You are on page 1of 7

The role of intermetallics in wetting in metallic systems

Pavel Protsenko
a
, Anne Terlain
b
, Vladimir Traskine
c
, and
Nicolas Eustathopoulos
a
*
a
Laboratoire de Thermodynamique et Physico-Chimie M etallurgiques, Inst. Nat. Polytechnique GrenobleENSEEG,
INPG, Domain Universitaire BP 75, 38402 Saint Martin d'H eresFrance
b
Service de la Corrosion et du Comportement des mat eriaux dans leur Environnement, CEA/Saclay, Gif Sur Yvette, France
c
Department of Colloid Chemistry, MSU, Moscow, Russia
Received 7 June 2001; received in revised form 19 September 2001
Abstract
Formation of intermetallic seems to improve strongly wetting of a solid metal by a liquid one. The aim
of this study is to identify the reasons of this improvement. For this purpose, wetting of Fe is determined
at the same experimental conditions for two liquid metals, one reactive (Sn), the other non-reactive
(Pb). 2001 Acta Materialia Inc. Published by Elsevier Science Ltd. All rights reserved.
Keywords: Reactive wetting; Interfaces; Contact angle; Spreading; Intermetallics
Introduction
Wetting of solid metals by liquid metals and alloys is important in numerous met-
allurgical processes, for instance in hot dip metallic coating of steels and in soldering in
microelectronics. In both examples wetting is accompanied by the formation of inter-
metallic compounds at the interface.
Since the classical study of Bailey and Watkins [1], it has been widely accepted that
intermetallic formation reactions at metalmetal interfaces play a major benecial role
in wetting in this kind of system. This role has been explained by the approaches of
Pask et al. [2] and Naidich [3] considering that the energy of reaction liberated at the
interface can increase the driving force of wetting. This interpretation seems to have
been veried by results obtained recently by Nakae et al. in Al/Ni systems [4,5]. Indeed,
good wetting (contact angle h 20) was found by these authors at 1173 K for the
liquid Al30wt.%Ni/solid Ni couple, in which Al reacts with solid Ni to form the Ni
2
Al
3
Scripta Materialia 45 (2001) 14391445
www.elsevier.com/locate/scriptamat
*
Corresponding author. Tel.: +33-4-76-826504; fax: +33-4-76-826767.
E-mail address: nikos@ltpcm.inpg.fr (N. Eustathopoulos).
1359-6462/01/$ - see front matter 2001 Acta Materialia Inc. Published by Elsevier Science Ltd. All rights reserved.
PII: S1359- 6462( 01) 01181- 2
intermetallic compound at the interface, while non-wetting (h 125) was found in the
corresponding equilibrium (i.e., non-reactive) liquid Al30wt.%Ni/solid Ni
2
Al
3
couple.
This result is very dierent from results of discriminant experiments carried out with
various metal/ceramic systems (reviewed in Ref. [6]) from which it has been concluded
that the steady contact angle in a reactive system is nearly equal to the contact angle in
the corresponding equilibrium system, i.e., to the contact angle on the reaction product.
As there is no reason for this conclusion to be valid only for metal/ceramic systems and
not for metal/metal ones, a contradiction would exist in the wetting behaviour of these
two types of system.
This aim of this study is to identify the origin of this contradiction, i.e., to identify the
reasons why intermetallic formation seems to improve wetting. For this purpose, the
spreading kinetics and the equilibrium contact angles on a metallic substrate will be
determined at the same experimental conditions (temperature, furnace, atmosphere. . .)
for two liquid metals, one reactive, the other non-reactive.
Experimental procedure
The Sn/Fe and Pb/Fe couples were selected to examine the above equations. Liquid
Sn and solid Fe at temperatures close to 673 K form several intermetallics [7]. Some
experiments were also performed using a Fe7.5wt.%Cr stainless steel. In Pb/Fe and Pb/
Cr systems there is no formation of intermetallic compounds. Moreover at 673 K the
solubility in these systems is in the ppm range [7]. Pb/Fe and Pb/FeCr couples can
therefore be considered as non-reactive. Wetting is studied by the ``dispensed drop''
method, which is derived from the classical sessile drop technique. Wetting is charac-
terised by measuring the contact angle and the linear dimensions of a drop deposited
in situ on the substrate, at the experimental temperature.
