Reciprocating wear mechanisms in a Zr-based bulk metallic glass

Page created by Rebecca Andrews
 
CONTINUE READING
Reciprocating wear mechanisms in a Zr-based bulk
           metallic glass
           H.W. Jin, R. Ayer, and J.Y. Koo
           Advanced Structural Materials Section, ExxonMobil Research & Engineering Company,
           Annandale, New Jersey 08801
           R. Raghavan and U. Ramamurtya)
           Department of Materials Engineering, Indian Institute of Science, Bangalore-560012, India

           (Received 20 July 2006; accepted 12 October 2006)

           The dry sliding friction coefficient ␮ and the wear volume loss W, in a
           zirconium-based bulk metallic glass (BMG) under high-frequency (50 Hz)
           reciprocating conditions, were investigated with the objective of assessing the influence
           of free volume and crystallization on the wear behavior of amorphous metals. The
           BMG samples were annealed either below the glass transition temperature Tg to induce
           structural relaxation and hence reduce the free volume that controls plasticity through
           shear-band formation or above Tg to crystallize the amorphous BMG prior to wear
           testing. Results show that the wear behavior of both the as-cast and relaxed glasses
           was dominated by the oxidation of the surface layers. A sharp transition in the contact
           electrical resistance complemented by a marked increase in ␮ was noted. This was
           attributed to the formation of a thick tribo film with high oxygen concentration and its
           subsequent delamination. The ␮ values, before as well as after the transition, in the
           relaxed glasses were similar to those for the as-cast alloy. However, a gradual decrease
           in W with annealing temperature was observed. A good correlation between W and
           nanohardness was noted, implying that the intrinsic hardness in the BMGs controlled
           the wear rate.

I. INTRODUCTION                                                        Two distinct deformation modes have been reported in
   Amorphous metals or metallic glasses were first syn-                amorphous alloys.5 At higher temperatures and low
thesized by rapidly quenching Au75Si25 alloy melts in                  stresses, the plastic deformation occurs through homoge-
1960.1 As a result of the extreme cooling rates (∼106 K/s)             neous (viscous) flow. At lower temperatures and higher
required to achieve a high degree of undercooling in the               stresses, the deformation is quite inhomogeneous, being
melt, metallic glasses could only be produced in thin                  concentrated in shear bands that form in orientations
ribbons (∼100 ␮m) or in powder form. Hence, despite                    close to the maximum resolved shear stress plane. Be-
their unique combinations of mechanical properties such                cause of these fundamental differences, the wear/friction
as high elastic limits, nearly theoretical strength com-               properties of amorphous alloys are also distinct from
bined with good toughness, the widespread application of               those of crystalline metals.
amorphous alloys has been significantly restricted. The                   Metallic glasses are generally expected to exhibit out-
development of bulk metallic glasses (BMGs) based on                   standing wear resistance due to their high hardness.
multicomponent alloy systems that remain amorphous                     However, a review of the literature shows conflicting
even when the cooling rates are low (∼1 K/s) rejuvenated               trends regarding the role of amorphous structure on the
interest in these alloys for practical applications.2–4 Due            wear performance. It has been reported that the Fe-based
to a lack of long-range order, these alloys exhibit quite              metallic glasses exhibit higher wear resistance than their
distinct deformation mechanisms and the absence of dis-                respective crystallized states.6,7 On the contrary, there
location-mediated slip, which is the major plastic defor-              are reports indicating that the wear performance of zir-
mation mechanism in conventional crystalline alloys.                   conium (Zr)-based BMGs are significantly inferior to
                                                                       that of the crystallized alloy8 and traditional crystalline
                                                                       alloys.9 Fu et al.10 have examined the unlubricated slid-
a)
 Address all correspondence to this author.                            ing characteristics of Zr-based BMGs and have observed
 e-mail: ramu@materials.iisc.ernet.in                                  that extensive material flow took place under sliding
DOI: 10.1557/JMR.2007.0048                                             conditions, indicating that plasticity at the contact plays

264                         J. Mater. Res., Vol. 22, No. 2, Feb 2007    © 2007 Materials Research Society
H.W. Jin et al.: Reciprocating wear mechanisms in a Zr-based bulk metallic glass

