Full text of DOI:10.1557/JMR.2003.0388

Selective modification of the tribological properties of
aluminum through temperature and dose control in oxygen
plasma source ion implantation
a)
M. Bolduc, B. Terreault, A. Reguer, E. Shaffer, and R.G. St-Jacques
´
Institut National de la Recherche Scientifique–Energie, Mat e´ riaux et T e´ l e´ communications
(INRS–EMT), Universit e´ du Qu e´ bec, Varennes, Qu e´ bec J3X 1S2, Canada
(Received 7 July 2003; accepted 2 September 2003)
Improvements in the tribological properties of pure aluminum and “aeronautical” alloy
1
7
2
AA7075-T651 were obtained by oxygen-ion implantation [(0.7 to 5) × 10 O/cm ,
0 keV] using our pulsed electron cyclotron resonance plasma source. This oxygen
3
plasma source ion implantation process produced oxide nanoprecipitates that enhanced
the hardness up to three times in the surface layer and caused reductions in the scratch
depths and the friction coefficients by similar factors. A spectrum of tribological
properties was obtained depending on temperature and ion dose. Temperature
measurement and control were obtained through an integrated thermocouple and by
changing the duty-cycle of the microwave source. The oxygen content and the
depth-resolved chemical composition were measured and optimized using x-ray
photoelectron spectroscopy (XPS) combined with Ar-ion etching. The tribological
properties were investigated by (i) depth-sensing nanoindentation for hardness and
Young’s modulus, (ii) scratching and scratch-depth measurement via atomic force
microscopy (AFM), and (iii) friction force measurements using AFM. Low-temperature
(ഛ160 °C) implantations with optimal O-ion doses produced, in both pure and alloyed
Al, an approximately 50-nm-thick, smooth, and extremely fine-grained metal–alumina
nanocomposite. The resulting surface was hard and stiff but nonbrittle and displayed
high scratch resistance and low friction. High-temperature (∼430 °C) implantation had
different effects on pure Al and AA7075. On pure Al, it produced a very hard but
brittle Al O layer for which yield points (displacement excursions) were observed at
2
3
critical load values in the nanoindentation force–displacement curves. On AA7075,
XPS chemical profiling revealed an effect of extreme Mg surface segregation and
complete Al surface depletion; MgO crystallites formed a rather rough but surprisingly
thick layer (>100 nm). The resulting AA7075 surface showed a hardness increase that
was substantial but slightly smaller than that obtained at low temperature.
I. INTRODUCTION
high residual stress at the Al/AlN interface. Oxygen-
ion implantation was then studied by the Sandia Labs
Aluminum components have numerous engineering
applications, such as aerospace materials. A promising
new application is microelectromechanical systems, es-
4
,5
group toward avoiding this undesirable effect by form-
ing a hard nanocomposite produced by oxide precipita-
tion. The synthesis of very fine oxide particles blocks
1
pecially high-speed low-actuation-voltage electrostatic
actuators, because of the high electrical conductivity of
aluminum. However, the poor performance of Al parts in
sliding contact due to their low surface strength is a
serious disadvantage. Ohira et al. investigated the im-
provement of Al surface mechanical properties by ion
implantation. They used nitrogen ions to form a hard
surface layer, but which delaminated easily due to the
6
dislocation motion and leads to surface hardening. How-
ever, ion-beam implantation is neither practical nor cost-
effective for most foreseeable applications. As an
alternative, our group initiated the use of oxygen plasma
source ion implantation (OPSII) with a low-pressure,
high-density electron cyclotron resonance (ECR)
2
,3
7
plasma. Observations indeed showed the formation of a
metal–oxide nanocomposite with enhanced mechanical
8
properties.
8
In that work on an Mg-containing Al alloy, we re-
a)
ported an extremely deep oxygen penetration and, con-
currently, observed a giant Mg surface segregation
Address all correspondence to this author.
e-mail: bolduc@inrs-emt.uquebec.ca
J. Mater. Res., Vol. 18, No. 12, Dec 2003
© 2003 Materials Research Society
2779
http://journals.cambridge.org
Downloaded: 13 Mar 2015
IP address: 130.56.64.29

M. Bolduc et al.: Selective modification of the tribological properties of aluminum
9
effect. This effect was confirmed in three different al-
for different times (5 min–24 h), and in different envi-
−4
loys containing magnesium (up to 4 wt.%). The oxide
was much thicker (>100 nm) than the incident ion range
∼30 nm), contrary to the case of pure Al, and it was
ronments (ultrahigh vacuum, 8 × 10 torr O , and am-
2
bient air). Others were heated up to 120 °C or 400 °C by
Ar plasma ion bombardment without oxygen.
(
almost entirely composed of MgO. Magnesium is known
to segregate to the surface of aluminum alloys under
high-temperature air oxidation due to its higher reactiv-
ity, but this effect is normally confined to a thin surface
The oxygen implantations were performed with our
ECR-PSII apparatus (Fig. 1). Our concept incorporates
several of the ideas of plasma immersion ion implanta-
1
7
tion (PIII), though it differs significantly in several as-
1
0–12
7,18
layer (<10 nm), below approximately 460 °C.
Ion
pects.
The plasma is generated by electron cyclotron
irradiation may also result in solute segregation in alloys.
Highly mobile point defects migrating away from their
production zone cause a transient enhancement of atomic
resonance at 2.45 GHz with the help of an approximately
875-gauss field produced by permanent magnets affixed
on the outside of the source. The plasma ions are ex-
tracted by a grounded grid and accelerated by a negative
dc high-voltage applied to the sample. This configura-
tion, in conjunction with the low-pressure environment
(ഛmtorr), allows monoenergetic implantation (e.g.,
30 keV). During the treatment, the target heats up rapid-
ly due to the high current density during the pulse
(>0.1 A cm ), producing a final equilibrium substrate
temperature. The adjustable duty-cycle of the micro-
wave source (0.2% to 2% for a pulse length of 110 ␮s)
is used to control the process temperature through a
thermocouple set inside the sample holder in contact
with the back of the sample. The materials studied were
1
3,14
diffusion (radiation-enhanced diffusion, or RED
).
These point defects can interact differently with under-
sized or oversized solutes and drag them along. This
leads to enrichment or depletion of some alloy compo-
nents, either around the peak of the radiation damage
profile or away from the peak. These effects were high-
lighted by experiments in which simultaneous argon–
oxygen ion irradiation in pure Al resulted in deeper O
−2
15,16
penetration.
