Pesquisar no resumo do documento
Fast Li@ Conducting Ceramic Electrolytes By Gin-ya Adachi," Nobuhito Imanaka, and Hiromichi Aono
In general, ions cannot migrate in a solid, since the constituent cations and anions are necessary to maintain the rigid skeletal structure. In some solids, only one ion species is able to migrate with a low energy barrier, and these have high ionicconductivities comparable to those ofmolten and aqueous electrolytes. Such solids are called "solid electrolytes".
In the nineteenth century ion migration such as that of Ag', OZe, and Fe was reported in some solids. Typical ion conducting structures are shown in Table 1. In the 1920s, Tubandt et al. proposed that Ag' ion conductivity in AgI was greatly enhanced by a phase transition from the low temperature p phase to the high temperature a structure.["'] Although the conductivity of the iodide at room tempera- ture is very low, it increases by more than three orders of magnitude at temperatures above the phase transition at 419 K.
Table 1. Typical ion conducting structures for solid electrolytes.
Structure Representative materials
Ion defect structure Two dimensional layered structure
Three dimensional network structure a-AgI type structure
Vitreous structure Li2S-sis2-Li3P04 system [LP]
0.92Zr02-0.08Y203 system [02e] Na2O-I 1A1203(P-alumina) p a m ] , Li3N[Lim] Na3Zr,Si2PO12(NASICON) pa'] Ag1(>419 K) [Agm], RbA&IS [Ag"], Rb4C~1617Cl13 [Cue]
The high conductivity of the a-AgI phase is obtained because only two Ag' ions are statistically distributed over 42 Ag' vacant sites to occupy around I' constituent ion per unit cell.[31 The stabilization of the a-form even at room temperature is one strategy to achieve a high Ag' conducting electrolyte. Since then, many Age ionic con- ducting materials with high conductivities at room tempera- ture have been inve~tigated.[~-~I In particular, RbAg41S shows an excellent Age ionic conductivity of 0.21 Scm-' at 298 K, which is higher than ca. 8 x lO-'S cm- for a 1.0 N NaCl aqueous solution at room temperature.[']
[*] Prof. G.-y. Adachi, Dr. N. Imanaka Department of Applied Chemistry Faculty of Engineering, Osaka University 2-1 Yamadaoka, Suita, Osaka 565 (Japan) Dr. H. Aono Department of Industrial Chemistry Niihama National College of Technology 7-1 Yagumo-cho, Niiharna. Ehime 792 (Japan)
A material with analogous structure, Rb4Cu1617Cl13 also shows an excellent Cu' ionic conductivity (0.34 Scm- ' at 298 K).[91 In the 1940s, Wagner reported the mechanism for 0' ion migration in zirconium dioxide (zirconia)."'] Oxygen defects are formed when Zr4@ ion sites in ZrOz are partially substituted for di- or tri-valent cations such as Ca2@ or Y3'. These defects lead to easier 02' migration in the solid. In addition, pure ZrOz has two phase transitions, from the low temperature monoclinic to the tetragonal structure, and finally to the cubic high temperature phase.
The cubic phase can be obtained by partial substitution of a Zr site by a di- or tri-valent cation having a larger ionic radius [e.g. Y3'(1.019 A), Ca''(1.12 A)] compared to Zr4' (0.84 A)."'] The cubic phase is then stable over a large temperature range, down to room temperature. This material is called "stabilized zirconia". Although stabilized zirconia has already been practically applied as an 0' gas sensor constituent at higher temperatures (> 900 K), it is still insulating at room temperature."']
In the mid-l960s, @-alumina (Na20 1 lA1203) was pre- pared as a fast Na' ion cond~c to r . "~ . '~~ p-alumina has a two-dimensionally layered structure and Na@ ions conduct between the layers. The ionic migration is limited between the layers due to the anisotropic conduction mechanism. In 1976, for the purpose of obtaining an isotropic structure, Goodenough and Hong designed the three-dimensional network structure with a suitable tunnel size for Na' migration, and named the Nal +xZr2SixP3-x012 material a s' 2uper ionic conductor (NASICON).['53'61 These Na' ionic conductors have relatively high conductivity (10- 3- 10- S cm- ') at room temperature, which is comparable to the conductivity of a liquid electrolyte.
