Journal of

Materials Chemistry A
View Article Online
PAPER View Journal | View Issue

Flame made nanoparticles permit processing of

Published on 01 August 2016. Downloaded by University of Michigan Library on 08/01/2018 [Link].

dense, flexible, Li+ conducting ceramic electrolyte

thin films of cubic-Li7La3Zr2O12 (c-LLZO)†
Cite this: J. Mater. Chem. A, 2016, 4,
12947

Eongyu Yi, Weimin Wang, John Kieffer and Richard M. Laine*

Ceramic electrolytes are proposed as key components in resolving challenges extant in developing next
generation, high energy density Li batteries by replacing liquid electrolytes to improve safety and
performance. Among numerous candidates, c-LLZO offers multiple desirable properties: high ionic
conductivities (0.1–1 mS cm
1), Li stability, a wide electrochemical operating window (6 V) and pH
stability (7–11.5). However, incorporation into prototype cells has yet to be demonstrated as c-LLZO
membranes at thicknesses <50 mm have not been achieved. Processing dense, thin films matching bulk
counterpart properties remains a very difficult target arising from energy and/or equipment intensive
sintering, Li volatilization, and contamination from substrates. We show that using metalloorganic derived
flame made nanoparticles can overcome these processing challenges resulting in a significantly reduced
energy input required for densification, 10–40 fold shorter dwell times at sintering temperatures,
compared to common solid state reaction derived c-LLZO. Furthermore, surface/volume ratios of the
films are determined to be a critical factor affecting final microstructures and phase compositions of the
sintered films. Through careful control of the processing variables, 10–15 grains thick, dense (94  1%)
Received 30th May 2016
Accepted 31st July 2016
c-LLZO thin (<30 mm), flexible films with high ambient ionic conductivities (0.2  0.03 mS cm
1) are
achieved using conventional casting–sintering of flame made nanoparticles. These c-LLZO membranes
DOI: 10.1039/c6ta04492a
greatly increase the selection of complementary cell components and simplify battery configurations
[Link]/MaterialsA broadening opportunities for cell designs.

batteries is complicated by compatibility issues with selected

Introduction cell chemistries resulting in decomposition or reaction.4,9
Lithium ion batteries (LIBs) continue to power modern To this end, ceramic electrolytes10 with ionic conductivities
society.1,2 However, newly emerging applications that require similar to liquid/separator couples6 (0.1–1 mS cm
1) are
bulk battery systems or low temperature operation mandate proposed to replace liquid electrolytes to further improve and/
new battery designs, electrochemistries, and materials to meet or stabilize cell chemistries while also reducing the cell foot-
target performance specications.3,4 Multiple next generation Li print by eliminating the peripherals.9 Ceramic Li+ superionic
battery designs have been proposed that offer 2–10 higher conductors are a class of materials that exhibit high Li+ ionic
energy storage per mass or volume compared to current LIBs, all conductivities but low electronic conductivities. Surprisingly,
using Li anodes due to its high theoretical capacity (3860 mA h ceramics with selected crystal structures show very high ionic
g
1).3,5 In terms of electrolytes, common commercial LIBs use conductivities, similar to those of liquid counterparts despite
lithium salts dissolved in organic carbonate liquids.6 However, offering a rigid skeletal structure, as they possess low activation
such electrolytes suffer from multiple drawbacks including energy conduction paths, i.e. an ion conduction highway.11,12
narrow operating windows based on thermal and electro- Oxides with garnet, perovskite, and NASICON structures in
chemical stability.7 They also suffer from poor performance particular have been engineered over the past several decades to
below ambient, and typically degrade at $60  C requiring the alter ion conduction properties by partially or fully substituting
use of thermal management peripherals to avoid catastrophic parent elements with aliovalent ions to modify Li+ site vacancy
failure.8 Furthermore, the potential utility of different anode concentration, conduction channel width, and crystallographic
and cathode materials in prototypical next generation Li site occupancy.13,14 In a broader perspective, grain boundary
properties have also been engineered by introducing secondary
functional phases or by optimizing processing and sintering
Materials Science and Engineering, University of Michigan, Ann Arbor, MI, 48109,
conditions to minimize grain boundary resistance arising
USA. E-mail: talsdad@[Link]
† Electronic supplementary information (ESI) available. See DOI:
from polycrystallinity.13,14 Efforts to optimize grain and grain
10.1039/c6ta04492a boundary properties have led to a greater understanding of

This journal is © The Royal Society of Chemistry 2016 J. Mater. Chem. A, 2016, 4, 12947–12954 | 12947
 View Article Online