This set of results suggested a synergy
between RED and Mg-oxidation-driven segregation.
This effect has important consequences for OPSII and
probably for many other ion implantation processes as
well. For instance, the deeper penetration may allow
thicker treated layers at low voltage, an obvious practical
and economic advantage, but segregation may give un-
desirable surface properties.
That earlier work was conducted without substrate
temperature control (except for the unpublished Ref. 16).
It was only deduced, from the postimplantation cooling
curves, that the temperature at the end of the treatment
had reached up to 425 °C. Obviously, substrate tempera-
ture is an extremely important parameter for both atomic
diffusivity and chemical reactivity. We therefore report
here new results obtained with a good control of the
temperature and which clarify several questions. To bet-
ter understand the oxidation and diffusion kinetics and
the tribological properties, comparative experiments of
oven oxidation were carried out. Argon plasma bombard-
ment was also used to identify possible purely physical
irradiation effects. A rather complete nanoscale tribologi-
cal characterization was carried out, not only in terms
of hardness, but also in terms of friction and wear. The
emphasis in this paper is the elucidation of the effects
of the temperature and the oxygen implantation profiles
on the tribological properties, with the goal of identifying
optimal treatments for specific applications.
FIG. 1. OPSII system: The ECR plasma source uses 2.45-GHz mi-
crowave pulses and permanent magnets in a multipole configuration.
The implantation chamber located below the ECR source is separated
from it by a grounded grid allowing ion extraction from the plasma.
The samples are mounted on a vertically movable holder on which a
negative dc high voltage up to 30 kV is maintained. Gas inlets and
turbomolecular pumping provide a controlled low-pressure environ-
ment. A thermocouple set inside the sample holder allows in situ
temperature measurements.
II. EXPERIMENTAL
To provide points of comparison regarding the tri-
bological properties and to assess independently the ef-
fects of temperature, unimplanted samples were
furnace-annealed at different temperatures (160–460 °C),
2
780
J. Mater. Res., Vol. 18, No. 12, Dec 2003
http://journals.cambridge.org
Downloaded: 13 Mar 2015
IP address: 130.56.64.29

M. Bolduc et al.: Selective modification of the tribological properties of aluminum
polycrystalline pure Al (1199-H18, 99.999% purity) and
an aeronautical heat-treated Al alloy (AA7075-T651)
the electron escape depth. The nominal XPS sensitivity
factors were used; they gave, for example, 57 at.% O in
alumina, affording good confidence in these factors.
High-resolution XPS can also provide binding informa-
tion, as shown in Fig. 3 for the Al 2p line profile, where
the oxide, with a 75-eV binding energy, is clearly dis-
tinguishable from the metallic Al peak at 72 eV. Back-
ground subtraction and peak fitting and integration give
the respective fractions of metal and oxide. AFM and
scanning electron microscopy with energy-dispersive x-
ray (SEM-EDX) analysis were used for morphological
and chemical surface characterization.
containing 2–3 wt.% Mg plus Zn, Cu, and Cr. The
2
samples were cut in 1 cm circular pieces and were pol-
ished mechanically with SiC and diamond slurries to a
mirrorlike aspect. They were cleaned by sonication in hot
acetone and methanol and were dried with a nitrogen jet.
The implantation process consisted of (i) a prebombard-
ment with 3-keV Ar ions to clean the surface, (ii) an
optional 30-keV Ar ion bombardment to heat it up to
approximately 430 °C in the case of high-temperature
implantation, and (iii) the 30-keV oxygen-ion implanta-
tion for 3–25 min, at variable duty-cycle as required to
maintain the temperature, up to doses in the range of (0.8
Continuous nanoindentation measurements were ob-
tained with a Hysitron Triboindenter XP (Minneapolis,
MN) featuring Berkovich three-sided pyramidal diamond
tips with curvature radii smaller than 300 nm. A series of
indentation forces in the 3 ranges of 100–450, 300–
1000, and 800–3000 ␮N, in steps of 50 or 100 ␮N, were
applied to each sample at 8 locations. The penetration
depth can be measured with an accuracy of a fraction of
nanometer due to an isolation chamber against external
vibrations. The hardness values were obtained from the
ratio of the applied load to the contact area between the
17
−2
to 4.0) × 10 O cm .
+
Based on plasma parameters, O2 ions are expected to
+
strongly dominate (∼90%) over O ions (∼10%), thus
95% of the ions have 15 keV per O atom. This is con-
firmed by the O implantation depth profile (at low tem-
perature and fairly low fluence), shown by squares in
Fig. 2, which closely agrees with that calculated by the
1
9
code STRIM (circles): projected range R ס 32 nm
p
+
and straggling ␴ ס 19 nm for 30-keV O2 ion in Al.
2
1
This and other depth profiles were determined by x-ray
indenter and the surface. An AFM image of indenta-
tions into unimplanted AA7075 is shown in Fig. 4(a).
Calibration procedures with sapphire were performed to
obtain a fit to the contact area as a function of the pen-
etration depth. From elastic contact theory, Young’s
modulus can be calculated using the Oliver and Pharr
+
photoelectron spectroscopy (XPS) and Ar ion etching
with a VG ESCALAB 2201-XL (Sussex, United King-
dom). The nominal etching rate of 2 nm/min was verified
by the very close agreement of the oxygen profile with
theory, cited above, and by the thickness of the native
2
0
22
oxide on pure Al. (This native oxide contribution was
subtracted in Fig. 2 to allow an easier comparison with
the simulation.) In addition, it was checked by atomic
force microscopy (AFM) that the roughness at the bot-
tom of the etching craters did not exceed a few nano-
meters; of the same order as the XPS resolution given by
(O&P) model and assuming that the Poisson ratio and
Young’s modulus of the diamond tip are 0.07 and 1140
2
2,23
GPa, respectively.
Only the data with penetration
depths greater than 20 nm were considered to minimize
2
3
the indentation size effect.
In addition, nanoscratch tests were performed by
means of an atomic force microscope with a Nanoscope
IIIa (Digital Instruments, Santa Barbara, CA) in the nano-
indentation mode using sharp Berkovich diamond tips
FIG. 2. Comparison between STRIM simulation and XPS profile of
oxygen implanted in AA7075 at low temperature (native oxide profile
subtracted out).
