/ SAND75-0094 Unlimited Release

Magma/Metal Compatibility Study: Compatibility off Metals in Molten Degassed Tholeiitic Basalt

Robert A. Sallach

SF 2900 QI7-73)

STER ; LV-JLT^TCD

LAND-75-0094

MAGMA/METAL COMPATIBILITY STUDY Compatibility of Metals In Molten Degassed Tholeiitic Basalt

R. A. Sallach Chemical Metallurgy Divlsic

ABSTRACT

-MDTICE' DtM. vrpitii w*t pttfuH) l i in it* nun I ol work

the LhHtd &uttt DUI i hi L'IDIV) S u m Fncriy lUmitlt ttid Dtrtfbpmeni AtfmtAiiindDn. nop *n> of UhtM tmpWjjm. KT »nt ol I hin (cnlllcluri, MtbcoMjirtnir or ihru i»ployrn. rtmtt* my wiii»niyb ripitu t*t utipbti), oi «iurmt my l-|*J Mbllllr DI ineunuMlit) rtu ihr *c*-vm>,i«}rpplmr*» of uictulMtt at in iflr,jrrT»ihin,*pp*JH*.pt^wci or rmoru dntloifd, CK irpmrmi Th*i ri* mr m t U i w

The compatibility of ten conmercially available alloys with molten

(1300°C) degassed tholeiitic basalt vas studied. The results can be

explained on t..e basis of 1) the relative free energies of formation of

pertinent oxides and 2) shifts in the chemical equilibrium among iron

species.

_i-.ur=:a \?>

Acknowledgments

The author would like to acknowledge the technical support of

F, E. Arellano, His work and assistance in the experimental portion o

this work was very important. In addition, the efforts of J, L. Rife

in the raicroprobe investigations are very much appreciated.

TABLE OF CONTENTS Page

Introduction 7 Experimental 6 Experimental Results 10

Iron 11 310 Stainless Steel, 12 kkG Stainless Steel 13 Nickel 13 Udimet 700 and Inconel 718 lU Molybdenum Ik

Tungsten 15 Tantalum 15 W3 3015 16 Palladium „ 16

Discussion 16 References 27

LIST OF TABLES Table I Compositions of Test Metals 9 II Composition of Tholeiitic Lava 10

5

LIST OP FIGURES

Fe/lava interface. Droplets appear to be farming. Dark region is lava. 19 Interface region of 310 SS/lava showing extensive reaction , 19 Electron microprobe data showing elemental distribution of the various phases at the 310 SS/lava interface. The grid lines define 25 M intervals, a) Fe, b) 0, c) Cr, d) Si, e) Mg, f) At 20

UU6 SS/lava interface. Compare to Figure 2 21 Nickel/lava interface. No oxide or other reaction product is visible. . 21 Electron microprobe data showing elemental distribution at the Ni/lava interface. Fe is only element present in both lava and metal phase. The grid lines define 15 (i intervals, a) Ni, b) Si, c) Fe, d) 0, e) At 22

Appearance of Udimet 700 and Inconel 7l8 after sectioning. (a) Udimet 700; (b) Inconel 7l8 23 M>lybdenum/lava interface. Note formation of porous layer 2*+ Tungsten/lava interface 2** Tantalum/lava interface. Molten Fe phase forms in lava. Dark gray area of metal contains oxygen 25 W 3015/lava interface 25 Electron microprobe data showing elemental distributions in the interior of the VC3015 alloy. Of the major metallic components only Hf is concentrated in the grain boundaries. The grid lines define 10 M intervals, a) Nb, b) Zr, c) Hf, d) Ti, e) w 26

Introduction

The extraction of thermal energy from magma undoubtedly has been con­

sidered before, but such a proposal was probably treated like early

considerations of space flights. However the present energy crises, which

is expected to persist in the future If not intensify, prompts a new, more

serious look at such proposals.

It is acknowledged that much more data are needed in the fields of

geology, geochemistry, metallurgy and engineering before one can thoroughly

assess the technical feasibility and economic competitiveness of power

generated by tapping a magma source, but present data does not indicate

that such an extraction of thermal energy is inherently impossible. Engi­

neering calculations made at Sandia suggest that the amount of energy

obtained in a single installation could be comparable to that obtained from

a nuclear reactor. Thus it is clear that successful application can bring

generous returns.