Two polycrystalline substrates were used: high purity iron (99.99%) and a Fe
7.5wt.%Cr martensitic stainless steel also containing 2%W as well as V (0.15%), Mn
(0.21%) and Si (0.10%). All the substrates polished up to 1 lm diamond paste, have an
average surface roughness R
a
of a few nm. High purity (99.999%) Sn and Pb were used.
Experiments were performed in a metallic furnace consisting essentially of a Mo
heater surrounded by Mo radiation shields located in a water-cooled stainless-steel
chamber. The chamber is tted with two windows enabling the illumination of the ses-
sile drop on the substrate. The experiment consists in heating the metal separately in
an alumina crucible. Once the experimental temperature is attained, the liquid is ex-
truded from the crucible, then the whole capillary introducer and the alloy droplet
descends so as to initiate contact between the lower surface of the droplet and the
substrate (Fig. 1). Thereafter, the capillary introducer rises and this movement induces
the complete transfer of the drop towards the substrate, then spreading continues in the
classical sessile drop conguration. The time-dependent change in linear dimen-
sions (base radius and height) and contact angle h of the drop are lmed by a video
camera connected to a computer enabling automatic image analysis. This device is used
to obtain the characteristics dimensions of the drop (drop base radius R and contact
1440 P. Protsenko et al./Scripta Materialia 45 (2001) 14391445
angle h) after 40 ms of contact between the liquid and the solid with an accuracy of 2
for h and 2% for R.
Experiments are carried out either under high vacuum (2 10
5
Pa) or in a static
atmosphere of helium puried by passing it through a bed of ZrAl getter before in-
troduction into the furnace.
Results and discussion
Fig. 2 shows the change with time of contact angles for pure Sn and Pb on Fe
substrates at 673 K under gettered He. In both cases, the initial contact angle h
0
is close
to 90 but while the contact angle for molten Pb does not change with time, for Sn, h
starts to decrease after about 1 s and reaches a nearly constant value h
F
= 25 in about
t
F
500 s. Experiments performed at 673 K with pure Pb and Sn on FeCr substrates
in high vacuum led to results similar to those obtained on Fe, i.e., to a constant contact
Fig. 2. Change with time of contact angles of pure Sn and Pb under gettered He at 673 K (direct heating at 673 K).
Fig. 1. Formation of a Pb drop by extrusion of the liquid through a capillary (left) and spreading on a pure Fe surface at
673 K in high vacuum (in this experiment before the drop formation, the furnace was heated to 1123 K in high vacuum).
P. Protsenko et al./Scripta Materialia 45 (2001) 14391445 1441
angle with time for pure Pb, h
0
= h
F
= 92 and to a contact angle that changes with
time from h
0
= 102 to h
F
= 32 for Sn. The only dierence in the case of Sn is an in-
crease in spreading time from t
F
500 s for pure Fe to t
F
10000 s for FeCr.
All the above results conrm the opinion given in the literature according to which a
metallic substrate is much better wetted by a reactive metal (here Sn) than by a non-
reactive one (Pb).
At the solidied SnFe interface of a sample cooled after 600 s at 673 K, a conti-
nuous layer of reaction product a few microns thick was found. (Fig. 3a). From EDS
microanalysis this layer was found to consist mainly of the FeSn
2
intermetallic. In view
of the small thickness of the reaction layer, it was not possible to check whether other
FeSn intermetallics grow between the FeSn
2
layer and the Fe substrate. Fig. 3b shows
that the reaction layer is also present outside the drop. The formation of an interme-
tallic layer was also found at the Sn/FeCr interface. No reactivity was observed at the
Pb/Fe interface which remains at at the scale of 0.1 lm.
The results of Fig. 2 can be explained using one of the following two hypotheses.
(i) The Fe surface at 673 K is nearly clean, i.e. metallic: In this case, the initial contact
angle h
0
90 observed for pure Sn on Fe would be the contact angle of Sn on a vir-
tually unreacted Fe surface, and the subsequent decrease in h with time would result
from the formation of intermetallics by reaction between Sn and Fe. This scheme is very
similar to that proposed previously [8] to explain the h(t) curves observed in reactive
metal/ceramic systems in which the reaction product is better wetted than the initial
substrate. This interpretation of the h(t) curve in the Sn/Fe system is unlikely because it
implies that the contact angle of Pb on clean Fe would be close 90 (Fig. 2). However,
from previous studies performed in highly reducing atmospheres, it is known that this
contact angle is in the range 4060 [9,10], values found also in the present study (see
later). Therefore, in the experiments of Fig. 2, the initial iron surface is not clean.