FIG. 3. High-resolution Al 2p line profiles, gathered at different
depths in AA7075 implanted with a dose of 1.2 × 10 O cm at low
temperature.
1
7
−2
J. Mater. Res., Vol. 18, No. 12, Dec 2003
2781
http://journals.cambridge.org
Downloaded: 13 Mar 2015
IP address: 130.56.64.29

M. Bolduc et al.: Selective modification of the tribological properties of aluminum
performed in ambient air (23 ± 1 °C, 35 ± 5% relative
humidity). These low loads produced scratches generally
shallower than approximately 50 nm. The individual
scratches can be imaged and the residual depth is found
from the cross sections of the AFM images; scratches
into unimplanted AA7075 are shown in Fig. 4(b). Note
that the surfaces had been polished to an approximately
5
-nm roughness, as measured by AFM, to facilitate these
measurements. As in the case of nanoindentation, applied
force versus penetration depth data (f–d curves) were
recorded. In the current study, the “parallel scan method”
2
5
was used. This method uses the fact that the friction
force also bends the cantilever along its longitudinal axis,
a mode for which the spring constant is well-known.
Thus, in AFM-nanoscratch testing (Fig. 5, right), the
cantilever-tip assembly is driven to (1) move toward
the sample, (2) contact and penetrate into the surface,
then (3) move parallel to the cantilever axis for a 2-␮m
length (with the sharp “prow” of the tip forward), and (4)
finally lift off the sample surface. Although the f–d
curves are not recorded during scratching itself, but only
at the beginning and end of the scratch while the tip is
moving into and then out of the sample, is it possible to
relate directly the difference in the vertical deflections
signals (Fig. 5, left) to the tangential forces (F ) created
T
during scratching. Assuming quasi-static forces acting on
the atomic force microscope cantilever-tip assembly dur-
ing scratching, a torque balance equation (M ס FN
×
d – F × d ) is used to obtain the tangential force from
1
T
2
2
5
the vertical displacement. By taking the difference in
vertical component signals between the load and unload
f–d curves, the plowing friction force is calculated [F ס
T
(
load – unload) × d /d , where d ס 265 ␮m and d ס
1 2 1 2
1
75 ␮m are the horizontal and vertical distances between
the probe tip and the cantilever end, respectively] and
used to estimate the friction coefficient (␮ ס F /
T
2
4,26
F ).
It was also assumed that the vertical de-
N
flection signal was not affected by a nonlevel sample.
Finally, to avoid lateral force components during scratch-
ing, a compensating lateral motion of the probe is applied
by the AFM software. In practice, perfect compensation
is difficult to achieve, so insuring good parallel load and
unload curves minimizes experimental uncertainties.
FIG. 4. AFM images of (a) nanoindentations with forces up to
000 ␮N and (b) nanoscratches at forces up to 30 ␮N on unimplanted
AA7075.
1
III. RESULTS AND DISCUSSION
A. Preliminary experiments: Oven oxidation and
argon plasma annealing
with an apex angle of 60° and which are mounted on a
very stiff stainless steel cantilever with a spring constant
Oxidations in air (35% relative humidity) were per-
formed at 430 °C for different times (5 min–24 h) and
for 5 h at different temperatures (350–460 °C). Figure 6
shows SEM micrographs of a pure Al sample (a) after
15 min in the furnace at 430 °C and (b) after 24 h in the
same conditions. Not unexpectedly, the oxide in (a) has
cracked under the high thermal stress after cooling to
2
4
of 200 N/m (manufacturer’s specifications ). The tip
curvature radius, determined from a scanning electron
microscope picture, was estimated at approximately
100 nm. The tip velocity during scratching was 1 ␮m/s.
A series of 5 measurements with normal loads (F ) rang-
N
ing from 10 to 30 ␮N with gradual steps of 5 ␮N were
2
782
J. Mater. Res., Vol. 18, No. 12, Dec 2003
http://journals.cambridge.org
Downloaded: 13 Mar 2015
IP address: 130.56.64.29

M. Bolduc et al.: Selective modification of the tribological properties of aluminum
FIG. 5. AFM force plot (left) for the diamond-tipped AFM cantilever on low-temperature-implanted AA7075. Cantilever-tip assembly (right) with
the corresponding laser deflection when the tip (1) approaches the sample, (2) contacts and penetrates into the surface, then (3) moves parallel to the
cantilever axis, and finally (4) lifts off the surface.
room temperature, and surface swelling is also observed.
On the other hand, after 24 h the surface (b) looks like a
freshly polished surface. This “healing” effect must be
due to a slow diffusion of Al atoms to the surface through
the grain boundaries and fissures in the oxide. Oxygen
diffusion along grain boundaries is faster than diffusion
of aluminum atoms on the surface or through grains.
Therefore, aluminum diffusion must be the limiting fac-
tor in the densification of the Al O surface. This is
2
3
consistent with the deconvoluted XPS O profile in Fig. 7,
which shows a flat top near the surface. Most importantly,
note that even after air oxidation for 24 h at 430 °C,
the oxide thickness is only 10 nm compared to 6 nm
for the native oxide. (The oxide thickness is here defined
as the depth at which the O concentration has dropped by
–
1/2
a factor e .) The O profiles do show a rather long tail
5% at 40 nm), which indicates that some O diffusion
(
does occur. [A long tail could be due to an artefact of
preferential Ar ion etching of the metallic Al as opposed
to the oxide during XPS analysis, but we do not think it
is the case here because the etching crater bottom was
checked to be smooth (by AFM) and compositionally
homogeneous (by SEM-EDX)]. In the case of the
samples annealed in ultrahigh vacuum (UHV) at 160 °C
or 430 °C during 90 min, no apparent fissures and very
thin O profiles without tail were observed, suggesting a
low level of oxidation.
Oxygen and Mg XPS depth profiles of AA7075 an-
nealed at 430 °C for different times in air or 5 h at
−4
different temperatures in 8 × 10 torr O are shown in
2
Figs. 8(a) and 8(b), respectively. High-temperature oxi-
dation is much more pronounced in the alloy than it was
in pure Al, and this oxidation is accompanied by intense
Mg segregation to the surface. The MgO layer is defi-
nitely thicker (i.e., ∼20 nm to ∼40 nm at the highest
temperatures) than the Al O layer on pure Al (Fig. 7). If
2
3
we now compare (a) and (b), the oxide thickness for an
O partial pressure of 150 torr (ambiant air) is only
FIG. 6. SEM micrographs of pure Al surface after (a) 15 min of air
oxidation at 430 °C and (b) 24 h in the same conditions.
