Additive Manufacturing 29 (2019) 100790

Contents lists available at ScienceDirect

Additive Manufacturing
journal homepage: www.elsevier.com/locate/addma

Full Length Article

Towards additive manufacturing of magnesium alloys through integration of T

binderless 3D printing and rapid microwave sintering
⁎ ⁎
Mojtaba Salehia, Saeed Maleksaeedib, Mui Ling Sharon Naib, , Manoj Guptaa,
a
Department of Mechanical Engineering, National University of Singapore, 9 Engineering Drive 1, Singapore, 117576, Singapore
b
Singapore Institute of Manufacturing Technology, 73 Nanyang Drive, 637662, Singapore

A R T I C LE I N FO A B S T R A C T

Keywords: 3D printing (3DP) is a two-step additive manufacturing technique (AM) in which additively manufactured green
Additive manufacturing parts in the first step are transformed into functional parts during the second step. 3DP could attract more
Binder jetting interest if a new window of opportunity for its first and second steps is opened. Here we use capillary-mediated
Capillary-mediated binderless 3D printing binderless 3DP as a novel method to additively manufacture green parts of Mg-5.06Zn-0.15 Zr powder. A unified
Magnesium alloys
perspective on the development steps of process parameters to obtain sufficient handling strength and a high
Hybrid microwave heating
level of dimensional accuracy in the green parts without compromising its chemical composition is established
by using a scanning electron microscope, X-ray micro-tomography, vibrational spectroscopy, and chemical
analysis. For the first time, microwave (MW) sintering is successfully used for densification of the green parts
with centimeter-scale dimensions in which the primary chemical composition of the Mg-Zn-Zr powder is re-
trieved from the green parts, resulting in a compositionally zero-sum AM process. It is found that swelling leads
to loss of shape fidelity during MW sintering of the green parts at temperatures ≥ 510 °C. As discussed in the
context of thermal and non-thermal effects, MW significantly reduced sintering time by a factor of three to four
times when compared to sintering in a conventional furnace. The results of this study suggest the notion of
capillary-mediated binderless 3DP as well as MW sintering as a potential alternative for the first and second steps
of 3DP, respectively.

1. Introduction infancy stage in comparison to the other metals (e.g., stainless steel,
titanium, and nickel-based superalloys) as their AM techniques have
Magnesium (Mg) as a biodegradable metal and the lightest en- been commercialized already.
gineering metal has a promising potential for usage in both biode- Unlike fusion-based AM processes which are a single-step AM
gradable medical implants and engineering applications. Thus, ever- technique, 3D printing (3DP), also regarded as binder jetting 3DP, is a
increasing attempts have been made to advance all aspects of Mg two-step AM process in which additively manufactured green parts in
technology. Developing new manufacturing technologies is of prime the first step are transformed into functional parts during the second
importance to enable broader applications of Mg alloys. Additive step. The mechanisms for binding powder materials that are currently
manufacturing (AM) is such an emerging advanced manufacturing used in the first step of 3DP can broadly be classified as adhesion and
process [1,2]. Selective laser melting is the more commonly in- reaction [12]. The former mechanism involves using a polymeric binder
vestigated AM method for Mg alloys among the other AM techniques, material either in a liquid or solid form to glue powder particles in
including laser additive manufacturing, wire-and-arc-based AM, and position throughout the designated area according to the sliced file
friction stir AM while the latter two techniques could potentially take layer-by-layer [13–15]. In the latter mechanism, powder particles serve
into account as a hybrid AM method for Mg alloys [3–10]. The chal- as reagents reacting with a selectively deposited solution such that the
lenges involved in the fusion-based AM of Mg alloys are largely asso- reactions' products bond together in the designated areas [16,17]. The
ciated with intrinsic properties of Mg such as high vapor pressure and second step of 3DP generally involves a binder removal treatment
great affinity to oxygen, leading to variations in microstructure, change (debinding process) followed by a densification process such as sin-
in chemical composition, the formation of porosity, oxidation, etc. tering [15,18,19], hot isostatic pressing [20], and infiltration with a
[3,4,7–9,11]. A study of current literature reveals that AM of Mg is in its second material [21]. Among these processes, sintering in conventional

⁎
Corresponding authors.
E-mail addresses: mlnai@simtech.a-star.edu.sg (M.L.S. Nai), mpegm@nus.edu.sg (M. Gupta).

https://doi.org/10.1016/j.addma.2019.100790
Received 27 May 2019; Received in revised form 7 July 2019; Accepted 7 July 2019
Available online 11 July 2019
2214-8604/ © 2019 Elsevier B.V. All rights reserved.
M. Salehi, et al. Additive Manufacturing 29 (2019) 100790

furnaces is the most prevalent densification approach. Table 1

3DP holds some distinctive advantages, namely high materials de- Chemical and physical characteristics of Mg powder.
position rate, lower capital and maintenance cost, operation in ambient Chemical composition1 Physical properties
atmospheric conditions, printing totally support-free and stress-free
structures, minimal materials wastage, full recycling of used powder, Element (wt.%) Density (g. cm−3)
etc. [12]. Thus, 3DP can play an essential role in AM provided that
Zn Zr Al Mg Other elements2 True Tap Apparent
development occurs in its first and second steps. In view of the first step 5.06 0.15 0.03 Bal < 0.01 1.8363 1.0604 0.9509
of the 3DP technique, binder related issues in terms of binder selection
1
and binder removal process are among the main formidable challenges. Determined by inductively coupled plasma - optical emission spectrometry.
2
Concern about possible reactions between binder and powder feed- Other elements tested were Fe, As, B, Be, Bi, Cd, Cr, Co, Hf, Mo, Nb, P, Pb,
stock, the residue left behind from the debinding process, and the need Se, Sb, Sn, Ta, Te, Ti, V, & W (< 0.01% each).
for a lengthy debinding process are the major limitations of using the
adhesive-based binding mechanism [18]. Apart from totally changing dispensing ink is regulated by an inkjet print head. In the present study,
the chemical composition of starting powder materials in the green 3DP of samples with different geometries was prepared under ambient
parts printed with the reaction-based mechanism, several other chal- conditions by an in-house system having a drop-on-demand inkjet print
lenges such as warping and dimensional change may confront in 3DP head. The ink used was a proprietary formulated solvent for Mg alloys
build cycle while the binder is reacting with the powder within the that neither contains suspended particles nor metal salts. All samples
powder bed [12]. These drawbacks restrict the usage of this mechanism were fabricated by using the same printing parameters including fixed
to a very limited number of materials [12], although no need for con- layer thickness of 100 μm; but using three distinct solvent saturation
ducting any debinding process is its great advantage. When it comes to levels (SSL) of 50%, 90%, and 150%. When the build cycle was com-
advancing the second step of the 3DP technique, the sintering knowl- pleted, the building envelope was fully dried for an hour to further
edge established in the powder metallurgy context could shed some evolve the interparticle bridges that were selectively created
light on possible directions of development. For example, employing an throughout 3D printed samples. This allows the extraction of samples
emerging non-conventional sintering method such as microwave (MW) from unbound powder in the powder bed. The extracted samples
heating would provide unique opportunities for the 3DP technique to printed with SSL 90% was sintered without delay.
shorten lead times and reduce energy consumption, given that the
greater efficacy of MW sintering is observed in processing of various 2.3. Sintering process
types of materials in comparison with conventional (CN) sintering
[22–24]. To the best of our knowledge, there is no study conducted to Using susceptor materials (microwave absorbers) resolves the issues
directly MW sinter 3D printed green parts. of heat instability and reproducibility that one may encounter in the
We introduced a new and facile approach for the first step of the direct MW heating. More details about the susceptor assisted micro-
3DP technique based on capillary-mediated assembly of powder parti- wave heating, also regarded as hybrid microwave heating, that provides
cles that neither need polymeric binders nor change the overall com- a practical solution for uniform heating of powder compacts can be
position of powder materials, and validated the concept on the basis of found in [27]. In the present investigation, MW sintering was per-
pure Mg powder as an alternative ambient-conditions AM method for formed on a 1.5 kW, 2.45 GHz multimode microwave (HAMiLab-
Mg alloys [25]. In the absence of a debinding-adapted profile during the V1500, Synotherm Co.) with an insulation-susceptor package providing
second step, the capillary-mediated binderless 3D printed Mg samples effective heating dimensions of 85 × 85 × 25 mm3. In this furnace, the
were densified by conducting a sintering process in a CN furnace in a temperature was measured using an infrared pyrometer (operating
similar way to the binder jet 3D printed samples [26]. The present study temperature > 250 °C) and recorded by a computer. All MW sintering
concentrates on two objectives. Firstly, in light of Mg-5.06Zn-0.15 Zr experiments were conducted in a controlled argon atmosphere while
alloy and several materials characterization techniques, investigations the furnace chamber was slightly pressurized to ensure that there is no
are focused on how to apply foundations of capillary-mediated bin- ingress of air into the chamber. The controlled atmosphere in the fur-
derless 3DP to fabricate green parts of a new powder material with nace before commencing the sintering process was obtained by evac-
sufficient handling strength and a high level of geometrical accuracy uating the chamber to a pressure < 10 Pa followed by purging with
without compromising the build time and chemical composition of the argon gas until the atmospheric pressure was reached. This process of
green parts. Secondly, for the first time, the feasibility of employing evacuation and purging was repeated three times to dilute the oxygen
MW sintering as a potential alternative for densification of the green level in the chamber. According to DSC results as per detailed discus-
parts is explored in comparison with CN sintering. sion in Section 3.2.1, green samples were sintered in temperature
ranging from 500 °C to 560 °C without any isothermal hold. For an in-
2. Materials and methods vestigation into the effect of isothermal dwelling at a set sintering
temperature, green samples were sintered at two different intervals of
2.1. Powder feedstock 1 h and 5 h. Constant power of 0.8 kW was used to heat up to a sintering
temperature with a ramp rate of ˜ 9 °C /min. After reaching sintering
Spherical Mg-Zn-Zr powder with a particle size ranging from 60 μm temperature, the samples were subjected to an isothermal dwell period
to 70 μm was used in the present study. The chemical composition of by regulating MW power using furnace’s software based on the tem-
the powder is shown in Table 1. Apparent, tap, and true density of the perature feedback signal from the pyrometer. The software then swit-
powder were determined in accordance with ASTM B527, ASTM B923, ched off the MW power at the end of a sintering profile and samples
and ASTM B212, respectively, and the results are presented in Table 1. were furnace cooled to room temperature.

