вход по аккаунту


In situ ESEM study of partial melting and precipitation process of AZ91D.

код для вставкиСкачать
Asia-Pac. J. Chem. Eng. 2007; 2: 493–498
Published online 13 September 2007 in Wiley InterScience
( DOI:10.1002/apj.087
Research Article
In situ ESEM study of partial melting and precipitation
process of AZ91D
Jenny Hao-Hsin Hung, Yu Lung Chiu,* Tianping Zhu and Wei Gao
Department of Chemical and Materials Engineering, The University of Auckland, Auckland, New Zealand
Received 6 December 2006; Revised 21 February 2007; Accepted 23 February 2007
ABSTRACT: In this paper, we report the results of the investigation of partial melting and precipitation processes of
an AZ91D magnesium alloy using an in-situ heating stage located in an environmental scanning electron microscope
(ESEM). For the first time, in-situ partial melting of the β phase (Al12 Mg17 ) was observed to initiate in the middle of
the large grain boundary β islands, at temperatures around 420 ◦ C. Upon cooling, areas where partial melting of the β
phase had occurred solidified into voids. Partial melting of β lamellae within the discontinuous eutectic β/α lamellar
structures has not been observed in the present study. Following a solution treatment and aging process, the β phase
exists as fine submicron precipitates.  2007 Curtin University of Technology and John Wiley & Sons, Ltd.
KEYWORDS: environmental scanning electron microscope; in-situ partial melting; AZ91D
Magnesium alloys have unique properties including
high strength-to-weight ratio, good castability, excellent
machinability, and high damping capacity (Ross, 1992;
Kramer, 2001; Housh et al ., 2002). The combination
of these unique properties makes magnesium alloys
suitable for many applications. It is forecasted that,
with continued research and development, the increase
in the use of magnesium alloys will be similar to
that of its competitors such as plastics and aluminium
(Mordike and Ebert, 2001). Among magnesium alloys,
AZ91D (Mg–9 wt% Al–1 wt% Zn) is the most popular
casting alloy, due to its excellent strength and stiffness
at ambient temperatures, good stability in atmospheric
conditions, excellent saltwater corrosion resistance, and
castability relative to other magnesium alloys (Kramer,
2001; Housh et al ., 2002).
Solid state phase transformation (second phase dissolving and precipitation) is an important mechanism
which when harnessed and understood, can be utilised
to improve the properties of materials to create highperformance and technologically useful forms. The traditional studies of phase transformation often rely on
post-mortem observations of microstructure and/or morphology (see for example (Porter and Easterling, 1992)).
The use of a heating stage in an environmental scanning
electron microscope (ESEM) provides the possibility of
*Correspondence to: Yu Lung Chiu, Department of Chemical and
Materials Engineering, The University of Auckland, Auckland, New
Zealand. E-mail:
 2007 Curtin University of Technology and John Wiley & Sons, Ltd.
viewing and recording solid state phase transformation
of materials in real time, with sub-micrometer resolution
(Exner and Weinbruch, 2004).
It is known that the morphology and distribution of β
phase (Al12 Mg17 ) have significant effects on the welding properties of AZ91D. For instance, Munitz et al .
(2001) showed that the liquation of the β phase resulted
in poor mechanical properties of the weld-base metal
interface. This was linked, in post-mortem analysis,
to the liquation, motion, and subsequent solidification
of the brittle β phase, which distributed continuously
along grain boundaries in the heat affected zone (HAZ).
Hence, it is of technical importance to understand the
partial melting process of the β phase during heattreatment of magnesium alloys. This study is focused
on the in-situ observation of the partial melting process
of the β phase in an ESEM, as an attempt to mimic
and therefore achieve an improved understanding of the
sub-solidus constitutional liquation which occurs in the
HAZ during welding processes. In-situ observation of
the precipitation processes of β phase from a solutiontreated matrix is also reported.
The AZ91D alloy used in the present study was cut from
a large sand-cast plate with the nominal composition
shown in Table 1. In order to fit into the heating stage,
discs of 1 mm in thickness and 4 mm in diameter were
prepared from the as-cast plate. For the partial melting
Asia-Pacific Journal of Chemical Engineering
Table 1. Nominal composition of AZ91D.
study, the as-cast samples were subjected to standard
metallographic preparation procedure with a final 1 µm
diamond paste polishing. The polished samples were
then either etched or electropolished for microstructure observation. The etching solution contains 4.2 g
picric acid, 10 ml H2 O, 10 ml acetic acid, and 70 ml
ethanol. Electropolishing was carried out in an electrolyte containing three parts 85% phosphoric acid and
five parts 95% ethanol with the applied voltage of 2 V.