Magma, by definition, is molten rock beneath the solid surface of the

earth. When magma Is forced from the depths and expelled onto the earth's

surface, lava is produced. Magma and lava are not identical materials for

some of the components of magma are volatilized as magma approaches the

earth's surface and the external pressure lessens. The primary volatile

substance is HpO but significant quantities of othpr gases—C0 p, CO, S0 p,

„, or H-S—may also be present. These volatiles, as a class, will total SO

only a few percent of the total magma mass. However, In the dissolved state

they would be expected to alter the chemical nature of U.e marina yva" i.:u.l&: ..

i:i regard to Its acidity and corrosive properties.

In regard to their common constituents magma and lava can be :iescri:.-:

as silicate melts in which, the cations are those of l*.he elements Co. Mg. /._.

Fe, Jfa, Ti, and K in an approximate descending order of concentrate o.-.-, ~": o

pivotal element is Fe. It can be present as Fe or Fe ; the ratio of

these two species is determined by the chemical potential or fugacity -A

oxygen in the magma. Of these elements Fe is also the one vj.ic!, is rncs:.

easily reduced to the metallic state.

Aside from the chemical environment, the materials use<i i:. a :nagma t-.ap '«^,o will experience long term exposure to temperatures of 600 c or more wit.:

excursions or short term exposure to temperatures in excess of 1000°C4

possibly even to 1300°C. At the same time the lithostatic pressure may he

as much as or even exceed 100 MN/m (1^,000 psi), p

A survey vas proposed where the compat ibi l i ty of various metals and

a l loys toward molten rock could be determined. I t was recognised t h a t a l l

the pe r t inen t var iab les could not simultaneously be reprou :ced in o;r

present experimental apparatus thus the survey was carr ied out under more

benign condit ions-- iamersion a t 1300°C In degassed lava with a surrounding 4 2

Ar atmosphere (1 .4 x 10 N/m or 2 p s i a ) .

This repor t describes the r e s u l t s of those t e s t s and discusses t-.e

poss ib le i n t e rp re t a t i ons and conclusions which can be reached.

Experimental

Ten candidate metals were selected for t h i s survey baseu on the i r >.r.cv

high temperature p roper t i e s and on t h e i r commercial a v a i l a b i l i t y . Ar.

eleventh, Pd, was included because of its scientific interest to related

studiea. The compositions of these eleven metals are listed ir. Table I,

TABLE I Compositions of Test Metals

Vt.

Iron 100 Fe 310 Stainless Steel 51 Fe - 25 Cr - SO Ni - 2 Mn - 1.5 Si Wvo Stair-less Steel 72 Fe - 25 Cr - 1.5 Mn - 1.0 Si Nickel 100 Ni Udinet 700 53 Hi - 18 Co - 15 Cr - 5 Mo - 3-3 Ti Inconel 718 53 Ni - 19 Cr - 18 Fe - 5-2 Nb - 3 M-Molybdenum 100 Mo Tungsten 100 V Tantalum 100 Ta WC 3015 ^9 Hb - 30 Hf - lh W - U.5 Ti - 1.5 Zr Palladium 100 Pd

The ten commercial metals were obtained in the form of tubing (1.27 OR

O.D.; O.63 cm I .D. ) from which specimens 1.2 cm in length vere cu t . The Pd

specimen was cut from sheet and had dimensions 1 cm x 1 cm x O.U mm.

The lava used in these t e s t s cara frou Hawaii and i s character ised as

t h o l e i i t i c b a s a l t . I t s composition ia presented in Table I I .

The metal specimens were placed in inuividuul alumina cruc ib les and J

covered with a quant i ty of coarse ly pulverized lava . The loaded crucibles

were put in a vacuum furnace which was ther evacuated and maintained a t

2 x 10 t o r r and ambient temperature for V. hours. The diffusion pucp was

turned on and evacuatiop continued for another 2<* hours a t 3 x 10 t o r r .