2
J. Mater. Res., Vol. 18, No. 12, Dec 2003
2783
http://journals.cambridge.org
Downloaded: 13 Mar 2015
IP address: 130.56.64.29

M. Bolduc et al.: Selective modification of the tribological properties of aluminum
−
4
doubled compared to that at 8 × 10 torr. This shows
that oxygen absorption is not the limiting factor in ther-
mal oxidation, but rather atomic diffusion. It is clear that
the alloy oxidation rate cannot be due to thermal diffu-
sion only (as in pure Al). Preferential Mg oxidation act-
ing as a chemical driving force and involving a strong
segregation has to be invoked. No evidence of Mg seg-
regation or enhanced oxidation was observed in alloy
samples annealed at lower temperature (<300 °C).
Our pure Al samples exhibit a bulk (500-nm depth)
hardness of 0.31 ± 0.02 GPa and a Young’s modulus of
68 ± 4 GPa when using a value of 0.33 for the Poisson
ratio. These values compare well with those found in
materials reference data (i.e., 0.3 GPa and 64 GPa, re-
2
7
spectively). In the case of AA7075, the bulk hardness
(1.45 ± 0.08 GPa) and Young’s modulus (84 ± 5 GPa)
are also similar to reference data (i.e., 1.5 GPa and
2
7,28
74 GPa).
During the nanoindentation measurements
on air-oxidized pure Al, large and sudden displacement-
excursions in the f–d curves were frequently observed. In
Fig. 9, such multiple yield points, also called “pop-in”
motions, forming a “staircase” may be observed. This
phenomenon is not observed in the freshly polished
samples or in those annealed at low temperature or in the
alloy. This anomaly is generally associated with disloca-
tion emission and/or a phase transition and/or an oxide
2
9
break-through event. Our observations are consistent
with other empirical findings and theoretical consider-
ations: (i) Yield points occur in oxide-covered materials
3
0
with low dislocation density. (ii) The hard oxide crust
3
1
arrests dislocation propagation and also spreads the in-
3
2
denter load over a wider area, giving an initially highly
elastic response. (iii) The material fails suddenly when
3
3
the yield stress of the substrate is reached locally; this
process can repeat as soon as the yield stress is again
exceeded at some point further away from the center of
the contact area. So, in our case, after mild annealing
only (160 °C), there are no yield points either in pure or
alloyed Al, probably due to the cold work resulting from
FIG. 7. XPS profiles of pure Al oxidized in air during 2 h or 24 h at
4
30 °C . The high-resolution spectra for Al were deconvoluted into
metallic and oxidized fractions (as in Fig. 3).
−
4
FIG. 8. XPS O (left) and Mg (right) profiles in AA7075: (a) air oxidized for different times at 430 °C and (b) oxidized in 8 × 10 torr O for
2
5
h at different temperatures.
2
784
J. Mater. Res., Vol. 18, No. 12, Dec 2003
http://journals.cambridge.org
Downloaded: 13 Mar 2015
IP address: 130.56.64.29

M. Bolduc et al.: Selective modification of the tribological properties of aluminum
polishing preventing condition (i) (this cold work is
Figure 11 displays typical nanoindentation f–d curves
for an alloy sample implanted at low temperature for five
increasing values of maximum load (on different spots);
two curves for an unimplanted sample are also shown.
One notes the excellent reproducibility of the measure-
manifested as a surface hardening, see Sec. III. B). In
contrast, the high-temperature oxidation of pure Al has
thoroughly annealed the dislocations. On the other hand,
the alloy is probably too hard to show yield points, even
under high-temperature oxidation. Finally, in the case of
UHV annealing of the pure Al, the oxide layer was prob-
ably too thin to detect any displacement-excursions.
Thus, yield points are only observed in high-temperature-
implanted or air-oxidized pure Al.
2
2
ments. In agreement with elasto-plastic contact theory,
the unimplanted (homogeneous) sample gives a power
n
relation for the loading curve: f ϰ d , with n ≈ 1.7. On the
other hand, the implanted sample gives very steep initial
slopes indicating a high hardness, but, for penetrations
deeper than 10 nm, there is a kink and a flattening of the
loading curves resulting in an almost linear behavior: this
is due to yielding of the softer substrate under the im-
planted layer. As a result, the measurement underestimates
To identify possible effects due solely to irradiation
damage, Al samples were “plasma-annealed” at tempera-
tures of 120 °C or 400 °C by prolonged Ar bombardment
(
25 min, 150 Hz) at 3 keV or 30 keV, respectively. This
is similar to the 3-keV Ar-ion sputter cleaning or 30-keV
Ar-ion heating normally used prior to the O-ion im-
plantations. In both cases, plasma etching reveals the
grain boundaries. However, no surface swelling, fissures,
and/or porosities were apparent and no nanohardness al-
teration occurred.
B. Low-temperature oxygen plasma implantation
Low-temperature implantations (∼160 °C) were per-
1
7
formed up to a fluence of roughly (0.8 to 2.0) × 10
−
2
O cm . The plasma was pulsed at 25 Hz with a 110-␮s
pulse length corresponding to a duty factor of 0.28%.
XPS O profiles of pure and alloyed Al are shown in
Fig. 10. These profiles include the implanted O and the
native oxide forming instantaneously after the treatment
when exposed to air. The evolution of the profiles with
the implantation dose may be observed up to a near-
1
7
2
saturated dose of 2.0 × 10 O/cm (triangles), which
FIG. 10. XPS depth profiles of oxygen implanted to different doses in
pure and alloyed Al at low temperature.
almost corresponds to stoichiometric Al O .
2
3
FIG. 11. Five individual force–distance curves from nanoindentations,
at successively greater depth and in different spots, in AA7075
implanted at low temperature and two f–d curves for unimplanted
(annealed) alloy.
FIG. 9. Two typical force–distance curves from nanoindentations into
pure Al air-oxidized 2 h at 430 °C. Yield points corresponding to large
penetrations for specific values of force may be observed.