Journal of Materials Chemistry A Paper

these materials resulting in high ionic conductivities of lms or plates such as Al2O3, ZrO2, Y:ZrO2 and even Li1.3Al0.3-
0.1–1 mS cm
1.15,16 However, additional qualications surface Ti1.7(PO4)3 Li+ conductors are commercially produced. In
when these have to be incorporated into a cell, such as Li metal particular, multi-layered ceramic capacitors produced
stability, thin lm form-ability, and ease of processing. commercially using tape casting are mass-produced by similar
Immense attention has been given to c-LLZO as it exhibits methods. Note that deposition28 or sputter29 derived c-LLZO
a combination of desirable characteristics as noted above.13,17 In lms are not considered here as they may nd use in 2-D thin
particular, due to higher safety standards required for bulk lm Li batteries but are unlikely to be useful for bulk battery
battery systems, recent interest has grown in incorporating systems.
Published on 01 August 2016. Downloaded by University of Michigan Library on 08/01/2018 [Link].

c-LLZO in all-solid-state lithium batteries (ASLBs) to construct

inherently safe cells, concomitantly obviating safety mecha-
Experimental
nisms related to LIBs.8 It may also be useful in assembling
aqueous Li cells as it shows structural stability over a wide pH Precursor synthesis
range.17 Indeed, a recent ARPA-E FOA indicates considerable Four types of precursors were synthesized in this study to serve
need for c-LLZO electrolyte lms <20 mm thick.18 Strangely, as a source of Li, Al, La, and, Zr, respectively. Lithium propio-
despite the interest, most prototype cells use relatively thick Li nate [LiO2CCH2CH3] and alumatrane [Al(OCH2CH2)3N] were
phosphate membranes (50–200 mm)9 which require Li stable synthesized as described elsewhere.27 Lanthanum isobutyrate
interfacial buffer layers to prevent reduction by Li metal.5,9 Net {La[O2CCH(CH3)2]3} was synthesized by reacting lanthanum
ionic conductivities at these thicknesses are far from optimal oxide (130 g, 0.4 mole) with isobutyric acid (530 g, 6 mole) in
and reduce their potential utility in Li batteries for bulk storage a 1 L round bottom ask equipped with a still head at 140  C in
applications. N2 atmosphere. Once transparent liquid was obtained, heat was
In order for c-LLZO to be used in actual cells, it must be removed and lanthanum isobutyrate crystallized on cooling
incorporated in thin lm forms preferably <50 mm.19 However, which was ltered off. Zirconium isobutyrate {Zr[O2-
no dense, thin c-LLZO lms with ionic conductivities equivalent CCH(CH3)2]2(OH)2} was synthesized by reacting zirconium
to those found in high density, bulk counterparts (>0.1 mS basic carbonate (86 g, 0.28 mole) with isobutyric acid (390 g, 4.4
cm
1) have been reported to date likely due to the energy mole) and isobutyric anhydride (350 g, 2.2 mole) in the same
intensive and rather problematic sintering processes involved manner as with lanthanum isobutyrate synthesis.
(Table S1, Fig. S1†).18 Normal sintering conditions are
1100–1250  C for 10–40 h.13,20 However, Li (as Li2O) volatilizes Powder synthesis
rapidly at these temperatures presenting exceptional challenges
Nanopowders of Li6.25Al0.25La3Zr2O12 composition with 50 wt%
in producing thin lms given their much higher surface/volume
excess lithium were synthesized by liquid-feed ame spray
ratios leading to faster Li loss.
pyrolysis (LF-FSP). Extra lithium is intended to compensate for
Multiple reports have assessed the materials' challenges in
its loss during sintering. Lithium propionate, alumatrane,
developing next generation Li batteries in terms of stability and
lanthanum isobutyrate, and zirconium isobutyrate were dis-
compatibility.4,21 However, one very important aspect is over-
solved in ethanol at selected molar ratios to give a 3 wt%
looked, i.e. material processing challenges. Unlike liquids or
ceramic yield solution. The precursor solution was aerosolized
polymers, oxides require high temperature sintering of powders
with oxygen into a quartz chamber where it was combusted with
to reach the high densities needed for optimal performance.
methane/oxygen pilot torches in an oxygen rich environment.
Thus, the selection of starting powder(s), processing chemis-
Resulting nanopowders were collected down-stream in rod-in-
tries, and sintering conditions are all crucial to obtaining the
tube electrostatic precipitators (ESP) operated at 10 kV. The
highest possible densities at the lowest possible energy input
LF-FSP process is detailed in our previous work.27,30,31
(cost). Furthermore, processing becomes even more difficult
when producing thin lms.
Numerous attempts to minimize processing conditions of c- Powder and lm processing
LLZO by introducing sintering aids22 or using micron-23 to nano- As-produced powders were rst dispersed in EtOH (200 proof,
particles24 have met with little success. In contrast, hot-pressing Decon Labs) with 2 wt% polyacrylic acid (Mn ¼ 2000, Sigma-
provides access to zfully dense pellets with superior bulk ionic Aldrich) dispersant, using an ultrasonic horn (Vibra cell VC-505,
conductivities.16,25,26 However, such an approach may be prob- Sonics and Materials, Inc.) at 100 W for 15 min. The suspension
lematic from a commercialization standpoint. Hence, there was let settle for 4 h to allow larger particles to settle. Super-
remains considerable need to rst, develop routes to c-LLZO natant was decanted and the recovered powder dried. Collected
thin lms, and second, in a practical and economical way. Here powder, polyvinyl butyral, benzyl butyl phthalate, acetone, and
we show a potential solution to both challenges by adopting two ethanol at selected wt ratios (Table S2†) were added to a 20 mL
proven methods of mass production; ame spray pyrolysis and vial and ball-milled with 3.0 mm diameter spherical ZrO2 beads
conventional casting–sintering.27 Fumed silica, nano-titania, for 12–24 h to homogenize the suspension. Suspensions were
and carbon black are produced by combustion synthesis at cast using a wire wound rod coater (Automatic Film Applicator-
kilotons/year quantities. Casting–sintering is one of the lowest- 1137, Sheen Instrument, Ltd). Film thicknesses were controlled
cost processing routes that converts ceramic powders to free by adjusting the gap between the rod and the substrate. Dried
standing, dense monoliths. For example, a myriad of ceramic green lms were manually peeled off the Mylar substrate, and