2.2. Three-dimensional printing process 2.4. Materials characterization methods

In essence, the amount of ink (i.e., solvent or binder) required for Differential Scanning Calorimetry (DSC) of the Mg-Zn-Zr powder
printing via inkjet-based 3DP techniques is defined by ‘saturation level’. was conducted in the range of 35–700 °C using heating rates of 5 °C/min
The saturation level is the ratio between the volume of ink and void and 10 °C/min. Thermo-gravimetric Analysis (TGA) of the green sam-
space existing among particles in a powder bed. The volume of void ples printed with SSL 90% (heating rate of 10 °C/min) was also con-
space is dictated by powder packing density, and the volume of ducted. The thermal analysis was performed on Netzsch STA 449 F3

2
M. Salehi, et al. Additive Manufacturing 29 (2019) 100790

under argon gas flow rate 50 mL min−1. Three test specimens were used to modulate 3D printability of a new powder system for obtaining
tested to ensure repeatability. The chemical compositions of the as-re- successful green components. According to the hysteretic nature of
ceived powder, the green samples printed with SSL 90% and SSL 150%, capillary bridges, the adhering liquid layers on adjacent powder parti-
and the green samples printed with SSL 90% after MW sintering at cles must connect with each other to enable liquid bridges, also re-
510 °C without an isothermal dwell time were analyzed for carbon, garded as capillary bridges, to form. The minimum amount of liquid
oxygen, and hydrogen elements by using combustion-infrared absor- required to satisfy the hysteretic criterion for a pile of equally size
bance (Eltra CS 800 Carbon/Sulfur Analyzer) and inert gas fusion- spherical particles with a packing density of ρ can be calculated from
thermal conductivity (Eltra ONH 2000 Oxygen/Nitrogen/Hydrogen the following equation [25]:
Analyzers), respectively. At least two test specimens were tested to
1/3
ensure accuracy. 3ρ ⎡ ⎛ ρm ⎞ ⎤ W0
Wf = -1⎥ + (ρ ≤ ρm)
(1 − ρ) ⎢ ⎝ ρ ⎠
⎜ ⎟
Given the absence of new XRD peaks in 3D printed green samples in (1 − ρ) (1)
⎣ ⎦
comparison with the as-received Mg powder, vibrational spectroscopy,
a powerful analytical tool for analyzing trace amounts of phases, poorly where Wf is SSL, ρm is the maximum packing density of a given powder
crystalline and amorphous materials, was used to examine the green bed, and W0 is the minimum amount of liquid needed for the formation
sample printed with SSL 90%. Fourier transform-infrared spectroscopy of liquid bridges when the packing density is maximum (i.e., ρm). Ac-
(FTIR spectroscopy) was conducted by placing a piece of the green cording to the spreading behavior of ink’s droplets in the powder bed,
sample on the diamond attenuated total reflection (ATR) crystal the droplets have both vertical and horizontal velocity components
(Bruker Vertex 80v vacuum spectrometer), and spectra using an such that the ink tends to flow more in the lateral direction than in
average of 32 scans were recorded over the 4000–400 cm−1 range with vertical one [28–30]. Putting the aforementioned concepts of capillary-
a resolution of 4 cm−1. Raman spectra (Renishaw InVia Raman mediated binderless 3DP into manufacturing perspective, 3DP of a re-
Microscope) were also collected from the green sample. Raman spectra latively simple geometry such as a pyramid allows investigating the
were excited by an argon ion laser beam (λ = 514 nm) and collected in interplay of all competing printing parameters qualitatively since the
the range between 115 and 4000 cm−1 at a resolution of 1.12 cm−1. rectangular base of a pyramid continually changes towards its vertex.
Three test specimens/locations were tested to ensure accuracy of vi- Furthermore, 3DP of a pyramid with distinct build orientations allows
brational spectroscopy. to examine the lateral and vertical spreading of solvent and optimally
Cross-sectional views of green and sintered samples were obtained adjust the SSL towards orientationally independent 3D printed green
using a scanning electron microscope (SEM) (JEOL JSM-6010). parts. Fig. 1a displays green pyramid shaped parts printed with the
Furthermore, the micro-computed tomography (μCT) scanning three different SSLs (i.e., 50%, 90%, and 150%), and two distinct build
(phoenix v|tome|x m) was performed on the green sample printed with orientations, namely upright and upside-down orientations, showing
SSL 90%. A green cylindrical sample (10 mm in diameter and 15 mm in clear geometrical differences in outcomes. The printed samples with
height) was rotated over 360° to acquire a sequence of 2D images which SSL 50% came out in a cone-like shape rather than a pyramid shape
were reconstructed to produce gray-scale images with 16-bit intensity irrespective of the build orientations. Printing with SSL 150% led to
(a true spatial resolution of 8.59 μm). Focused-ion beam milling (ZEISS asymmetrical features and larger sizes of the samples compared to the
Crossbeam) was conducted on a pair of particles from the green sample input CAD file regardless of the build orientations, while the bottom
printed with SSL 90% to examine the cross-sectional view of an inter- part of the upright oriented sample was warped. The printed samples
particle bridge area. In order to protect the bridge area against spurious with SSL 90% exhibited virtually the same shapes for both the build
sputtering in the course of ion milling, a ribbon of 4 μm thick platinum orientations. In a similar way, the crucial effects of binder saturation
was initially coated on the bridge, and then the pair was cut by using level on binder jetting 3DP have been observed [30–32].
gallium-ion beam. In view of Eq. 1 for calculation of the minimum amount of solvent,
The green density of sample printed with SSL 90% was calculated W0 should be established experimentally; however, it could be esti-
from the measured mass of six cuboids of 15 mm × 15 mm × 15 mm mated in the range of 3 × 10−4 to 2.4 × 10-2, according to validated
and their Vernier caliper-measured dimensions. The density of sintered simulation-based and experimental studies [33,34]. Furthermore, the
samples was measured using Archimedes' principle. In view of the re- tap density of powder which is the tightest configuration of free-flowing
action between porous Mg samples with water, oil was used as the pore- particles can be considered as ρm. Since there is almost no compaction
filling liquid media. Three test specimens were tested to ensure con- involved while powder feedstock is being spread over the powder bed,
sistency and accuracy. each outspread layer could be packed in a close condition to the free-
flowing state of particles (i.e., the apparent density). Taking the mea-
3. Results and discussion sured apparent density for the Mg-Zn-Zr powder as the packing density
of particles in the powder bed and assuming that W0 equals to 2.4 × 10-
2
3.1. Capillary-mediated binderless three-dimensional printing of green parts , Eq.1 estimates the SSL at 16.91%. This value is considered as the
lowest threshold of SSL enabling capillary-mediated binderless 3DP of
3.1.1. Effect of the solvent saturation level on dimensional accuracy the Mg-Zn-Zr powder. In the 3DP technique, the preferred layer thick-
3DP build cycle of green parts needs to be meticulously designed to ness is about three times the particle size to facilitate powder feed-
attain desired parts’ features before proceeding with post-printing stock’s flow [13]. This means that each outspread layer of the Mg-Zn-Zr
processes (i.e., the second step of 3DP). Depositing an adequate amount powder composes of about three particles on the top of each other.
of solvent to each powder layer spread over a powder bed is of para- Thus, greater amounts of solvent than that of the lowest threshold value
mount importance for successful 3DP of green parts in light of our need to be used to assure fast penetrations of solvent along the vertical
previous work on capillary-mediated binderless 3DP of pure Mg alloy and lateral directions. As a result of triggering powder-solvent inter-
[25]. As discussed in forthcoming sections, the adequate amount of actions, a greater amount of solvent results in the formation of a larger
solvent ensures the formation of capillary bridges between neighboring number of capillary bridges and more solid interparticle bridges in both
powder particles, governs the capillary-mediated interactions between inter- and intra-layer. Vertical spreading of solvent is a vital factor in
the powder and solvent, and determines the integrity and dimensional determining the interface bond between layers that directly affects the
accuracy of green parts. Moreover, the sinterability of fabricated green mechanical integrity of green parts. When the SSL is low, inadequate
parts during subsequent sintering step is directly affected by the solvent spreading of solvent -especially along an interlayer- leads to partial
amount due to Mg’s extreme sensitivity to O content [26]. A funda- layer displacements in 3DP build cycle due to the shear forces imparted
mental framework of capillary-mediated binderless 3DP [25] can be during outspread of a new powder layer. In addition to this macro-

3
M. Salehi, et al. Additive Manufacturing 29 (2019) 100790

result of the gravitational force. As illustrated in Fig. 1b, the center of

gravity of the upright oriented sample is downwards while the excess
solvent in an upper layer permeated towards the bottom layers (the
vertical spreading effect). On the other hand, the center of gravity of the
upside-down oriented sample is upwards, whereas the vertical
spreading effect is still the same as the upright oriented sample. In view
of these geometrical differences observed in the printed pyramids with
SSL 150%, the impact of gravitational force should be considered when
it comes to developing a process-based model for inkjet-based AM
technologies.