an important role in the wear/friction characteristics of         Oliver-Pharr method,15 to extract the elastic modulus and
the BMGs.                                                         the hardness. Microhardness measurements were per-
   It is well known that plasticity in amorphous metals is        formed using a Vickers diamond indenter and a 500-g
controlled by the free volume. For example, it has been           load with a dwell time of 15 s.
shown that free-volume reduction [through structural re-
laxation anneals below the glass transition temperature
(Tg)] leads to severe embrittlement due to the reduced            B. Wear testing
propensity for shear-band formation.11,12 Also, the crys-            A high-frequency reciprocating rig (HFRR) instru-
tallization induces embrittlement in a tough, as-cast,            ment (PCS Instruments, London, UK) with “ball-on-
amorphous alloy.13,14 Thus, both structural relaxation            plate/disc” geometry was used for measuring the friction
and crystallization may yield different deformation and           coefficient and wear volume loss (W). The 6-mm-
flow behavior. However, a systematic study that explores          diameter WC ball (Salem Specialty Balls, Salem, MA)
the relationship between the plasticity and its wear char-        and the 1.5-mm-thick BMG specimens were fixed tightly
acteristics has not yet been reported in the literature.          in specimen holders that have appropriate recesses ma-
Such understanding will facilitate the design of BMGs             chined into them. The upper specimen holder has a load-
from a wear resistance point-of-view. Therefore, the ob-          ing pin to which a dead weight load was applied. An
jective of the present study was to conduct a detailed            electromagnetic vibrator drives the upper specimen
study on the friction and wear characteristics of a Zr-           holder back and forth via a pushrod. A linear variable
based BMG in which the propensity for shear banding               differential transducer (LVDT) attachment to this allows
was controlled systematically by annealing. Nanoinden-            for monitoring of the motion of the pushrod, and adjusts
tation was used to characterize the changes in intrinsic          the frequency and amplitude of the vibrator. A piezoelec-
plasticity in the BMG due to annealing and was corre-             tric force transducer connected to the lower block mea-
lated with wear rates.                                            sures the friction force generated between the ball and the
                                                                  specimen as they slide relative to each other. Prior to the
                                                                  wear tests, all of the samples were cleaned with toluene,
II. MATERIAL AND EXPERIMENTS                                      acetone, and hexane successively to have a clean surface.
                                                                  The initial arithmetic average roughness of the samples,
A. Heat treatment and characterization                            after these treatments and estimated using the profilome-
   A 14-mm-diameter rod of the Zr-based BMG, with a               ter, was ∼0.8 ␮m, whereas it is approximately 艋2 ␮m
commercial designation LM-1 was obtained from Liq-                after wear testing. Note that the latter would be signifi-
uidmetal Technologies Ltd. (Lake Forest, CA). The                 cantly affected by the presence of wear debris and oxide
nominal chemical composition of the alloy (in at.%) was           flakes that form during the wear process.
Zr42Ti15.5Cu14.5 Ni3.5Be24.5. Discs of ∼1.5 mm thickness             Unless otherwise mentioned, all of the wear tests were
were electrodischarge-machined from the rod. The pri-             conducted with an applied load of 100g for 2 h at a
mary objective of this work is to examine how structural          frequency of 50 Hz. The sliding distance for each pass
relaxation changes the wear characteristics of the amor-          was kept constant at 2 mm. A resistance heater attached
phous alloys. For this purpose, the BMG discs were an-            to the lower specimen holder allowed for control of the
nealed for 1 h in a flowing argon environment at 400,             temperature through a thermocouple that was located
469, 531, and 687 K, which correspond to 0.62, 0.73,              close to the specimen. All of the tests reported in this
0.83, and 1.07 times the Tg (643 K), respectively. The            article were conducted at 303 K.
last temperature, being above the Tg, induces crystalliza-           The HFRR instrument is capable of measuring the
tion of the BMG, whereas the first three temperatures of          friction coefficient, temperature, and the electrical con-
annealing induce structural relaxation. The as-cast and           tact resistance ⍀ between the specimen and the ball. The
annealed samples, polished to a 0.5-␮m finish, were               electronics unit applies a bias voltage through the ball,
characterized by optical microscopy, x-ray diffraction            measures ⍀, and scales it between 0% and 100%. When
(XRD) and differential scanning calorimetry (DSC).                the ⍀ was zero, metal-to-metal contact exists. When
   The variation in mechanical properties, specifically           there was no electrical contact between the specimen and
elastic modulus and hardness, of the BMG with the heat            the ball, ⍀ ⳱ 100%. Thus, ⍀ gives a qualitative indica-
treatment temperature were assessed by nanoindentation            tion of the presence or absence of an oxide film.
experiments using a Hysitron Tribometer (Hysitron Inc,               W was measured by scanning the wear scars using a
Minneapolis, MN) with a Berkovich tip. All of the tests           laser profilometer with a depth resolution of 1 ␮m and a
were conducted with a maximum load of 9 mN, with a                spatial resolution of 30 ␮m. For each case, at least five
prescribed loading rate of 0.36 mN/s with a 10-s hold             repeat experiments were conducted, and the average val-
time at the peak load. The loads, P, versus the depth-of-         ues are reported herein. Scanning electron microscopy
penetration, h, curves were analyzed by employing the             (SEM) of the wear scars was conducted to understand the

                                            J. Mater. Res., Vol. 22, No. 2, Feb 2007                                     265
H.W. Jin et al.: Reciprocating wear mechanisms in a Zr-based bulk metallic glass

wear mechanisms. Energy-dispersive spectroscopy was
performed, where necessary, to identify the nature of the
scales formed on the surface.