.': -3 furnace temperature then was ra ised to 300°C for 3 hours fur fur ther

-ut^asstng. Without lowering the temperature the furnace was flushed

r e p e t i t i v e l y with pur i f ied argon; the f i n a l backf i l l brought the pressure 2 l. 1* ka/m - Temperature was ra i sed stepwise during the next h hours *.n

r':e t e s t temperature of 1300°C and maintained there for LOO hours. Vh'jn

•he furnace had cooled, the specimens were mounted, sect ioned, and p'.lish'jd

for metallographic examination. Some of the mounted specimens wen-* alSw wx

aalned with the scanning e lec t ron microscope and with the e l ec ' ron

rr.icroprobe.

TA3LF I I Compoaition of Tho l e i i t i c lava

Wt.*

sio 2 53 Ti02 2.1* A1 20 3 13 FeOx 9 MnO MgC T Cao 11 N a 2 ° 2.5 K ?0 .5

Experimental Results

The experimental r e s u l t s for each oe*al or a l loy w i l l be presented i: .-

i i v i i u a l l y . Cocnents germane t o a p a r t i c u l a r metal or a l loy are in t h i s

se^: i^n; general t rends and co r re l a t ions are aiscisseti in the next sec t ion .

Iron - When the specimen was sectioned, polished and examined with the micro­

scope, no adnerent product layer was found. However, an iron oxide film or

layer would be expected to dissolve in the molten lava. Thermodynamics

would predict an oxidation/reduction reaction between metallic iron and

ferric cations to form additional ferrous cations. How this reaction would

affect the appearance of the metal/lava interface is not known.

It is apparent that the interface has changed. The metal/lava inter­

face is irregular. DropletB of metal occur at or near the interface

(Figure 1). The fact that several of the droplets are attached to the bulk

metal suggests that all the droplets were derived from the original metal.

The formation of droplets at a temperaturo considerably below the

melting point of pure iron indicates that an alloying reaction must have

occurred. An analysis of the droplets with the electron microprobe in­

dicated however that only iron was present; no detectable amounts of the

major elements of the lava were present in the metal phase.

T^e binary phase diagrams of iron and other elements were reviewed

and elements which form interstitial solutions appeared to be thofle most

likely to induce r.elting. Alloys of carbon and phosphorus with iron are

completely molten at l^OO C when only 3 and 6 weight percent, respectively,

has been dissolved. These facts prompted another examination with the

microprobe, wherein these elements were sought. But again the results

were negative.

At the present t ime, no explanat ion has b-r-::: : \ ^- : :* v :- ;' r:?ji' \ n

of a l i q u i d , predominantly i r o n , phase . I t ^eunu: cur'aL;, \) nm. t!v.

phenomenon i s a s soc ia ted with the l a / a phase or i t s f , n s t iw-':;*:, a':::-;,

as described l a t e r , s imi l a r phases can be produced v-^r, '.-•."• w.r n .: -_i:--;

brought in contact with molten l ava .

310 S t a i n l e s s S t e e l - This a l l o y reac ted ex tens ivo ly v l ' h \.\ •- \-:i:;^:;*s : ^F-.\ ';

as shown in Figure 2. The e l e c t r o n microprobe was uso.: * LiL-ntLfy t :

var ious phases . Some of the e le i renta l d i s t r i b u t 1-. ne Lji ' !.•: '..nt^ri'uo*: :*'-:. ~

a re presented in Figure 3 . A chromium oxide l aye r i s o i -•;:.* , \r\ :•;

Cr nO-,. Next t o t h i s and a l so i n t e r spe r sed in the a:/a-j-jr:t .a u i. - rhr^:-.

v h i ' h i s predominantly i ron and vhich appears zc h:r. •-• :.-er. \. :r

experimental temperature . Again the molten appear an.*': } 1:-" ' • v- •; ';:

a l l o y i n g re r : *>n.

Inc.- v • <% found on the metal s i ae of the 1: r mi\:t >:: : . ; . ' v \ r ,

The majo J - these have a composition s imi l a r t *:.s-. :" :•: _'* t .

