How do rewritable DVDs work? · How do rewritable DVDs work? Optical properties of phase-change materials W. Wełnic,1,2 Silvana Botti,2,3,4 L. Reining,2,4 and M. Wuttig1 1I. Physikalisches

How do rewritable DVDs work?Optical properties of phase-change materials

W. Wełnic,1,2 Silvana Botti,2,3,4 L. Reining,2,4 and M. Wuttig1

1I. Physikalisches Institut IA, RWTH Aachen University, Germany

2Laboratoire des Solides Irradiés, CNRS-CEA-Ecole Polytechnique, Palaiseau, France

3Centre for Computational Physics, University of Coimbra, Portugal

4European Theoretical Spectroscopy Facility

June 27, 2007 – Queen’s University, Belfast

Silvana Botti (Paris) How do rewritable DVDs work? Belfast 1 / 30

Outline

1 Introduction to Phase Change Materials (PCMs)

2 Open questions for Phase Change Materials optimizationLocal structural changes upon amorphizationOptical spectra: from crystalline to amorphous

3 Answers to the open questions


Introduction to Phase Change Materials (PCMs)

Outline






Need for next generation memories

Phase Change Materials (PCMs) are already employed in optical datastorage, e.g. in rewritable DVD’s.The resistivity contrast could be used in phase change random accessmemories.



Writing, reading and erasing a DVD



Reversible phase transitions in PCMs

GeSbTe alloys:pronouncedstructuraldifferencesrapidtransitions( 10-100 ns)large opticaland electriccontrastalreadyapplied butstill not wellunderstood



Optical spectra of semiconductors

GaAs: absorption in crystallineand amorphous phase similarGeSbTe alloys: strong opticalcontrast

GeSbTe alloys are suitable for data storage!


Open questions for Phase Change Materials optimization

Outline






Local order of GeSbTe alloys in the amorphous phase

Ge in redTe in blue

Experiment: a fraction of Gebecomes tetrahedrallycoordinatedSuch a phenomenon hasnever been observed forconventional covalentsemiconductorsContradiction to continuousrandom network modelCorrelation between localorder and optical properties?

Kolobov et al. Nature Materials 3, 703 (2004)



Open problems for PCMs optimization:

1 Can a change in local order explain the strong optical contrast?2 Is this contrast due to a change in the electronic energy levels?3 Can we tune the optical contrast?


Open questions for Phase Change Materials optimization Local structural changes upon amorphization

Crystalline structures of GeSbTe alloys

Rocksalt - octahedral coor. Spinel - mixed coor. Chalcopyrite - tetrahedral coor.

Te in gray, Sb in blue, Ge in red

Rocksalt structure relaxes in slightly distorted state



DFT total energies

Spinel (mixed coordination) suitable model for amorphous short rangeorder

volume increase 5-10%energy difference 30 meVincrease of bulk modulus

W. Welnic et al. Nature Materials 5, 56 (2006)



From experimental observation to our models

If all Ge atoms were in tetrahedral positions the volume increasewould be ≥ 30% (experiment: 5-10%)Models for amorphous GeTe:

64 atoms supercelloctahedral coordination except for 2, 4 or 8 Ge with tetrahedralshort range order

Models for amorphous Ge1Sb2Te4:56 atoms supercelleither 4 or 8 Ge with tetrahedral short range order

All models still exhibit long rang order!


Open questions for Phase Change Materials optimization Optical spectra: from crystalline to amorphous

Optical absorption: approximations within DFT

v

c

unoccupied states

occupied states

�2(ω) = 24π2

ΩNkω2limq→0

1q2

∑v ,c,k

∣∣mv ,c,k∣∣2 δ(�ck − �vk − ω)mv ,c,k = 〈c |q · v| v〉

Independent-particle → GW → e-h interaction



Linear response functions within TDDFT

V

Vind

ext

macroscopic and microscopic

macroscopic

χ̄ = χ0 + χ0 (v̄ + fxc) χ̄

εM = limq→0

[1 − v0(q)χ̄00(q, ω)]

Independent-particle (v̄ = 0, fxc = 0)RPA (fxc = 0)

TDLDA (fxc =dV LDAxc

dρ δ(r − r′))



GeTe: experiment vs. RPA calculation

0 1 2 3 4 5 6Energy [eV]

20

40

60

80

ε 2

GeTe rocksaltGeTe amorphous

0 1 2 3 4 5 60

20

40

60

ε 2

GeTe rocksaltGeTe amorphous

EXPERIMENT

THEORY

Main features of the opticalcontrast well reproducedSpectroscopy onpolycrystalline thin films ofGeTePresence of Ge vacancies inthe sampleOur models still exhibit longrange order

