Mg Alloy Dev Guided by Thermo Calculations

MAGNESIUM ALLOY DEVELOPMENT GUIDED BY THERMODYNAMIC CALCULATIONS

Joach~m Grobner. Dmytro Kevorkov, Rainer Schmid-Fetzer

Technical University of Clausthal. Institute of Metallurgy, Robert-Koch-Str. 42

D-38678 Clausthal-Zellerfeld. Germany

Abstract

In tradit~onal alloy development. expenmental investigations mith

many different alloy compositions are performed. The selection

criter~a for multicomponent alloying elements and the~r

cornpos~tions become diffuse in a traditional approach.

Computational thermochemistry can provide a clear g ~ ~ i d e l ~ n e for

such selections and helps to avoid large scale experiments with

less promising alloys. Thus. ~t is a powerful tool to c ~ ~ t down on

cost and time during development of Mg-alloys. As an example.

me report on recent developments of nem, creep resistant alloys

that show about 100 times less creep than the best con~mercial

alloys. Also outlined are the methods we use in our long-term

project of construction of the necessary therrnodynam~c database

for several prom~sing alloying elements like Al. Li. SI. Mn. Ca.

Sc. Y. Zr and Rare Earths. using the Calphad method cornb~ned

~ ~ t h key experiments.

Introduction

There is an urgent need for the development of new or improved

magnesium alloys if \4e want to fully exploit the potent~al of this

t'asc~nating lightue~ght mater~al that also offers excellent

cxstah~llty, ma china hi lit^ and b~o-compat~bility. Experiments on a

technological scale f o r preparation and testing o f new a l l q s are

\er> eupensl\e and time consuming. In vie^ of the huge number

oi' posvhle allo) components. compositions and processing

parameters. one \\auld like to hale at least an "educated guess" in

which direction to go. In this report we want to s h o ~ that

thermodynamic calculations can provide much more than that.

Computational thermochemistry is a modem tool that supplies

quantitative data to guide the development of alloys. It enables the

calculation of rnulticomponent phase diagrams and the tracking of

individual alloys during heat treatment or solidification by

calculation of phase distributions and phase compositions. These

are the basic data to understand and control the behavior of any

novel or modified Mg-alloy. Large-scale experiments for new

multicomponent alloys can then be focused on most promising

alloys identified in that approach. Long-term experiments with

less promising alloys can be avoided. Thus, it is a pom'erful tool to

cut domn on cost and time during development of magnesium

alloys.

Database Develournent

In the core of this method a h ~ g h quality thermodynamic database

for Mg-alloys is needed. Development of a reliable

thermodynamic database for multicomponent alloys requlres a

combination of experiments and computational thermochenlistrq

with data Prom alloy application. Since numerous binary and

ternary subs)stenis have to be treated and validated \v~th key

experiments before multicomponent alloys can be calculated

reliably. t h ~ s development becomes a long-term project. In our

group at the TLI Clausthal a thermodynamic database for seberal

alloying elements like Al. Li. Si. Mn, Ca. Sc. Y. Zr and Rare Earth

elements (Fig. I ). has been under construction for more than fi \e

bears and is still ongoing.

Magnwum Technologj 2001 Edited by J Hr?n

TMS (The Mnerais. Metals & Materials Society). 2001

I Creep resistance - Density reduction I ' \

Fig. I: Currently selected alloying elements in the Mg-database.

To create the thermodynamc phase descriptions the Calphad

method is used. The principle of this method IS shoan in Fig. 2

Thermodynamic Modelling

Fig 2 Schemat~c approach of database development and the Imks to

appl~catlons

In a multicomponrnt, niult~phase system each phase is described in a

suitable model for the G h b s energy. The parameters are optimi~ed

by avmlable expenmental data integrating both thernwdynamic and

phase equil~bnurn data. Tn_ls data set is used to calculate stable and

neta astable equi l~hr~a for process simulat~on or direct applmt~ons.

The calculat~ons are ~ m p r o ~ e d cont~nuously by new espcnrnental

data coming from apphcations nnd are checked hy key evperlments

to improve their r e l ~ a b ~ l ~ t y . An example for process simulat~on 1s

the coupl~ng u ~ t h 3D-sol1d1ficat1on n~odel~ncp. Example\ for d~rect

applicat~ons are given in the followng chapters.

In Mg-system5 the exper~mental l ~ t e r ~ t u r e data \+ere \ery q w s e

In both. phase d~agram data and the rmod~nam~c proprrt1e5

Therefore, seheral exper~mental methods are b a n g applied in our

group to produce a sufficient exper~mental database (Fig. 3) .