3.1.2. Effect of the saturation level on powder’s interaction with solvent

In contrast to the constructive impact of depositing greater amounts
of solvent than that of the calculated lowest-threshold SSL on the build
cycle, the SSL should be minimized such that the chemical composition
of starting raw powder materials remains less degraded in green parts.
Such compositional changes in green parts can be estimated by the ratio
between the volume of interparticle bridges and powder particles that
gives a dimensionless quantity as following [25]:

(1-ρ) sin2β
V * = Wf
2ρ (2)

where V* is the bridge per particle volume, and? ? is the interparticle

bridges angle. As can be deduced from Eq. 2, the higher the SSL, the
greater the bridge per particle volume. Thus, a higher SSL causes the
chemical composition of 3D printed green parts to vary to a greater
extent as compared to the starting raw powder, making the subsequent
sintering process of the green parts likely more challenging. Further-
more, any compositional change in the sintered parts can be considered
as process contaminants deteriorating parts’ functionality.
In view of this contradiction in the choice of SSL, there has to be a
trade-off between printing parts with high dimensional accuracy and
Fig. 1. Binderless 3D printed Mg-5.06Zn-0.15 Zr in the green condition, a) minimal compositional change. To address this challenge in the current
pictorial representations of the green pyramids printed with different SSLs and work, the solvent content, which is equivalent to a required SSL, was
build orientations (indicated by arrows) and b) schematic illustrations of sol- deposited into each layer of the Mg-Zn-Zr powder followed by
vent spreading behavior in printed pyramid having two distinct build orienta- spreading a new layer of the powder to satisfy absolutely the hysteretic
tions. (For interpretation of the references to colour in this figure legend, the criterion in both inter- and intra-layer. Taking the lowest-threshold SSL
reader is referred to the web version of this article). into accounts, an excess amount of solvent in the previous layer was
eliminated before dispensing the solvent to the newly outspread layer.
distortion during the build cycle, either delamination or break of green Fig. 2a and b show the cross-sectional and longitudinal SEM micro-
parts during the de-powdering process could occur. Capillary-mediated graphs of the green sample printed with SSL 90%, respectively. As
binderless 3DP with SSL less than 35% was experimentally possible; shown, solid interparticle bridges derived from capillary-mediated
however, severe layer displacements during the build cycle made the bridging created interdigitate network of powder particles in both inter-
process go out of control. With increasing the SSL to 50%, the layer and intra-layers. μCT results in Fig. 2c and d clearly illustrate uniform
displacement issue was resolved. However, the corners showed rounded particle packing throughout the green sample, while interlayer zones
edges and overall the pyramid printed with SSL 50% provided a gran- were observed in binder jet 3D printed parts [19,35]. This uniformity is
ular finish due to eroding away during the depowdering step, leading to attributed to the appropriate spreading of the solvent along both ver-
geometrical inaccuracy (Fig. 1a). When the SSL is high, the extensive tical and lateral directions, obtaining uniform green parts for the sin-
lateral permeation causes the solvent to seep out of the designated print tering step. Fig. 3a displays exemplary funicular bridges (i.e., trimer,
area in each 2D layer. As a result of the vertical spreading, the residual tetrahedron, pentamer, etc.) whereby three or more particles are in
solvent in an upper layer could end up supplying excessive solvent to contact. Taking several relations developed for calculations of the co-
lower layers in case of printing successive layers with an excess of SSL. ordination number of particulates into accounts [36], the coordination
This causes dimensional errors and a relatively rough surface finish that number for the green sample with packing fraction of 0.55 is estimated
were observed in the pyramids printed with SSL 150% irrespective of in the range of 5.3 to 6.4. This is in good agreement with Fig. 3a in
the build orientation. In addition, asymmetries of faces and edges of the which the indicated particles are linked to five or six adjacent particles.
upright and upside-down built pyramids with SSL 150% have clearly
shown the interplay between the build orientation and SSL. In parti- 3.1.3. Conservation of raw powder’s chemical composition in green parts
cular, the bottom segment of the upright built sample was warped, and In addition to the choice of proper printing parameters, the che-
its size was larger than that of the upside-down built counterpart mical composition of green parts could remain more intact if the solvent
(Fig.1a). Since the center of gravity for the pyramid geometry is the is judiciously formulated in the light of powder material’s chemical
only parameter that varies with changing build orientations, it can be composition. The conceptual illustration of capillary-mediated binder-
postulated that the gravitational force could also contribute to the less 3DP is depicted in Fig. 3b. When the solvent is projected over a
overall velocity of solvent spreading and, in turn, solvent permeation layer of Mg powder, liquid bridges form spontaneously in the inter-
behavior. This is because the external pressure during the build cycle particle necks upon meeting the liquid layers adhering to the outermost
originates from the weight accumulation of the printed segment as a surfaces of each Mg particles. Meanwhile, interactions between the

4
M. Salehi, et al. Additive Manufacturing 29 (2019) 100790

Fig. 2. SEM micrographs and μCT images of the green samples printed with SSL 90%, a) SEM image taken from parallel to the layer stacking direction, b) SEM image
taken from perpendicular to the layer stacking direction, c) μCT shadow projection taken from parallel to the layer stacking direction, and d) μCT shadow projection
taken from perpendicular to the layer stacking direction. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version
of this article).

Fig. 3. Formation of capillary-mediated solid

interparticle necks among Mg particles, a) SEM
micrograph of funicular bridges in the green
sample printed with SSL 90%, b) conceptual
framework illustration for capillary-mediated
binderless 3DP of Mg alloys, c) SEM micro-
graph of a pendular capillary-mediated inter-
particle bridge, and d) SEM micrograph of fo-
cused ion beam milled bridge in c. (For
interpretation of the references to colour in this
figure legend, the reader is referred to the web
version of this article).

5
M. Salehi, et al. Additive Manufacturing 29 (2019) 100790

MgO film, unavoidably present on the outermost surface of all Mg 790 cm−1 are assigned to magnesium carbonate trihydrate [50]. Mag-
particles, and the solvent are initiated in the liquid bridges as well as in nesium hydride exhibits a broad transmittance band around wave-
the adhering layers far from the interparticle necks. According to si- number of 560 cm−1 and a flattop one in the range of 1000 cm−1 to
mulation studies on the dynamic evolution of capillary bridges [37,38], 1300 cm−1 [51,52]. The weak band at 467 cm−1, the one around wa-
the liquid contents in the adhering layers drain into the interparticle venumber of 496 cm−1, and the high-intensity band at 3698 cm−1
necks during a slow process of liquid flow. The driving force for this correspond to magnesium hydroxide in the FTIR spectrum [47,53,54].
autogenous liquid transfer process originates from the pressure differ- Finally, FTIR bands consistent with magnesium oxide are observed at
ences in these two zones. The liquid transfer continues until either the 416, 430, and 443 cm−1 due to Mg–O stretching vibration [54–56]. If
pressure difference becomes negligible or the liquid layers rupture. ZnO was formed, it would exhibit its characteristic IR vibrations at
Thus, the products of solvent’s superficial interactions with powder will 450 cm−1 [57] or 457 cm−1 [58] and its dominant Raman band at
largely accumulate throughout the interparticle necks, providing an 437 cm-1 as a very sharp and intense peak [59–61]. The vibrational
autogenously derived in-situ binding agent to connect powder particles spectroscopy results (Fig. 4) rule out the possibility of reactions be-
together. The nucleation and growth of these products along both tween the Zn content and the solvent as neither FTIR peaks nor Raman
vertical and horizontal directions resulted in solid interparticle necks bands corresponding to ZnO were detected. Comparing the obtained
among the neighboring Mg-Zn-Zr particles as can be seen in Fig. 3c and results in this study on the Mg-Zn-Zr powder with our previous work on
d. The comparative percentages of Mg, O, and C elements obtained by pure Mg powder [25] indicates that the same four compounds con-
EDS analysis (not present here) for the bridge in Fig. 3c implied that the stituted from the solvent’s interactions with the pure Mg powder as
interparticle neck could be composed of magnesium carbonate trihy- well. This suggests that the conversion of unavoidable MgO film which
drate (MgCO3.3H2O) as expected from the designed interactions of the exists on the surfaces of Mg alloys powder can successfully be deployed
Mg powder and the solvent. However, a thermodynamic calculation for binderless 3DP of Mg alloys in spite of Mg’s high reactivity and high
indicates that the Zn content in the Mg-Zn-Zr powder will react with the affinity to oxygen.
solvent to form ZnO if the MgO film is damaged [39]. Formation of ZnO The chemical analyses of the as-received Mg-Zn-Zr powder and the
in green samples is not desirable because it would consume the pow- green parts printed with two distinct SSL of 90% and 150% are dis-
der’s Zn content while it cannot be decomposed during sintering pro- played in Fig. 5, showing the marginal compositional change in terms of
cess due to its very high decomposition temperature (i.e., 1977 °C). hydrogen, carbon, and oxygen levels for the green parts as compared to
Vibrational spectroscopy results of the green sample printed with SSL the raw powder. Comparisons in terms of C, H, and O contents in the
90% are shown in Fig. 4. As labeled in Fig. 4a, magnesium hydride green samples revealed that the solvent reacted to a larger extent with
shows several sharp or broad bands in the Raman spectrum at 120, 268, the surface of particles in the printed sample with SSL 150% than that
365, 555, 953 and 1270 cm−1 and their relative intensity and positions of SSL 90%. A greater extent of the products of solvent-powder inter-
matches well with reported Raman shift for magnesium hydride powder actions in the SSL 150% sample was also realized via comparing SEM
[40]. The most intense Raman shift for magnesium carbonate trihydrate images (not present here) for these two SSLs. Owing to use of the same
is a band around 1098 cm−1 corresponding to the symmetric stretching set of parameters other than the SSL to print both samples, the higher
mode of CO3-2 unit which overlapped with Raman active bands of solvent content in a printed sample using SSL 150% led to a larger
magnesium hydride [41,42]. Likewise, magnesium oxide exhibits a quantity of reagents and longer lasting reactions, thereby forming a
strong broad Raman brand near 618 cm−1 that can be seen as a larger amount of the products in the resulting green sample. However,
shoulder in the Raman spectrum in Fig. 4a [43,44]. The characteristic the difference in O contents for the two SSLs was negligible which
Raman band at 3650 cm−1 belongs to magnesium hydroxide [45,46]. signifies that the solvent is effectively preserved in interparticle necks
The ATR-FTIR vibrational spectrum in Fig. 4b shows that bands con- within the porous structure of magnesium carbonate trihydrate to
sistent with magnesium carbonate trihydrate are at 386 and 855 cm−1 conserve the overall oxygen content in the green parts. This preserva-
due to the CO3-2 out-of-plane bending mode [47], 1405 and 1515 cm-1 tion of solvent in the porous structure of interparticle necks can also be
due to antisymmetric C − O stretching mode vibrations [42,48], and realized via Eq. 2. Substituting the following values in Eq. 2; the cal-
1660 cm−1 due to the OeH bending mode of water molecules [41,48]. culated lowest-threshold SSL (i.e., 16.91%) as Wf; the measured ap-
The bands observed outside the FTIR fingerprint region in the range of parent density for the Mg-Zn-Zr powder (i.e., 0.5178) as ρ; and the
1950 cm-1 to 2550 cm−1 are usually assigned to the asymmetric OCO interparticle bridges angle of 45°, yields the bridge per particle volume
stretching mode of magnesium carbonate trihydrate and its second 3.94%. This value is larger than the total compositional changes ob-
overtone of the symmetric OCO stretching mode [49]. In addition, the served in the green samples printed with either SSL 90% or 150%
low-intensity vibrational features observed in the range of 620 cm−1 to (Fig. 5). This is because the assumption for deriving Eq. 2 is that the

