Indian Journal of Pure & Applied Ph ysics Vo l. 37. Novemher 19')9. pp. X23-X27

Thermal crystallization behaviour of Ge-Te-Se glasses

K V Acharya*. S Asokan ** & T S-Panchapagesan+

"' Department of Electronic Science and University Science Instrumentation Centre. Bangalore University . Banga lore 56

"'''' Department of Instrumentation. Indian Institute of Science. Bangalore 560 01 2

+Materials Research Centre. Indian Institute of Science. Bangalore 560 012

Received 20 August 1999: revised 29 October 1999: accepted I November 199')

GeloTe90_xSe, (50 :0; x :0; 70) and Ge2oTeso_xSe, (x = 30. 50) glasses have been prepared by melt-quenching. The rhenn,i1 crystallization behaviour of these samples has been studied by Differential Scanning Calorimetry (DSC). in order to characteri se these glasses for memory-threshold switching applications. It is found that GelO Te90_x Sex glasses have higher thermal stability and are more stable against devitri ricmioll . These samples may be suitable for threshold switching devices. GelO Teso_xSex glasses. on the other hand . phase separate on heating and exhibit a double stage crystallization. Based on this. it can be expected that Ge20Teso_xSex samples will show memory behaviour. The activation energy for thermal crystallization of a representati ve GelOTe40 Se50 glass belonging lO the GCloTe90_xSex series has been found by the Kissinger's method to be 0.92 eV. The value of the acti vation energy obtained also indicates that GeIOTe<Jo_xSe, sampl es are less prone to devi triri cati on and more sui table for threshold behaviour.

1 Introduction Differential Scanning Calorimetric (DSC) studies

he lp in understanding the glass transition and crysta l­li zation behaviour of amorphous solids. Further, the

thermal parameters of chalcogenide glasses , obtained from DSC experiments are useful for characterizing

these materials for diffe rent applications, especially for electrical switching appl ications I. For example, the vari­

ation of e lectrical switching f ie lds with composition of chalcogenide glasses is found to be direct ly re lated to

the composition dependence of crystalli zation tempera-, . . tu res-. Also, there eX Ists a correlatIon between the glass transiti on temperature and the switching field s of chal­

cogenide glasses'- Easy resettability of the memory state has been found to be intimate ly linked to the glass

forming abil ity of chalcogenides4 (estimated from ther­

mal parameters) . The thermal stability of chalcogenide glasses (given by the separation between the glass tran­si tion and crystalli za ti on temperatures) is found to be a dec iding fac tor, in dete rmining whether the sample will ex hibit melllory or thres ho ld behaviour5 Further, as the memory (irrevers ibl e) switching in g lassy cha\co­genides is thermal in nature, in vo lving the formation of conducting crystalline channe ls in the materia(', data on

the acti vation energy for thermal crysta ll ization is useful in selec ting a suitabl e mate ri al for memory app lications.

This paper deals with the differential scanning ca l­ori metric studies on the glass transition and the crystal­I ization behaviour of Ge-Te-Se glasses . In the Ge-Te-Se system, bulk, homogeneous glasses can be obtained in the composition range 10 to 30 atom percent of Ge, for varying Se and Te proportions7

.x, Preparation and ther­

mal characterization of Ge-Te-Se glasses of a few indi­vidual compositions have been reported earlier'J·III. In the present work, DSC stud ies have been undertaken on Ge-Te-Se glasses of different compositions , wi th an idea to find the su itab il ity of these materials for different switch ing app lications.

2 Experimental Details

Bulk , semiconducting Ge loTe90_xSe, (50 :s; x ::; 70), Ge2oTexo_xSe, (x= 30, 50) glasses have been prepared by the melt-quenching method . Constituent e lements in desired proportions are sealed in evacuated quartz am­poules ( 10-6 torr). The sealed ampoul es are heated to

temperatures around I 100°C, in a tubular furnace. The ampou le containing the Illolten e lements are rotated continuously at 10 rmp for about 48 hours, in order to homogenize the melt. Subsequentl y, the ampoules are quenched in a trou gh containing NaOH + ice-water mixture. The glassy nature of the samples is confirmed by X-ray diffraction. A Philips diffractometer with Cu Ka radiation is used for the X-ray studies.