Experimental Methods

Sample Preparation Levltatlon Meltlng Arc Meltlng Electron Beam Meltlng Reaction Slnterlng

Sample Analysis ThermalAnalys~s (DTA,STA) X-ray Dlffractornetry ( X R D , 25-1500" ) Electron Mlcroscopy (SEMIEDXIWDX) Metallography, Opt~cal Mlcroscopy

Fig. 3: Experimental methods.

Our ao rk focuses on ternary and quaternary systems for

~mprovement of creep reslstance [ I ] , thermal s t a b ~ l ~ t ) zpra?

forrmng (Mg-Mn-Sc-RE. Mg-AI-Ca-RE) and density reduct~on

(M~-LI-AI-SI. Mg-AI-Ca-Si) [2. 3, 4. 5 . 61 The thermodynamc

and technical application of the temarj Al-Mg-Sc 15 alreddb

discused In l~terature [7, 81

Here the quaternary systems Mg-Mn-Sc-Gd and Mg-Mn-Sc-Y are

shown as example for the selection of new creep resistant alloys

uslng computational thermochem~stry.

First generation of creep resistant ternary hIg-Mn-Sc alloys

In~est~gat ions started w t h binary Mg-Sc alloys. Scandium \\as

chosen for precipitation hardenmg because of its large solub~lity

In (Mg) and the retrograde solubility at louer temperatures. The

b~nary phase d~agram had to be re-investigated (91. it is s h o ~ n in

Fig. 3. Ho\+ever, hinary MgSc precipitates form Lery s lo~. l !

dur~ng ageing and improve the mechanical properties on14

slightly because of their incoherent interface. Therefore. Mn uas

added as second alloy~ng element. The precipitatmn of Mn&

\+as predicted by thermodynamic calculat~ons Mn&

precip~tatmns form coherently and uere found very useful for

~rnproi ing creep reslstance and hardness. Neu, MgSc l5Mn l or

MgSc6Mnl alloys shou about 100 tlmes better creep resistance

than the best commercial IY E43 alloy at 350°C and 30 hIPa [ I ]

In spite of the good properties of t h ~ s first generation of h,lg-hln-

S c alloys. the high cost of S c add~t lon ( 6 wt.'? S c o r more)

~ n ~ t ~ a t e d a search for a second generation by i n v e s t ~ g a t ~ n g

quaternary systems.

At this point the questlon arose. "ho14 to ldent~t:, pronilslng a l l q

cand~dates from all these calculated diagrams." What phase

diagram features are related lo what alloy processing steps .' Whal

is needed is a list of beneficial phahe diagram feature\. d r r ~ ~ e d

from the relevant alloy processing steps. The most Important

polnts are g l w n In Table I

0 l I I I I I I

0 2 0 4 0 6 0 80 1 0 0 M g at . O 6 S c

s c

FIE 1 Thc b ~ n a ) Mg-SL \)\tern atter [9]

A d d ~ t ~ o n a l a l loyng elements Gd. Y and Zr were considered for

t h ~ \ purpose to achieve a sufficient quantity of suitable

preclpltatcs to Improve rncchan~cnl propertics u u n g n mlnlmum

(>I' eupensi\e alloy element add~t ion . These elerncnt comb~nat ions

hlg-MII-(Sc. Gd. Y. Z r ) f o ~ m a i a n e t y o l quatcrnar) systems and

thin tho\c there is a huge amount o f p o s s i h i l ~ t ~ e ~ to select allog

compo\itions. Therefore phase d~ayrai i i and other cillculatlons

\\?I-e performed to ~ d e n t ~ f ) p ro in i~ ing c a n d d a t t s

Ulo\ selection in the 51~-\In-Sc-(;d ,\stem

5 0 G d 0.2 0.4 0.6 5 0 G d

1 O M n W t . % S C 1 O V n 9 4 0 M g 9 3 0 M g 0 0 S c 1 osc

F I ~ . 5: Phase d ~ a g r a m section n ~ r h constant I n t i; Mn. 5 U I

Table 1: Beneficial phase equilibrium features and their

relevance for (5Ig)-alloy processing

I Phaw diagram feature I Kele\ance for alloy

Larpe t.nougti c Alp) 51nglc- Enable\ Iiomogeni/ation plia\e field ailnealing

For an alloy ~vith I wt.Q Mn. 5 nt.% Gd and 0 8 ut. '? Sc

(indicated by arrow In Fig. 5 ) equ~ l ib r~um phase amoilnts during

sol~d~fication and heat treatment are piken In Fig. 6. At the

I ~ q u ~ d u s temperature 650°C. prlrnary (Mg) is formed and

consumes the melt totally up to the sohdus polnt at 6 1 9 T At

590°C the first prcclpltate Mn2Sc starts forrmnp, which can he

seen In the enlarged Insert in Fig. 7. Large amounts of the second

preclpltate Gdhlg, start forming at 325'C. This alloy fulfills all

features illustrated In Table 1 and was classified as ver)