(ii) The iron surface at 673 K is oxidised: Indeed, as found by XPS analysis, after
polishing the iron surface is covered by an oxide layer of about two nanometers.
Moreover, at 673 K the oxygen partial pressure P
O
2
needed to reduce this oxide lm is
as low as 10
30
Pa. Such low values of P
O
2
can hardly be obtained without a special,
Fig. 3. (a) SEM micrograph of a Sn/Fe sessile drop specimen heated at 673 K for 600 s. The reaction layer was found to
be the FeSn
2
intermetallic and (b) optical micrograph taken from above showing a FeSn intermetallic layer several tens
of microns width extending outside the drop.
1442 P. Protsenko et al./Scripta Materialia 45 (2001) 14391445
highly reducing gas mixture associated with heat treatments at higher temperatures [9].
Note that contact angles close to 90 are closer to values typical for non-reactive metal/
oxide couples than for non-reactive metal/metal systems [11, p. 202].
If the iron surface at 673 K is oxidised, a change in furnace atmosphere and/or a
prior heat treatment at a higher temperature would aect the surface chemistry and, as
a result, the h
0
value of Sn. This is conrmed by the results of Fig. 4 in which it can be
seen that a prior heat treatment at 1123 K in gettered He reduces the initial contact
angle from h
0
90 to h
0
60. Thereafter, the contact angle decreases with time and
tends towards the value h
F
28 which is very close to the h
F
value observed in the rst
experiment. A further decrease in the initial contact angle to the value h
0
44 was
observed when all the experiments, i.e., the heat treatment at 1123 K, and the subse-
quent drop deposition and spreading on Fe at 673 K, were carried out under high
vacuum. As in the other experiments, the contact angle decreases again and tends to-
wards nearly the same limit.
The results of Fig. 4 conrm that the surface of Fe substrates heated directly at 673 K
was indeed oxidised. As suggested in Ref. [11, p. 187], for oxidised surfaces the eect of
intermetallic formation, which is possible by diusion of reacting components through
the thin oxide layer, is to disrupt the oxide layer and to create in situ a clean surface of
intermetallic (Fig. 5).
At 673 K, in the absence of an interfacial reaction, the oxide layer is permanent
barrier to wetting, as in the Pb/Fe couple (Fig. 2). Deoxidation of Fe by heating at 1123
K under high vacuum leads to contact angles of Pb at 673 K of 4050 (an example is
shown in Fig. 1) which are in the expected range.
The above results and conclusions explain the results obtained by Nakae et al. [4,5]
for Al30wt.%Ni liquid alloys on Ni an Ni
2
Al
3
substrates in a HeH
2
reducing at-
mosphere. In this atmosphere Ni surfaces can be easily deoxidised at 1173 K.
Fig. 4. Changes with time of contact angle observed for pure Sn on pure Fe at 673 K for three dierent conditions: direct
heating to 673 K in gettered He (a) after heat treatment at 1123 K in gettered He (b) and after heat treatment at 1123 K
in high vacuum (c).
P. Protsenko et al./Scripta Materialia 45 (2001) 14391445 1443
In contact with the Al30wt.%Ni alloy, solid Ni reacts with Al to form the Ni
2
Al
3
intermetallic at the interface. The contact angle of 20 observed in this case is close to
the contact angle of the alloy on the reaction product, i.e., on the intermetallic. On the
other hand, nickel aluminates are known to be oxidisable materials, their surface being
covered by thermodynamically very stable aluminium oxide layers. Indeed, by assuming
that this oxide is Al
2
O
3
, the equilibrium P
O
2
value for the reaction:
2Al 3=2O
2
Al
2
O
3
at 1173 K is of the order of 10
34
Pa, a value too low to be attained even using dry
hydrogen atmospheres. Deoxidation of liquid AlNi alloys or solid Ni
2
Al
3
intermetallic
can in fact occur by reaction between Al and Al
2
O
3
with the formation of volatile Al
2
O
[12], according to the reaction:
4Al Al
2
O
3
3Al
2
O
However, at 1173 K, this reaction is eective only in high vacuum, not in a reducing gas
atmosphere [13] such as that employed by Nakae et al. Clearly the high contact angle
h 125 observed by these authors for the non-reactive Al30wt.%Ni/Ni
2
Al
3
couple is
for an oxidised Ni
2
Al
3
substrate.