12948 | J. Mater. Chem. A, 2016, 4, 12947–12954 This journal is © The Royal Society of Chemistry 2016
 View Article Online

Paper Journal of Materials Chemistry A

cut to selected sizes. Green lms were uniaxially pressed in (International Crystal Laboratories) were ground, with 1 wt% of
between stainless steel dies at 80–100  C with a pressure of 50– the sample to be analyzed. The sample chamber was purged
70 MPa for 5–10 minutes using a heated bench top press with N2 prior to data acquisition in the range 4000–400 cm
1
(Carver, Inc) to improve packing density. with a precision of 4 cm
1.
Electrochemical impedance spectroscopy. AC impedance
Film sintering data were collected with a broadband dielectric spectrometer
Green lms were placed in between graphite foils and heated to (Novocontrol technologies) in a frequency range of 10 MHz to
selected temperatures and dwell times in N2 (100 mL min
1). 0.1 Hz at
35 to 85  C in increments of 10  C. Concentric gold
Published on 01 August 2016. Downloaded by University of Michigan Library on 08/01/2018 [Link].

Graphite foil was used to avoid possible contamination or electrodes, 3 mm in diameter, were deposited using a SPI
reaction with common ceramic crucibles. The resulting lms sputter coater on both surfaces of the lms using a deposition
had carbon deposits that were removed by reheating to 700–800 mask. “EIS spectrum analyser” soware was used for extracting