Fig. 4. Vibrational spectroscopy of the green Mg-Zn-Zr samples printed with SSL 90%, a) Raman spectrum, and b) ATR-FTIR spectrum.

6
M. Salehi, et al. Additive Manufacturing 29 (2019) 100790

MW sintering.

3.2.1. Selection of temperature range for sintering

Solid state sintering of uncompacted Mg powder as in green parts is
kinetically unfeasible since the diffusion of Mg atoms through the un-
interrupted oxide film is extremely hindered [26]. This is because the
diffusion coefficient of Mg atoms through the MgO is several orders of
magnitude lower in comparison to its self-diffusion coefficient [26]. In
this regard, a possible way of disrupting the MgO film in green Mg parts
is to use liquid phase sintering (LPS). According to the mechanisms for
supersolidus LPS, the liquid phase forms both within the grain and
along the grain boundaries of powder particles when sintering tem-
perature reaches above the solidus line [62]. This nucleation of liquid
phase disintegrates the particles. Given that the Mg particles in green
parts are surrounded by MgO film, the forming liquid phase is enclosed
inside each individual Mg particle. With increasing sintering tempera-
ture and forming more liquid pockets within each particle, the MgO
Fig. 5. Chemical analysis of the as-received Mg-Zn-Zr powder together with
green and sintered samples (printed with SSL 90%). (For interpretation of the film could rupture due to increasing internal pressure inside the parti-
references to colour in this figure legend, the reader is referred to the web cles. The rupture of particles turns the part into liquid phase in a solid
version of this article). matrix, initiating the sinter densification as a result of formations of
paths for accelerated diffusion of Mg atoms through liquid bridges
forming between neighboring Mg particles [26]. To determine an ap-
solvent content in the interparticle neck fully reacts with powder par-
propriate range of temperature for LPS, DSC was conducted on Mg-
ticles. Rather, the interactions stop upon formations of magnesium
5.06Zn-0.15 Zr powder and the curves are shown in Fig. 6a. The liquid
carbonate trihydrate in the green parts as it insulates the solvent from
volume fraction against temperature for Mg-5.06Zn-0.15 Zr alloy,
MgO film. Furthermore, this hydrated magnesium carbonate does not
which is calculated by thermodynamic software, is also plotted in
react with the solvent upon its formation, according to thermodynamic
Fig. 6b. Virtually both the DSC and calculation results identified the
calculations. All in all, the porous structures of MgCO3.3H2O along with
same liquidus temperature of ˜ 634 °C; conversely, there is an incon-
its insulating and stability allow the existing solvent content in the
sistency between the solidus temperature acquired through the calcu-
interparticle necks to partially react with the Mg-Zn-Zr powder while
lation (˜ 341 °C) and the one by DSC. This discrepancy is originated
the hysteretic criterion of the liquid bridge in terms of the minimum
from the fact that DSC is conducted on the powder surrounded by MgO
liquid content was already satisfied. This remaining solvent preserved
film. Furthermore, the onset of a transformation obtained from the
within the interparticle necks could be slowly vaporized during the 1-h
results of thermal analysis is dependent on the heating rate as indicated
holding step. In addition to taking these drastic measures to minimize
in Fig. 6a, whereas the calculation is based on thermodynamic equili-
the overall compositional change in the green parts, the products of
brium. Taking both the calculation and DSC results into account, the
powder’s interactions with solvent is determined such that all products
range of sintering temperature starting from 500 °C which corresponds
thermally decompose into their primary constituents in the raw Mg
to the liquid fraction of ˜ 10 vol. % was selected for MW sintering. It
powder during a sintering step as discussed in Section 3.2.2.
should be noted that the optimal amount of liquid phase for sintering of
die-pressed powder metallurgy compacts is between 20 vol. % and
3.2. Microwave sintering of green parts 40 vol. % [62], and the optimum amount of liquid phase for sintering of
3D printed Mg parts in a CN furnace under the constant flow of Ar gas
Given that geometrical accuracy combined with minimal composi- was found to be 22 vol.% [26].
tional change obtained in the green sample printed with SSL 90%,
samples printed with 90% SSL were used for further investigations into

Fig. 6. Temperature dependence behavior of Mg-5.06Zn-0.15 Zr powder/alloy, a) DSC curves of the powder performed under two heating rates of 5 °C/min and
10 °C/min, and b) liquid fraction of the alloy as a function of temperature calculated by using thermodynamic software. (For interpretation of the references to colour
in this figure legend, the reader is referred to the web version of this article).

7
M. Salehi, et al. Additive Manufacturing 29 (2019) 100790

263 °C–459 °C, respectively [41,63,64]. Furthermore, the TGA results

showed that the overall mass loss occurred in the range of 50–440 °C is
0.44%. This overall mass loss is in good agreement with the chemical
analysis results obtained for the green sample printed with SSL 90% in
terms of the sum total of C, H, and O elements (Fig. 5).
Taking the above-mentioned measures during 3DP build cycle to
retain the composition of green parts intact (Section 3.1.3), in combi-
nation with the complete decompositions of the products of powder’s
interaction with solvent into their initial constituents in the raw Mg-Zn-
Zr powder during the MW heating, brings about a compositionally zero-
sum AM process. Such zero process contaminants were also attained by
CN sintering of capillary-mediated binderless 3D printed Mg parts in a
tube furnace under constant flow of Ar gas [25,26], indicating that the
green samples fabricated by this ex-situ binder free approach can
readily proceed with sintering in both MW and CN furnaces in the
Fig. 7. The SEM micrograph of green Mg-5.06Zn-0.15 Zr sample printed with
absence of any debinding step. These flexibilities inimitably aid the
SSL 90% after MW heating to 500 °C with no isothermal dwell time. (For in- capillary-mediated binderless 3DP in lead-time and environmental
terpretation of the references to colour in this figure legend, the reader is re- sustainability. It should be noted that the presence of polymeric binders
ferred to the web version of this article). in binder jet 3DP method prevents green parts from processing with
MW during a single-step sintering process. Thus, the green part fabri-
cated by binder jet 3DP approach was thermally debound and partially
3.2.2. Microwave’s interactions with green parts: Retrieval of primary
sintered before proceeding with MW sintering [65]. This is because a
chemical composition
thermal debinding profile should be adjusted to facilitate pyrolysis
Fig. 7 shows an SEM micrograph of the MW sintered sample at
process in a manner that the decomposition rate of polymeric binders
500 °C without an isothermal dwell, and the corresponding chemical
matches with the transport rate of pyrolysis’s gaseous by-products to
analysis results in terms of C, H, and O are illustrated in Fig. 5. These
prevent green parts from distortion and cracking. This usually is
results demonstrated that all products of powder’s interaction with the
achieved at a slow heating rate. In a lack of relevant data, the literature
solvent that superficially formed during 3DP build cycle were decom-
on powder injection molding process having usage of binder materials
posed, and the primary composition of raw Mg powder was totally
in common with binder jet 3DP could be beneficial for the sake of
retrieved from the green sample during MW heating in the absence of a
comparison. As an example, in powder injection molding of 430
debinding step, resulting in a compositionally zero-sum process with
stainless steel and after performing one step of solvent debinding, it
zero process contaminants. In order to investigate decompositions of
took two hours for thermal debinding in MW while thermal debinding
the green sample further, TGA was performed on the green sample with
in CN furnace required 12 h [66], suggesting that an MW assisted de-
heating rate matching with that of MW’s sintering ramp rate (i.e., ˜
binding process though saves time is still a time-consuming step. In
9 °C/min). The TGA result (Fig. 8) indicated that the mass loss started
powder injection molding of Mg alloys as another example, the com-
from 50 °C, prolonged continuously up to 300 °C, and then followed by
bined solvent and thermal debinding steps took up to 20 h [67–69],
an additional mass loss between 320 and 400 °C. These thermal de-
indicating the unique advantage of the binderless 3DP approach de-
compositions correspond to the decomposition behavior of the afore-
monstrated here for Mg alloys.
mentioned products of Mg-Zn-Zr powder’s interaction with the solvent
(Section 3.1.3). More specifically, the three-step dehydration reactions
3.2.3. Effect of sintering temperature on shape fidelity
of MgCO3.3H2O occur between 55 and 270 °C followed by the dec-
Maintaining shape fidelity of parts is a formidable challenge for LPS.
arbonization of anhydrous MgCO3 to MgO in the range 300 °C–442 °C
On the one hand, greater amounts of liquid formed at higher sintering
[41]. The dehydrogenation of MgH2 to Mg and dihydroxylation of Mg
temperature result in enhanced densification rate during LPS. On the
(OH)2 to MgO take place in the range of 330 °C–404 °C and
other hand, larger fractions of liquid phase cause less skeletal stability
in parts during LPS, leading to distortion, swelling, dimensional in-
accuracy, etc. Loss of shapes, distortion, and collapse in LPS due to the
formation of extensive liquid phase have been observed for binder jet
3D printed parts made from stainless and nickel-based superalloys
[70–72]. In a similar way to binder jet 3DP as a two-step AM method,
all dimensional deviation in the capillary-mediated binderless 3DP is
virtually dictated by the second step. In view of the importance of net
shape/near-net shape fabrication in AM, sintering temperature should
be determined such that a green part only shrinks, preferably isotropic
shrinkage. Fig. 9 illustrates macrographs of MW sintered samples
without isothermal dwell time at four different temperatures, namely,
500 °C, 520 °C, 540 °C, and 560 °C. As shown, the liquid phase in the
sample sintered at 520 °C pulled out to the surface forming small
spheres while more liquid phase exuded with increasing sintering
temperatures. Swelling in compacts of various powder made of Al, Ag,
and Mg alloys during LPS have been observed previously [73–75].
Swelling causes not only undesired shape loss of parts but also a macro-
scale variation in chemical composition. This compositional variation is
anticipated from the phase diagram whereby the composition of cor-
responding liquid and solid phases are different from the nominal
Fig. 8. TGA of the green Mg-5.06Zn-0.15 Zr sample printed with SSL 90% that composition of an alloy along the ranges of temperature applicable to
performed under heating rate of 10 °C/min. LPS. Swelling during super-solidus LPS of 3D printed Mg parts stems

