journal of the mechanical behavior of biomedical materials 21 (2013) 1 –8

Available online at www.sciencedirect.com

www.elsevier.com/locate/jmbbm

Research paper

Synthesis, microstructure and mechanical properties

of porous Mg–Zn scaffolds

Z.S. Seyedraoufin, Sh. Mirdamadi

School of Metallurgy and Materials Engineering, Iran University of Science and Technology, 16846-13114 Tehran, Iran

art i cle i nfo ab st rac t

Article history: Magnesium alloys have been intensively studied as biodegradable implant materials, as
Received 16 June 2012 their mechanical properties render them promising candidates for bone tissue engineering
Received in revised form applications. In the present work, porous Mg–4 wt% Zn and Mg–6 wt% Zn scaffolds were
19 January 2013 prepared using a powder metallurgy process. The effects of the porosity and Zn content on
Accepted 23 January 2013 the microstructure and the mechanical properties of the fabricated scaffolds were studied.
Available online 8 February 2013 The above mentioned fabrication process involved sequential stages of mixing and

Keywords: compression of Mg and Zn powders with carbamide materials as space-holder particles

Magnesium followed by sintering the green compacts at different temperatures below the melting

Zinc point of Mg. The results indicate that the porous Mg–Zn specimens with a porosity and

Porous materials pore size of approximately 21–36% and 150–400 mm, respectively, could have enhanced

Porosity mechanical properties comparable with those of cancellous bone. In addition, an increase

Mechanical properties in the amount of Zn in the applied alloy gives rise to a significant refinement of

Sintering magnesium grain size and an improvement in the mechanical properties, such as the
compression strength, of the porous Mg–Zn specimens. Furthermore, according to the
results, the porous Mg–Zn alloy could be considered one of the most promising scaffold
materials for hard tissue regeneration.
& 2013 Elsevier Ltd. All rights reserved.

1. Introduction been investigated for tissue engineering of bone (Wen et al.,

In recent decades, great interest in porous scaffold substi-
tutes has stemmed from their applicability in bone tissue
engineering applications (Zhuang et al., 2008). The scaffold
provides the necessary support for cells to proliferate and
maintain their differentiated function, and its architecture
defines the ultimate shape of the new bone. The ideal bone
substitute material should be osteoconductive, biodegrad-
able, and strong enough to fulfil the required load-bearing
functions. Several scaffold materials, including hydroxyapa-
tite (HA) and natural polymers (e.g., collagen and chitin), have
2001; Zhuang et al., 2008). The major limitation of these
porous materials is, however, their inadequate mechanical
properties (Zhuang et al., 2008). For instance, the brittle
nature of the porous bioactive ceramics or the very low
strength and Young’s modulus of the porous polymers –
mainly lower than those of real human bones – have severely
limited the applicability of these materials in load-bearing
applications (Wen et al., 2004). Therefore, it is crucial to
develop new scaffold materials with both a porous structure
similar to natural bones and excellent mechanical properties.
Furthermore, scaffold materials that undergo a complete

n
Corresponding author. Tel.: þ98 21 77459151; fax: þ98 21 77240480.
E-mail address: zahraseyedraoufi@iust.ac.ir (Z.S. Seyedraoufi).