III. RESULTS
A. Structural relaxation
   Inspection of the polished cross section of the as-cast
alloy revealed that the BMG was not fully amorphous,
but contained crystalline regions with needle-shaped and
dendritic particles in the amorphous matrix. The volume
fraction of these crystalline particles was very small
(
H.W. Jin et al.: Reciprocating wear mechanisms in a Zr-based bulk metallic glass

TABLE I. Summary of the DSC experiments.

                                 Enthalpy at the                  Glass transition              Enthalpy at the           Peak crystallization
   Annealing                     glass transition,                 temperature,                 crystallization,             temperature,
temperature (°C)                    ⌬Hg (J/g)                        Tg (°C)a                      ⌬Hx (J/g)                   Tx (°C)a
As-cast                             6.79 ± 0.09                     370.3 ± 1.8                   80.7 ± 2.2                  464.4 ± 1.6
127 (0.62 × Tg)                     6.71 ± 0.25                     370.6 ± 1.7                   84.3 ± 5.4                  464.4 ± 1.9
196 (0.73 × Tg)                     6.42 ± 0.12                     369.6 ± 1.2                   83.9 ± 0.8                  463.5 ± 0.3
258 (0.83 × Tg)                     5.93 ± 0.03                     372.4 ± 1.6                   85.0 ± 1.6                  463.3 ± 1.0
414 (1.07 × Tg)                         …                               …                             …                           …

Values given as mean ± SD.
a
 Inflection temperature.

                                                                              (profuse shear-band activity) mode. However, Greer
                                                                              et al.18 have suggested that the smoothening of the P-h
                                                                              curves was an instrumental artifact rather than a transi-
                                                                              tion in flow behavior.
                                                                              C. Friction and wear properties
                                                                                 Figure 3 shows a typical plot of the ⍀ and the friction
                                                                              coefficient ␮ as a function of time of the reciprocation of
                                                                              the WC ball on the as-cast BMG disc. Note that ⍀ re-
                                                                              flects the electrical resistance of the contact, therefore
                                                                              zero or low ⍀ means metal-to-metal contact, whereas a
                                                                              high ⍀ value can be inferred to be due to the formation
                                                                              of a thick, nonconducting tribolayer. The details of this
                                                                              transition are discussed in Sec. III. D. As seen from Fig.
                                                                              3, the ␮ value initially was low (∼0.62), but after about
                                                                              10 min, a sharp transition was seen, with ␮ increasing to
                                                                              about 1. Further, the noise in the ␮ variation with time
                                                                              increased markedly. A concomitant transition in ⍀, from
FIG. 2. Typical load, P-h curves obtained on the as-cast and annealed
                                                                              a near-zero value to 100% was also noted. This behavior
alloys. The inset highlights the reduced susceptibility for serrated flow     was observed in all of the different annealing conditions
(a result of the shear-band formation and propagation) with increasing        tested. Average values of ␮ before and after the transition
annealing temperature.                                                        as well as the average transition times tT are listed in
                                                                              Table II. Here, tT is defined as the time at which the ⍀
from 6.91 GPa in the as-cast alloy to 8.49 GPa in the                         crosses 50% for the first time.
alloy annealed at 531 K, a marked increase (to 10.8 GPa)                         The data presented in Table II show that the friction
occurred in the crystallized alloy.                                           coefficient after the transition ␮a was, within the experi-
   The P-h curves obtained during the nanoindentation of                      mental scatter, independent of the heat treatment with a
amorphous alloys generally tended to be serrated, as seen                     value of ∼0.95. This was also true for the friction coef-
from the response of the as-cast alloy shown in Fig. 2. It                    ficient before the transition ␮b for the structurally relaxed
has been well established that the serrations are a result                    glasses (i.e., those annealed at temperatures below the
of the nucleation and propagation of shear bands, the                         Tg), with a value of ∼0.62. Likewise, the tT values were
carriers of plasticity in amorphous metals. Since defor-                      also similar. In the case of the crystallized BMG, a
mation through shear banding is inhomogeneous in na-                          slightly higher value for ␮b (∼0.70) was noted. However,
ture, each shear-band activity is associated with a dis-                      the tT in this case was much smaller (20 s) when com-
crete displacement burst, which resulted in a serration in                    pared to the tT of ∼600 s measured in the as-cast and
the loading part of the P-h curve.16 With increasing an-                      relaxed glasses. This implies that the transition in the
nealing temperature, we noted that the P-h curves got                         crystallized glass was essentially instantaneous.
smoother, with the fully crystallized alloy showing no                           A typical profile, measured using a laser-scanning pro-
serrations at all, which is highlighted through the inset in                  filometer, of the wear scar produced after 2 h of testing
Fig. 2. Schuh et al.17 have argued that P-h curves of                         is shown in Fig. 4(a). As seen, the wear scar is symmetri-
amorphous alloys are smoother at higher loading rates                         cal and uniform. A line across the middle of the scar [Fig.
because of a transition in flow behavior from a hetero-                       4(b)] indicates a cylindrical groove. More importantly, it
geneous (discrete shear-band activity) to homogenous                          shows the edges to be flat, implying that there is no

                                                     J. Mater. Res., Vol. 22, No. 2, Feb 2007                                               267
H.W. Jin et al.: Reciprocating wear mechanisms in a Zr-based bulk metallic glass

TABLE II. Summary of the indentation, friction, and wear experiments.