X24 INDIAN J PURE APPL PHYS. VOL 37, NOVEMBER 1999

A Stan ton-Redcro ft Differential Scanning Cal orime­ter i>. used for the thermal in ves ti ga ti ons. About 20-2.'i mg of samp le is tak en in quartz c ru cibl es and alumina is u>.cd as the reference material. DSC studi es arc also undertaken as a functi on of heatin g rate on a repre­

sc ntat i ve samp le. to eva I uate the act i vat ion energy of crys tal I izati on.

3 Results and DisclIssion

. t I Class transition and cr~'s tallization hehaviour

The Di ffe renti al Scann ing Ca lorillletri c traces of

Ge lOT e,)().,Se, (50 S.r S 70) glasses , obttl ined ti t a heat i ng rat e of 15 deg/min , are shown in Figs I and 2. lt can be

seen fro m th ese Figs I and 2 that a ll the GeloTe9().xSex glasses studied exh ibit one glass tran siti on reac ti on on

heatin g. However. the sa illpi es show different crys tal­li zation behaviour , dependin g on the co mpos iti on;

GCIOTc~()Se~O sample ex hibits one exothermi c crystal­li zati on reaction and an endotherillic Ille lting peak above th e g lass transiti on (Fi g . I). On the oth er hand ,

GeloTc:1()Sc(,() and Gc loTc20Se70 sa mples do not ex hibit ~lIl y crysta lli zati on on heatin g (Fig. 2). These sa mples

melt direc tl y, after the glass tran siti on. Fi g. 3 shows the differenti al sca nnin g calorimetric trace o r a repre­sentati ve GelOTe20Se70 glass taken with a higher sensi ­ti vit y (heating rate = 15 deg/Illin) . It is ev ident frolll Fig. 3 that even with hi gher sensiti vity, it has not been poss ibl e to detect anv crvstallization in thi s Illaterial.

i ° ~ I o "0 C

W

TemperoturE'(°c )

Fig . I - DSC thermogram or GeloT c4oSeso g lass. Heatin g ratc = 15 dcg/min

ln the case of Ge loTe:;oSe(,O sample. the me lti ng occ u rs

around 2R5°C. For GeloTe2()Se70 sa illp le, th e Ille lting

sta rt s just around 625°C and it has not bce n possih le tn observe the melting compl etely as the temp rature lim it of the DSC in strument in the high sensiti vity Illode is

650°C. It is also interestin g to note that the glass transi­tion reaction in GeloTe20Se70 sample is sharper than that for other glasses.

The thermal behav iour of Ge2oTex().xSe, (.\" = 30, 50) glasses is d ifferent frolll that or GeloTe,)(}.,Se, glasses . Ge20Te'iOSe l() and Ge20Te ;OSe:,(} sa mple s ex hihlt tW() crys talli zati on reac ti ons. between the g lass tran sitioll and me lting reacti ons. Fig. 4 shows the DSC trace or the representat i ve Ge20 Te,oSe~() sa mple .

Telluride glasses whi ch ex hibit two crysta lli zat ion reac ti ons are usua ll y round to show two glass tran sit ions also. To observe the second glass transition, the sall1 plc :-,

~ w

° "0 C

W

~

GelD Te)J5e60

13 - l So/min

T9 = l07°C

T emperatu r e (OC)

Fig . 2 - Dirrercn ti al Scanning Ca lorimctri c traccs 0 1"

GCIOTc20SC70 alld GCIOT c:1oSc(,() g lasscs t,lkc ll

at 15 dcg/llli n

i o x

W

° "0 C

W

~

GIlO Te20Se70

n = 15°/min

Se nsi ti vi\y ~ lO O mV ifui! scole

Tg =113 .8°C

~ OL-J-~60--L--1L20~--1~80--L-~~~0 300

Temperat ure (DC)