C for 1–4 h in O2. Films were placed on MgO plates during total resistance. Equivalent circuit consisting of (RtotalQtotal)-
carbon removal. (Qelectrode) was used. R and Q denote resistance and constant
phase element, respectively. SEM fracture surface images were
Thermal etching taken to measure sample thicknesses.
Density measurements. Final sintered lm densities were
Sintered lms were manually fractured and heated to 910  C for determined by Archimedes method using ethanol.
0.2 h in N2, placed in between graphite foils. Since fracture
provides a uniform two-dimensional surface, no grinding/pol-
ishing was necessary. Average grain sizes were calculated by the Results and discussion
lineal intercept method. Below, we rst discuss the characterization of liquid-feed ame
spray pyrolysis (LF-FSP) produced LLZO powders and then
Characterization efforts to transform the powders into dense LLZO lms.
X-ray diffraction. Measurements were carried out using Nanopowders of Li6.25Al0.25La3Zr2O12 composition with 50
a Rigaku Rotating Anode Goniometer (Rigaku Denki., LTD., Tokyo, wt% excess Li were produced using LF-FSP (Fig. 1a). Al3+ doping
Japan). Scans were made from 10 to 70 2q, using a scan rate of 5 stabilizes the high ionic conductivity cubic phase,26 and excess
min
1 in 0.01 increments and Cu Ka radiation (1.541 Å) operating Li is used to balance losses incurred during sintering. Literature
at 40 kV and 100 mA. Scan rate of 1 min
1 in 0.01 increments indicates that sintering 1–2 mm thick pellets requires 10–15
was used for as-produced powder mixed with a-Al2O3 internal wt% excess Li,24,32 50 wt% was chosen for our studies.
standard. The Jade program 2010 (Version 1.1.5 from Materials The resulting spherical nanoparticles offer a narrow size
Data, Inc.) was used to determine the presence of crystallographic distribution (Fig. 1b and S2†). The BET N2 adsorption derived
phases, wt fraction, and to rene lattice constants. Following specic surface area was 16 m2 g
1 with corresponding average
reference les were used; c-LLZO (PDF# 04-018-3158), t-LLZO particle sizes of z90 nm.
(PDF# 01-080-6140), La2Zr2O7 (PDF# 01-070-5602), La2O3 (PDF# 04- While LF-FSP normally generates pure oxides, a mixture of
008-8233), Li2CO3 (PDF# 98-000-0473), a-Al2O3 (PDF# 98-000-0174). Li2CO3 and off-stoichiometric La2Zr2O7 was detected by XRD
N2 adsorption. Specic surface areas (SSA) were obtained (Fig. 1c). The combustion byproducts H2O and CO2 and high
using a Micromeritics ASAP 2020 sorption analyzer. Samples ame temperatures (>1500  C), either accelerate decomposition
(400 mg) were degassed at 300  C/5 h. Each analysis was run at or prevent LLZO from forming.33 For off-stoichiometric

196  C (77 K) with N2. The SSAs were determined by the BET La2Zr2O7, LF-FSP provides access to phases or compositions
multipoint method using ten data points at relative pressures of stable at high temperatures due to rapid quenching.30 XRD peak
0.05–0.30. SSA was converted to average particle sizes (APS) shis were conrmed and new lattice constants were rened
using the equation APS ¼ 6/(SSA  r). The net density (r) of the using a-Al2O3 as an internal standard corroborating off-stoi-
as-produced powder was approximated by rule of mixtures. The chiometry as detailed in Fig. S3.† TGA and FTIR further conrm
density of stoichiometric La2Zr2O7 was used as an estimate. the product is fully decomposed LLZO. TGA shows an endo-
Scanning electron microscopy. Micrographs were taken thermic mass loss starting at 700  C, ascribed to Li2CO3
using a FEI NOVA Nanolab system (FEI company). Powder melting with a loss of CO2 ending at 950  C (Fig. 1d). Mass loss
samples were used as is, sintered lms were fractured for matches theory for heating fully decomposed LLZO incorpo-
imaging. All samples were sputter coated with gold using a SPI rating excess lithium. FTIR shows nC]O for carbonates (865,
sputter coater (SPI Supplies, Inc.). 1400–1600 cm
1) as well as nM–O (<600 cm
1) (Fig. 1e).
Thermogravimetric analysis. Q600 simultaneous TGA/DSC Previous attempts to produce nano-LLZO also resulted in
(TA Instruments, Inc.) was used to observe thermal decompo- decomposed products likely due to its instability under the
sition of as-produced powders and green lms. Samples synthesis condition.24,34 Powders showed non-uniform sizes
(15–25 mg) were loaded in alumina pans and ramped to 1000  C and shapes which were aggregated, or lost nano-scale on
at 10  C min
1 under constant air ow at 60 mL min
1. calcination. These morphological features are all detrimental
Fourier transform infrared spectroscopy. FTIR spectra were to sintering, hindering uniform packing of the green body,
taken using a Nicolet 6700 Series FTIR spectrometer (Thermo likely limiting access to target ceramic shapes with optimal
Fisher Scientic, Inc.). Optical grade, random cuttings of KBr properties.

This journal is © The Royal Society of Chemistry 2016 J. Mater. Chem. A, 2016, 4, 12947–12954 | 12949
 View Article Online

Journal of Materials Chemistry A Paper

Published on 01 August 2016. Downloaded by University of Michigan Library on 08/01/2018 [Link].

Fig. 1 (a) Schematic of LF-FSP apparatus. LF-FSP is an iteration of aerosol combustion synthesis. Metalloorganic precursors dissolved in ethanol
are aerosolized and combusted in a controlled fashion generating nanoparticles recovered in electrostatic precipitators. (b) SEM (c) XRD (d) TGA
(e) FTIR of the as-produced nanopowders.