For the aging study, a block of the as-cast sample was
first solution-treated at 413 ◦ C in a protective nitrogen
atmosphere inside a tube furnace for 4 days. Complete
dissolution of the β phase was achieved and the sample subsequently water-quenched. The quenched sample
was then prepared similarly as those for partial melting
experiments and was electropolished for in-situ studies.
In-situ heating was carried out in an FEI Quanta
200F ESEM fitted with a heating stage that can be
run up to 1000 ◦ C with water-cooling. A high temperature gaseous secondary electron detector (GSED) with a
pressure-limiting aperture, mounted directly above the
specimen on the heating stage was used for electronimaging. The specimen chamber was first purged in
high vacuum mode and a helium atmosphere of 2 Torr
using the GSED for imaging was used during experiments. The heating schedule used for the partial melting
study is shown in Table 2. The sample was quickly
heated up to 370 ◦ C at a ramping rate of 20 ◦ C/min. The
fast heating was an attempt to simulate the fast-heating
of the welding process and to minimise possible oxidation during heating. Also, it has been noted that at
temperatures below 370 ◦ C, no structural change would
be expected to occur during the heating of an AZ91
alloy (Padfield, 2004). Beyond 370 ◦ C, the heating rate
was decreased to 5 ◦ C/min and further down to 3 ◦ C/min
when the temperature was above 460 ◦ C. The slow heating rate is required at these temperatures, where phase
transformation would be expected, in order to facilitate
imaging and analyses. For the aging experiments, the
Table 2. Partial melting heating schedule.
(◦ C)
(◦ C/min)
 2007 Curtin University of Technology and John Wiley & Sons, Ltd.
specimen was heated at a ramping rate of 20 ◦ C/min to
the target temperatures, i.e. 216 ◦ C and 320 ◦ C, which
are within the α-β two-phase region for the alloy system studied. It was then allowed to age at the target
temperature while the surface topography was closely
Partial melting
Figure 1 shows the morphology change during the partial melting of an etched specimen. Two different types
of β phase exist – large islands along grain boundaries and fine lamellar structures. When the sample was
heated, no noticeable morphology change was observed
at temperatures lower than 420 ◦ C (Fig. 1(a)). The initial melting of the β phase occurred when the temperature reached 420 ◦ C, and started with a micro-crack
in the middle of the large-grain boundary β phase,
as marked with white dashed lines in Fig. 1(b). It is
noteworthy that these observed partial melting sites
(micro-cracks) are in fact directly correlated to the location of grain boundaries where the β phase islands
are situated. As the temperature increased, the crack
opened up as more of the β phase became liquefied and expanded. Other parts of the specimen however, remained intact (Fig. 1(c), (d)). The discontinuous eutectic β lamellar structures remained unchanged
throughout the whole in-situ experiment which differs
from the literature (e.g. (Padfield, 2004)) and previous
post-mortem observations in the heat treatment experiments by the present authors, where the lamellar structure was found to rapidly dissolve at temperatures above
370 ◦ C.
Figure 2 shows the surface morphology of a specimen which was cooled to room temperature after a
partial melting experiment where the sample was heated
to 500 ◦ C. As indicated by the arrow, there was retention of the β lamellar structure after about 1.5 h holding at temperatures higher than 403 ◦ C. One plausible
explanation for the presence of the lamellar precipitates in the present study is that a thin oxide layer
may have formed on the sample surface as a result of
etching and was obstructing the observation of small
area phase-change underneath. As the melting temperature of this oxide layer (presumably magnesium
oxide) is high, it remains intact, preserving the original surface topography imprint of the specimen. It
Asia-Pac. J. Chem. Eng. 2007; 2: 493–498
DOI: 10.1002/apj
Asia-Pacific Journal of Chemical Engineering
Figure 1. Partial melting experiment of the as-cast sample, polished and etched.