These i n c l u s i o n s are found only a short d i s tance i;v . :o : . - ' , " , . . •: *m

1^0 jl from the chromium oxide l a y e r . The i n p l i ^ * : . : : 1: ' . : . ' ~:-. ;•;. L:.^.

me ta l / l ava i n t e r f a c e was i n t e r i o r t^ i t s pro sent *. rr i r. *-_: - ;* *r : r niur.

oxide and a molten ir.-n phase, producrs of 2 r ; t . : t i :i : " . . . r . *.

a l l o y and the lava , accreted a t the surface v i t : • - ;; r v.: n. >:i ; : ;: a~<":

mainta in ing a p o s i t i o n near or a t t he in ter face* ;• •• urp&r-jn: r-;u^:i .r .

i s ' h e reduc t ion of the : t s so lved i r on ox i i e s by C:J-OCIUE: n-^ta,. ^ T T r " —•>

evidence i s the lack of any i ron spec ies in the l a \ a aft-ir c t s r l u ' i . n .:" t:.-.

t e s t .

hkb S ta in less Steel - This a l loy a l so reacted extensively wit,? the formation

:f a thick chrocdurr. jxidu l ay t r at t:h*j ncral/l.ava interface ' r ^ ' i r ' j -*) •

Although a s imi lar r ' :Gc 4 : :n s c a r r e d GE vi t i , the ^"^ a l l oy , *•; •:;••:: . / . - : -•'

d i f ferences , with vV^ a i l j y , i l l u s i o n s of lava vithir. *:.- Irv.-irL:' f : -

metal are no" found. The i ron-containing phase aces net acc^etr* a t V:'.- i r / c r -

face but i s found e n t i r e l y within the adjacent lava. In addi t ion , -his

phase has a more angular and extended appearauce as i f i t has n^t b^en m-:, ^r.,

Tlickel - From the opt ica l examination cf the sectioned specimen, t-nt w;ul :

cjnclude that uc reac t ion occurred. The metal/ lava in ter face i s snev i an,:

ve i l -def ined with no react ion product v i s i b l e (Figure 5)-

Nevertheless the in te r face region was examined with the e lect ron rticr -

probe. The r e s u l t s (Figure *i) show a c lear separat ion of a l l elements save

^•ne. Nickel i s observed in the metal phase only; aluminum, s i l i c o n , and

oxygen are only In the lava. Only i ron i s found in b : th phases aft^-r t?e

t e s t . Further aeasurements with the microprohe show tha t there i s a iron

concentrat ion gradient in to the metal phase. One concludes tha t ir~n atoms

are produced at the metal/ lava interface and diffuse in to the niey.el metal.

The only ^reasonable source cf these i ron atoms i s the d ispropor t i j?nati;n

a t the metal / lava in terface of ferrous ions in to f e r r i c ions which remain in

the lava and i ron atoms which then diffuse in to the nickel o e t a l .

The reverse reaction would be expected if iron atoms had unit activity.

3ut apparently the chemical activity of iron in. the nickel is sufficiently

IJW that the disproportionation is allowed*

udirnet TOO and Inconel TL^ " These alxoys shew the same behavior. There is

apparently no chemical interaction with the lava. However, the experimental

temperature was high enough that the soiidus temperatures of the alloys were

exceeded. As a result the alloys suffered what might be termed "loss of

structural integrity.11 Their physical appearance is seen in Figure 7.

Wo further examination of these alloys were made.

Molybdenum ~ The interface between molybdenum and lava show evidence of an

interaction (Figure 8. A porous layer is generated which, however, is no":

an oxide. Hather it is part of the metallic phase since no boundary can be

discerned between the bulk metal and the porous layer*

The electron microprobe shows that, as in the case of nickel, iron atoms

are diffusing into the molybdenum metal. The source of the iron atoms is

again attributed to the disproportlocation of ferrous cations present, in the

r.ol::en leva.

The fact that no molybdenum oxide phase is observed does net exclude

?li oxidation reactions. Molybdenum oxides are coiroarahle in stability to

tie iron oxiies and thus can coexist at the Low oxygen activity cf the melt.