1751-6161/$ - see front matter & 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.jmbbm.2013.01.023
2 journal of the mechanical behavior of biomedical materials 21 (2013) 1 –8
biodegradation process after a proper time span in a human
body are even more desirable (Wen et al., 2004; Zhuang et al.,
2008).
Stainless steels, cobalt-based and titanium alloys have
been widely studied and applied clinically in hard tissue
implants, specifically in load-bearing applications, due to
their high strength, ductility and good corrosion resistance
(Ng et al., 2009; Wanga et al., 2009; Yang et al., 2009). However,
they cannot degrade spontaneously, and a second surgical
procedure is usually needed to remove metal implants from
the body after the tissues have completely healed (Ng et al.,
2009; Wanga et al., 2009; Yang et al., 2009). In addition, after
the healing stage, the presence of a permanent bone plate
has the potential to result in a number of adverse effects,
most seriously the occurrence of osteoporosis in the neigh-
bouring bone tissues due to a mismatch in elastic modulus,
creating stress shielding (Ng et al., 2009; Wanga et al., 2009;
Ye et al., 2010). For young patients, permanent bone plates
also restrict bone growth. Repeated surgeries increase both
costs and patient morbidity (Xin et al., 2008). Magnesium has
been recently recognised as a very promising biomaterial for
bone implants because it has excellent mechanical properties
and is biodegradable and bioresorbable (Khanra et al., 2010;
Kirkland et al., 2012; Li et al., 2008; Müller et al., 2007; Wen
et al., 2001, 2004; Xin et al., 2008, 2011; Zhuang et al., 2008).
Magnesium is the fourth most abundant cation in the human
body (1 mol/adult) and is found mainly in bone tissue, and
stabilising DNA and RNA structures (Khanra et al., 2010).
Magnesium alloys enjoy low densities and high strength-to-
weight ratios (Krawiec et al., 2011; Wen et al., 2004). Porous
magnesium has the potential to serve as a degradable scaf-
fold for bone substitute applications (Atrens et al., 2011; Wen
et al., 2004; Zhuang et al., 2008). Some studies have shown
that the dissolved magnesium ions may promote bone cell
attachment and tissue growth on the implants (Xin et al.,
2008).
However, as a biomedical material, the high corrosion rate
of magnesium alloys has been observed to result in high
concentrations of Mg ion and hydrogen gas release. It has
been found that different Mg alloys, such as AZ91, AZ31,
AZ31, WE43 and LAE442, might be prone to localised corro-
sion in in vitro and in vivo tests (Khanra et al., 2010). In
addition, because the yield strength of pure magnesium is
approximately 20 MPa, much lower than that of human long
bone (106–133 MPa), it is essential to take relevant measures
to control its corrosion rate and to improve its mechanical
properties. Zinc not only improves the corrosion resistance
but also enhances the mechanical properties of magnesium
alloys (Gu et al., 2010; Kaya et al., 2006; Yin et al., 2008; Zhang
et al., 2009). Moreover, zinc is recognised as a highly essential
element for humans (Zhang et al., 2009). It has been shown
that Mg–Zn alloys possess the highest capacity for precipita-
tion hardening due to segregation of an intermediate phase
that ensures a combination of high strength and ductility
(Kaya et al., 2006). Gu et al. (2010) showed that Zn may
improve the plasticity of a Mg–Zn–Ca alloy, and Yin et al.
(2008) reported that the addition of Zn can effectively
increase the strength of a Mg–Zn–Mn alloy.
Wen et al. (2001) and Zhuang et al. (2008) produced porous
magnesium scaffolds and investigated the mechanical
properties of the specimens with respect to different physical
properties. Until now, there has been no study on either the
production of porous Mg–Zn scaffolds or the ways in which
the mechanical properties and corrosion resistance of the
magnesium scaffolds may be improved. In the present work,
porous Mg–4 wt% Zn and Mg–6 wt% Zn specimens have been
fabricated using a powder metallurgy process. The microstruc-
tures and the mechanical properties of the specimens contain-
ing different zinc amounts and porosities are investigated.
2. Materials and methods
Pure magnesium (purity Z99%, particle size r0.1 mm) and
pure zinc powder (purity Z99.8%, particle size r45 mm)
purchased from Merck were utilised as the starting materials.
Carbamide (CO(NH2)2) particles purchased from Merck were
employed as the space-holder agent. The particle size of the
spacer agent material was in the range of 200–400 mm with a
purity of 99.9%. After mixing the starting materials with the
space-holder particles, porous Mg–4 wt% Zn and Mg–6 wt%
Zn specimens were prepared through a powder metallurgy
process. The mixtures of magnesium and zinc powders were
prepared based on two different zinc amounts, 4 and 6 wt%,
while the carbamide particles were thoroughly added to the
above specimens with variable volume contents of 15%, 25%
and 35%. Details of the fabricated samples are listed in
Table 1.
The mixed powders were uniaxially pressed at a pressure
of 100 MPa into green compacts with dimensions of 10 mm in
diameter and 15 mm in length. Differential thermal analysis
(DTA) and thermal gravimetry analysis (TGA) (DTA–TGA:
Shimadzu, TGA-50, Japan) were conducted under an argon
atmosphere to determine the temperatures at which the
carbamide particles were removed from the samples and
when the reactions between Mg and Zn occurred during the
sintering process. The green compacts were subsequently
heat treated to burn out the spacer particles and to sinter the
porous Mg–Zn specimens separately in a tube furnace under
an argon atmosphere. The heat treatment process consisted
of two steps: first, heating up to 250 1C at a rate of 3.751/min
and staying at that temperature for 4 h, and then heating up
to 500, 550, 565 and 580 1C at a rate of 81/min and staying at
the final temperature for 2 h. The sintering temperature was
set slightly above the temperature at which DTA–TGA peaks
appeared, and after the sintering stage, the specimens were
furnace cooled down to room temperature.
Pore sizes and morphologies of the porous magnesium
specimens were observed using optical microscopy. Total
Table 1 – Details of the fabricated samples.
Zn (wt%) Carbamide (vol%) Sample
4 15 A1
4 25 A2
4 35 A3
6 15 B1
6 25 B2
6 35 B3
 journal of the mechanical behavior of biomedical materials 21 (2013) 1 –8 3