Figure 1 shows the temperature dependence of typical ionic conducting solid electrolytes. One of the promising applications of the fast ion conducting solid electrolytes is in all-solid-state rechargeable batteries. Up to now, excellent conductivities at room temperature have been reported for various kinds of ionic conductors. However, the Ag' and Cu' ionic conductors mentioned above are still unsuitable for use for practical battery use, because of a very low decomposition potential (0.5-0.7 V) and deliquescence in humid air. In addition, a high energy density cannot be obtained because both Ag and Cu have a high mass number compared to Na. However, Na metal applied as an anode for a battery is very reactive to oxygen in the atmosphere even at room temperature. Another disadvantage of the all- solid-Na battery is that Na metal is inferior to Li because of a lower energy density.
Adv. Mater. 1996,8, No. 2 0 VCH Verlagsgesellschafl mbH. 0-69469 Weinheim, 1996 093.S-9648/96/0202-0127 $lO.O0+.2S/O 127
G.-y. Adachi, N . Imanaka, H. AonolFast Lie Conducting Ceramic Electrolytes ADVANCED MATE RIALS
T/"C IMH) 500 300 2:
'8 10.2 Y
, \ lo-"o.5 1.0 1.5 2.0 2.5 3.0 3.5
Fig. 1. Temperature dependence of the ionic conductivity for typical solid electrolytes.
Table 2 lists energy densities and electromotive forces for typical batteries. The theoretical energy density of the lithium battery (ca. 2000 W h kg- ') is about 10 times higher than that of the widely commercialized nickel-cadmium or
Table 2. Theoretical energy densities and electromotive forces obtained for rechargeable batteries on the market.
Battery Energy density EMF ( W . h. kg-') (V)
Lead storage battery 180 2.0 Lithium battery 2000 3.0
Nickel-cadmium battery 200 1.3 Nickel-hydrogen battery 400 1.4
a lead-acid storage batteries. Furthermore, the existence of toxic elements, like cadmium and lead, in those batteries has become a serious problem from an environmental point of view, resulting in the nickel-hydrogen battery rapidly entering the market. It offers a relatively high energy density and non-toxicity of the constituent materials. However, its theoretical energy density is still ca. one fifth that of the rechargeable lithium battery.
The lithium battery has the highest electromotive force (ca. 3.0 V) of the batteries described above, and is the most promising due to its high energy density with high cell voltage. Elemental Li has the smallest mass number of the metals and a very negative standard electrode potential. The lithium batteries on the market utilize some organic solvents such as LiC104 dissolved in a propylene carbonate (PC) as electrolyte. However, leakage of the organic electrolyte, its freezing at lower temperatures, and ignition at higher temperatures remain problems. In addition, some side
Gin-ya Adachi was born in Osaka, Japan, in 1938 and earned his undergraduate degree at Kobe University. He then obtained his Ph.D. from Osaka University in 1967. He joined the faculty at Osaka University, where he is now Professor. His main research interests concentrate on inorganic materials containing rare earth elements.
Nobohito Imanaka was born in Kawanishi, Hyogo, Japan, in 1958. He received his B.E. and M.E. in applied chemistry from Osaka University. He then obtained a Ph.D. from Osaka University. He has been on the faculty at Osaka University since 1988. His main researchjelds include rare earths and functional materials such as solid electrolytes and chemical sensors.
Hiromichi Aono was born in Imabari, Ehime, Japan, in 1963, and received his B.E. in industrial chemistry from Ehime University in 1986. He has been on the faculty in the Department of Industrial Chemistry at Niihama National College of Technology since 1986. He obtained a Ph.D. from Osaka University in 1994 for "Studies on Li' ionic conducting solid electrolytes of the NASICON-type structure ". His main interests are the electrical properties of solid electrolytes and their applications.
0 VCH Verlagsgesellschaft mbH. 0-69469 Weinheim, 1996 0935-9648/96/0202-0128 $10.00+.25/0 Adv. Marer. 1#6,8, No. 2
G.-y. Adacki, N . Imanaka, H . AonolFast Li' Conducting Ceramic Electrolytes
reactions might occur because all the ions in a liquid electrolyte easily migrate.