In contrast, in this study spherical nanoparticles with narrow Single-phase c-LLZO was observed at 1080  C/1 h (Fig. 2g)
size distributions are processed directly to green lms using with some closed porosity per the SEM fracture surface image
chemical reaction and uniform packing to drive densication at (Fig. 2a and b). On heating to 1090  C/1 h, traces of La2Zr2O7
lower temperatures. For convenience, the initial nanopowder is form (Fig. 2g). Trans-granular fracture surfaces reveal very high
labeled LLZO although it is a decomposed mixture. relative densities (Fig. 2c and d) and suggest good-to-excellent
Green lms were produced by conventional ball-milling and grain boundary contact. Trans-granular fracture dominates
casting. LLZO nanopowders were ball-milled with dispersant, mechanical behavior with increasing relative densities.25
binder, plasticizer, and solvents per Table S2.† Aer 12–24 h of 1100  C/1 h sintered samples exhibit noticeable secondary
ball-milling, the homogenized suspension was bar cast on phase peaks corresponding to La2Zr2O7 and La2O3 (Fig. 2g). The
Mylar sheets. Compared to common pellet compaction studies overall microstructures look similar other than distinct micro-
wherein nanoparticle agglomerates result in pores aer sinter- structural features appearing near the surface, likely initiation
ing, wet processing breaks down agglomerates during ball- of secondary phase formation as Li volatilizes (Fig. 2e and f and
milling and gives uniform particle packing within the polymeric S6†). Single phase c-LLZO thin lms (<30 mm) with high relative
matrix, benecial to sintering.35 It is important to formulate densities (94  1%) form on sintering at 1090  C/1 h. These
a stable system such that particles do not occulate during lms are the very rst examples that meet the challenges cited
drying. TGAs of green lms were run to conrm the solids above.
loading prior to sintering (Fig. S4†). Sintering times are greatly reduced from 10–40 h to 1 h, and
Given there is a limited reservoir of Li in the starting LLZO at lower temperatures compared to common solid-state reac-
nanopowder (50 wt% excess), densication must be complete tion methods.13,20 This result likely arises from combined effects
before it volatilizes completely, preferably resulting in single- of uniform particle packing, nanoparticles' high surface ener-
phase c-LLZO. Hence, a short dwell time of 1 h was selected, gies, and chemical reactions of the constituent materials
targeting optimal densication. Indeed, extended dwell below including liquid phase sintering promoted by Li2CO3. Off-stoi-
the sintering temperature generated c-LLZO but with low chiometric La2Zr2O7 does not seem to be a determining factor
relative densities. Green lm thicknesses were xed at 45  since similar starting powders form in sol–gel syntheses and do
2 mm for this set of experiments since it is also a design not dramatically enhance sintering behavior.34
parameter (see below). Tetragonal-LLZO (t-LLZO) was rst Based on Fig. 2 we conclude that c-LLZO thin lms' micro-
observed at 800  C as expected due to excess Li (Fig. S5†), structures and phase compositions are very sensitive to sinter-
known to drive formation of t-LLZO even with Al3+ doping.26 ing temperature; correlated to Li loss rates. Hence another set of
Further increases in temperature drive Li loss and transform t- experiments was run with the sintering temperature xed at
LLZO to c-LLZO. 1090  C/1 h but with different green lm thicknesses. Given Li is

12950 | J. Mater. Chem. A, 2016, 4, 12947–12954 This journal is © The Royal Society of Chemistry 2016
 View Article Online

Paper Journal of Materials Chemistry A

Published on 01 August 2016. Downloaded by University of Michigan Library on 08/01/2018 [Link].

Fig. 2 Characterization of the LLZO films sintered at selected temperatures. SEM fracture surface images of films sintered at (a) and (b) 1080  C/1
h (c) and (d) 1090  C/1 h and (e) and (f) 1100  C/1 h. (g) XRD of sintered films.