J. Mater. Res., Vol. 18, No. 12, Dec 2003
2785
http://journals.cambridge.org
Downloaded: 13 Mar 2015
IP address: 130.56.64.29

M. Bolduc et al.: Selective modification of the tribological properties of aluminum
the intrinsic hardness of this implanted zone. The
the surface is obviously already work hardened in this
case, probably due to polishing. In pure Al, the effect of
smoothness of the penetration curves indicates that the
implanted layer remains reasonably ductile in spite of its
enhanced hardness. For most applications this is highly
preferable to a mechanical response of the type shown in
Fig. 9.
1
7
2
dose is weak above the critical dose (0.8 × 10 O/cm ).
In the case of the alloy, an optimum nanohardness en-
hancement (∼3×) is clearly observed at an intermediate
17
2
dose of 1.2 × 10 O/cm (circles). This intermediate dose
corresponds to approximately 35 at.% peak O concentra-
tion. It suggests the formation of an Al–Al O nanocom-
2
2
The hardness values obtained by the O&P procedure
are shown as a function of penetration depth in Figs. 12(a)
and 12(b) for pure Al and AA7075, respectively. In both
materials, remarkable improvements may be observed in
comparison with unimplanted samples that were an-
nealed in UHV at the same temperature. The implanted
material compares well with hard steel (e.g., AISI 304L
2
3
posite, as confirmed by two other pieces of evidence. The
first is the XPS Al 2p line profile that was already shown
in Fig. 3, in which oxide formation (75 eV binding en-
ergy) within a metallic aluminum matrix (72 eV) is
clearly visible. The second is the SEM observation of
nanometer-size grains, as seen in Fig. 13. Note that these
grains could clearly be resolved only at the highest dose;
at the optimal (intermediate) dose, the grains are likely to
be even smaller than the approximately 10 nm seen here.
In this connection, the root-mean-square (rms) surface
roughness measured by AFM does not exceed 5 nm in all
low-temperature implanted samples. Finally, the increases
in Young’s modulus (Fig. 14), though measurable, are ex-
tremely modest.
27,28
stainless steel
). Moreover, finite element simulation
of the penetration of an indenter into a hard layer on top of
3
4
softer material (e.g., Knapp et al. ) has shown that the
peak value obtained by the O&P procedure is in fact
representative of the hardness of the whole implanted
layer and even slightly underestimates it by typically
20%. Thus, the peak values shown in Fig. 12 represent
lower bounds to the intrinsic hardness of the implanted
layers. In pure Al, this nanohardness value at 30 nm is
enhanced from 1.2 ± 0.2 GPa to 2.6 ± 0.6 GPa. Note that
To summarize the hardness results, Fig. 15(a) shows
the surface hardness (20–40 nm) as a function of the
dose. The saturation or decrease at high dose will be
discussed later, but one can already recall that the highest
dose corresponds to 75–80% oxide (Fig. 10).
The nanoscratch tests show the effect of hardening on
friction and wear. The scratch depths obtained by AFM
imaging of the cross sections of the scratches are plotted
as a function of the normal force in Fig. 16 for the alloy.
Note that the scatter in the data does not exceed 8%.
Actually, for 10 ␮N, the scratch depth after implantation
is of the same order as the surface roughness. Note that
it was also assumed that wear of the tips was negligible
FIG. 12. Nanoindentation hardness versus penetration depth obtained
by the O&P procedure for low-temperature implantations: (a) pure Al;
(
b) AA7075. Data for unimplanted but UHV-annealed samples are
FIG. 13. SEM micrograph of pure Al implanted at low temperature
1
7
−2
also shown for comparison.
with a saturated dose of 2 × 10 O cm
.
2
786
J. Mater. Res., Vol. 18, No. 12, Dec 2003
http://journals.cambridge.org
Downloaded: 13 Mar 2015
IP address: 130.56.64.29

M. Bolduc et al.: Selective modification of the tribological properties of aluminum
for loads ഛ30 ␮N. By taking the slope from the fits to
is responsible for a scatter in the measurements, which is
reflected in the error bars given here, but it does not bias
the data, we can state that the implanted samples are
several times more scratch-resistant than the unimplanted
Al alloy (1.7 nm/␮N vs. 4 nm/␮N). All scratch depths
are smaller than 60 nm in implanted samples while the
unimplanted alloy shows a scratch depth in excess of
3
5
the friction coefficient on average. Friction indeed is
clearly reduced by the treatment. Moreover, the factor of
reduction is very neatly correlated with the increase in
the surface hardness [Fig. 15(a)]. For pure Al, both prop-
erties vary little with the dose, and the factor is of the
order two. For the alloy, at the common optimal dose of
120 nm at the maximum load. The wear depth is reduced
by a factor of three at the optimal dose. At first sight, the
scratch penetration (with loads ഛ30 ␮N) appears much
deeper than expected from extrapolation of the nanoin-
dentation data (at loads ജ100 ␮N). This is due to the
sharper diamond tips used in scratching, which have ap-
proximately 2 to 3 times smaller curvature radii than
those used in indentation. Another point is that the pen-
etration of the sharp tip prow increased continuously in
the course of the first 100–200 nm of the scratch, and
then stabilized; it is the final steady-state value that is
plotted in Fig. 16. The scratch depths are plotted as a
function of dose, for three values of force, in Fig. 15(c).
Figure 15(b) presents the nanofriction coefficients as a
function of the implanted dose for pure and alloyed Al.
Note that these coefficients were found to be independent
of the load. The initial penetration increase at high load
1
7
2
1.2 × 10 O/cm , the gain factor for both properties is
of the order three. In Fig. 15, one is struck by the obvious
correlation between hardness, friction coefficient, and
scratch depth. The last two are reduced by almost exactly
the same factors as the hardness is increased. It seems to
be a classic example of the effect of hardness on friction
and wear.
The precise mechanism by which these tribological
improvements are obtained is not necessarily simple, but
must be connected with dispersion strenghthening due to
precipitates that are impenetrable by dislocations
FIG. 15. Shown as a function of implanted dose: (a) the surface hard-
ness (<40 nm) of both pure and alloyed Al; (b) the friction coefficient
measured by nanoscratch testing of pure and alloyed Al; (c) the scratch
depth in alloyed Al for different normal loads. The data for unim-
planted UHV-annealed samples are also shown at “zero dose.”
FIG. 14. Nanoindentation Young’s modulus versus penetration depth
(
(
O&P procedure) for low-temperature implantations: (a) pure Al;
b) AA7075. Data for unimplanted but UHV-annealed samples are
also shown for comparison.