                    Reduced        Maximum                        Micro-Vickers                     Friction        Friction
   Annealing        modulus,         depth,      Nanohardness       hardness        Transition     coefficient    coefficient
temperature (°C)    Er (GPa)       hmax (nm)a       (GPa)            (VHN)         time, tT (s)   before tT, ␮b   after tT, ␮a   W (×106 ␮m3)
      As-cast      106.2 ± 0.6     254.5 ± 2.5      6.9 ± 0.2      753.4 ± 7.3     550 ± 131      0.62 ± 0.02     0.93 ± 0.05     60.6 ± 50.1
       127         110.3 ± 3.2     250.6 ± 5.9      7.1 ± 0.4      752.8 ± 8.9     650 ± 151      0.63 ± 0.01     0.95 ± 0.01     80.2 ± 25.6
       196         104.5 ± 4.6     250.1 ± 6.7      7.4 ± 0.4      740.6 ± 4.4     605 ± 105      0.62 ± 0.02     0.95 ± 0.01     58.9 ± 15.3
       258         104.2 ± 0.6     242.7 ± 0.8      8.5 ± 0.16     740.8 ± 4.7     517 ± 221      0.61 ± 0.01     0.96 ± 0.05     40.0 ± 19.9
       414         119.2 ± 3       220.0 ± 3.4     10.8 ± 0.7      902.2 ± 18.8       20           0.7 ± 0.03     0.95 ± 0.02      5.4 ± 1.0

Values given as mean ± SD.
a
 Depth at the maximum load.

                                                                         confirmed that the scales observed indeed contained
                                                                         large concentrations of oxygen [Fig. 6(d)] with the pres-
                                                                         ence of a strong oxygen peak, whereas the spectrum
                                                                         obtained from other areas did not show an oxygen peak
                                                                         [Fig. 6(c)]. Among the alloying elements present in the
                                                                         BMG, Zr, titanium (Ti), and beryllium (Be) are prone to
                                                                         easy oxidation. Fu et al.10 reported similar layers con-
                                                                         taining large amounts of oxygen due to sliding wear in
                                                                         air, whereas they were absent when tested in a vacuum.
                                                                         The tribolayer produced in air may indeed be some
                                                                         mixed oxide; however, further studies on its chemical
                                                                         composition or structure (amorphous or crystalline) were
                                                                         not performed in this study. Hence, it is referred to sim-
                                                                         ply as a tribolayer.
                                                                            The tribolayers appear to crack extensively, with the
                                                                         periodically spaced cracks perpendicular to the recipro-
                                                                         cating ball direction, as shown in Fig. 7(a). These
                                                                         cracked films eventually peel off, due to the action of the
FIG. 3. Typical variation of ⍀ and ␮ as a function of time. Note that
⍀ ⳱ 0% implies no tribolayer, whereas ⍀ ⳱ 100% implies a com-            tangential stress applied by the sliding WC ball, exposing
plete loss of contact.                                                   the unoxidized BMG. Such a peeled-off region is shown
                                                                         in Fig. 7(b), with EDAX values obtained from this region
                                                                         showing the absence of oxygen. Figure 7(b) also indi-
pileup of material (due to plastic flow of the material
                                                                         cates that some bulk material being chipped off along
underneath the WC ball) around the edges. The W during
                                                                         with the scale (i.e., the tribolayer when it peels off) does
the wear process is estimated by a three-dimensional in-
                                                                         not debond cleanly along the oxide/bulk interface.
tegration process and is listed in Table II. Dry sliding
                                                                            To investigate the microscopic origins of the step-like
wear in most metallic materials was observed to obey
                                                                         transition in ⍀ and ␮ variations with time (Fig. 4), wear
Archard’s law, W ⳱ K(SN/H), where S is the total sliding
                                                                         tests were interrupted after 300 s (recall that the transition
distance, N is the normal load, and K is the wear coeffi-
                                                                         time is around 600 s), and the resulting scars were ex-
cient. The least-square fit of the above equation is used
                                                                         amined. The low- and high-magnification images ob-
for the data plotted in Fig. 5. While considerable scatter
                                                                         tained from these scars are shown in Figs. 8(a) and 8(b),
exists in the data (especially in the as-cast alloy), a gen-
                                                                         respectively. Clearly, no scales could be seen, although
eral trend of a decrease in W with an increase of nano-
                                                                         indications of the early stages of tribolayer formation
hardness (annealing temperature) can be seen.
                                                                         could be seen. Also, the surface morphology was much
                                                                         smoother, indicating that the mechanism of material re-
D. Wear mechanisms                                                       moval was different in this regime. Sliding wear tests
   The worn surfaces were examined by SEM to under-                      conducted in vacuum by Fu et al.10 show that a soft layer
stand the microscopic reasons for the wear of the BMG.                   develops during sliding, with a low friction coefficient,
Figure 6(a) is a lower magnification image of the wear                   and that the wear rate is lower than when sliding in air.
scar in the as-cast BMG, showing abrasive wear marks.                    These observations, in conjunction with those made in
A higher magnification image [Fig. 6(b)] shows that the                  the present study, indicate that the micromechanism of
worn surfaces were partially covered with tribofilms. En-                wear in the BMG is abrasive wear, with the high-oxygen-
ergy dispersive x-ray spectroscopy (EDAX) analyses                       containing tribolayers acting as the abrasive medium