lost at the surface, the surface/volume ratio becomes an densities as secondary phases La2Zr2O7 (6.04 g cm
3) and La2O3
important parameter. Surface/volume ratio dictates lithium loss (5.98 g cm
3) have higher densities compared to c-LLZO
rate and drastically changes at lm thicknesses below 50 mm, (5.10 g cm
3) and La2Zr2O7 has high sintering resistance.36
indicating nal sintered product properties are sensitive to Thicker lms also gave low relative densities suggesting that
green lm thicknesses (Fig. 3d). For simplicity sake, it is t-LLZO does not sinter at this temperature whereas c-LLZO does.
assumed there is no porosity. Note that 40 mm thick lm's The sintered lms' (in optimal range) electrochemical
surface/volume ratio is 20 greater than for 1 mm thick pellet. properties were measured. Nyquist plots show semi-circles at
As such, different sintering conditions are required for samples high frequencies followed by a spike at low frequencies, typical
with different thicknesses. Changes in surface/volume ratio of ion conductors with blocking electrodes (Fig. 4a). Room
exist with varying lateral dimensions for thick samples but temperature conductivities of 0.2  0.03 mS cm
1 and activa-
converge to the same values as thickness is reduced. tion energies of 0.35  0.01 eV (33.8  0.7 kJ mol
1) were ob-
As hypothesized, thicker green lms showed t-LLZO and tained (Fig. 4a and b), in good agreement with high density
thinner green lms La2Zr2O7 and La2O3 secondary phases when pellet counterparts.20,22,26 No noticeable grain boundary resis-
sintered (Fig. 3e) using the same schedule. Li poor, optimal, and tance was observed due to high relative density and superior
Li rich ranges in terms of green lm thicknesses were deter- grain boundary contact. In comparison, c-LLZO lms with low
mined based on the phase composition and microstructure relative densities showed grain boundary resistance and lower
(Fig. 3a, b, c and e and S7†). Li poor lms show low relative ionic conductivities (Fig. S8†).

This journal is © The Royal Society of Chemistry 2016 J. Mater. Chem. A, 2016, 4, 12947–12954 | 12951
 View Article Online

Journal of Materials Chemistry A Paper

Published on 01 August 2016. Downloaded by University of Michigan Library on 08/01/2018 [Link].

Fig. 3 Effect of green film thicknesses on the final microstructure and phase composition at fixed sintering condition (1090  C/1 h). SEM fracture
surface images of sintered (a) 22 mm (b) 45 mm and (c) 73 mm thick green film. (d) Surface/volume ratio plot. (e) Phase composition changes with
film thicknesses although all of them were heated to the same condition of 1090  C/1 h.

Ionic conductivities of 0.008  0.001 to 1.5  0.1 mS cm
1 inaccessible using common liquid electrolytes due to decom-
were measured at
35 to 85  C (Fig. 4a and b) demonstrating position,8 ammability,8 or freezing.37 We anticipate that still
a wide operating temperature window, including temperatures higher operating temperatures leading to faster charge/

Fig. 4 Electrochemical properties of sintered LLZO films. (a) Nyquist plots of sintered films. Thickness and electrode area are taken into account.
Note the units are kU cm. (b) Ionic conductivities of c-LLZO rises with temperature showing Arrhenius trend.

12952 | J. Mater. Chem. A, 2016, 4, 12947–12954 This journal is © The Royal Society of Chemistry 2016
 View Article Online

Paper Journal of Materials Chemistry A

Published on 01 August 2016. Downloaded by University of Michigan Library on 08/01/2018 [Link].

Fig. 5 Translucent, flexible sintered c-LLZO films. (a) Photograph of c-LLZO film sintered within the optimal range. “c-LLZO” printed on the
background is visible due to low thicknesses (<30 mm). Films show decent flexibility, property inherent to thin ceramics. The sintered film is
roughly 2  2 cm2. (b) and (c) Thermally etched fracture surfaces of sintered films.

