DNA Barcoding:

Barc

oder

Barc

odin

g

Organism is identified as belonging to a particular species

Sci. Am. p. 82-88 (October 2008)

Spectral Barcoding vs. DNA Barcoding:

Barc

oder

Barc

odin

g

Organism is identified as belonging to a particular species

Sci. Am. p. 82-88 (October 2008)

Spectral Barcoding vs. DNA Barcoding:

Barc

oder

Barc

odin

g

Organism is identified as belonging to a particular species

QD is identified as belonging to a group of QDs with common structural motifs.

?

Vladan Mlinar and Alex Zunger, PRB 80, 035328 (2009).

How does the Spectral Barcoding work?

Spectral barode:

Vladan Mlinar and Alex Zunger, PRB 80, 035328 (2009).

How does the Spectral Barcoding work?

Artificial Intelligence QD library

Deterministic links between

structures and spectral marker(Distilling rules from library)

Spectral barode:

Spectral barcoding

procedure

Vladan Mlinar and Alex Zunger, PRB 80, 035328 (2009).

How does the Spectral Barcoding work?

Artificial Intelligence QD library

Deterministic links between

structures and spectral marker(Distilling rules from library)

Structure

Structural Motifs:

• h = 2 – 3nm

• Xav(In) = 75-80%

RESULT: a set

of QD structural

motifs!

Spectral barode:

Spectral barcoding

procedure

Vladan Mlinar and Alex Zunger, PRB 80, 035328 (2009).

Spectral Barcoding: Data-mining of the library

Structure

QD structure is discretized into a set of Ns=5 structural motifs, each taking up one of

Nv possible values:

Motifs: Shape b (nm) h (nm) XIn (%) profile

Trun.Cone 12 2.0 50 Homog.

Trun. Pyr. 18 3.0 60 Linear

Lens 20 3.5 70

Elong.

Lens [110]

23 4.0 80

Elong.

Lens [110]

25 5.0 90

Elong.

Lens [100]

30 6.0 100

Spectral Barcoding: Data-mining of the library

Structure

QD structure is discretized into a set of Ns=5 structural motifs, each taking up one of

Nv possible values:

Motifs: Shape b (nm) h (nm) XIn (%) profile

Trun.Cone 12 2.0 50 Homog.

Trun. Pyr. 18 3.0 60 Linear

Lens 20 3.5 70

Elong.

Lens [110]

23 4.0 80

Elong.

Lens [110]

25 5.0 90

Elong.

Lens [100]

30 6.0 100

Bayesian Data Reduction Algorithm:

• Training: Testing how each structural motif and its corresponding values influences the

barcode

• Result: Identifies the set of structural motifs that are responsible for a given spectral

barcode sequence.

Spectral Barcoding: Consistency test!

Vladan Mlinar and Alex Zunger,

PRB 80, 035328 (2009).

Spectral Barcoding: Consistency test!

Vladan Mlinar and Alex Zunger,

PRB 80, 035328 (2009).

Spectral Barcoding: Consistency test!

Vladan Mlinar and Alex Zunger,

PRB 80, 035328 (2009).

Spectral Barcoding: Consistency test!

Validation!

Vladan Mlinar and Alex Zunger,

PRB 80, 035328 (2009).

Question: How does the deduced structure relates to the “real structure”?

Spectral Barcoding: Why is it important?

Quantum Dot

growth

Structural Characterization by X-STM

Single-dot Spectroscopy

Antonio Badolato

(ETH Zurich, Switzerland)

Theory

Collaboration with

three experimental

groups!

Many body

pseudopotential

calculations

Calculated spectra

Spectral Barcoding: Why is it important?

Quantum Dot

growth

Structural Characterization by X-STM

Single-dot Spectroscopy

Antonio Badolato

(ETH Zurich, Switzerland)

M. Bozkurt, J. M. Ulloa, & P. M. Koenraad

(TU Eindhoven, The Netherlands)

Theory

Collaboration with

three experimental

groups!

Many body

pseudopotential

calculations

Calculated spectra

Spectral Barcoding: Why is it important?

Quantum Dot

growth

Structural Characterization by X-STM

Single-dot Spectroscopy

Antonio Badolato

(ETH Zurich, Switzerland)

M. Bozkurt, J. M. Ulloa, & P. M. Koenraad

(TU Eindhoven, The Netherlands)

Theory

M. Ediger & R. J. Warburton

(Heriot-Watt University, UK)

Collaboration with

three experimental

groups!

Many body

pseudopotential

calculations

Calculated spectra

Spectral Barcoding: Why is it important?