porosity (P) of the porous specimens was measured using testing of samples with dimensions of F10 mm  15 mm. The
gravimetry according to the following equation (Zhuang et al., tests were performed with a SANTAM (STM-20, Iran) testing
2008): machine at room temperature at a rate of 0.3 mm/s. Each
result was taken as the mean value of testing on five samples.
P ¼ ð1r=rs Þ  100% ð1Þ

where rs is the density of the magnesium–zinc specimen

evaluated via the immersion method and r is the apparent
3. Results and discussion
density of the porous magnesium–zinc specimen, which can be
measured by the weight-to-volume ratio of the porous specimen.
3.1. DTA–TGA analysis
Microstructural and elemental analyses were conducted
Fig. 1 illustrates the DTA–TGA curves of the A2 samples.
using scanning electron microscopy (SEM: TSCAN-VEGA,
As can be observed, the DTA analysis demonstrated an
China) and energy dispersive X-ray spectroscopy (EDS:
endothermic peak at E140 1C, which is accompanied by a
TSCAN-VEGA, China), respectively. An X-ray diffraction tech-
8.38% mass decrement.
nique (XRD: JDX-8030, Jeol, Japan) with a Cu-Ka source was
Such a decrease in the mass is interpreted as the removal
used to obtain diffraction patterns of the sintered samples.
of the spacer particles from the specimen. Furthermore, the
The compressive strength of the porous Mg–4 wt% Zn and
6.98% mass decrement from E500 1C to E580 1C indicates the
Mg–6 wt% Zn specimens was measured using compression
occurrence of the sintering process in this temperature range.
According to a phase diagram of the Mg–Zn binary alloy
system (Deng et al., 2005) and considering the data repre-
sented by the DTA–TGA curves, in order to achieve the
optimum sintering temperature, consolidation of the speci-
mens was initiated at variable sintering temperatures of 500,
550, 565 and 580 1C under an argon atmosphere.

3.2. Microstructure of porous specimens

Using Eq. (1), the porosity volume fractions of the three classes
of A1 (B1), A2 (B2) and A3 (B3) specimens are E21, E29 and
Fig. 1 – DTA/TGA curves of the Mg–4 wt% Zn powder E36 vol%, respectively. Fig. 2 shows the optical micrographs of
containing 25 vol% carbamide. the A1, A2 and A3 specimens, sintered at 550 1C. It is clear that

Fig. 2 – Optical micrographs of the A1, A2 and A3 specimens, sintered at 550 1C, with different porosity amounts: (a) 21%,
(b) 29% and (c) 36%.
4 journal of the mechanical behavior of biomedical materials 21 (2013) 1 –8

the pore sizes of the A1, A2 and A3 specimens are smaller than
200 mm, 100–300 mm and 200–400 mm, respectively. Visual char-
acterisation of the pores revealed that two types of pores, open
interconnected macropores and small isolated micropores,
were distinguishable within the microstructure of the porous
specimens. Some of the micropores are formed as a result of
volume shrinkage of the magnesium and zinc powders during
the sintering process. In addition, the number of isolated pores
increases with a decrease in the volume fraction of the micro-
structural porosities.
Fig. 3 shows the optical micrographs of the A2 samples
sintered at 500, 550, 565 and 580 1C. As observed in the figure,
an increase in the sintering temperature leads to an incre-
ment in the grain size. This result implies that 550 1C should
be considered the optimum sintering temperature because
the grain growth effect would result in an obvious degrada-
tion of the mechanical properties.
Fig. 4 shows the optical microstructures of the A1 and B1
samples sintered at 550 1C. As shown in the figure, increasing
the Zn content of the synthesised alloys produces a signifi-
cant reduction in the grain size of the matrix. Such an
observation indicates that the addition of Zn could effectively
refine grain size and improve mechanical properties.
3.3. XRD analysis
X-ray diffraction patterns of the A2 samples sintered at
different temperatures are shown in Fig. 5. As observed in
the patterns, Mg peaks were detected in all of the samples,
and MgO was detected as the minor phase in the patterns.