(a) Original structure of β phase in bright contrast, sitting on grain boundaries;
(b) Micro-crack initiated from middle of the β phase at 420 ◦ C, dashed lines; (c) and
(d) Melting and expansion of β phase creating a lump that gives a brighter contrast.
The lamellar structure has not changed in the process.
has been later found during partial melting experiments
of electropolished samples that this oxide layer helps
to protect against further oxidation at high temperatures.
From the post-mortem examination of the partially
melted samples it was observed that there were voids
dispersed throughout the surface, an example of which
is located inside the oval marked in Fig. 2. These
voids correspond to the sites where β phase has
partially melted. The formation of the voids, as a loss
in structural integrity, partially explains the property
degradation of the PMZ after welding, and is consistent
with the theory of a mobile network of liquid β phase
travelling along grain boundaries as proposed by Munitz
et al . (2001).
Figure 3 shows the surface of a partially melted
specimen cooled to room temperature. Four energy
dispersive X-ray (EDX) analyses of the α phase were
carried out in the areas α1 to α4 . By calculating the
Al/Mg weight percent ratio, it was found that the
Figure 2. Post-mortem SEM image of a partially
melted sample. The void located inside the circle
has formed due to the melting and subsequent
solidification of β phase. The arrow indicates an
area where the β lamellar precipitates are retained.
Figure 3. SEM image of an etched, then partial,
melted sample. The partially melted area is
located inside the oval. Dashed lines indicate grain
boundaries and arrows indicate the general flow
direction of liquefied β phase.
 2007 Curtin University of Technology and John Wiley & Sons, Ltd.
Asia-Pac. J. Chem. Eng. 2007; 2: 493–498
DOI: 10.1002/apj
Asia-Pacific Journal of Chemical Engineering
aluminium content in α1 and α4 (i.e. 0.39 and 0.29
respectively) was much higher than that in α2 and α3
(0.09 and 0.10 respectively). This suggests that the
melted β phase has perhaps significantly diffused along
the grain boundaries (dashed lines) and/or along the
surface in the general direction indicated by arrowheads.
Surface topography of an electropolished magnesium
alloy can be revealed by the preferential dissolving
of the α matrix, as the β phase is relatively cathodic
(Vander Voort, 2004). A fresh surface, relatively free
of oxide layer, is shown in Fig. 4(a). At around 370 ◦ C
(Fig. 4(b)), the specimen surface started to be oxidised,
a rippling of the sample surface can be seen. Oxidation
was observed to initiate from the middle of the α
grain. As the temperature increases, oxidation proceeds
towards the lamellar α/β structure and β island leaving
behind a wrinkled surface (Fig. 4(c), (d)).
Aging experiment
Two kinds of in-situ aging experiments were carried
out on solution-treated and then electropolished AZ91D.
Figure 5 shows the microstructure of the solutiontreated sample where no β precipitates can be observed.
EDX analysis showed the existence of Al4 Mn and
Al6 Mn particles, as shown more clearly in Fig. 6(a),
(c). Aging was carried out in the ESEM. One sample
was aged at 216 ◦ C for 330 min and the other at 320 ◦ C
for 80 min. In both experiments, no significant surface
morphology change was observed. Only the grain
boundaries gradually became faintly visible as aging
proceeded, possibly due to the effects of thermal etching
(Fig. 5(b)). The aged specimens were then mechanically
polished, and chemically etched for observation using
an optical microscope and electropolished for scanning
electron microscope (SEM) observations (Fig. 6).
Figure 4. Partial melting experiment of electropolished specimen showing oxidation
(seen as wrinkling) of the surface above 370 ◦ C.
Figure 5. (a) SEM image of a solution treated and then electropolished sample prior
to aging. The two particles were identified as Al4 Mn (the large one) and Al6 Mn (the
small one) by EDX. (b) SEM image of the same position as (a) after aging at 320 ◦ C
for 80 min. The visibility of the grain boundary (marked by the arrows) has been
 2007 Curtin University of Technology and John Wiley & Sons, Ltd.
Asia-Pac. J. Chem. Eng. 2007; 2: 493–498
DOI: 10.1002/apj
Asia-Pacific Journal of Chemical Engineering
Figure 6. β precipitates after aging at 216 ◦ C for 330 min (a) optical microscope
image, (b) SEM image, and aging at 320 ◦ C for 80 min (c) optical microscope image,
(d) SEM image.