8
M. Salehi, et al. Additive Manufacturing 29 (2019) 100790

90% is 55 ± 1. This value is comparable with the range of values

obtained for the green parts printed with binder jetting technology such
as Ti (45.5–58.3 %) [15], stainless steel (53–58 %) [79], and nickel-
based superalloy 625 (50%) [35]. Given that interparticle forces acting
between powder particles and lack of external force for compaction of
powder particles while being spread over the powder bed, the obtain-
able density value in the green sample was expected to be the range of
the apparent density to tap density [12]. In this regard, the relative
density of green samples is between the measured values for the ap-
parent (51.78%) and tap density (57.75%) of Mg-5.06Zn-0.15 Zr
powder. In line with Fig. 3a, the coordination number of Mg particles in
the green sample is higher than that of the free-flowing condition of
powder (i.e., the apparent density) and lower than that of tightest
packing configuration for non-compacted powder (i.e., the tap density).
Fig. 9. Pictorial representations of printed Mg-5.06Zn-0.15 Zr cylinders having The results of density measurement also indicate that the powder
centimeter-scale dimensions after MW sintering without socking time at various packing configurations in the sintered sample without dwell time is still
temperatures. (For interpretation of the references to colour in this figure le- lower than the tap density. Particle rearrangements occurred in the
gend, the reader is referred to the web version of this article). samples sintered for 1 h led to denser packing configuration than that of
the tapped powder. Comparing density results obtained in MW with
from the presence of oxide film on the outermost surface of Mg particles results of 3D printed Mg-Zn-Zr samples sintered in the CN furnace
which leads to insufficient wetting of forming liquid phase with the (Table 2), clearly demonstrates that the MW sintering significantly re-
particles, causing an excess liquid content to pull out under the effect of duces the sintering time by a factor of three to four times. Such accel-
the surface tension [26]. In view of the root cause of swelling, one way eration of densification kinetic during sintering under MW when com-
to deal with the observed swelling problem could be reducing MgO film pared to CN furnaces was also reported earlier for Cu [80] and Ni [81].
by addition of sintering aids. The sintering aids should contain an ele- The heating mechanism in MW fundamentally differs from CN
ment with a greater affinity for oxygen such as calcium and yttrium heating. In CN heating, heat transfers from an external source to sam-
[76,77]. These elements react with the MgO film, reduce the MgO to ples through convention, conduction, and radiation. In MW heating,
Mg, and form their own oxide during LPS while the overall oxygen samples themselves absorb MW energy, and heat is directly generated
content still remains the same. In this regard, the capillary-mediated through the volume of samples. In addition to this thermal effect in MW
binderless 3DP as a compositionally zero-sum process offers a great heating, an interaction between the alternating electromagnetic field
advantage in conserving O content for the further ease of the sintering with samples results in non-thermal effects, also known as MW effects.
step. Accordingly, the driving force induced by MW is not merely by a
In view of our earlier investigation on sintering of green Mg-Zn-Zr thermal effect as CN heating, but the non-thermal phenomenon con-
parts in an electrical-resistance furnace, swelling occurs at the liquid tributes significantly leading to unexpected kinetic enhancements in
fraction ˜ 24 vol. % [26]. In the present study, swelling during MW MW [82,83]. Despite the fact that the root cause of the kinetic accel-
sintering happened at much lower liquid content of ˜ 11 vol. %. Vir- eration under MW sintering is originated from the electromagnetic
tually the sintering approach is the only difference between the current field, its physical explanation is still a matter of debate. For example,
study and our previous work, and thus the reason for this observation enhancements in sinter neck growth of Ni powder in MW over the si-
should be associated with the root causes of swelling (i.e., insufficient milar sample processed by CN sintering was attributed to formation of
wettability between MgO and molten Mg). Studies into wetting of liquid in the vicinity of interparticle neck regions (while the sintering
molten Mg droplets by MgO substrates at various time durations up to temperature recorded was ˜ 550 °C below the melting point of Ni), re-
30 min revealed that the contact angle between the molten Mg and sulting in a remarkable decrease of activation energy for diffusion
MgO decreases as time proceed [78]. In view of much slower rate of under MW sintering compared to a CN counterpart [81]. In contrast, no
liquid phase formation during LPS of the green Mg parts in the CN such presence of liquid phase nor variation in activation energy was
furnace (i.e. ˜ 1 h isothermal dwell time at the corresponding tem- observed in MW sintering of Cu while an anomalous neck growth rate
perature to realize liquid fraction ˜ 24 vol. % based on our previous occurred [80]. A thermodynamic study of die-pressed compacts made
work) in comparison with the MW furnace in the present study (nil of Al2O3 and ZnO powder in both MW and CN furnaces argued that the
holding time), the contact angle in the CN heating -which is a measure anomalous densification is not the effect of change in activation energy
of the wettability- could be lower. Accordingly, the wettability between [84]. The microstructural analysis could give an insight as to how the
the MgO film and liquid Mg in the CN furnace could be enhanced with accelerated densification observed in the present study proceeded.
time, resulting in accommodation of a greater amount of the liquid According to the aforementioned mechanisms for the formation of solid
phase in the parts without swelling. interparticle bridges during the 3DP build cycle in Section 3.1.3, the
oxide film in the vicinity of the bridge becomes thinner due to its
progressive interaction with the solvent to form MgCO3.3H2O. Fur-
3.2.4. Effect of isothermal dwell time on physical properties: Thermal and thermore, the decomposition of MgCO3.3H2O and MgH2 causes the
non-thermal phenomena release of gaseous by-products resulting in the formation of some pores
In view of the vital importance of the shape fidelity in additively and cracks in the MgO product or the pre-existing MgO, in addition to
manufactured parts, sintering temperature of 510 °C was chosen for the flaws that already exist in the MgO film given its non-compact
further investigation. Fig. 10 shows SEM micrographs of the MW sin- nature. Hence, the interparticle neck areas are particularly prone to
tered samples at three different isothermal time intervals of 0 h, 1 h, rupture than the surface of particles when an adequate liquid phase
and 5 h. The corresponding values for density of these samples along forms within each particle during LPS process. A burst of Mg particles
with the density of the green sample are shown in Table 2. As a result of from the neck regions is in line with SEM examinations of MW sintered
improved mass transport, SEM images illustrated that the diameter of samples (Fig. 10). Such a burst of particles and seepage of the interior
sinter necks between contacting Mg particles enlarges with increasing liquid phase into the surface of Mg particles are a direct consequence of
dwell time and the density of sintered samples. As can be seen in the thermal effect happening in both the MW and CN heating during
Table 2, the relative density of the green Mg sample printed with SSL LPS process. In view of lower sintering temperature in MW sintering

9
M. Salehi, et al. Additive Manufacturing 29 (2019) 100790

Fig. 10. SEM micrographs of 3D printed samples with SSL 90% after MW sintering at 510 °C for various isothermal time intervals, a) 0 h, b) 1 h, and c) 5 h.