Fig. 3 - DSC thermogram or GC loT c 2oSC7n g L1s~ at highcr sensiti vi ty (hcatin g ratc = ') deg/min)

... f

ACHARY A et (fl. : Ge-Te-Se GLASSES 825

i l w I � I

o

GE'20T�SE'so fl = ISoClmin

t 19 =101.6°C

lOa

Tel = 365°C

200 300 400 Temperatu rE' (oe )

500

Fig. 4 - Differential Scanning Calorimetric trace of Ge:wTe:1oSeso sample. Heating rate = 15 deg/min

are heated to the end of first cry stal l i zat ion reaction,

quenched to ambient and reheated I I . G el0Te30Se50

glasses also exhibits the phenomenon of double glass

transition and double stage crysta l l izat ion, when they

are quickly cooled to room temperature and reheated i n

a DSe experiment l 1 . The present DSe results indicate that the glass tran-

sition temperatures of GeJOTeqo-xSex g lasses (50 S x s 70) i n creases w i t h t he compos i t i o n x. A l so ,

Ge lo Te30Se6() and Ge l oTe20Se71l g lasses do not crysta l l i ze

on heat ing. Even for the Ge lOTe40Seso sample, the crys­

tall ization temperature is comparat ively h igher ( 1 98°C).

Also, these samples have a h igh melting temperature. It

can be concl uded from the above that the GeIOTe90-xSe.

samples ( 50 s x s 70), have a better thermal stab i l ity and

hence w i l l be more usefu l for threshold switching appl i ­

cations . GeloTexo-xSex (x = 30, 50) glasses, which phase sepa­

rate on heat ing and crysta l l i ze at lower temperatures

( I OOOe) are more su itable for memory switch ing app l i ­

cations . In th is context, i t i s in teresting to note glasses

as Al-Te and Ge-Te, which phase separate and exhibit a

double stage crysta l l ization, also show memory switch-. 3. 1 1 . 1 3 IIlg

3.2 Activation energy for crystallization

DSe stud i es have been u ndertaken on a repre­sentat i ve Ge lOTe�o Se,\() g lass at d ifferent heat ing rate, to evaluate the act ivation energy for crysta l l i zation . Figs 5 and 6 give the differential calorimetric thermograms of Ge loTe40Se50 sample, taken at 1 5 , 20, 25 and 3() deg/min heali ng rates . The act ivation energy for crystal l i zation Ec can be estimated using the Kissinger' s method I.!, from the fol lowing relation :

d [ I n ( �/Te 2) ]/d [ l iT, 1 = - E, IR where R is the gas constant; Te the crysta l l i zat ion

temperature; � i s the heat ing rate. The plot of In ( �/T, 2) versus l iTe i s a straight l i ne, the s lope of which g ives the activation energy of the crystal l i zation ( E,.). Fig. 7 shows

the variation of In (BIT. 2) with 1 000/T. for a reore-

i a x

U1 a

u c

U1

�

o

GE'lO TE'L,()SE'50 fJ = 1 5°/min

GE'1O TE'LjJSE'50 fJ = 20o/min

60 120 180 Temperature (oe )

240 300 360

Fig. 5 - DSC thermogram of GeloTe40Se�iO glass taken al 1 5 deg/min and 20 deg/min heating rates

826 INDIAN J PURE APPL PHYS, VOL 37, NOVEMBER 1999

GE'lOTef.OSE'50 Tc = 20aoC

.rJ=2So/'Tl in

i 1 0 )(

w

0 Gel:) Tel,()SE'50 Tc =21l.oC u

c ill fl =30'"/min

I ,

21.0 300

Temp era ture ( 0C)

Fig. 6 - DSC thermogram or GelOTqOSe50 glass taken al 25 cleg/min and 30 cleg/min heati ng rates

sentati ve GeloTe40Seso sampl e. The crystalli zation ac ti­

vati on energy of thi s sample has been es timated fro m

Fig. 7 as 0.92 eV. The crystalli zation acti vati on energy

of other glasses in the Ge loTe')().xSex series, hav ing

hi gher Se content.. will be even hi gher. Memory sw itch­

ing glasses such as As-Te, usuall y have the crystal-

I·· . . . I 11'10 1 0'" V Izat lon acti vati on energy In t le range-" . -.. _' e .