J. Mater. Res., Vol. 18, No. 12, Dec 2003
2787
http://journals.cambridge.org
Downloaded: 13 Mar 2015
IP address: 130.56.64.29

M. Bolduc et al.: Selective modification of the tribological properties of aluminum
3
∼
0 keV was used for sample heating (∼15 min, up to
430 °C) prior to the O-ion implantations. There were
two motivations for these experiments. The first is that in
some PBII systems, the temperature of the high-voltage
sample stage is not controlled and may rise to high tem-
peratures due to the large current. The second reason is
that a remarkable effect was found at high temperature in
the Mg-containing alloys AA2024, AA6061, and
8
,9
AA7075 : the O profiles extended much deeper than
the ion range and, concurrently, the alloying element
Mg (<4 at.% in the bulk) was strongly segregated, up
to ∼45 at.% Mg, in the first ∼100 nm from the surface.
The XPS Al 2p and Mg 2p line profiles, as illustrated in
Fig. 17, showed Al to be mostly metallic and Mg oxi-
1
8
dized. This phenomenon has been explained by an un-
usually strong synergy between RED and the chemical
drive provided by the higher oxidation enthalpy of Mg
compared to Al. As mentioned, the deeper penetration
could conceivably allow thicker treated layers at lower
voltage.
FIG. 16. Scratch depths in low-temperature-implanted or annealed
AA7075, obtained by AFM imaging of the cross-sections of the
scratches, are shown as a function of the applied normal force.
6
(
Orowan mechanism ). One may estimate the resulting
increase in yield stress, ⌬␴ , by using the formula-
Y
34,36,37
tion:
1
/2
⌬
␴ ס 0.84 K G b ln(d/b) / [2␲␭ (1−␯) ] ,
Y
where G is the shear modulus (26 GPa for pure Al), ␯ the
Poisson ratio (0.33), b the dislocation Burgers vector
(
0.28 nm for {111} slip planes), d the precipitate diam-
eter (∼5 nm), and ␭ the distance between the precipitates
a few nanometers for 50% oxide); K is factor of order
(
two to three, which accounts for the averaging of the
orientations of the glide planes with respect to the shear
stress. Then one obtains ⌬␴ ∼2.5 GPa. Using the com-
Y
mon rule-of-thumb H ∼3 ␴ , which is confirmed by both
Y
38
34
theory and simulation, it is seen that this model gives
H ∼7.5 GPa. This is the right order of magnitude for the
hardening effect in optimal conditions, recalling also that
the O&P procedure gives a lower bound for H. The satu-
ration of the properties at high dose (in pure Al) or even
their degradation (in the alloy) can be due to two alterna-
tive causes. First, grains with sizes in the few nano-
meter range are too small to contain dislocations,
therefore dislocation blocking becomes meaningless,
and, furthermore, the grain boundaries may start sliding,
3
9
which softens the material. Second, when the precipi-
tates have come to occupy the greater part of the volume,
precipitate coalescence and effective size increase are not
4
0
surprising; this again leads to hardness loss.
C. High-temperature implantation
FIG. 17. High-resolution Al 2p and Mg 2p XPS line profiles acquired
For high-temperature O-ion implantations (∼430 °C),
we used a repetition rate of 150 Hz (3 min–25 min) corre-
sponding to a duty-cycle of 1.65%. Ar ion bombardment at
at several different depths in AA7075 implanted at 430 °C for 15 min,
17
−2
corresponding to a dose of 3 × 10 O cm . Arrows point to the
peaks corresponding to the metal or to the stoichiometric oxide.
2
788
J. Mater. Res., Vol. 18, No. 12, Dec 2003
http://journals.cambridge.org
Downloaded: 13 Mar 2015
IP address: 130.56.64.29

M. Bolduc et al.: Selective modification of the tribological properties of aluminum
In preliminary experiments, in which alloy samples
had been gradually heated up to 400, 430, or 460 °C
during (and not prior to) O plasma implantation, it was
2
found that the profiles widened abruptly at 460 °C. It
indicated a change in regime, and this temperature is dan-
gerously close to the dissolution temperature (∼460 °C) of
the MgZn hardening precipitates produced during the
2
3
7
heat treatment in AA7075T651. Therefore, we settled
on 430 °C as a temperature giving at the same time a
strongly enhanced diffusion and Mg segregation, but be-
low the alloy degradation temperature. The resulting O
and Mg profiles obtained at different ion doses are shown
in Figs. 18 and 19 for pure Al and AA7075, respectively.
They may be compared to air oxidation profiles (Figs. 7, 8),
and low-temperature implantation profiles (Fig. 10). The
oxide layer thickness in high-temperature air oxidation
does not exceed 10 nm in pure Al and 40 nm in the alloy
even after several hours—though strong Mg segregation
does take place in the alloy. The low-temperature im-
planted profiles are basically consistent with the incident
19
ion range, with a little widening at the highest dose.
The high-temperature-implanted pure Al definitely
shows widening, due to RED; this last interpretation is
supported by experiments involving simultaneous im-
1
5,16
plantation of Ar with O2.
The high-temperature-
implanted AA7075 displays the most remarkable effects
1
8
due, as mentioned, to the RED-segregation synergy. In
that last work, it was shown that the oxidation kinetics
FIG. 19. Dose-dependant XPS O and Mg profiles of high-
temperature-implanted AA7075.
1
/2
is diffusive-like (distance ∼ time ), and the effective
diffusion coefficients have been estimated to be >2 ×
−
12
2
−1
−9
2
−1
10
cm s and >2 × 10 cm s for O and Mg,
observe that the O saturation composition is 45 at.% in
the alloy. From the surface to a depth of approximately
80 nm, Mg and O account for about 90 at.% of the total
composition, the remaining being Al and some C. The O
and Mg profiles are not only very wide, but also closely
correlated up to great depths, suggesting that MgO is
formed throughout this layer. This is confirmed by the
high-resolution Mg 2p and Al 2p line spectra, shown in
Fig. 17, which indicate that Mg is totally oxidized while
Al is strongly depleted and mostly metallic. Scanning
electron microscopy shows faceted grains of approxi-
mately 100-nm size (Fig. 20), and AFM reveals con-
comitant surface roughening (∼20 nm rms).
respectively. These correspond to an enhancement factor
3
of 10 compared to air oxidation and to a factor of 10
compared to pure Al oxidation by implantation. One may
The surface morphology of high-temperature-
implanted pure Al was examined by means of SEM and
AFM [Fig. 21(a) and 21(b), respectively]. Very small
porosities (10–50 nm) surrounded by fine grains (100–
2
00 nm) were observed on all the samples, and the den-
sity of these pores increases with the implantation time
and leads to a roughening of the surface. Note that these
features are not visible at the bottom of the XPS craters
(∼120 nm). By comparison, only very fine nanoprecipi-
tates (∼10 nm; Fig. 13) were observed on the surface of
pure Al samples implanted at low temperature and no
apparent fissures, swelling, or porosities were found.