(510 °C) in the current study than that of the samples conventionally shells can disrupt connectivity between powder particles and even
sintered (573 °C) as indicated in Table 2, the diffusion rate should be electrically insulate the metal cores from one another, resulting in a
much lower in MW sintering when the thermal effect is solely taken into powder compact with lower loss dielectric that could affect microwave
account. Calculation of the coefficient of total diffusivity, that is, the sintering significantly [85–88]. The eddy currents generated by the MW
collective diffusivities of the solid and liquid phases existing at each field in the shell can be one order of magnitude higher than that of the
temperature, shows that the total diffusivity for those 3D printed ZK core of particles, and thus the oxide shell heats up rapidly such that the
powder that were conventionally sintered at 573 °C is superficial temperature of particles might be considerably higher than
1.46 × 10−12 m2/sec while the total diffusivity for the MW sintering in the overall temperature which is measuring externally [89]. According
this study at 510 °C is 2 × 10-13 m2/sec. This indicates that the diffusion to simulation-based investigations [85,88], faster and more stable
rate in the MW sintering at 510 °C would be 7.3 times lower than those heating rate, more stabilized and greater microwave power absorption,
3D printed samples conventionally sintered at 573 °C. Given the re- and greater densifications are obtainable in metal powder by virtue of
markable shorter isothermal dwell time in MW (Table 2), much lower even a very thin shell (e.g., 1 nm). As per discussion in Section 3.1.3,
density would be anticipated compared to those conventionally sin- the products of solvent’s superficial interactions with the Mg powder in
tered samples if the thermal effect is solely taken into account. Inter- the adhering layers are transferred and accumulated within the inter-
estingly, the accelerated kinetics of densification observed in MW sin- particle bridges during the 3DP build cycle (Fig. 11a). These superficial
tering over CN sintering (Table 2) indicates a substantial contribution of products subsequently decompose into their primary reagents over the
the non-thermal effects to the densification of the sintered sample under initial course of sintering process as discussed in Section 3.2.2, forming
MW. Such a contribution made by the electromagnetic field could be an aggregate of powder particles in which the particles are connected to
realized via the observed microstructural characterizations as follows. each other through interparticle bridges made of MgO while each in-
Interaction of MW with metals is different from ceramics. Metal dividual particle consists of a Mg metal core enveloped in a very thin
powder with an oxide film on its outermost surface can be considered as MgO film as depicted in Fig. 11b. Continuing with the MW sintering
a core-shell structure consisting of a conductive metallic core and di- process, the electromagnetic field can be surpassed to some extent at
electric ceramic shell. Both the core and shell absorb microwave en- the interparticle bridge regions which transforms into heat and produce
ergy. Simulation and experimental studies showed that the amount of a temperature gradient. This exterior gradient from the Mg metal cores
electromagnetic energy absorbed in the dielectric shell can be com- to the bridges could enable the mass transport between adjacent par-
parable to that absorbed in the metal core [85–88]. Moreover, the oxide ticles to anomalously promote as a result of the accelerated diffusion

Table 2
Density of the green and sintered Mg samples in MW and CN furnaces.
Sample conditions Density (g/cm3) Relative density (%) Ref

Green sample printed with SSL 90 % 1.00 ± 0.02 55 ± 1 The present study
MW sintered at 510 °C for 0 h 1.03 ± 0.02 56.25 ± 0.80 The present study
MW sintered at 510 °C for 1 h 1.09 ± 0.01 59.35 ± 0.79 The present study
MW sintered at 510 °C for 5 h 1.15 ± 0.01 62.87 ± 0.42 The present study
3D printed ZK powder-green sample 1.02 55.49 [26]
3D printed ZK powder- sintered sample in a CN tube furnace at 573 °C for 5 h 1.08 59.27 [26]
3D printed ZK powder- sintered sample in a CN tube furnace at 573 °C for 20 h 1.15 63.11 [26]

10
M. Salehi, et al. Additive Manufacturing 29 (2019) 100790

Fig. 11. Graphical representation of super-so-

lidus LPS MW sintering of capillary-mediated
binderless 3D printed Mg-Zn-Zr samples, a)
green samples in which the particles are con-
nected via the capillary-mediated bridges
made of MgCO3.3H2O, b) early stage of MW
sintering in which all products of the powder’s
interactions with solvent fully decompose into
either Mg or MgO, the MgO film encapsulating
Mg metal cores breaks, and liquid bridges form
among particles in a similar way to the CN
sintering process (the thermal effect of MW
sintering), c) later stage of MW sintering in
which the electromagnetic field is locally sur-
passed at the interparticle bridge regions,
transformed into heat, and produced a tem-
perature gradient, resulting in accelerated dif-
fusion of Mg atoms between adjacent particles
(the non-thermal effects of MW sintering), and
d) printed sample with SSL 150% after MW
sintering at 510 °C for 1 h in which sintering
behavior of indicated particles with red arrows
are clearly different from the adjacent parti-
cles, supporting the non-thermal effects. (For
interpretation of the references to colour in this
figure legend, the reader is referred to the web
version of this article).

Fig. 12. SEM micrographs of the samples printed with two different SSLs after MW sintering at 510 °C for 1 h, a) printed sample with SSL 90%, and b) printed sample
with SSL 150%.

rate of Mg atoms through the oxide bridges while the Mg core tem- MW sintering at 510 °C for 1 h. As can be seen, the diameter of sinter
perature can sustain close to the overall sintering temperature necks in the sample printed with SSL 150% is larger compared to the
(Fig. 11c). As shown in Fig. 11d, sintering behavior of the indicated one printed with SSL 90%, supporting the aforementioned hypothesis
particles with red arrows clearly differed from the rest of adjacent about the contribution of the non-thermal effects to densify the green
particles. Interestingly for the indicated particles, sinter necks were not samples under MW sintering, which is not the case for CN sintering.
developed, and particle shapes were not deformed in comparison with Inability to match the anomalous sintering kinetic under MW sintering
adjacent particles. This suggests the significance of the non-thermal with the kinetics of CN sintering in terms of densification rate and sinter
effects given that the indicated particles had minimal interparticle necks growth rate was previously observed for Ti [90], Cu [80], and Ni
bridges in the green sample as can be seen from Fig. 11d. In order to [81] powder.
confirm this postulation of substantial contributions of the interparticle
bridges to the sinter necks growth, MW sintering was performed on the
samples printed with SSL 150%. The SSL 150% sample would expect to 4. Conclusions
have a greater MW absorption in terms of the non-thermal effects, given
that the solvent reacted with the powder to a greater extent in this A comprehensive perspective on the process development con-
sample than that of the SSL 90% (Fig. 5). Fig. 12 shows SEM micro- siderations and the underlying principles for the capillary-mediated
graphs of the samples printed with two SSLs of 90% and 150% after binderless 3DP of a new powder material were systematically described
based on Mg-5.06Zn-0.15 Zr alloy. For the first time in the 3DP