Both the results obtained by Nakae et al. for the AlNi/Ni
2
Al
3
system and the results
of the present study for the Sn/Fe and Pb/Fe systems show that the eect of interfacial
reactions on wetting of a metallic substrate is strong when this substrate is oxidised. As
set out above, this eect is mainly due to the in situ formation of clean surfaces of
intermetallic compounds which are wettable by liquid metals. For clean surfaces of
metallic substrate the eect of intermetallic formation on wetting is probably not zero
but it is rather limited. In the present study, despite the great dierence in reactivity
existing for the Pb/Fe and Sn/Fe systems, the dierence in the minimum contact angles
observed at 673 K for pure Pb and Sn is only 15 (40 against 25 respectively).
Results given on Figs. 2 and 4 are for times typical of a wetting experiment, i.e., 10
20 min. Note that for longer times, the contact angle on polycrystalline substance starts
Fig. 5. Schematic description of the region close to the triple line for an oxidised metal: (a) in the case of a non-reactive
liquid metal B-solid metal A couple and (b) in the case of formation of a layer of intermetallic compound AB
n
at the
interface. The isolated particles come from breaking up of the oxide layer.
1444 P. Protsenko et al./Scripta Materialia 45 (2001) 14391445
to decrease again due to the contribution of grain boundary grooves formed in contact
with the liquid metal 11, p. 399. Moreover, during prolonged wetting experiments in
reactive systems, diusion of a component of the liquid into solid or over the solid
surface [14], can produce a signicant decrease in the drop volume leading to a chance
in the macroscopic contact angle from an advancing to a receding conguration. These
long-term phenomena go beyond the aims of the present study.
Conclusions
In low temperatures, usual metals are covered by oxide lms non-wetted by liquid
metals. For thin oxide lms wetting is strongly improved by intermetallic formation
reactions leading to replacement of the oxidised surface by a clean surface of an in-
termetallic compound. For this reason the nal degree of wetting in systems in which
intense reactions occur at the interface is less sensitive to environmental factors than in
non-reactive systems.
For clean solid metal surfaces, the formation of intermetallics at the interface seems
to have a benecial but rather limited eect on wetting. These results invalidate the
long-standing claim that the free energy of reaction liberated at the interface during
intermetallic formation can increase the driving force for wetting. However, interme-
tallic layers can have a considerable eect on the mechanical behaviour of the system.
References
[1] Bailey, G., & Watkins, H. (19511952). J Inst Metals 80, 57.
[2] Aksay, I., Hoye, C., & Pask, J. (1974). J Phys Chem 78, 1178.
[3] Naidich, Y. (1981). Prog Surf Memb Sci 14, 353.
[4] Nakae, H., & Goto, A. (1998). In N. Eustathopoulos & N. Sobczak (Eds.). Proceedings of 2nd High Temperature
Capillarity Conference, 29 June2 July 1997 (p. 29). Foundry Research Institute, Cracow, Poland.
[5] Nakae, H., Hane, T., & Sudo, T. (2000). Proceedings of 3rd High Temperature Capillarity Conference, Kurashiki,
Japan, 1922 November 2000, in press.
[6] Eustathopoulos, N. (1998). Acta Mater 46, 2319.
[7] Massalski, T. B. (Ed.). (1990). Binary Alloy Phase Diagrams, 2nd edition. ASM International.
[8] Landry, K., & Eustathopoulos, N. (1996). Acta Mater 44, 3923.
[9] Ebrill, N. (2000). Ph.D. Thesis, University of Newcastle, Australia.
[10] Popel, S., Kozhurkov, V., & Zakharova, T. (1971). Zashchita Metall 7, 421 (English translation).
[11] Eustathopoulos, N., Nicholas, M., & Drevet, B. (1999). Wettability at High Temperatures, Pergamon Materials
Series (p. 187, 202, 399). Oxford: Elsevier.
[12] Laurent, V., Chatain, D., Chatillon, C., & Eustathopoulos, N. (1988). Acta Metall 36, 1797.
[13] Merlin, V., & Eustathopoulos, N. (1995). J Mater Sci 30, 3619.
[14] Wang, X. H., & Conrad, H. (1995). Metal Mater Trans A 26A, 459.
P. Protsenko et al./Scripta Materialia 45 (2001) 14391445 1445

You might also like