promising for further alloy development. Similar features are

ohserved for an alloy with even less scand~um (0 3 wt.G Sc) .

In h c t , first results for this alloy shows a creep resistance similar

to the prevlous ternary high-scandium Mg-Mn-Sc alloys, about

100 times better than best commercial WE33 alloy at 350cC and

30 MPa as detailed later.

temperatures. In the whole ranye ot primary cr) \ ta l l~/a t~oi i oi

(My) the secondary phase is Mnl lY. The des~red LIn,Sc

preclpltate f'orms only at temperatures he lo^ 500'C. wh~cti

decrease ~ ~ t h decreas~ng Sc c o n m t .

0 1 I I I

0 0 2 0 4 0 6 0 8 1 0

P h a s e amoun t [mo l l

Fig. 6 : Phase amounts of MgMnIGdiSc0.8 alloy

Alloy selection in the hIg-Mn-Sc-Y system

In the Mg-Mn-Sc-Y system heveral ~er t ica l \ectlon\ In the I-anyer

01'0-1.5 L I ~ . ' / ; Mn. 0-10 wt.'; Sc and 0-10 ut . ' i Y \\ere stud~ed

F I ~ 7 ihous a T-x sectlon of a calculated cjuaternar> phase

e q i ~ ~ l ~ h r i a ~11th constant I u t C; M n . 5 \\t.':; 1' from 0- I \ i t . ' ; Sc.

A large one-pha\e ticld o f ( M g ) and se\e~-a1 d~ffcrcnt ro l~d p h a m

\table at Iouer ternpcratures can be seen, like in the con-espond~ng

d~agrani f'or (id. Diftercnces cornpal-cd to the Gd \)stern can he

ohwr\cd concernln? the htablc so l~d phases and the forniat~on

Fig. 7: Phase d~ayrani section ulth constant I ~ t . 7 4 hln. 5 L I ~ ' i Y

from 0- 1 w t . 5 Sc.

200

MgSc 0

liquid

Mg24Y5

0 06 0.08 0 1

0 0 2 0 4 0 6 0 8 10

Phase amount [moll

Fig 8 . Phase anioilnts of LIgLln I1'5Sc0.8 allo!

Pol- an alloy \ \ ~ t h I \\ t ' i ivln. 5 i \ t r'i 'r' and 0.8 M t ' i Sc

I ~ n d ~ c a i e d b) arrou In Fi?. 7 ) phaw amounts durlng soiid~licat~on

ai. gl\cn In F I ~ . 8. Xr [he 11qu1du\ temperature 644°C. prirnar!

I 1 1 ~ ) 1s formed and consumes the ~ ~ i e l t totally up to the sol~dus

po~nt at 613-C. The Mn12\r' phase would onl! be stable In the

\()I111 \tat? In a high temperature range from 605 to 322°C and

n i~ ! not form at all du r~ng fast cool~ng. The estahllshcd Mn2Sc

and the ncu hlg24'r'5, eien In \cry large amount, could form by

agtlng in a favorable temperature range. T h ~ s makes

LlyhlnI Y5ScO.S also a promis~ng alloy.