Fig. 3 – Optical micrographs of the A2 specimens sintered at different temperatures: (a) 500, (b) 550, (c) 565 and (d) 580 1C.

Fig. 4 – Optical micrographs of the porous Mg–Zn specimens sintered at 550 1C with different Zn amounts: (a) 4 wt% and
(b) 6 wt%.
 journal of the mechanical behavior of biomedical materials 21 (2013) 1 –8
Fig. 5 – XRD patterns of the A2 specimens sintered at the
temperatures of 500, 550, 565 and 580 1C.
In fact, the presence of the magnesium oxide peaks indicates
that MgO was most likely formed during the preparation and/
or sintering process; however, MgO is biologically a non-toxic
oxide (Kannan and Singh, 2008). The increased intensity of
the MgO peaks is a consequence of sintering at higher
temperatures and is a further evidence that 550 1C is the
optimum sintering temperature. However, XRD results could
not reveal the presence of any other possible phases.
3.4. SEM analysis
SEM micrographs of the porous A2 samples sintered at 550
and 580 1C are illustrated in Fig. 6. According to the EDS
analysis results (which are listed in Table 2) of points ‘‘A’’ and
‘‘B’’ (marked with an  in Fig. 6), precipitates detected in the
grain boundaries of the samples sintered at 550 and 580 1C
might be Mg7Zn3 and MgZn, respectively.
5
Fig. 6 – SEM micrographs of the A2 specimens sintered at
different temperatures: (a) 550 and (b) 580 1C.
Table 2 – EDS analysis results of the porous A2 samples.
Temperature (1C) Mg content (at%) Zn content (at%)
550 18.73 19.91
580 16.67 43.04
The SEM results along with the Mg–Zn binary alloy phase
diagram (Deng et al., 2005) indicate that the Mg7Zn3 and
MgZn phases can act as local cathodic agents under atmo-
spheric conditions (Krawiec et al., 2011) and are formed at
550 1C and 580 1C, respectively. The XRD and SEM results
indicated that the amount of precipitates (Mg7Zn3 and MgZn)
in the samples is very low because their corresponding peaks
could not be detected by XRD. EDS elemental mappings of the
Mg–4 wt% Zn samples with 21 vol% porosity and sintered at
550 1C are shown in Fig. 7. From the figure, it can be inferred
that Zn is uniformly distributed in the Mg matrix. Micro-
structures of the samples, sintered at 550 and 580 1C (Fig. 6),
show that the grain size of the matrix increased at the higher
sintering temperature, reinforcing that 550 1C could be the
optimum sintering temperature, as previously mentioned.
6 journal of the mechanical behavior of biomedical materials 21 (2013) 1 –8

Fig. 7 – (a) SEM micrograph of the A1 sample sintered at 550 1C, (b) and (c) elemental mapping images of (a).

Fig. 8 – Compressive strength of the porous Mg–Zn specimens sintered at different temperatures with different Zn amounts:
(a) 4 wt% and (b) 6 wt%.