Their absence might be the result of a reaction sue;*, as

in which rx separate MO phase is formed. Molybdenum is not ietected ir.

the lava vi-;, the micrcvrobe; this shows that t; e extant, of ar.y such reactitr.

is small in a static system. Izi a dynamic system w>*sre netal arrl lava mv/e

relative to each other such reactions :nay have to be considered since tr.e

cumulative effect may become significant-

Tungsten - In Figure 9 the tungsten/lava interface can be viewed. There is

no product phase visible. There is , however, evidence cf an etching prtcess--

the small angular intrusions into the metal tody. This ::i£ht be explained by

reactions such as

2 3'soln 2 soln so±r.

Further experimentation and Eeasurements would be needed to verify this

hypothesis but are not plamed at present-

lantalum - This refractory netal does react with lava- In figure 10 the white

:.hase at the lower edge has been identified -with the microjrobe as iron.

Its rounded features suggest; that i t had been welter*. As in the ire:, lest,

no alloying elements are found.

Two phases are present in the metal. T'r.e interior, white-appearing

phase is the original tantalum metal. The grey phase, lying near the metal/

lava interface contains both tantalum and oxygen. Or. the basis of elemental

intensities observed with the electron nicroprobe i t does net appear to be

Ta ?0 s; a Lower oxide or metal with interstit ially disserved oxygen is a more

likely choice.

Tantalum is a more reactive metal than ircr.j thus i t should reduce the

dissolved iron oxides. In addition tantalus, can dissolve cxyger. in te r s t i t i a l . / .

15

The evidence indicates that both processes are occurring.

V.'C 3015 - This a l l oy r eac t s extensively with the lava. An apparently puro.

but molt-jn iron phase i s found in the lava and there i s evidence oi" a surface

oxide layer (Figure I I ) . The l a t t e r i s not "chick, espec ia l ly in iro-.-iparls-.-:

to the amount of i rcn phase produced. This fact suggests that much czyger.

has diffused into the i n t e r i o r of the metal. When the i n t e r i o r region i s

examined with the electro.-, mieroprobe, hafnium, is found to be concentrated

in the grain boundaries, while the other elements--niobium, tungsten,

t i tanium, and zirconium—remain dispersed in the metal phase 'see Fig-ore 12).

In the o r ig ina l a l l oys , hafnium i s uniformly dispersed. The oxygen d i s t r i ­

bution pa t t e rn i s not wel l resolved, but one could infer tha t oxygen occurs

r.ore often in the grain boundaries. This inference suggests that Vafr.iurr.

oxide i s p r e c i p i t a t i n g in the grain boundaries—a view which would oe con­

s i s t e n t with thermodynamic data on the r e l a t i v e free energies of formati;r. of

these metal oxides.

palladium - The o r ig ina l mater ia l was in sheet form. During the t e s t t: e

rr.etah spheroidized. Sir.ce the t e s t temperature was ~20C°C below t r e melting

point , a l loying was immediately suspected, indeed the e lect ron micropro-c-

showed t ha t much iron ie in the metal phase.

The explanation is the .-same as t ha t offered in the case of n ickel—that

i s , the a c t i v i t y of iron in a palladium a l loy is so low that d i s p r o p o r t i o n a t e

of ferrous ion is made poss ib l e .

Discussion

The in te rac t ions tha t are observed can be corre la ted on the basis of two

—.-xcerts. F i r s t , any chemically ac t ive eieinent w i l l be oxidized. Since free

0 gas i s not present , oxidation i s accomplished by the reduction of the

f e r r i c cat ion i n i t i a l l y and l a t e r , since lava i s aot present in excess,

by the reduction of ferrous cat ions- Chemically act ive elements are ti.oso

elements for which the free energy of formation of t h e i r oxides are more

negative than t ha t of e i t h e r i ron oxid- . Conversely, elements having

oxides whose free energy of formation are l e s s negative than these of iror.

oxides do not oxidize in molten lava .

Other in t e rac t ions are poss ib le and can be understood in terms of the

equil ibrium between iron metal and the ferrous and f e r r i c ca t ions :

o 3+ P+ Fe + 2 F e J - 3 Fe

Generally, the equil ibrium l i e s toward the r i g h t hand side of the equation

and one expects t o find major amounts of only two of the species a t any one

t ime—either (Fe° and Fe ' ) or (Fe and Fe^ )> However the equilibrium can

be offset t o the l e f t side of the equation by lowering the chemical a c t i v i t y

e i t h e r of the two species on the l e f t . In these systems t h i s i s most

r e a d i l y accomplished by a l loying Fe with another metal phase as in the case

of n ickel or palladium.