11
M. Salehi, et al. Additive Manufacturing 29 (2019) 100790

technology, MW sintering was effectively employed as the second step parts in Binder Jetting process, Addit. Manuf. 20 (2018) 1–10.
for densification of the 3D printed green Mg-5.06Zn-0.15 Zr parts. The [15] E. Wheat, M. Vlasea, J. Hinebaugh, C. Metcalfe, Sinter structure analysis of titanium
structures fabricated via binder jetting additive manufacturing, Mater. Des. 156
following conclusions are drawn from the present study: (2018) 167–183.
[16] A. Khalyfa, S. Vogt, J. Weisser, G. Grimm, A. Rechtenbach, W. Meyer,
1 While SSL should be reduced to minimize the compositional change M. Schnabelrauch, Development of a new calcium phosphate powder-binder system
for the 3D printing of patient specific implants, J. Mater. Sci. Mater. Med. 18 (5)
in the green parts, a considerably higher SSL than that of the lowest (2007) 909–916.
threshold value is required for the formation of adequate numbers of [17] T. Wang, R. Patel, B. Derby, Manufacture of 3-dimensional objects by reactive inkjet
capillary-mediated bridges in both inter- and intra-layers of the printing, Soft Matter 4 (12) (2008) 2513–2518.
[18] E. Stevens, S. Schloder, E. Bono, D. Schmidt, M. Chmielus, Density variation in
green parts. Green parts printed with SSL 90% provide the best re- binder jetting 3D-printed and sintered Ti-6Al-4V, Addit. Manuf. 22 (2018) 746–752.
sults in the present study. [19] I. Rishmawi, M. Salarian, M. Vlasea, Tailoring green and sintered density of pure
2 MgCO3.3H2O is autogenously derived in interparticle bridges ser- iron parts using binder jetting additive manufacturing, Addit. Manuf. 24 (2018)
508–520.
ving as in-situ binding agent to connect the Mg-Zn-Zr powder par-
[20] A. Yegyan Kumar, Y. Bai, A. Eklund, C.B. Williams, The effects of Hot Isostatic
ticles together in green parts while its porous structure allows pre- pressing on parts fabricated by binder jetting additive manufacturing, Addit. Manuf.
servation of the raw powder’s chemical composition during the 3DP 24 (2018) 115–124.
build cycle. Only marginal compositional changes in the green [21] Z.C. Cordero, D.H. Siddel, W.H. Peter, A.M. Elliott, Strengthening of ferrous binder
jet 3D printed components through bronze infiltration, Addit. Manuf. 15 (2017)
samples when compared to the raw Mg-Zn-Zr powder are observed. 87–92.
3 In the absence of debinding-adapted MW sintering profile, the initial [22] M. Oghbaei, O. Mirzaee, Microwave versus conventional sintering: a review of
chemical composition of starting Mg-Zn-Zr powder is retrieved from fundamentals, advantages and applications, J. Alloys. Compd. 494 (1) (2010)
175–189.
the green parts, resulting in a compositionally zero-sum AM process. [23] D. El Khaled, N. Novas, J.A. Gazquez, F. Manzano-Agugliaro, Microwave dielectric
4 In the temperature range investigated for MW sintering of the green heating: applications on metals processing, Renew. Sustain. Energy Rev. 82 (2018)
parts, the optimum liquid content required to break the MgO film 2880–2892.
[24] R.R. Mishra, A.K. Sharma, A review of research trends in microwave processing of
while avoiding the swelling effect is established at 510 °C corre- metal-based materials and opportunities in microwave metal casting, Crit. Rev.
sponding to a liquid fraction of ˜ 10 vol. %. Solid State Mater. Sci. 41 (3) (2016) 217–255.
5 While the thermal effect contributes lesser towards densification of [25] M. Salehi, S. Maleksaeedi, S.M.L. Nai, G.K. Meenashisundaram, M.H. Goh,
M. Gupta, A paradigm shift towards compositionally zero-sum binderless 3D
the green parts in MW sintering than CN sintering, the non-thermal printing of magnesium alloys via capillary-mediated bridging, Acta Mater. 165
effects originated from the interaction between the interparticle (2019) 294–306.
bridges and electromagnetic field make a substantial contribution to [26] M. Salehi, S. Maleksaeedi, M.A.B. Sapari, M.L.S. Nai, G.K. Meenashisundaram,
M. Gupta, Additive manufacturing of magnesium–zinc–zirconium (ZK) alloys via
densification during MW sintering. As such, MW sintering sig-
capillary-mediated binderless three-dimensional printing, Mater. Des. 169 (2019)
nificantly reduces the sintering time by a factor of three to four 107683.
times when compared to CN sintering. [27] M. Bhattacharya, T. Basak, A review on the susceptor assisted microwave proces-
sing of materials, Energy 97 (2016) 306–338.
[28] K. Lu, M. Hiser, W. Wu, Effect of particle size on three dimensional printed mesh
References structures, Powder Technol. 192 (2) (2009) 178–183.
[29] Z. Zhou, F. Buchanan, C. Mitchell, N. Dunne, Printability of calcium phosphate:
[1] B.P. Conner, G.P. Manogharan, A.N. Martof, L.M. Rodomsky, C.M. Rodomsky, calcium sulfate powders for the application of tissue engineered bone scaffolds
D.C. Jordan, J.W. Limperos, Making sense of 3-D printing: creating a map of ad- using the 3D printing technique, Mater. Sci. Eng. C 38 (2014) 1–10.
ditive manufacturing products and services, Addit. Manuf. 1-4 (2014) 64–76. [30] H. Miyanaji, S. Zhang, A. Lassell, A. Zandinejad, L. Yang, Process development of
[2] S.A.M. Tofail, E.P. Koumoulos, A. Bandyopadhyay, S. Bose, L. O’Donoghue, porcelain ceramic material with binder jetting process for dental applications, JOM
C. Charitidis, Additive manufacturing: scientific and technological challenges, 68 (3) (2016) 831–841.
market uptake and opportunities, Mater. Today 21 (1) (2018) 22–37. [31] H. Miyanaji, S. Zhang, L. Yang, A new physics-based model for equilibrium sa-
[3] N.A. Zumdick, L. Jauer, L.C. Kersting, T.N. Kutz, J.H. Schleifenbaum, D. Zander, turation determination in binder jetting additive manufacturing process, Int. J.
Additive manufactured WE43 magnesium: a comparative study of the micro- Mach. Tools Manuf. 124 (2018) 1–11.
structure and mechanical properties with those of powder extruded and as-cast [32] M. Vaezi, C.K. Chua, Effects of layer thickness and binder saturation level para-
WE43, Mater. Charact. 147 (2019) 384–397. meters on 3D printing process, Int. J. Adv. Manuf. Technol. 53 (1) (2011) 275–284.
[4] S. Gangireddy, B. Gwalani, K. Liu, E.J. Faierson, R.S. Mishra, Microstructure and [33] M. Scheel, R. Seemann, M. Brinkmann, M. Di Michiel, A. Sheppard, B. Breidenbach,
mechanical behavior of an additive manufactured (AM) WE43-Mg alloy, Addit. S. Herminghaus, Morphological clues to wet granular pile stability, Nat. Mater. 7
Manuf. 26 (2019) 53–64. (2008) 189.
[5] H. Takagi, H. Sasahara, T. Abe, H. Sannomiya, S. Nishiyama, S. Ohta, K. Nakamura, [34] M.M. Kohonen, D. Geromichalos, M. Scheel, C. Schier, S. Herminghaus, On capillary
Material-property evaluation of magnesium alloys fabricated using wire-and-arc- bridges in wet granular materials, Phys. A Stat. Mech. Its Appl. 339 (1) (2004) 7–15.
based additive manufacturing, Addit. Manuf. 24 (2018) 498–507. [35] A. Mostafaei, S.H.V.R. Neelapu, C. Kisailus, L.M. Nath, T.D.B. Jacobs, M. Chmielus,
[6] Y. Li, J. Zhou, P. Pavanram, M.A. Leeflang, L.I. Fockaert, B. Pouran, N. Tümer, Characterizing surface finish and fatigue behavior in binder-jet 3D-printed nickel-
K.U. Schröder, J.M.C. Mol, H. Weinans, H. Jahr, A.A. Zadpoor, Additively manu- based superalloy 625, Addit. Manuf. 24 (2018) 200–209.
factured biodegradable porous magnesium, Acta Biomater. 67 (2018) 378–392. [36] R.M. German, Coordination number changes during powder densification, Powder
[7] C. Liu, M. Zhang, C. Chen, Effect of laser processing parameters on porosity, mi- Technol. 253 (2014) 368–376.
crostructure and mechanical properties of porous Mg-Ca alloys produced by laser [37] M. Wu, S. Radl, J.G. Khinast, A model to predict liquid bridge formation between
additive manufacturing, Mater. Sci. Eng. A 703 (2017) 359–371. wet particles based on direct numerical simulations, Aiche J. 62 (6) (2016)
[8] K. Wei, Z. Wang, X. Zeng, Influence of element vaporization on formability, com- 1877–1897.
position, microstructure, and mechanical performance of the selective laser melted [38] M. Wu, J.G. Khinast, S. Radl, Liquid transport rates during binary collisions of
Mg–Zn–Zr components, Mater. Lett. 156 (2015) 187–190. unequally-sized particles, Powder Technol. 309 (2017) 95–109.
[9] Y. Yang, P. Wu, X. Lin, Y. Liu, H. Bian, Y. Zhou, C. Gao, C. Shuai, System devel- [39] C.W. Bale, E. Bélisle, P. Chartrand, S.A. Decterov, G. Eriksson, A.E. Gheribi, K. Hack,
opment, formability quality and microstructure evolution of selective laser-melted I.H. Jung, Y.B. Kang, J. Melançon, A.D. Pelton, S. Petersen, C. Robelin, J. Sangster,
magnesium, Virtual Phys. Prototyp. 11 (3) (2016) 173–181. P. Spencer, M.A. Van Ende, FactSage thermochemical software and databases,
[10] S. Palanivel, P. Nelaturu, B. Glass, R.S. Mishra, Friction stir additive manufacturing 2010–2016, Calphad 54 (2016) 35–53.
for high structural performance through microstructural control in an Mg based [40] L. Matović, N. Novaković, S. Kurko, M. Šiljegović, B. Matović, Z. Kačarević Popović,
WE43 alloy, Mater. Des. 65 (1980-2015) (2015) 934–952. N. Romčević, N. Ivanović, J. Grbović Novaković, Structural destabilisation of MgH2
[11] M. Salehi, S. Maleksaeedi, H. Farnoush, M.L.S. Nai, G.K. Meenashisundaram, obtained by heavy ion irradiation, Int. J. Hydrogen Energy 34 (17) (2009)
M. Gupta, An investigation into interaction between magnesium powder and Ar gas: 7275–7282.
implications for selective laser melting of magnesium, Powder Technol. 333 (2018) [41] M.C. Hales, R.L. Frost, W.N. Martens, Thermo-Raman spectroscopy of synthetic
252–261. nesquehonite - Implication for the geosequestration of greenhouse gases, J. Raman
[12] M. Salehi, M. Gupta, S. Maleksaeedi, M.L.S. Nai, Inkjet Based 3D Additive Spectrosc. 39 (9) (2008) 1141–1149.
Manufacturing of Metals, Materials Research Forum LLC, Millersville, PA, USA, [42] J.T. Kloprogge, W.N. Martens, L. Nothdurft, L.V. Duong, G.E. Webb, Low tem-
2018. perature synthesis and characterization of nesquehonite, J. Mater. Sci. Lett. 22 (11)
[13] S. Maleksaeedi, G.K. Meenashisundaram, S. Lu, M. Salehi, W. Jun, Hybrid binder to (2003) 825–829.
mitigate feed powder segregation in the inkjet 3D printing of titanium metal parts, [43] A.K. Sharma, A. Kvit, J. Narayan, Growth of single crystal MgO on TiN/Si hetero-
Metals 8 (5) (2018) 322. structure by pulsed laser deposition, J. Vac. Sci. Technol. A Vac. Surf. Films 17 (6)
[14] H. Miyanaji, N. Momenzadeh, L. Yang, Effect of printing speed on quality of printed (1999) 3393–3396.
[44] S.J. Khambátá, Raman Spectrum of magnesium oxide, Proc. Phys. Soc. Sect. A 69