The utilization of an organic polymer solid electrolyte is a way to overcome solvent leakage. The maximum conduc- tivity found for polymeric solid-Lie ion conductors such those based on 2-(2-methoxyethoxy)ethylglycidylether (MEEGE)/ethylene oxide (EO) copolymers containing LiN(S02CF3)2 is around 10-5-10-4Scm-' at room temperature.['q This value is still about two orders of magnitude lower than that of the organic solvent.
Recently, Angel1 et al. reported that rubbery Lie electrolytes, where a small amount of a polymer, e.g. polypropylene oxide (PPO), is mixed with a LiC104- LiC103 based salt, exhibits a conductivity higher than lop4 S cm- ' even at room temperature."*] However, these organic polymers cannot be used at temperatures below their glass transition temperature (Tg) because of the reduction in conductivity.
If the organic solvent in the lithium battery can be replaced by inorganic Lie-conducting solid materials, the following advantages are expected:
0 No leakage of the liquid electrolyte 0 Broad operating temperature range, since inorganic solid
electrolytes do not freeze, boil, or ignite in an applied situation
0 Excellent charge-discharge cyclic properties since no side reactions occur and only one kind of carrier ion migrates
0 Long life-time because of little self discharge
In this article we review fast Lie-conducting ceramic solid electrolytes mainly from the electrical property and crystal structure viewpoints.
2. Li@-Ion Conducting Non-Oxide Based Ceramics
The disposable cell employing a LiI solid electrolyte was originally commercialized as an all dry cell for use in pacemakers for heart disease patients in the early 1970s. In this cell, Li metal and an iodine complex (poly-2- vinylpyridine . n12) electrodes were utilized as anode and cathode, respectively. A thin LiI electrolyte layer is formed between two electrodes from the direct chemical reaction (Li + $I2 + LiI). The discharge current density for the cell is about 10 ,uAcm-2, since the pure LiI shows a consid- erably lower conductivity at room temperature (ca. ~ o - ' s c ~ - ' ) . [ ' ~ ]
With an increase of the LiI electrolyte thickness on discharge, the current density monotonically decreases. Although this cell has long-term stability and reliability,[201 the current density is not high enough for use as the cell for other electrical devices such as pocket calculators and portable phones. Table 3 shows the conductivities for LiI- based electrolytes. The introduction of Ca2% ion to the LiI
Table 3. The conductivities at 298 K for LiI-based electrolytes.
Sample o(S.cm-') Ref.
pure LiI 1 lo-' 1191 LiI-2 mol% CaI, 2 x 10-~ [211 LiI-35 mol%Al,O, 4 x 1 0 - ~ [221 LiI-40 mol%A120, 1 x 1 0 - ~ ~ 9 1
crystal lattice induced cation defects and the Li@ conduc- tivity is greatly enhanced by 2 mol % Ca12 doping.12'] However, the high conductivity could not be sustained, and it decreased with time.
In 1973, Liang reported that the conductivity of LiI is considerably enhanced by the addition of insulating A1203 powder.[201 The maximum conductivity, 4 x 10- S cm- ' at 298 K was obtained for a composition of LiI-35 mol-% A1203.[221 The added A1203 makes a mixture with LiI. The conductivity enhancement was almost proportional to the total surface area of the dispersed A1203 particles (particle size is less than 0.3 pm). This result is mainly ascribed to the increase of the charge carrier concentration in a diffuse space-charge layer formed near the charged surface of the A1203 particles. The conductivity was enhanced by the Lie ion defects and/or the interstitial Lie ions in the space- charge layer. The high conductivity does not change with time in the case for the A1203-added system.
Batteries using a LiI-Al203 composite electrolyte have been p r o p o ~ e d . " ~ , ~ ~ , ~ ~ ~ In these cells, a well-blended mixture of LiI and A1203 was pressed for use as the electrolyte. Although its current density is 1-10 times higher than the 10 pAcmP2 of the cell with the LiI solid electrolyte, it is still too low for applications in some appliances. However, the mixing of insulating powder with the electrolyte to form a composite is one of the key technologies used to enhance the conductivity.