3.5. Mechanical properties
Figs. 8 and 9 illustrate the compressive strength and Young’s
modulus values of the porous Mg–4 wt% Zn and Mg–6 wt% Zn
samples as a function of porosity at different sintering
temperatures. The compressive strength and Young’s mod-
ulus of the porous samples decreased with an increase in the
volume fraction of porosities at all sintering temperatures.
What is evident in the figures is that the values of the
compressive strength and Young’s modulus for the porous
Mg–6 wt% Zn samples are higher than that for Mg–4 wt% Zn.
The other conclusion that can be drawn from these results is
the remarkable grain size decrement in the specimens as a
result of an increase in the Zn content. On the other hand, Zn
mainly resolves in primary magnesium when the Zn content
is 1–2 wt%, which can improve the strength of the alloy by
 journal of the mechanical behavior of biomedical materials 21 (2013) 1 –8
Fig. 9 – Young’s modulus of the porous Mg–Zn specimens sintered at different temperatures with different Zn amounts:
(a) 4 wt% and (b) 6 wt%.
Table 3 – Summary of mechanical properties of various porous materials and natural bone.
Porous materials Porosity (%) Pore size (mm)
Porous magnesium 36–55 200–400
Porous polymers 58–80 300
Porous HA 50–77 200–400
Porous composite bioglass 77–80 100
(20–50 wt þPLLA)
Natural bone – –
solid-solution strengthening (Yin et al., 2008). However, when
the Zn content is 4 or 6 wt%, such as in the specimens of the
current research, many Mg–Zn phases would most likely be
precipitated from the matrix. Such an effect is expected to
enhance the strength by a dispersion strengthening mechan-
ism (Yin et al., 2008). Therefore, Zn can act as a constituent to
refine the microstructure of the matrix and to improve the
mechanical strength. In addition, it is shown that the highest
compressive strength and Young’s modulus values occurred
in the specimens sintered at 550 1C. Therefore, these mechan-
ical property results are more evidence that confirms that
550 1C is the optimum sintering temperature. The major
problems of the implant materials in use today are related to
the high Young’s modulus of metallic materials and the low
mechanical properties of highly porous ceramics and poly-
mers. These problems can be solved by selecting a porous
Mg–Zn scaffold with the appropriate porosity. Although the
compressive strength and Young’s modulus of the porous Mg–
Zn specimens decrease as porosity increases, the mechanical
properties of these specimens are closer to those of natural
bone (Table 3; Zhuang et al., 2008). Therefore, it is suggested
that porous Mg–Zn is a promising scaffold material for hard
tissue regeneration.
4. Conclusions
Porous Mg–4 wt% Zn and Mg–6 wt% Zn specimens have been
fabricated by a powder metallurgy process. According to the
results, the optimum sintering temperature to synthesise
the porous Mg–Zn specimens is E550 1C. The mechanical
7
Flexure (F) and compressive (C) Modulus (GPa)
strength (MPa)
14.1–27 (F), 15–31 (C) 3.6–18.1
12.4–56.2 (F), 2.7–11 (C) 0.05–1.2
2.1–7.2 (F), 1.2–17.4 (C) 0.12–7
1.5–3.9 (F), 0.42 (C) 0.14–0.26
1.8–150 (F), 2–180 (C) 0.1–20
properties of the porous specimens with porosities of 19–36%
and with pore sizes of approximately 150–400 mm were
investigated by compressive testing. The results indicate that
the mechanical properties (compressive strength and Young’s
modulus) of the porous Mg–Zn specimens improve with a
reduction in the volume fraction of porosities. Moreover,
increasing the amount of Zn could refine the magnesium
grain size and, as a result, could improve the mechanical
properties of the porous Mg–Zn alloy. However, compared to
porous bioactive ceramic and polymeric scaffolds, the porous
Mg–Zn specimens have more appropriate mechanical proper-
ties that are closer to those of natural bone. Therefore, the
porous Mg–Zn metals have the potential to serve as degradable
implants for bone substitute applications.
r e f e r e nc e s
Atrens, A., Liu, M., Abidin, N.I.Z., 2011. Corrosion mechanism