268                                                J. Mater. Res., Vol. 22, No. 2, Feb 2007
H.W. Jin et al.: Reciprocating wear mechanisms in a Zr-based bulk metallic glass

               FIG. 4. (a) Topography of the wear scar and (b) a line scan across the scar showing the profile of the scar.

                                                                       any features and appeared to be very smooth. Most im-
                                                                       portantly, no tribolayers could be seen. EDAX spectra
                                                                       obtained from regions marked 1, 2, and 3 in Fig. 9(b)
                                                                       were identical. An example is shown in Fig. 9(c).

                                                                       IV. DISCUSSION
                                                                          A gradual reduction in the wear rate with increasing
                                                                       annealing temperature can be seen in Fig. 5. This reduc-
                                                                       tion in wear rate was due to the gradual increase in the
                                                                       hardness with the annealing temperature. Van den Beu-
                                                                       kel and Sietsma19 developed a model that described the
                                                                       functional form of the DSC curves based on the free-
                                                                       volume theory. According to their model, for every su-
                                                                       percooled liquid, there exists an equilibrium free-
                                                                       volume, vfe, at a given temperature. In the as-cast alloy,
                                                                       an excess amount of free volume, vf > vfe, is frozen-in
                                                                       due to the nonequilibrium processing conditions, which
FIG. 5. Variation of the W measured during high-frequency-             during continuous heating in a DSC experiment annihi-
reciprocating wear tests, as a function of H of the BMG.               lates and approaches vfe. The reduction in vf gives rise to
                                                                       heat release ⌬H, when the glass sample is heated in a
(because of their high hardness and low toughness, the lat-            DSC. It was experimentally verified by Slipenyuk and
ter facilitating comminution of the films once they peel off).         Eckert20 that ⌬Hg ⬀ ⌬vf. Since the free volume gets
   The SEM images obtained from the BMG samples that                   reduced during the structural relaxation anneals, ⌬Hg
were annealed at temperatures below the Tg were very                   gradually decreases with increasing annealing tempera-
similar, and were consistent with the ⍀, ␮, and W data                 ture (Table I), implying a systematic reduction in the
given in Table II, indicating that the wear mechanisms do              ability of the BMG to accommodate plastic deformation.
not change significantly with relaxation heat treatments               The reduction in the serration step sizes with increasing
below the Tg. Wear scars in the crystallized BMGs (i.e.,               annealing temperature indicates that, at the nanoscale,
the samples annealed at 687 K) showed a different wear                 increasing the structural relaxation annealing tempera-
morphology altogether. The low- and high-magnification                 tures leads to a reduction of free volume. This reduction
SEM micrographs obtained from the wear scars in the                    of free volume diminishes the ability of the BMG to
crystallized alloy are shown in Fig. 9. In contrast to those           accommodate plastic deformation, which is consistent
shown in Figs. 7(a) and 7(b), respectively, which were                 with earlier reports on annealing induced embrittlement
obtained at the same magnifications but in the as-cast                 in BMGs.11,12
alloy, wear scars in the crystallized alloy are devoid of                 Interestingly, the microhardness of the BMG appeared

                                                J. Mater. Res., Vol. 22, No. 2, Feb 2007                                      269
H.W. Jin et al.: Reciprocating wear mechanisms in a Zr-based bulk metallic glass

FIG. 6. (a) A lower magnification SEM image of the wear scar in the as-cast BMG, showing abrasive wear marks. (b) A higher magnification
image showing a tribolayer formed on the wear surface. EDAX spectrums obtained from (c) the region marked 1 and (d) the region marked 2 in
(b). Note the presence of a strong oxygen peak in the latter, confirming that the scale indeed contains a large amount of oxide.