XS-2 < XT

-2 < X-1 < XX0 < X0 sequence

in measured spectra from each and

every QD studied in the ensemble is

kept.

Quantum Dot

growth

Structural Characterization by X-STM

Single-dot Spectroscopy

Antonio Badolato

(ETH Zurich, Switzerland)

M. Bozkurt, J. M. Ulloa, & P. M. Koenraad

(TU Eindhoven, The Netherlands)

Theory

M. Ediger & R. J. Warburton

(Heriot-Watt University, UK)

Collaboration with

three experimental

groups!

Many body

pseudopotential

calculations

Calculated spectra

Spectral Barcoding: Why is it important?

XS-2 < XT

-2 < X-1 < XX0 < X0 sequence

in measured spectra from each and

every QD studied in the ensemble is

kept.

Quantum Dot

growth

Structural Characterization by X-STM

Single-dot Spectroscopy

Antonio Badolato

(ETH Zurich, Switzerland)

M. Bozkurt, J. M. Ulloa, & P. M. Koenraad

(TU Eindhoven, The Netherlands)

Theory

M. Ediger & R. J. Warburton

(Heriot-Watt University, UK)

Collaboration with

three experimental

groups!

Many body

pseudopotential

calculations

Calculated spectra

• Exciton energies

• XS-2 < XT

-2 < X-1 < XX0 < X0

sequence

?V. Mlinar, G. Bester, &

A. Zunger (NREL)

Vladan Mlinar et al., PRB 80, 165425 (2009).

XSTM→Theory→Spectroscopy Fails to Close Loop!

Structure

• Exciton Energies:

Calculated: 1.05 -1.12 eV

Measured: 1.08-1.09 eV

Vladan Mlinar et al., PRB 80, 165425 (2009).

XSTM→Theory→Spectroscopy Fails to Close Loop!

Structure

• Spectral Hard Rules:

All five XSTM deduced Model QDs

violate Spectroscopic Hard rules!

EXP. XS-2 < XT

-2 < X-1 < XX0 < X0

Model 1 XS-2 < X0 < XX0 < X-1 < XT

-2

Model 2 XS-2 < X0 < XX0 < XT

-2 < X-1

Model 3 X0 < XX0 < XS-2 < X-1 < XT

-2

Model 4 X0 < XS-2 < XX0 < X-1 < XT

-2

Model 5 XS-2 < XX0 < X0 < X-1 < XT

-2

Vladan Mlinar et al., PRB 80, 165425 (2009).

Structural motifs underlying Spectral Hard Rule:

Spectral barcoding

Procedure

INPUT:

Vladan Mlinar et al., PRB 80, 165425 (2009).

Structural motifs underlying Spectral Hard Rule:

Spectral barcoding

Procedure

INPUT:

OUTPUT:

Primary structural

Motifs

1. Height (h)

2. Base-length (b)

3. Average In

composition (XIn) Vladan Mlinar et al., PRB 80, 165425 (2009).

Spectroscopy→Theory→Structure closes the Loop!

Spectroscopy→Theory→Structure closes the Loop!

• More than one dot can be constructed!

• Spectral Hard Rules are satisfied by

the construction!Vladan Mlinar et al., PRB 80, 165425 (2009).

Conclusions:

Spectral Barcoding: Procedure for deciphering structural motifs from the multiexcitonic spectra

• We established missing structural basis for QD spectroscopy

• We offer spectroscopically-derived structural motifs that combined with

X-STM measurements give more realistic QD structure.

Thank you for your attention!

Vladan Mlinar et al., PRB 80, 165425 (2009).

Vladan Mlinar and Alex Zunger, PRB 80, 035328 (2009).

Basic Paradigm of Spectroscopy of Molecules

Structure

• To understand the spectra one must know the structure

(hence symmetry) of the molecule

• Structure-spectra relationship in molecules has historically been

facilitated by the accumulated knowledge on electronic and vibrational

spectral fingerprints of specific groups making up the molecules

• Deliberate design of molecules with given properties

Spectroscopic vs. Geometrical QD size:

Can we construct a model QD that has geometrical size as extracted from XSTM, but

spectroscopic size as deduced by spectral barcoding?

XSTM deduced Model QDs:

Model 1 Model 2

Model 5

Model 3 Model 4

• Truncated cone

• No wetting layer

• Truncated pyramid

• No wetting layer

• Truncated pyramid

• No wetting layer

• Ellipsoid

• No wetting layer

• Truncated cone

• Includes wetting layer