In 01-der k-r checA if the amount of MnJc could he raised h j

Increasing the manganese content to 1.5 u t ', u e can examine

F I ~ 9 J'or the alloy MpMnl.SY.5Sc0.8. Thls alloy I S d~squalilied

LIncc Mn2Sc forms a j the prlrnar) phase from the I ~ q u ~ d . clearly

wen In thc Inset of F I ~ . 9. (Mg) fornis only secondar!. and e \en

Lln,:\r' limns as a terliary phase du r~ng solld~fication. Such a

nilcrwtructure cannot be "repalred" h! anneal~ng

cn 0 0 0 2 0 0 4 0 0 6 0 0 8 0 . 1

0 1 I I I

0 0 2 0 4 0 6 0 8 1 0

P h a s e a m o u n t [ m o l l

i l lo, \election in the \lg-hln-'l -%I. \\stern

u~bstlt~ltlon can he srudled In Flg. 10 for scandli~ni frt'c ;lllo)<

iuth 1 ~ 4 t . q ' Mn. 4.5 i 4 t . q Y and 0-1 u t . ' i Zr. T h ~ s phase

d~agrarn sectlon sho\<s a ver) steeply rlslng I ~ q u ~ d u s lint. l000'C

are reached for less than 0 .1 wt.% Zr. Moreover. and actually the

reason for the steep liqu~dus line. a huge primary crystali~zat~on

tield. L + Mn2Zr. stretches over the entire composition range In

Fig. 10 Only for extremely small Zr-add~tions. not d~sccrnible In

Fig. 10. is a prlmary (hlg) sol~diiicat~on expected The reason for

thls destructi\e phase dlagram f e a t ~ ~ r e I S the extremely high

thermodynamic stabihty of .MnzZr in cornparlaon to the other

phases. Since yttrlurn does not play a significant role In that part.

the only \\,a)' to dimrn~sh the L+Mn2Zr prlmar} ficld ~~l.ould he ;I

drastic reduct~on of the manganese content. But this \\auld also

drastically reduce the amount of beneficial Mn-conta~n~ng

preclpltates. As a result. the entire quaternary Mg-Mn-Y-Zr alloy

\) ztem 1s disqualified.

F I ~ . 10: Phase d l a ~ r a m sectlon wtli con,tant I \ \ t c; Xln. 5 \ \ t

1' lrom 0- I \ 4 t 3 Zr.

- 2 -

111t goal ofdrclaticall)~ rrduclng the Sc-contc~il c m he ac~cc~rnpl~hcd

More cletails of alloy selection In the q~iate~nary s) \ kms 112-kln-

.Sc. Gd. Y. Zrj \ \11l he publishecl \eon [ 101

L t ( M g ) + M n ' Z r

Comparison to commercial creep resistant Mg alloys Alloy preparation and creep resistance measurements

In order to assess the results on the creep resistance of the selected

new alloys we have to choose a commercial alloy as a benchmark.

The history of commercially developed creep resistant alloys is

briefly summarized in Fig. 11.

Alloy Composition (mass%) Year Zn A g Y R E

- most of these alloys conlain as gram refmer about 0 7% Zr - R E , the inf luence decreases In the order Nd (Pr) > C e > La

Fig. 1 1: Commercial creep resistant Mg-alloys

Alloys like AE42 or AS21, known to be more creep resistant as

standard AZ or AM alloys, are not included in the listing of Fig.

I I since they rank below the ZE alloys. The current end point of

commercial alloys is given by the WE series containing Y and

Rare Earths. In fact, the maximum stress that can be tolerated at

200°C for 100 h and 0 . 2 6 elongation is highest for the WE series

as seen in Fig. 12. The alloy WE43 with T6 heat treatment was

chosen as the benchmark alloy.

185 181 Lord 100 h, 200'C

WE54 18 WE43 16 OE22 18 ZE41 T5 A281 T4

Alloy

Fig. 12: Creep properties of commercial magnesium alloys.

The most promising alloy compositions identified by the

thermochemical calculations were prepared by squeeze casting by

our project partners within the Thrust Research Project SFB 390.

"Magnesium Technology", at the TU Clausthal. Yield strength

and creep rates of these alloys were measured in the as cast

condition and also after heat treatment [I I ] . It is clearly seen from

Fig.13 that the secondary creep rate of our first generation (high-

scandium) alloys is better by a factor of 100 compared to the

WE43 benchmark alloy.

T=350°C, -30 MPa 1 .SlI

time [s] Fig. 13: Creep curves of first generation (high scandium) alloys

compared to the WE43 benchmark [I 11.

The first generation of alloys showed a strong annealing response

due to the formation of the Mn,Sc precipitates. The existence of

the MnzSc precipitates was confirmed by X-ray diffraction. SEM

and TEM investigations and energy dispersive X-ray

microanalysis. A micrograph showing finely dispersed Mn&

precipitates in a MgSc6Mnl alloy after T5 treatment is given in

Fig. 14.

Conclusions

Fig. 14: Micrograph of MgSc6Mnl T5 alloy.

The second generation (low-scandium) alloys were selected from

the most promising candidates identified by the thermodynamic

calculations given in the previous sections.