to be relatively insensitive to the relaxation anneals (with           pear to alter the wear performance/friction coefficient
Vickers hardness number data ranging between 740 and                   significantly. One possibility that needs consideration is
753), whereas a marked rise was seen in the crystallized               the role of localized frictional heating and flow soften-
alloy, similar to that seen in nanohardness. The origin for            ing24 of the BMG. If the temperature rise at the surface
the differences in the nanohardness and microhardness of               layer due to vigorous scratching of the BMG against the
the alloys that were annealed below the Tg is discussed as             hard surface is significant, it could lead to a transition
follows. The deformation underneath an indenter is in-                 from a heterogeneous to a homogeneous mode of
trinsically stable (i.e., the contact area always increases            deformation.25 This in turn can drastically alter the
unless it is interrupted by cracking in the case of brittle            wear performance. However, the average sliding speed
materials being indented with sharp indenters) with an                 (∼0.2 mm/s) is not high enough to give large temper-
increasing depth of penetration of the indenter. The stress            ature increases from friction and hence to alter defor-
singularity at the tip of the Vickers indenter induces plas-           mation mechanisms.
tic deformation through shear banding, even in the an-                    The experimental results and observations indicate
nealed material, instead of cracking.21 It is possible that            that the predominant wear mechanism in the as-cast and
the stress singularity at the indenter tip is mitigated                structurally relaxed Zr-based BMG investigated in this
through the homogeneous deformation of the glass,                      work was oxidation of the alloy. This is not surprising
which is induced by the viscoplasticity that is generally              considering the high chemical affinity of Zr, Ti, and Be
dominant at low strain rates, followed by shear-band nu-               with oxygen to form respective oxides. The formation of
cleation. The high stresses that prevail under the Vicker’s            high-oxygen-containing tribolayers and their subsequent
indenter (estimated to be equal to 35.8%) could promote                debonding under the action of the reciprocating WC ball
shear banding.22 While the nanoindentation is highly                   constitutes the main mechanism of wear in the as-cast
sensitive to relatively small changes (reflected in the                and structurally relaxed alloys. Then, the tT is considered
P-h curves as well as in the nanohardness), microin-                   to be the time required for sufficiently thick tribolayers to
dentation measures the response of the BMG averaged                    form. The residual stress in the tribofilm (that arises due
over a significantly larger volume and hence is insensi-               to the differences in the molar volumes of the oxide layer
tive to the underlying subtle changes in micromecha-                   and that of the substrate that is oxidized) increases with
nisms of deformation.                                                  the increasing tribolayer thickness. This, coupled with
   It has been proposed by previous studies10,23 that the              the applied frictional force, appears to determine the
wear/friction properties could be affected by plastic de-              process of delamination.
formation (i.e., shear-banding propensity), since the wear                An interesting observation in the present study is that
process essentially involves deformation. Based on the                 the crystallized alloy exhibited minimal scale formation,
present investigation, structural relaxation does not ap-              while the as-cast and structurally relaxed alloys showed

270                                             J. Mater. Res., Vol. 22, No. 2, Feb 2007
H.W. Jin et al.: Reciprocating wear mechanisms in a Zr-based bulk metallic glass

                                                                       FIG. 8. (a) Low-magnification and (b) high-magnification images of
                                                                       the wear scars obtained after 300 s (before the transition) from the
                                                                       as-cast BMG.

                                                                       has been suggested that the formation of crystalline
                                                                       Zr2Cu and Zr2Al was responsible for improving the
                                                                       oxidation resistance due to their higher oxidation
                                                                       resistance and promotion of the development of Al2O3
                                                                       and oxides of copper. These results indicate that the
                                                                       crystalline phase of Zr2Cu is much more resistant to
                                                                       oxidation than the amorphous state. Paljevia et al.28 re-
                                                                       ported that the oxidation reaction onset temperature of
                                                                       ZrAl2 is ∼700 K.
FIG. 7. (a) SEM micrograph showing extensive cracking of the tri-
bolayer formed on the wear surface. (b) Micrograph showing a region       Following annealing above the Tg, the oxidation resis-
where the layer has peeled off. (c) EDAX spectrum from the circled     tance remarkably improved with crystallization so that
region of (b) showing the absence of oxygen.                           film formation was rarely observed, possibly due to its
                                                                       insignificant amount. As the result of their superior
extensive scale formation. The oxidation behaviors of                  oxidation resistance, the crystallized alloys were worn
Zr-based amorphous alloys have been reported by sev-                   without the intervention of the oxidation under testing
eral researchers.26–29 It has been reported by Wu et al.27             conditions in the present study. However, there are contra-
that the oxidation rate of Zr–Al–Cu metallic glass speci-              dicting reports on the effect of crystallization on oxida-
mens was higher than that of annealed samples above the                tion. Kai et al.29 reported that the oxidation rate of Zr–
Tg and that the oxidation resistance of the specimen de-               Cu–Al–Ni BMG is lower than that of the crystalline alloy
pended on the amount of crystalline precipitates. It                   composed of Zr2Cu, Zr2Ni, and ZrAlCu2 at temperatures

                                                 J. Mater. Res., Vol. 22, No. 2, Feb 2007                                              271
H.W. Jin et al.: Reciprocating wear mechanisms in a Zr-based bulk metallic glass

                                                                        above 698 K. It is noteworthy that the oxidation resis-
                                                                        tance of Zr-based amorphous alloys strongly depends on
                                                                        the chemistry as well as the degree of crystallinity.
                                                                        Therefore, the effect of crystallization/chemistry on the
                                                                        oxidation behavior of the Zr-based BMG should be con-
                                                                        sidered when these alloys are designed for use under
                                                                        wear/friction circumstances where optimized stability
                                                                        against oxidization is required. Although it is possible for
                                                                        the surface layers of the crystalline BMG to reamorphize
                                                                        during sliding wear (which in turn oxidize in a manner
                                                                        similar to that of amorphous alloys), it does not appear to
                                                                        be the case in the present context. The combination of
                                                                        relatively low sliding velocities and loads may be the
                                                                        reason for this.