discharge rates will be accessible in Li batteries using ceramic assembled cells. Subsequent goals will be to reduce nal
electrolytes. thicknesses, sintering temperatures, and to reduce excess Li
Sintered <30 mm thick lms with 2  2 cm2 lateral dimen- contents. Reduced sintering temperatures will retard Li2O loss
sions were produced that are translucent at high relative rates, widening the optimal processing window.
densities, and exible (Fig. 5a). With decreasing thicknesses, Films processed here show good exibility, suggesting
the strain accompanying bending reduces such that even casting–sintering of nanoparticles may be an alternate route to
ceramics become exible below certain thicknesses.38 Average process batteries requiring mild exibility. Reduction in lm
grain sizes (AGSs) are 2.4  0.4 mm, which translates to lms thicknesses would be necessary to further improve exibility, to
10–15 grains thick (Fig. 5b and c). AGSs are 40–80 times smaller incorporate active material layers, as well as to improve attain-
compared to solid state reaction23 derived c-LLZO but similar to able energy densities. On the other hand, it is easy to envision
hot pressed.25,26 the use of the lms produced here in aqueous Li cells or solid
A further comment is necessary. Additional benets from state Li–air cells to replace the commonly used thick,
these Li+ conducting thin lms is that they offer mechanical commercial Li metal phosphate membranes (50–200 mm) which
properties that allow them to ex such that roll-to-roll pro- has narrower pH stability window (7–10 vs. 7–11.5).9,17
cessing and easier handling are permitted.18 Flexibility in Recent reports on Ga doped LLZO shows ambient ionic
ceramics extends from two attributes: (1) the need to have conductivities as high as 1.3 mS cm
1.40,41 A following paper will
dense, ne grained materials in the nal lm to ensure that any discuss processing of Ga doped LLZO at thicknesses <30 mm.
cracks must propagate via a tortuous path that absorbs the
energy driving propagation and (2) the absence of surface aws
that initiate cracking.39 A more rigorous denition based on Acknowledgements
Griffith micro-crack theory states that insufficient mechanical We are grateful for support of a signicant portion of this work
force can be imparted to the lm or ber to initiate crack by NSF through DMR 1105361 and DMR 116058. We also thank
formation. Such constraints mandate the use of nanopowders the University of Michigan Energy Institute (UMEI) and Michi-
to access ne grained, exible, thin and yet dense <30 mm lms gan Translational Research and Commercialization (MTRAC)
as demonstrated in this work. program for support to purchase chemicals, solvents, furnace
and LF-FSP reactor components. We would especially like to
thank Professor J. Sakamoto and Dr T. Thompson for their
Conclusions encouragement and useful discussions.
In conclusion, we have overcome processing difficulties
involved in c-LLZO by reactive sintering carbonate and oxide Notes and references
nanoparticle mixtures and wet processing to minimize
agglomerate formation. The rst examples of c-LLZO thin lms 1 M. Armand and J.-M. Tarascon, Nature, 2008, 451, 652–657.
(20–30 mm thick, similar to commercial polymer separators6) 2 B. Dunn, H. Kamath and J.-M. Tarascon, Science, 2011, 334,
with properties similar to bulk materials are demonstrated. 928–935.
Surface/volume ratios were determined to be a critical factor in 3 P. G. Bruce, S. A. Freunberger, L. J. Hardwick and
sintering due to Li2O volatility. Cell assembly and testing can J.-M. Tarascon, Nat. Mater., 2012, 11, 19–29.
now be done at prototype levels, and the lm thicknesses ach- 4 Y. Shao, F. Ding, J. Xiao, J. Zhang, W. Xu, S. Park, J.-G. Zhang,
ieved here potentially enable higher energy storage densities in Y. Wang and J. Liu, Adv. Funct. Mater., 2013, 23, 987–1004.

This journal is © The Royal Society of Chemistry 2016 J. Mater. Chem. A, 2016, 4, 12947–12954 | 12953
 View Article Online

Journal of Materials Chemistry A Paper

5 W. Xu, J. Wang, F. Ding, X. Chen, E. Nasybulin, Y. Zhang and 23 L. Cheng, J. S. Park, H. Hou, V. Zorba, G. Chen,
J.-G. Zhang, Energy Environ. Sci., 2014, 7, 513–537. T. Richardson, J. Cabana, R. Russo and M. Doeff, J. Mater.
6 H. Lee, M. Yanilmaz, O. Toprakci, K. Fu and X. Zhang, Energy Chem. A, 2014, 2, 172–181.
Environ. Sci., 2014, 7, 3857–3886. 24 R. Djenadic, M. Botros, C. Benel, O. Clemens, S. Indris,
7 B. Scrosati and J. Garche, J. Power Sources, 2010, 195, 2419– A. Choudhary, T. Bergfeldt and H. Hahn, Solid State Ionics,
2430. 2014, 263, 49–56.
8 P. G. Balakrishnan, R. Ramesh and T. Prem Kumar, J. Power 25 Y. Kim, H. Jo, J. L. Allen, H. Choe, J. Wolfenstine, J. Sakamoto
Sources, 2006, 155, 401–414. and G. Pharr, J. Am. Ceram. Soc., 2016, 99, 1367–1374.
Published on 01 August 2016. Downloaded by University of Michigan Library on 08/01/2018 [Link].