applicable to biodegradable magnesium implants. Journal of
Materials Science and Engineering B 176, 1–28.
Deng, C.J., Wong, M.L., Ho, M.W., Yu, P., Ng, Dickon H.L., 2005.
Formation of MgO and Mg–Zn intermetallics in an Mg-based
composite by in situ reactions. Composites Part A 36, 551–557.
Gu, X., Zheng, Y., Zhong, S., Xi, T., 2010. Corrosion of, and cellular
responses to Mg–Zn–Ca bulk metallic glasses. Biomaterials 31,
1093–1103.
Kannan, M.B., Singh, R.R.K., 2008. Evaluating the stress corrosion
cracking susceptibility of Mg–Al–Zn alloy in modified-
simulated body fluid for orthopaedic implant application.
Scripta Materialia 59, 175–178.
8 journal of the mechanical behavior of biomedical materials 21 (2013) 1 –8
Kaya, A., Eliezer, D., Ben-Hamu, G., Golan, O., Na, Y.G., Shin, K.S.,
2006. Microstructure and corrosion resistance of alloys of the
Mg–Zn–Ag system. Metal Science and Heat Treatment 48,
50–54.
Khanra, A.K., Jung, H.C., Yu, S.H., Hong, K.S., Shin, K.S., 2010.
Microstructure and mechanical properties of Mg–HAP
composites. Journal of Materials Science 33 (1), 43–47.
Kirkland, N.T., Birbilis, N., Staiger, M.P., 2012. Assessing the
corrosion of biodegradable magnesium implants: a critical
review of current methodologies and their limitations. Acta
Biomaterialia 8, 925–936.
Krawiec, H., Stanek, S., Vignal, V., Lelito, J., Suchy, J.S., 2011. The
use of microcapillary techniques to study the corrosion
resistance of AZ91 magnesium alloy at the microscale.
Corrosion Science 53, 3108–3113.
Li, Z., Gu, X., Lou, S., Zheng, Y., 2008. The development of binary
Mg–Ca alloys for use as biodegradable materials within bone.
Biomaterials 29, 1329–1344.
Müller, W.D., Nascimento, M.L., Zeddies, M., Córsico, M., 2007.
Magnesium and its alloys as degradable biomaterials.
Corrosion studies using potentiodynamic and EIS
electrochemical techniques. Journal of Materials Research 10
(1), 5–10.
Ng, W.F., Wong, M.H., Cheng, F.T., 2009. Cerium-based coating for
enhancing the corrosion resistance of bio-degradable Mg
implants. JMaterials Chemistry and Physics 119, 1–5.
Wanga, Y.M., Wang, F.H., Xu, M.J., Zhao, B., Guo, L.X., Ouyang, J.H.,
2009. Microstructure and corrosion behavior of coated AZ91
alloy by microarc oxidation for biomedical application.
Applied Surface Science 255, 9124–9131.
Wen, C.E., Mabuchi, M., Yamada, Y., Shimojima, K., Chiho, Y.,
Asahina, T., 2001. Processing of biocompatible porous Ti and
Mg. Scripta Materialia 45, 1147–1153.
Wen, C.E., Yamada, Y., Shimojima, K., Chino, Y., Hosokawa, H.,
Mabuchi, M., 2004. Compressibility of porous magnesium
foam: dependency on porosity and pore size. Materials Letters
58, 357–360.
Xin, Y., Huo, K., Tao, H., Tang, G., Chu, P.K., 2008. Influence of
aggressive ions on the degradation behavior of biomedical
magnesium alloy in physiological environment. Acta
Biomaterialia 4, 2008–2015.
Xin, Y., Hu, T., Chu, P.K., 2011. Degradation behaviour of pure
magnesium in simulated body fluids with different
concentrations of HCO3. Corrosion Science 53, 1522–1528.
Yang, J.X., Cui, F.Z., Yin, Q.S., Zhang, Y., Zhang, T., Wang, X.M.,
2009. Characterization and degradation study of calcium
phosphate coating on magnesium alloy bone implant in vitro.
IEEE Transactions on Plasma Science 37, 1161–1168.
Ye, X., Chen, M., Yang, M., Wei, J., Liu, D., 2010. In vitro corrosion
resistance and cytocompatibility of nano-hydroxyapatite
reinforced Mg–Zn–Zr composites. Journal of Materials Science:
Materials in Medicine 21, 1321–1328.
Yin, D., Zhang, E., Zeng, S., 2008. Effect of Zn on mechanical
property and corrosion property of extruded Mg–Zn–Mn alloy.
Transactions of Nonferrous Metals Society of China 18,
763–768.
Zhang, E., Yin, D., Xu, L., Yang, L., Yang, K., 2009. Microstructure,
mechanical and corrosion properties and biocompatibility of
Mg–Zn–Mn alloys for biomedical application. Materials
Science and Engineering C 29, 987–993.
Zhuang, H., Han, Y., Feng, A., 2008. Preparation, mechanical
properties and in vitro biodegradation of porous magnesium
scaffolds. Materials Science and Engineering C 28, 1462–1466.