The sample aged at 216 ◦ C (Fig. 6(a), (b)) showed
predominantly discontinuous lamellar precipitation
along the grain boundaries, while the sample aged at
320 ◦ C (Fig. 6(c), (d)) showed continuous precipitation
throughout the whole of the α grains. This is consistent with literature (Roberts, 1960; Padfield, 2004).
The discontinuous precipitates can be slightly larger,
individual precipitate measures up to 20 µm in length
and about 200 nm wide, while continuous precipitates
in Fig. 6(c), (d) measure up to about 15 µm in some
cases. Both types of aging resulted in a wide range of
precipitation sizes. While the sample aged at 216 ◦ C
consists predominantly of discontinuous lamellar precipitations, as can be seen more clearly from the SEM
image (Fig. 6(b)), the rest of the grain outside the discontinuous lamellar precipitation are filled continuously
with fine sub-micron sized β precipitates.
In-situ studies of the partial melting and precipitation
of β phase in an AZ91D alloy have been performed. It
was observed for the first time that the initiation of the
melting of the β phase occurs at around 420 ◦ C, appearing as a ‘micro-crack’ through the large β phase islands
running along grain boundaries. The partial melting of
the β phase resulted in voids in the final microstructure
after cooling down to room temperature, attributed to
the mechanical performance-degradation of the partially
melted material. The electropolished specimen prepared
in the present study has been found to be unsuitable
for in-situ partial melting study owing to severe oxidation during heating. It has been found that aging at
 2007 Curtin University of Technology and John Wiley & Sons, Ltd.
216 ◦ C generated the microstructure consisting of predominantly discontinuous lamellar β precipitates along
grain boundaries with the rest of the grain filled continuously with β precipitates of sub-micron size. At the
higher aging temperature of 320 ◦ C, continuous precipitation became dominant. Although aging of AZ91D
in ESEM has been successfully performed, the in-situ
observation of the β phase precipitation process has
not been achieved. The fact that partial melting of β
phase within the α/β lamellar structure has not been
observed and the failure to detect β phase precipitation in ESEM require other characterisation techniques
with less emphasis on surface oxidation but comparable
space resolution.
The work described in this paper was partially supported
by an IIOF grant (Contract No. UOAX0601) from the
Foundation for Research, Science and Technology, New
Zealand. The technical assistance from Ms Catherine
Hobbis and Dr Bryony James is gratefully acknowledged.
Exner HE, Weinbruch S. Scanning electron microscopy. ASM
Handbooks Online, Vol. 9. ASM International: Ohio, 2004.
Housh S, Mikucki B, Stevenson A. Selection and application of
magnesium and magnesium alloys. ASM Handbooks Online, Vol.
2. ASM International: Ohio, 2002.
Kramer DA. Magnesium and magnesium alloys. Kirk-Othmer
Encyclopedia of Chemical Technology. John Wiley and Sons, Inc:
New York, 2001.
Asia-Pac. J. Chem. Eng. 2007; 2: 493–498
DOI: 10.1002/apj
Mordike BL, Ebert T. Mater. Sci. Eng. A 2001; 302: 37–45, DOI:
Munitz A, Cotler C, Stern A, Kohn G. Mater. Sci. Eng. A 2001; 302:
68–73, DOI: 10.1016/S0921-5093(00)01356-3.
Padfield TV. Metallography and microstructure of magnesium and
its alloys. ASM Handbooks Online, Vol. 9. ASM International:
Ohio, 2004.
 2007 Curtin University of Technology and John Wiley & Sons, Ltd.
Asia-Pacific Journal of Chemical Engineering
Porter DA, Easterling KE. Phase Transformation in Metals and
Alloys. Taylor and Francis: London, 1992; 243.
Roberts CS. Magnesium and Its Alloys. Wiley: New York, 1960; 119.
Ross RB. In Metallic Materials Specification Handbook. Chapman
and Hall: London, 1992; 211.
Vander Voort GF. Chemical and electrolytic polishing. ASM
Handbooks Online, Vol. 9. ASM International: Ohio, 2004.
Asia-Pac. J. Chem. Eng. 2007; 2: 493–498
DOI: 10.1002/apj
Без категории
Размер файла
977 Кб
process, az91d, melting, precipitation, stud, partial, esem, situ
Пожаловаться на содержимое документа