                                                                        V. SUMMARY
                                                                           An experimental investigation into the high-frequency,
                                                                        dry-sliding wear behavior of a Zr-based BMG was con-
                                                                        ducted. The BMG was systematically annealed at differ-
                                                                        ent temperatures (both below and above the Tg), and the
                                                                        role of structural relaxation on the wear resistance of the
                                                                        BMG was investigated. On the basis of the experimental
                                                                        results the following conclusions can be drawn.
                                                                           (1) The wear behavior of both the as-cast and relaxed
                                                                        glasses was dominated by the oxidation of the surface
                                                                        layers. Formation of the oxide layers and their subse-
                                                                        quent peel-off constituted the main mechanism of wear
                                                                        for the experimental conditions of this study. A sharp
                                                                        transition in the oxide film thickness complemented by a
                                                                        sharp increase in ␮ after ∼600 s were observed.
                                                                           (2) With increasing annealing temperature (below the
                                                                        Tg), the nanohardness of the BMG increased. This hard-
                                                                        ness increase was proposed to be due to the reduced
                                                                        propensity for shear-band formation, caused by the an-
                                                                        nihilation of the free volume present in the BMG. A
                                                                        reasonable correlation between W and nanohardness was
                                                                        identified.
                                                                           (3) The crystallized alloy has considerably higher
                                                                        wear resistance and also higher hardness. In contrast to
                                                                        the as-cast and structurally relaxed alloys, the wear
                                                                        mechanism was predominantly abrasive wear in the crys-
                                                                        tallized alloy. The formation of oxide scales was insig-
                                                                        nificant, indicating superior oxidation resistance in crystal-
                                                                        lized alloy to the as-cast and structurally relaxed glasses.

                                                                        ACKNOWLEDGMENTS
                                                                           The authors acknowledge S. Bellamare, H. Jun,
                                                                        J. Krylowski, C. Marzinsky, A. Schwartzman, T. Smith,
                                                                        and M.J.N.V. Prasad for their assistance with different
FIG. 9. SEM micrographs of the wear scars in the crystallized alloy     experiments reported in this article. We are grateful to
showing (a) lower magnification and (b) higher magnification
                                                                        Professor S. Suresh of Massachusetts Institute of Tech-
views. EDAX spectrums obtained from regions marked 1, 2, and 3
were identical, as shown in (c). Compare these images with those        nology and Dr. M. Webster of ExxonMobil Research and
obtained at the same magnification and shown in Figs. 7(a) and 7(b),    Engineering Company (EMRE) for their advice and sup-
respectively.                                                           port. U. Ramamurty acknowledges the support he has

272                                               J. Mater. Res., Vol. 22, No. 2, Feb 2007
H.W. Jin et al.: Reciprocating wear mechanisms in a Zr-based bulk metallic glass