By the 1970s, the highest conductivity in a Lie-ion conductor investigated was in single crystal Li3N,[251 which has a two-dimensional layered structure. Although the conductivity perpendicular to the c axis is as high as 1.2 x 10- S cm- ' at room temperature, that parallel to the c-axis is about two orders of magnitude lower. In this structure, the ionic conduction is limited through the anisotropic two-dimensional planes. In addition, much effort and time is necessary to produce a single-crystal form compared with the polycrystal one. Polycrystalline materials are investigated in order to obtain an isotropic Lie ion-conducting pellet easily.r263271
The maximum conductivity achieved for the Li3N polycrystal is 4 x 10-4Scm-' at 298 K. One of the most serious problems is that Li3N has a very low theoretical decomposition potential of 0.445 V, and that the max- imum EMF output of the battery using the Li3N electrolyte is thus limited to below the decomposition potential. Since then, the new Li3N-LiI systems have been reported. Most of these have higher decomposition voltages and a cubic crystal structure, but the conduc- tivity, ca. 1 x 1 0 - ~ s c r n at room temperature, is more
Adv. Mater. 1996,8, No, 2 0 VCH Verlagsgesellschaft mbH, 0-69469 Weinheim, 1996 093S-9648/96/0202-0129 $ lO.O0+.2S/O 129
G.-y. Adachi, N . Imanaka, H . AonolFast Li' Conducting Ceramic Electrolytes ADVANCED MATE RIALS
than three orders of magnitude lower than that of the Li3N single crystal.[281
In order to enhance the conductivity, LiOH was added to the Li3N-LiI (1:2) mixture. The conductivity for quasi- ternary Li3N-LiI-LiOH (1:2:0.77) is 9.5 x lop4 S cm- ' at 298 K, and its decomposition voltage is 1.6 V at 298 K which is appreciably higher than the 0.445 V of Li3N.[291 This quasi-ternary electrolyte is one of the promising candidates for practical applications. However, this material is not suitable for application as a lithium battery constituent because of its strong hygroscopic properties. Once the stable mono-hydrated LiOH . H 2 0 is formed, it cannot be dehydrated even by heating at the melting point of the mixture.
Lithium sulfide glasses have been investigated as candi- dates for high Lie conducting materials since the 1980s. The Li2S-based glasses, such as the Li2S-P2S5-LiI,[301 Li2S- B2S3-LiI,[311 or Li2S-SiS2-LiI[321 systems, have a high conductivity, ca. 10-3Scm-' at 298 K. However, it has been pointed out that these glasses with LiI are unstable in contact with Li metal used as an electrode.[331
Kondo et al. observed that Li3P04-doped Li2S-SiS2 glass is very stable against electrochemical reduction in contact with Li[341 during cyclic voltammetry. The electrical charge of the reduction reaction during a cycle is about 10-20 % larger than that of the oxidation reaction for a 0.6Li2S- 0.4SiS2 and 30 % LiI added 0.6Li2S-0.4SiS2 sample. This explicitly indicates that the electrolyte is not stable in contact with Li metal. However, the electrical charge of the reduction and oxidation reactions is almost equal in the case of the 3-5 % Li3POtdoped Li2S-SiS2 sample.
Oxygen doping by Li3PO4 addition in the glass network formation is effective in increasing stability towards Li metal. The maximum value, 6.9 x Scm- at 298 K was obtained for 0.58Li2S-0.39SiS2-0.03Li3P04 and the number of the carrier Li' ions increases on Li3P04 addition. This glass was prepared such that the molten material was quenched by dropping into liquid N2.
The conductivity is expected to be enhanced with the increase of the Li2S content in the glass. However, the range of concentrations at which glass formation takes place is too narrow to obtain a glass with high Li2S concentration. The glass-formation region can be expanded by a twin roller technique, which offers a very high cooling rate. A higher conductivity, 1.5 x 10-3Scm-', was obtained for an Li2S- rich sample of 0.63Li2S-0.36SiS2-0.01Li3P04.[3s1
The high conductivity and chemical stability towards the lithium metal anode identifies the Li2S-SiS2-Li3P04 system as the most suitable candidate electrolyte for rechargeable lithium batteries.[361 One of the most serious problems is deliquescence of these materials in ambient air. Further- more, SiS2 in the glass decomposes to Si02 and generates poisonous H2S gas when the glassy electrolyte absorbs water. It is therefore essential that sample preparation and the fabrication of cells of this type are done in a dry environment.