FIG. 18. XPS depth profiles of oxygen implanted to different doses in
pure Al at high temperature.
J. Mater. Res., Vol. 18, No. 12, Dec 2003
2789
http://journals.cambridge.org
Downloaded: 13 Mar 2015
IP address: 130.56.64.29

M. Bolduc et al.: Selective modification of the tribological properties of aluminum
Figure 22 shows extremely stiff f–d curves on the
high-temperature-implanted samples of pure Al, suggest-
ing a very significant nanohardness improvement—at
least at small depth. However, in two of the measure-
ments, catastrophic yielding takes place at a depth of
approximately 40 nm. This behavior indicates that a
brittle layer was formed, suggesting a continuous Al O
2
3
surface, contrary to the low-temperature treatment giving
a polycrystalline Al matrix with Al O nanoprecipitates.
2
3
At shallow depth (<40 nm), high values of nanohardness
up to 4 GPa) and Young’s modulus (up to 150 GPa) can
(
be deduced form the curves but, beyond this point, the
mechanical properties fall abruptly to reach finally the bulk
values. Oxide film fracture is assumed to be the principal
cause of the excursions observed on Fig. 22. Note that
only pure Al, oxidized at high temperature, in air or by
implantation, exhibits this yielding phenomenon. Never-
theless, there is a quantitative difference with the case of the
air oxidation (Fig. 9). In agreement with Ref. 30, the thick
FIG. 20. SEM micrograph of high-temperature-implanted AA7075 at
17
−2
a saturated dose of 4 × 10 O cm
.
(
∼75 nm) implantation-induced oxide only fails at a
depth of ∼40 nm compared to ∼10 nm for the thin oven-
produced oxide. A single displacement excursion is ob-
served instead of multiple “staircase” ones. The critical
load at which the excursion first occurs increases with
the oxide film thickness at a rate of the order 4 ␮N/nm
(250 ␮N for 60 nm). An oxide elastically stiffer than
metal will tend to repell free dislocations in the metal
3
0
away from the oxide/metal interface. However, the
presence of surface roughness, dislocations, and possible
inclusions have also an influence on the critical load at
4
1
which a discontinuity appears on the f–d curves. It is
expected that variations in the critical load will occur from
place to place on the same sample, as seen in Fig. 22.
For the alloy implanted at high temperature, on the
other hand, the f–d curves are “well-behaved” and a sub-
stantial nanohardness enhancement is observed (Fig. 23),
FIG. 21. The surface morphology of high-temperature-implanted pure
17
−2
Al at a dose of 2.5 × 10 O cm , examined by means of (a) SEM and
b) AFM, shows small porosities (10–50 nm) surrounded by fine
grains (100–200 nm).
(
FIG. 22. Typical force–distance curves from five nanoindentations
into pure Al implanted at high temperature to a dose of 2.5 × 10 O cm .
17
−2
2
790
J. Mater. Res., Vol. 18, No. 12, Dec 2003
http://journals.cambridge.org
Downloaded: 13 Mar 2015
IP address: 130.56.64.29

M. Bolduc et al.: Selective modification of the tribological properties of aluminum
change. A very hard but brittle surface is obtained in pure
Al, although this surface cracks catastrophically for pen-
etrations of the order 40 nm, it may still prove useful for
some applications. In the alloy, magnesium segregates to
the surface to an extraordinary extent, and a deep O
penetration is observed. This leads to the formation of a
thick surface layer of mostly MgO (>100 nm), but the
composite that is formed is not quite as hard as the low-
temperature Al-Al O nanocomposite.
2
3
ACKNOWLEDGMENTS
The authors thank J. Oberste-Berghaus and J. Sapieha
from the École Polytechnique (Montréal) for access to
the Hysitron Triboindenter and their colleague Barry
Stansfield for advice on plasma science and technique.
They are also grateful for the financial support of Na-
tional Science and Engineering Research Council of
Canada (NSERC) of Canada and Fonds Nature et Tech-
nologie-Qu e´ bec (NATEQ) and Centre Qu e´ b e´ cois de Re-
cherche et D e´ veloppement de l’Aluminium of Qu e´ bec
FIG. 23. Nanoindentation hardness versus penetration depth in high-
temperature-implanted AA7075. Data for unimplanted but UHV-
annealed samples are also shown for comparison.
(
CQR&DA) of Québec.
though it is slightly inferior to that obtained at low tem-
perature at optimal dose [Fig. 12(b)]. The high-
temperature treatment has also affected the bulk
properties (loss of 20% of the initial bulk hardness). This
may be due to the proximity to the dissolution tempera-
REFERENCES
1. V.T. Srikar and S.M. Spearing, Sens. Actuators A 102, 279
(2003).
2
. S. Ohira and M. Iwaki, Mater. Sci. Eng. 90, 143 (1987).
3. S. Ohira, M. Iwaki, and K. Hiei, Nucl. Instrum. Methods B 32, 66
1988).
ture (∼460 °C) of the MgZn hardening precipitates pro-
2
4
2
duced during the commercial heat treatment. We also
note a wide scatter in the data, indicated in Fig. 23 by the
error bars. We attribute this to the heterogeneity of
the surface, which is characterized by grains (probably
MgO), with sizes comparable to the probe tip radius and
penetration depth, embedded in the residual Al. Unfor-
tunately, the enhanced O penetration has not caused a
hardness enhancement at greater depth than that found in
low-temperature experiments.
(
4
. R.S. Bourcier, S.M. Myers, and D.H. Polonis, Nucl. Instrum.
Methods B 44, 278 (1990).
5. J.C. Barbour, D.M. Follstaedt, and S.M. Myers, Nucl. Instrum.
Methods B 106, 84 (1995).