received from ExxonMobil Research and Engineering                        15. W.C. Oliver and G.M. Pharr: An improved technique for deter-
Company for a summer visit to their Corporate Strategic                      mining hardness and elastic modulus using load and displacement
                                                                             sensing indentation experiments. J. Mater. Res. 7, 1564 (1992).
Research Labs, which facilitated the research reported in
                                                                         16. C.A. Schuh, T.G. Neih, and Y. Kawamura: Rate dependence of
this article. Constructive comments and suggestions for                      serrated flow during nanoindentation of a bulk metallic glass.
the overall improvement of the article by Dr. S.                             J. Mater. Res. 17, 1651 (2002).
Karthikeyan as well as anonymous reviewers are grate-                    17. C.A. Schuh, A.C. Lund, and T.G. Nieh: New regime of homoge-
fully acknowledged.                                                          neous flow in the deformation map of metallic glasses: Elevated
                                                                             temperature nanoindentation experiments and mechanistic mod-
                                                                             eling. Acta Mater. 52, 5879 (2004).
REFERENCES                                                               18. A.L. Greer, A. Castellero, S.V. Madge, I.T. Walker, and
                                                                             J.R. Wilde: Nanoindentation studies of shear banding in fully
 1. W. Klement, R.H. Willens, and P. Duwez: Noncrystalline struc-            amorphous and partially devitrified metallic alloys. Mater. Sci.
    ture in solidified gold-silicon alloys. Nature 187, 869 (1960).          Eng., A 375–377, 1182 (2004).
 2. A. Inoue: Bulk amorphous and nanocrystalline alloys with high        19. A. Van Den Beukel and J. Sietsma: The glass transition as a free
    functional properties. Mater. Sci. Eng., A 304–306, 1 (2001).            volume related kinetic phenomenon. Acta Metall. Mater. 38, 383
 3. S. Schneider: Bulk metallic glasses. J. Phys.: Condens. Matter 13,       (1990).
    7723 (2001).                                                         20. A. Slipenyuk and J. Eckert: Correlation between enthalpy change
 4. W.H. Wang, C. Dong, and C.H. Shek: Bulk metallic glasses.                and free volume reduction during structural relaxation of
    Mater. Sci. Eng., R 44, 45 (2004).                                       Zr55Cu30Al10Ni5 metallic glass. Scripta Mater. 50, 39 (2004).
 5. F. Spaepen: A microscopic mechanism for steady state in homo-        21. M.N.M. Patnaik, R. Narasimhan, and U. Ramamurty: Spherical
    geneous flow in metallic glasses. Acta Metall. 25, 407 (1977).           indentation response of metallic glasses. Acta Mater. 52, 3335
 6. M. Anis, W.M. Rainforth, and H.A. Davies: Wear behavior of               (2004).
    rapidly solidified Fe68Cr18Mo2B12. Wear 172, 135 (1994).             22. U. Ramamurty, S. Jana, Y. Kawamura, and K. Chattopadhyay:
 7. T. Gloriant: Microhardness and abrasive wear resistance of me-           Hardness and plastic deformation in a bulk metallic glass. Acta
    tallic glasses and nanostructured composite materials. J. Non-           Mater. 53, 705 (2005).
    Cryst. Solids 316, 96 (2003).                                        23. R. Bhowmick, R. Raghavan, K. Chattopadhyay, and U. Ramamurty:
 8. G. Li, Y.Q. Wang, L.M. Wang, Y.P. Gao, R.J. Zhang, Z.J. Zhan,            Plastic flow softening in a bulk metallic glass. Acta Mater. 54,
    L.L. Sun, J. Zhang, and W.K. Wang: Wear behavior of bulk                 4221 (2006).
    Zr41Ti14Cu12.5Ni10Be22.5 metallic glass. J. Mater. Res. 17, 1877
                                                                         24. X. Fu, M.L. Falk, and D.A. Rigney: Sliding behavior of metallic
    (2002).
                                                                             glass: Part II. Computer simulations. Wear 250, 420 (2001).
 9. P.J. Blau: Friction and wear of a Zr-based amorphous alloy under
    dry and lubricated conditions. Wear 250, 431 (2001).                 25. C.A. Schuh and T.G. Neih: A survey of instrumented indentation
10. X. Fu, T. Kasai, M.L. Falk, and D.A. Rigney: Sliding behavior of         studies on metallic glasses. J. Mater. Res. 19, 46 (2004).
    metallic glass: Part I. Experimental investigations. Wear 250, 409   26. S.K. Sharmal, T. Strunskus, H. Ladebusch, and F. Faupel:
    (2001).                                                                  Surface oxidation of amorphous Zr 65 Cu 17.5 Ni 10 A l7.5 and
11. P. Murali and U. Ramamurty: Embrittlement of a bulk metallic             Zr46.75Ti8.25Cu7.5Ni10Be27.5. Mater. Sci. Eng., A 304–306, 747
    glass due to sub-Tg annealing. Acta Mater. 53, 1467 (2005).              (2001).
12. U. Ramamurty, I.M.L. Lee, J. Basu, and Y. Li: Embrittlement of       27. Y. Wu, T. Nagase, and Y. Umakoshi: Effect of crystallization
    a bulk metallic glass due to low temperature annealing. Scripta          behavior on the oxidation resistance of a Zr–Al–Cu metallic glass
    Mater. 47, 107 (2002).                                                   below the crystallization temperature. J. Non-Cryst. Solids 352,
13. N. Nagendra, U. Ramamurty, T.T. Goh, and Y. Li: Effect of                3015 (2006).
    crystallinity on the impact toughness of a La-based bulk metallic    28. M. Paljevia: High-temperature oxidation behaviour in the Zr-Al
    glass. Acta Mater. 48, 2603 (2000).                                      system. J. Alloys Compd. 204, 119 (1994).
14. J. Basu, N. Nagendra, Y. Li, and U. Ramamurty: Microstructure        29. W. Kai, H.H. Hsieh, T.G. Nieh, and Y. Kawamura: Oxidation
    and mechanical properties of partially-crystallized la-based bulk        behavior of a Zr–Cu–Al–Ni amorphous alloy in air at 300–425 °C.
    metallic glass. Philos. Mag. 83, 1747 (2003).                            Intermetallics 10, 1260 (2002).

                                                   J. Mater. Res., Vol. 22, No. 2, Feb 2007                                               273
You can also read