Thus those elements which are t r u l y ine r t t o degassed molten lava are

those in which there i s a l imi ted s o l u b i l i t y of i ron and whose oxides have

f ree energies of formation l e s s negative than tha t of the i ron oxides. These

c r i t e r i a give a r e s t r i c t e d choice of elements. However, i t does suggest t ha t

a l loys which contain a ce r t a in amount of iron and which otherwise contain no

chemically ac t ive elements as defined previously in t h i s sec t ion—that such

a l loys would be i n e r t t o molten degassed lava .

It is convenient to picture the molten degassed lava as a solvent and

-he gases which could be present as solutes. In this view the above results

involve aetal/solvent interactions solely. Of next concern are the yetal/

solute reactions "Which might occur and their effect on the lifetime of

materials emplaced in magma.

Some tests are planned in which "the high temperature, high pressure

conditions of magma are simulated. But because of the experimental dif­

ficulty of such tests, an alternative way to survey these metal/solute reactions would be desirable.

Metal/gas reactions are much easier to study. If the metal/solute

interactions are identical whether or not the solvent (molten lava) is

present, then a survey becomes much simpler and easier* In addition "the

extensive literature on metal/gas reactions becomes a helpful source of

data. This hypothesis v i l l be tes ted by appropriate future experiments.

. V • L - -

: • • > : ? - ' - . 100 M

Figure 1. Fe/lava interface. Droplets appear to be forming. Dark region is lava.

t, '•*?.>; * * H x

•; (M

^ t o >

Figure 2. Interface region of 310 SS/lava showing extensive reaction.

"Iron"

19

''- ' " — / . i

f

Figure 3. Electron microprobe data showing elemental distrihition of the various phases at the 310 SS/lava interface. The grid lines define 25 M intervals, a) Fe, b) 0.. c) Cr, d) Si, e) Mg, f) At.

100 ii

•°---•'"- %,-. ^ 1 * £ * « *

Figure *K kk6 SS/lava interface. Compare to Figure 2.

1 0 0 M

Figure 5. Nickel/lava interface. No oxide or other reaction product is visible.

X •*

Figure 6. Electron microprobe data showing elemental distributions at the Ni/lava interface. Fe is only element present in both lava and metal phase. The grid lines define 15 h intervals, a) Ni, b) Si, c) Fe, A) 0, e) M-.

22

. . . =r ^" f - ' ^

h h

Appearance of Udinet 700 and Inconel 718 after sectioning, (a) Udinet 700, (b) Inconel 718.

!'****- J-i .4

L •"

' tr r_

k * • P /

h r • T • ; Hi F • * t

F r F '

r ' .-

* c:.^

I " " J ' 100 ** tf . 1 1 •

••••'•-• - " . f e t , , ! £ A . ^ < K « a !

Figure 8. Molybdenum/lava interface. Note formation of porous layer.

1 0 0 M \ '

Figure 9- Tungsten/lava interface.

2h

Oxygen " Con­taining Phase

Oxide

Lava

"Iron*

Figure 10. Tantalum/lava interface. MbItan Fe phase forms in lava. Dark gray area of metal contains oxygen.

'I Alloy

Oxide

Iron'

I*T&

Figure 11. UC3015/lava interface.

•KM* * yt ^

Figure 12. Electron microprobe data showing elemental distributions in the interior of the WC3015 alloy. Of the major metallic components only Hf is concentrated in the grain boundaries. The grid lines define 10 h intervals, a) Nb, b) Zr, c) Hf, d) Ti, e) v

26

References

1, C. W. Young, "A Proposa l to ' I n v e s t i g a t e a ?i w Energy Source: The Di rec t Magma Tap , " SLA-73-^50 ( Janua ry X^k).

2. D. W. Mottern and M. J . Davis (unpubl i shed p r o p o s a l ) . These exper iments 'TOj.-i r-r.-' l l r . mr)pr t.h^ Hi >"~>pti ^n "-f1 ^ , ,T *- ~" - t o e 4 '..'ore underway ai. the time of his death.