6. E. Orowan, Symposium on Internal Stresses in Metals and Alloys
(
Institute of Metals, London, U.K., 1948), p. 451.
7
. D. Popovici, M. Bolduc, B. Terreault, A.H. Sarkissian,
B.L. Stansfield, and R.W. Paynter, J. Vac. Sci. Technol. A 17,
1996 (1999).
8
9
. M. Bolduc, D. Popovici, and B. Terreault, Surf. Coat. Technol.
138, 125 (2001).
. M. Bolduc, D. Popovici, and B. Terreault, Nucl. Instrum. Methods
IV. CONCLUSIONS
B 175/177, 452 (2001).
1
1
1
0. S.O. Saied and J.L. Sullivan, J. Phys.: Condens. Matter 5, A165
1993).
1. C.R. Werrett, D.R. Pike, and A.K. Bhattacharya, Surf. Interface
Anal. 25, 809 (1997).
We have shown that OPSII can be used to harden the
surface of pure and alloyed aluminum through a very
quick process (∼10 min) by producing an Al/Al O
(
2
3
nanocomposite with improved hardness (up to three
times). Moreover, a reduction in the friction coefficient
and in the wear depth by the same factor was obtained. It
is supposed that the oxide precipitates block dislocation
motion and lead to surface hardening through the
Orowan mechanism. The treatment is achieved through
an optimal control of the implant dose and temperature.
The excellence of the nanotribological properties so ob-
tained makes the process attractive for the fabrication of
micro- and nanoelectromechanical systems. If the tem-
perature is allowed to rise above those normally reached
under optimal conditions (e.g., above 400 °C), the results
2. S.C. Lim, D.H. Kim, J.S. Kim, C.H. Lee, and E.P. Yoon, Mater.
Sci. Technol. 13, 859 (1997).
1
1
1
3. R.A. Johnson and N.Q. Lam, Phys. Rev. B 13, 4364 (1976).
4. R.C. Piller and A.D. Marwick, J. Nucl. Mater. 71, 309 (1976).
5. M. Bolduc, D. Popovici, B.L. Stansfield, and B. Terreault, Surf.
Coat. Technol. 156, 162 (2002).
1
1
6. M. Bolduc, B. Terreault, and E. Shaffer, presented at PBII2003
(submitted to Surf. Coat. Technol., 2003).
7. J.R. Conrad, J.L. Radtke, R.A. Dodd, and F.J. Worzola, J. Appl.
Phys. 62, 4591 (1987).
1
1
8. M. Bolduc and B. Terreault, Appl. Phys. Lett. 82, 895 (2003).
9. J.P. Biersack and J.F. Ziegler, STRIM-2000 <http://www.
research.ibm.com>.
20. R.K. Hart, Proc. R. Soc. A 236, 68 (1956).
J. Mater. Res., Vol. 18, No. 12, Dec 2003
2791
http://journals.cambridge.org
Downloaded: 13 Mar 2015
IP address: 130.56.64.29

M. Bolduc et al.: Selective modification of the tribological properties of aluminum
2
1. D. Tabor, The Hardness of Metals (Clarendon Press, Ox-
31. M. Pang and D.F. Bahr, J. Mater. Res. 16, 2634 (2001).
32. K.L. Johnson, Contact Mechanics (Cambridge University Press,
Cambridge, 1985).
33. W.W. Gerberich, S.K. Venkataraman, H. Huang, S.E. Harvey, and
D.L. Kohlstedt, Acta. Metall. Mater. 43, 1569 (1995).
34. J.A. Knapp, S.M. Myers, D.M. Follstaedt, and G.A. Petersen,
J. Appl. Phys. 11, 6547 (1999).
ford, 1951).
2
2. W.C. Oliver and G.M. Pharr, J. Mater. Res. 7, 1564 (1992).
3. E.W. Gerberich, W. Yu, D. Kramer, A. Strojny, D. Bahr,
E. Lilleoden, and J. Nelson, J. Mater. Res. 13, 421 (1998).
4. Digital Instrument Inc., Support Note No. 225, Rev. F (Digital
Instruments, Inc., Santa Barbara, CA, 1998).
2
2
2
2
5. J-A. Ruan and B. Bhushan, J. Tribol. 116, 378 (1994).
6. B. Du, M.R. Vanlandingham, Q. Zang, and T. He, J. Mater. Res.
35. T.F. Page, W.C. Oliver, and C.J. McHargue, J. Mater. Res. 7, 450
(1992).
1
6, 1487 (2001).
36. P.B. Hirsch and F.J. Humphreys, Proc. R. Soc. London, Ser. A
318, 45 (1970).
2
2
2
7. Metals Handbook, 8th ed. (ASM, Metals Park, OH, 1961), Vol. 1,
p. 917.
37. J.D. Embury, Metall. Trans. A 16, 2191 (1985).
38. D. Kramer, H. Huang, M. Kriese, J. Robach, J. Nelson, A. Wright,
D. Bahr, and W.W. Gerberich, Acta Mater. 47, 333 (1998).
39. S. Zhang, D. Sun, Y. Fu, and H. Du, Surf. Coat. Technol. 167, 113
(2003).
40. R. Hauert and J. Patscheider, Adv. Eng. Mater. 5, 247 (2000).
41. W.W. Gerberich, D.E. Kramer, N.I. Tymiak, A.A. Volinsky,
D.F. Bahr, and M.D. Kriese, Acta. Mater. 47, 4115 (1999).
42. H.J. Weiss, Phys. Status Solidi A 129, 167 (1992).
8. Smithells Metals Reference Book, 6th ed., edited by E.A. Brandes
(
Butterworths, London, U.K., 1983), Chapter 22.
9. S.G. Corcoran, R.J. Colton, E.T. Lilleodden, and W.W. Gerberich,
in Thin Films: Stresses and Mechanical Properties VI, edited by
W.W. Gerberich, H. Gao, J-E. Sundgren, S.P. Baker (Mater. Res.
Soc. Symp. Proc. 436, Pittsburgh, PA, 1997), p. 159.
0. D.F. Bahr, D.E. Kramer, and W.W. Gerberich, Acta. Metall.
Mater. 46, 3605 (1998).
3
2
792
J. Mater. Res., Vol. 18, No. 12, Dec 2003
http://journals.cambridge.org
Downloaded: 13 Mar 2015
IP address: 130.56.64.29