The comparat ively larger acti vat ion energy (0.92 e V) of

Ge ioTe .IOSe 50 g lass also corroborates the idea that

Ge loTet)(l.,Se, glasses are less likely to show memory

switc hing and more su ited fo r threshold applicati ons.

8.8 GE',:) Tet,oSE'50

9.0 0

0

Nu e 9.2

~

c: 0

- 9.4

~J 9.6

9·8 2.02 2.04 2.06 2.08 2.10 2~2

looo/Te (-1<:')

Fig, 7 - Ki ssinger's plot [In (~/ Tl' " ) versus I OOO/Tl' I or Ge IOTe~()Seso glass

4 Conclusion

Bulk , semiconducting GeIOTe')().xSex (50 $ x $ 70) and Ge20 Te8().,Sex (x = 30, 50) glasses have been prepared by melt-quenching. Di ffere nti a l Scanning Calorimetric studies on the thermal crystalli zat ion behaviour ind icate that GeIOTe<)o.xSex glasses have higher thermal stability and are more stable aga inst dev itrificati on. These sa m­ples may be more suitabl e fo r threshold sw itchin g appli­ca ti ons. The hi gher ac ti va ti on energy fo r therma l crysta lli zat ion ofGeIOTe~oSe5() glass (0.92 eV) al so sup­ports the idea. Ge2D Tes(l.xSex glasses, on the other haneL phase separate on heatin g and ex hi bi t a double stage crystalli zation. It may be ex pected thai these samples will show memory behav iour.

References 1 Fri tzsche H in AII/orphous (lilt! Liquid Sell/icondllctors. Ed.

2

3

4

Tauc, (Plenum Press. New York ). 1974. p. 3 13 ,

Titus S S K. Chatterjee R. Asokan S & KUIll;lr A. P/lI 'S Nel' II 48 ( 1993) 14650.

Pr;lk ash S. Asokan S & Ghare D B. Sell/i, 'o//(/ Sci T"dl. (J ( 1994) 14X4.

Prakash S. Asokan S & Ghare D B . .I /'h.1's Jj II/i/ i/ /' /t.1's . :2~ (1996) 2()04.

M urugavel S & A~okall S . .I Matl'r Res . 1:1 (19~X) 2<JX2.

ACHARYA et al.: Ge-Te-Se GLASSES 827

Binenslock A I. Bales C W & Spicer W. Ed S/il:'cial l.rs lle Oil

AII10 llilIOlIS Sell1 icollduc tor DC' I'ices. IEEE Trails Elec troll 0 1'­I'in ' s. ED- 20 ( 1973).

7 Muir.l & Cashmen R J . .I O/i Soc AlI1erica. 57 ( 1967) I .

R Pazin A V . Obrazlsov A A. Zubori sovc & Balmakov MD. I II00X Mater . 13 ( 1977) 15H I .

9 Ohrazl sov A A. !lIorg Mut('/'. H ( 1972) 374.

I () Sarraeh D J & Dc Neui"vil le .l P . .I NOII -Cnst Solids. 27 (1978) 193.

II Asokan S & Gopal E S R. ReI' Solid Stute Sci. 3 ( 1989) 273.

12 Acharya K V & Asokan S. (to be published).

13 Babcnskas E. Balcvi cius E S. Cesnys A. Poskus A & Sik lOrov

N. J NOIl-Cryst Solids . 90 ( 1987) 60 ~ .

14 Ki ssinger H E. J Res NBS. 57 ( 1956) 2 179.

15 Tit us S S K . Asokan S & Gopal E S R. Solid State COII/ll/WI.

83 ( 1992) 745.