9 F. Li, H. Kitaura and H. Zhou, Energy Environ. Sci., 2013, 6, 26 E. Rangasamy, J. Wolfenstine and J. Sakamoto, Solid State
2302–2311. Ionics, 2012, 206, 28–32.
10 P. Knauth, Solid State Ionics, 2009, 180, 911–916. 27 E. Yi, W. Wang, S. Mohanty, J. Kieffer, R. Tamaki and
11 N. Kamaya, K. Homma, Y. Yamakawa, M. Hirayama, R. M. Laine, J. Power Sources, 2014, 269, 577–588.
R. Kanno, M. Yonemura, T. Kamiyama, Y. Kato, S. Hama, 28 J. Tan and A. Tiwari, ECS Solid State Lett., 2012, 1, Q57–Q60.
K. Kawamoto and A. Mitsui, Nat. Mater., 2011, 10, 682–686. 29 D. J. Kalita, S. H. Lee, K. S. Lee, D. H. Ko and Y. S. Yoon, Solid
12 L. J. Miara, S. P. Ong, Y. Mo, W. D. Richards, Y. Park, State Ionics, 2012, 229, 14–19.
J.-M. Lee, H. S. Lee and G. Ceder, Chem. Mater., 2013, 25, 30 T. R. Hinklin, J. Azurdia, M. Kim, J. C. Marchal, S. Kumar and
3048–3055. R. M. Laine, Adv. Mater., 2008, 20, 1373–1375.
13 V. Thangadurai, S. Narayanan and D. Pinzaru, Chem. Soc. 31 C. R. Bickmore, K. F. Waldner, R. Baranwal, T. Hinklin,
Rev., 2014, 43, 4714–4727. D. R. Treadwell and R. M. Laine, J. Eur. Ceram. Soc., 1998,
14 Y. Ren, K. Chen, R. Chen, T. Liu, Y. Zhang, C.-W. Nan and 18, 287–297.
B. Vyas, J. Am. Ceram. Soc., 2015, 98, 3603–3623. 32 H. Imagawa, S. Ohta, Y. Kihira and T. Asaoka, Solid State
15 J. C. Bachman, S. Muy, A. Grimaud, H. H. Chang, N. Pour, Ionics, 2014, 262, 609–612.
S. F. Lux, O. Paschos, F. Maglia, S. Lupart, P. Lamp, 33 C. Galven, J. Dittmer, E. Suard, F. Le Berre and
L. Giordano and Y. Shao-Horn, Chem. Rev., 2016, 116, 140– M.-P. Crosnier-Lopez, Chem. Mater., 2012, 24, 3335–3345.
162. 34 I. Kokal, M. Somer, P. H. L. Notten and H. T. Hintzen, Solid
16 T. Thompson, A. Shara, M. D. Johannes, A. Huq, J. L. Allen, State Ionics, 2011, 185, 42–46.
J. Wolfenstine and J. Sakamoto, Adv. Energy Mater., 2015, 5, 35 D. Hotza and P. Greil, Mater. Sci. Eng., A, 1995, 202, 206–217.
1500096. 36 R. Vassen, X. Cao, F. Tietz, D. Basu and D. Stover, J. Am.
17 C. Ma, E. Rangasamy, C. Liang, J. Sakamoto, K. L. More and Ceram. Soc., 2000, 83, 2023–2028.
M. Chi, Angew. Chem., Int. Ed., 2015, 54, 129–133. 37 Y. Ji, Y. Zhang and C. Y. Wang, J. Electrochem. Soc., 2013, 160,
18 Advanced Research Projects Agency-Energy (ARPA-E), U.S. A636–A649.
Department of Energy. DE-FOA-0001478. Integration and 38 L. J. Bonderer, P. W. Chen, P. Kocher and L. J. Gauckler,
optimization of novel ion conducting solids (IONICS). J. Am. Ceram. Soc., 2010, 93, 3624–3631.
2016, ([Link] 39 Y. Liu, Z.-F. Zhang, J. Halloran and R. M. Laine, J. Am. Ceram.
19 B. D. McCloskey, J. Phys. Chem. Lett., 2015, 6, 4581–4588. Soc., 1998, 81, 629–645.
20 R. Murugan, V. Thangadurai and W. Weppner, Angew. 40 C. Bernuy-Lopez, W. Manalastas Jr, J. M. Lopez del Amo,
Chem., Int. Ed., 2007, 46, 7778–7781. A. Aguadero, F. Aguesse and J. A. Kilner, Chem. Mater.,
21 J.-M. Tarascon and M. Armand, Nature, 2001, 414, 359–367. 2014, 26, 3610–3617.
22 N. Janani, C. Deviannapoorani, L. Dhivya and R. Murugan, 41 D. Rettenwander, G. Redhammer, F. Preishuber-Pugl,
RSC Adv., 2014, 4, 51228–51238. L. Cheng, L. Miara, R. Wagner, A. Welzl, E. Suard,
M. M. Doeff, M. Wilkening, J. Fleigh and G. Amthauer,
Chem. Mater., 2016, 28, 2384–2392.

12954 | J. Mater. Chem. A, 2016, 4, 12947–12954 This journal is © The Royal Society of Chemistry 2016