Experimental results obtained in RWTH Aachen University



GeTe: effect of distorsion and Ge vacancies

0 1 2 3 4 5 6E [eV]

20

40

60

80

100

120

ε 2

c-GeTe (undistorted)c-GeTe (distorted)c-GeTe (Ge

V)

a-GeTe

Distorsion: no more 6equivalent neighboursGe Vacancies:

Drude peak wellreproducedIntensity of the mainpeak decreases

Distorsion andvacancies reduce thenumber of Ge-Te bonds

Calculations within RPA approximation



TDLDA, GW & BSE

TDLDA and RPA give the same resultBetter agreement with experiment using many-body calculations?

0 1 2 3 4 5 6Energy [eV]

20

40

60

80

100

120

140

160

180

ε 2

RPAIPGW-RPABSEEXP

GW shifts gap by 0.15 eVSmall excitonic effectcompensate for GW correctionRPA is enough for ourpurposes!



GeSbTe: experiment vs. RPA calculation

0 1 2 3 4 5 6Energy [eV]

20

40

60

80

ε 2

Ge1Sb2Te4 rocksaltGe1Sb2Te4 am. md. 1Ge1Sb2Te4 am. md. 2

0 1 2 3 4 5 60

20

40

60

ε 2

Ge1Sb2Te4 rocksalt

Ge1Sb2Te4 amorphous

EXPERIMENT

THEORY

again main features of theoptical contrast wellreproducedwe conclude that a change inlocal order can explain thestrong optical contrast

Experimental results obtained in RWTH Aachen University




1 Can a change in local order explain the strong optical contrast?2 Is this contrast due to a change in the electronic energy levels

(JDOS)?3 Can we tune the optical contrast?



Optical absorption

v

c

unoccupied states

occupied states

�2(ω) = 24π2

ΩNkω2limq→0

1q2

∑v ,c,k

∣∣mv ,c,k∣∣2 δ(�ck − �vk − ω)mv ,c,k = 〈c |q · v| v〉 velocity matrix elements

JDOS/ω2 =1

Nkω2∑v ,c,k

δ(�ck − �vk − ω)



Origin of the optical contrast

No significant changes inJDOSWith constant matrix elementsone wouldn’t expect anyoptical contrastOptical contrast is determinedby changes in transition matrixelements



Understanding the optical contrast

0.4 0.5 0.6 0.7 0.8overlap

0

200

400

600

800

num

ber

of tr

ansi

tions

Ge1Sb2Te4 rocksaltGeTe rocksaltGe1Sb2Te4 amorph.GeTe amorphous

Changes in local structure cause changes in spatial dispersion ofwavefunctionsOverlap changes → matrix elements change



Contribution from distorsion and vacancies

Decrease of coordination and inclusion of vacancies: reduce thenumber of Ge-Te bondsDecrease of coordination and distorsions: reduce the overlap ofthe wavefunctionsOptical properties in PCM can be tuned by modifying thesecontributions!


Answers to the open questions

Outline






Conclusions

Calculation of optical properties of PCMs from first principlesOptical contrast in PCMs understood!Change in local order origin of optical contrastOptical contrast governed by changes in spatial overlap ofwavefunctions which lead to smaller transition matrix elementsThe model explains how to change the optical contrast by tuningthe number of Ge-Te bonds (vacancies, distortions, composition)



Acknowledgments

Ground state and GW calculations: www.abinit.org/

TDDFT for periodic systems: theory.polytechnique.fr/codes/

BSE for periodic systems: theory.polytechnique.fr/codes/

The authors acknowledge funding by the NANOQUANTA NoE, theBMWA, and the EU “Marie Curie Host Fellowship”


www.abinit.org/theory.polytechnique.fr/codes/theory.polytechnique.fr/codes/


Thanks to W. Welnic, L. Reining and M. Wuttig!

W. Welnic, S. Botti, L. Reining, and M. Wuttig,Origin of the optical contrast in phase change materialsPhys. Rev. Lett. 98, 236403 (2007).

W. Welnic, M. Wuttig, S. Botti, and L. Reining,Local atomic order and optical properties in amorphous andlaser-crystallized GeTeto be published


Introduction to Phase Change Materials (PCMs)Open questions for Phase Change Materials optimizationLocal structural changes upon amorphizationOptical spectra: from crystalline to amorphous


Documents

How do rewritable DVDs work? · How do rewritable DVDs work? Optical properties of phase-change materials W. Wełnic,1,2 Silvana Botti,2,3,4 L. Reining,2,4 and M. Wuttig1 1I. Physikalisches