The creep curves for these low-scandium quaternary alloys are

shown in Fig. 15. Again, an almost 100 times lower creep rate is

achieved at 35OoC and 30 MPa compared to the best commercial

alloy WE43. The best alloy MgGdSMnlSc0.3 with the smallest

elongation after 1 5 . 1 0 ~ ~ contains only 0.3 w t . 8 of expensive Sc.

T=35OoC, -30 M P a

'O time [ lo4 s] l5

F I ~ 15: Creep curves of second generation low-scandium alloys

compared to the WE43 benchmark.

This I S substantial progress compared to the first generation of

alloys with 6 or 15 wt.% Sc. Further work on the selected alloys

is in progress.

Focused magnesium alloy development is now possible

using the powerful tool of thermodynamic calculations.

Alloy compositions with promising possibilities of alloy

microstructure design can be selected by means of

thermodynamically calculated phase diagrams, phase

amount charts and solidification curves. Most importantly,

element combinations and compositions with unwanted

properties can be recognized before starting large-style

experiments, thus reducing the experimental effort to a

reasonable volume. The next step, the experimental study

of mechanical properties of identified promising alloys has

shown excellent results both in the first generation

(ternary) and second generation (quaternary) of new creep

resistant alloys.. Obviously, these experiments cannot be

replaced by thermodynamic calculations. However,

considering the huge number of less promising alloy

combinations that could have been selected from

multicomponent systems, the focused alloy development

following this approach avoids a waste of time and effort.

Acknowledgement

This work is supported in the thrust Research Project SFB 390:

Magnesium Technologya by the German Research Council

(DFG).

References

1. F. von Buch. B.L. Mordike, A. Pisch. R. Schmid-Fetzer.

"Properties of Mg-Mn-Sc-Alloys", Material Scie. & Ene. A 263

(1999). 1-7.

2. U. Draugelates. B. Bouaifi, S. Steinbom. R. Schmid-Fetzer. A.

Pisch, "Simulation of Laser beam welding of Magnesium alloys".

First Israeli international Conference on Magnesium

Science&Technology. Dead Sea, Israel. November 10- 12. 1997.

3. .A. P~sch. R . Schmld-Fet7er. F. \ . Buch. B L. Mordrhe. P.

Juchniann. F.-W Bach. H. Haferkanip.: "Computer aded deslgn

of novel Mg-allo\s". In M a ~ n e s ~ u n i Alloys and the11

Applrcations. (B.L. hlordtke. K.L. Kralner ieds.) . Inteniat~onal

Magnesium Conference - Wolfsburg. April 28-30, 1998.

Ll.'crkstoff-lnfc~r~nattonsgesellschaf rnbl I. Frankfurt ). 18 I - 186.

4. A. Pisch. R. Schmid-Fetrer. G. Cacclamanl. P. R~ani. A.

Saccone. R . Ferro. "Mg-rlch phase equ~ l~br i a and the rmod~nam~c

assessment of the Mg-Sc system". Z Metallkd. 89 ( 7 ) (1998).

171-177.

5 . A. Pisch. R. Schrnld-Fetm. "Experimental Stud) and

Thermodynanilc Assessnient of Phase Equilihria in the Mn-Sc

System". Z. Metallkd. 89 (10) 11998). 700-703.

6. G. Cacciamanl. P. Riani. A . Saccone. R . Ferro. A P M ~ R.

Schn~id-Fct~er . "Thermodynamic measurements and a5sescrnent

of the AI-Sc system". Intermetall1c5 7 ( 1999). 101-108.

7 . J . Cirohner, R. Schmid-Fctzcr. A. P~sch. G. Cacc~aman~. P Rrmi.

N Parodi. G. Bormne. A. Saccone. R. Ferro, "Expenmental

In\ectigations and Thennodynamic Calculation In the AI-Mg-Sc

Slstern", Z. Metallkd. 90 ( 8 ) (1999). 72-880.

S A , Pi5ch. J . Grohner. R. Schmid-1:et~er. "Applrcat~on of

Computat~onal Thcrmocheni~st~-y to Al- and Mg-alloy p~mcessing

\\rth Sc additions", Material Scie Er En<. A289 (3000). 123-139

9 .A Prwh. R Sclim~d-Fetter. Ci Cacc~amanr. P. R~ani . A

Saccone. R . Fen-o. "My-rlch phase e q u ~ l ~ b r ~ a and thermodynamrc

:is\cssmcnt ofthc Llg-Sc \>\teni", Z. Metallkd 89 ( 7 ) ( 1998).

171-177.