12
M. Salehi, et al. Additive Manufacturing 29 (2019) 100790

(5) (1956) 426. sintering of MIM/CIM parts, Ceramic Transactions (2010) 259–270.
[45] T.S. Duffy, C. Meade, F. Yingwei, M. Ho-Kwang, R.J. Hemley, High-pressure phase [67] M. Wolff, J.G. Schaper, M. Dahms, T. Ebel, K.U. Kainer, T. Klassen, Magnesium
transition in brucite, Mg(OH)2, Am. Mineral. 80 (3-4) (1995) 222–230. powder injection moulding for biomedical application, Powder Metall. 57 (5)
[46] P. Dawson, C.D. Hadfield, G.R. Wilkinson, The polarized infra-red and Raman (2014) 331–340.
spectra of Mg(OH)2 and Ca(OH)2, J. Phys. Chem. Solids 34 (7) (1973) 1217–1225. [68] M. Wolff, J.G. Schaper, M. Dahms, T. Ebel, R. Willumeit-Römer, T. Klassen, Metal
[47] M. Jönsson, D. Persson, D. Thierry, Corrosion product formation during NaCl in- injection molding (MIM) of mg-alloys, Minerals Metals Materials Series (2018)
duced atmospheric corrosion of magnesium alloy AZ91D, Corros. Sci. 49 (3) (2007) 239–251.
1540–1558. [69] J.G. Schaper, M. Wolff, B. Wiese, T. Ebel, R. Willumeit-Römer, Powder metal in-
[48] X. Wu, H. Cao, G. Yin, J. Yin, Y. Lu, B. Li, MgCO3·3H2O and MgO complex na- jection moulding and heat treatment of AZ81 Mg alloy, J. Mater. Process. Technol.
nostructures: controllable biomimetic fabrication and physical chemical properties, 267 (2019) 241–246.
J. Chem. Soc. Faraday Trans. 13 (11) (2011) 5047–5052. [70] P. Nandwana, A.M. Elliott, D. Siddel, A. Merriman, W.H. Peter, S.S. Babu, Powder
[49] J.S. Loring, C.J. Thompson, C. Zhang, Z. Wang, H.T. Schaef, K.M. Rosso, In situ bed binder jet 3D printing of Inconel 718: densification, microstructural evolution
infrared spectroscopic study of brucite carbonation in dry to water-saturated su- and challenges⋆, Curr. Opin. Solid State Mater. Sci. 21 (4) (2017) 207–218.
percritical carbon dioxide, J. Phys. Chem. A 116 (19) (2012) 4768–4777. [71] A. Mostafaei, E.L. Stevens, E.T. Hughes, S.D. Biery, C. Hilla, M. Chmielus, Powder
[50] R.L. Frost, S.J. Palmer, Infrared and infrared emission spectroscopy of nesquehonite bed binder jet printed alloy 625: densification, microstructure and mechanical
Mg(OH)(HCO3)·2H2O–implications for the formula of nesquehonite, Spectrochim. properties, Mater. Des. 108 (2016) 126–135.
Acta A. Mol. Biomol. Spectrosc. 78 (4) (2011) 1255–1260. [72] T. Do, P. Kwon, C.S. Shin, Process development toward full-density stainless steel
[51] P.K. Soni, A. Bhatnagar, M.A. Shaz, O.N. Srivastava, Effect of graphene templated parts with binder jetting printing, Int. J. Mach. Tools Manuf. 121 (2017) 50–60.
fluorides of Ce and La on the de/rehydrogenation behavior of MgH2, Int. J. [73] E. Crossin, J.Y. Yao, G.B. Schaffer, Swelling during liquid phase sintering of
Hydrogen Energy 42 (31) (2017) 20026–20035. Al–Mg–Si–Cu alloys, Powder Metall. 50 (4) (2007) 354–358.
[52] X. Wang, L. Andrews, Infrared spectra of magnesium hydride molecules, complexes, [74] P. Burke, C. Petit, V. Vuaroqueaux, A. Doyle, G.J. Kipouros, Processing parameters
and solid magnesium dihydride, J. Phys. Chem. A 108 (52) (2004) 11511–11520. and post-sintering operations effects in magnesium powder metallurgy, Can. Metall.
[53] R.L. Frost, J.T. Kloprogge, Infrared emission spectroscopic study of brucite, Q. 50 (3) (2011) 240–245.
Spectrochim. Acta A. Mol. Biomol. Spectrosc. 55 (11) (1999) 2195–2205. [75] E. Biguereau, D. Bouvard, J.M. Chaix, S. Roure, On the swelling of silver powder
[54] N.C.S. Selvam, R.T. Kumar, L.J. Kennedy, J.J. Vijaya, Comparative study of mi- during sintering, Powder Metall. 59 (5) (2016) 394–400.
crowave and conventional methods for the preparation and optical properties of [76] P. Burke, J. Li, G.J. Kipouros, DSC and FIB/TEM investigation of calcium and yt-
novel MgO-micro and nano-structures, J. Alloys. Compd. 509 (41) (2011) trium additions in the sintering of magnesium powder, Can. Metall. Q. 55 (1)
9809–9815. (2016) 45–52.
[55] P.B. Devaraja, D.N. Avadhani, S.C. Prashantha, H. Nagabhushana, S.C. Sharma, [77] M. Wolff, T. Ebel, M. Dahms, Sintering of magnesium, Adv. Eng. Mater. 12 (9)
B.M. Nagabhushana, H.P. Nagaswarupa, Synthesis, structural and luminescence (2010) 829–836.
studies of magnesium oxide nanopowder, Spectrochim. Acta A. Mol. Biomol. [78] H. Fujii, S. Izutani, T. Matsumoto, S. Kiguchi, K. Nogi, Evaluation of unusual change
Spectrosc. 118 (2014) 847–851. in contact angle between MgO and molten magnesium, Mater. Sci. Eng. A 417 (1)
[56] K. Krishnamoorthy, J.Y. Moon, H.B. Hyun, S.K. Cho, S.-J. Kim, Mechanistic in- (2006) 99–103.
vestigation on the toxicity of MgO nanoparticles toward cancer cells, J. Mater. [79] H. Miyanaji, N. Momenzadeh, L. Yang, Effect of powder characteristics on parts
Chem. 22 (47) (2012) 24610–24617. fabricated via binder jetting process, Rapid Prototyp. J. 25 (2) (2018) 332–342.
[57] N. Lepot, M.K. Van Bael, H. Van den Rul, J. D’Haen, R. Peeters, D. Franco, [80] D. Demirskyi, D. Agrawal, A. Ragulya, Neck growth kinetics during microwave
J. Mullens, Synthesis of ZnO nanorods from aqueous solution, Mater. Lett. 61 (13) sintering of copper, Scr. Mater. 62 (8) (2010) 552–555.
(2007) 2624–2627. [81] D. Demirskyi, D. Agrawal, A. Ragulya, Neck growth kinetics during microwave
[58] K. Sowri Babu, A. Ramachandra Reddy, C. Sujatha, K. Venugopal Reddy, sintering of nickel powder, J. Alloys. Compd. 509 (5) (2011) 1790–1795.
A.N. Mallika, Synthesis and optical characterization of porous ZnO, J. Adv. Ceram. [82] A. de la Hoz, Á. Díaz-Ortiz, A. Moreno, Microwaves in organic synthesis. Thermal
2 (3) (2013) 260–265. and non-thermal microwave effects, Chem. Soc. Rev. 34 (2) (2005) 164–178.
[59] A. Umar, C. Ribeiro, A. Al-Hajry, Y. Masuda, Y.B. Hahn, Growth of highly c-Axis- [83] S. Horikoshi, R.F. Schiffmann, J. Fukushima, N. Serpone, Microwave Chemical and
Oriented ZnO nanorods on ZnO/Glass substrate: growth mechanism, structural, and Materials Processing: a Tutorial, Springer, Singapore, 2017.
optical properties, J. Phys. Chem. C 113 (33) (2009) 14715–14720. [84] F. Zuo, A. Badev, S. Saunier, D. Goeuriot, R. Heuguet, S. Marinel, Microwave versus
[60] A. Umar, B. Karunagaran, S.H. Kim, E.K. Suh, Y.B. Hahn, Growth mechanism and conventional sintering: estimate of the apparent activation energy for densification
optical properties of aligned hexagonal ZnO nanoprisms synthesized by non- of α-alumina and zinc oxide, J. Eur. Ceram. Soc. 34 (12) (2014) 3103–3110.
catalytic thermal evaporation, Inorg. Chem. 47 (10) (2008) 4088–4094. [85] K.I. Rybakov, I.I. Volkovskaya, Electromagnetic Field Effects in the Microwave
[61] X. Cai, Y. Cai, Y. Liu, H. Li, F. Zhang, Y. Wang, Structural and photocatalytic Sintering of Electrically Conductive Powders, Ceramics International, 2018.
properties of nickel-doped zinc oxide powders with variable dopant contents, J. [86] V.D. Buchelnikov, D.V. Louzguine-Luzgin, G. Xie, S. Li, N. Yoshikawa, M. Sato,
Phys. Chem. Solids 74 (9) (2013) 1196–1203. A.P. Anzulevich, I.V. Bychkov, A. Inoue, Heating of metallic powders by micro-
[62] R.M. German, Sintering: From Empirical Observations to Scientific Principles, waves: experiment and theory, J. Appl. Phys. 104 (11) (2008) 113505.
Elsevier Science & Technology Books, 2014. [87] M.M. Mahmoud, G. Link, M. Thumm, The role of the native oxide shell on the
[63] X. Xiao, Z. Liu, S. Saremi-Yarahmadi, D.H. Gregory, Facile preparation of [small microwave sintering of copper metal powder compacts, J. Alloys. Compd. 627
beta]-/[gamma]-MgH2 nanocomposites under mild conditions and pathways to (2015) 231–237.
rapid dehydrogenation, J. Chem. Soc. Faraday Trans. 18 (15) (2016) 10492–10498. [88] K.I. Rybakov, V.E. Semenov, S.V. Egorov, A.G. Eremeev, I.V. Plotnikov, Y.V. Bykov,
[64] R. Trittschack, B. Grobéty, P. Brodard, Kinetics of the chrysotile and brucite de- Microwave heating of conductive powder materials, J. Appl. Phys. 99 (2) (2006)
hydroxylation reaction: a combined non-isothermal/isothermal thermogravimetric 023506.
analysis and high-temperature X-ray powder diffraction study, Phys. Chem. Miner. [89] W. Liu, F. Xu, Y. Li, X. Hu, B. Dong, Y. Xiao, Discussion on microwave-matter in-
41 (3) (2014) 197–214. teraction mechanisms by in situ observation of "Core-Shell" microstructure during
[65] S. Tarafder, V.K. Balla, N.M. Davies, A. Bandyopadhyay, S. Bose, Microwave-sin- microwave sintering, Materials Basel (Basel) 9 (3) (2016).
tered 3D printed tricalcium phosphate scaffolds for bone tissue engineering, J. [90] C. Manière, G. Lee, T. Zahrah, E.A. Olevsky, Microwave flash sintering of metal
Tissue Eng. Regen. Med. 7 (8) (2013) 631–641. powders: from experimental evidence to multiphysics simulation, Acta Mater. 147
[66] P. Veronesi, C. Leonelli, G. Poli, L. Denti, A. Gatto, Microwave rapid debinding and (2018) 24–34.

13