3. Oxide Materials
The Li' conducting oxides are another promising class of materials for application in the all-solid-state lithium batteries, because the oxide can be handled easily even in humid air. Until the 1970s, the Lie room-temperature conducting properties for the oxide based electrolytes had not been reported.
One candidate was Li3P04 with an orthorhombic structure (yII-Li3P04 Pmnb). The oxygens in the Li3PO4 lattice are hexagonal close packed, and cations of 3L1' and Pse fully occupy the sites with four-fold oxygen coordi- nation.13'] For Li3PO4 doped with Li4SiO4, i.e. the L ~ ~ + ~ P ~ - x ~ i , ~ 4 system, a penta-valent P" ion was partially substituted by two cations of Si4@ and Li'. In this case, an excess of Li' ions cannot occupy the cation sites coordinated by four oxygens, because the number of total cations is more than that of the cation sites. The Lie ions exist in migratory spaces between the M04 (M =cation) tetrahedrons in order to keep a charge balance of cations and anions.
Figure 2 shows the conductivity values for the Li3+*P1 -x- Si,04 system. Although pure Li3PO4 exhibits low conduc- tivity, the conductivity was greatly enhanced by partial LidSi04 doping.[3s1 This conductivity enhancement mainly comes from the occupation of second sites by Li' ions with an increase in the number of cations. Straight lines were obtained in the u T - 1/T relationship in Figure 2. The activation energy for Lie migration can be calculated from the Arrhenius equation (Eq. l), where T, go, E, and k denote the absolute temperature, pre-exponential factor, activation energy for ionic migration, and Boltzmann's constant, respectively.
aT = a. exp(-E/kT) (1)
300 200 100 I I
\ 2.0 1
1.0 1.5 2.0 2.5 3.0 3.5 io3rr (K-')
Fig. 2. n T - l/Trelationship for the Lis+xPI -,Six04 system
130 0 VCH Verlagsgesellschaft mbH. 0-69469 Weinheim. 1996 0935-9648/96/0202-0l30 $ lO.O0+.25jO Adv. Mater. 1996,8, No. 2
G.-y. Adachi, N . Zmanaka, H . AonolFast Li' Conducting Ceramic Electrolytes ADVANCED MATERIALS
Lie ions migrate with a low activation energy when the slope of UT - 1/T has a low value. The activation energy for Li' migration was also decreased by the Li4Si04 doping. The maximum conductivity and the minimum activation energy for Li@ ion migration were obtained for x = 0.5. In 1977, Hong reported a new yII-Li3P04-type material, i.e. Liz + 2xZnl - ,Ge04, named as a 2' zuper ionic conductor (LISICON).[391 In general, PO4 and Si04 tetrahedrons form a skeleton-type structure, which has cavities for ion migration. LISICON and NASICON were discovered during the search for an ion-conducting skeleton-type structure.
The Lie conductivity of LISICON (x=O.75) is high enough at elevated temperatures, e.g. 1.3 x 10- S cm- at 573 K. However, the conductivity for LISICON is less than
S cm- * at room temperature. Rodger et al. studied the various yrI-Li3P04-type materials, i.e. Li3 +,Al - ,B,04 (A = P,V,As; B = Si,Ge,Ti).[401
The Lie-ion conductivity of an ionic conductor is the most important property to consider with respect to its application in an all-solid-state lithium battery. To clarify the relationship between the electrical properties and the crystal structure, the conducting properties of the bulk component should be discussed. The total conductivity (u) of the polycrystalline electrolyte is obtained from the reciprocal of the summation of the bulk (b) resistivity and that for the grain boundary (pg& which can be determined individually by a complex impedance analysis (Cole-Cole
The total conductivity was almost the same as the bulk conductivity in the case of the y11-Li3P04-type
The conductivity for y11-Li3P04-type materials was hardly influenced by the grain boundary layer. The maximum conductivity is obtained at around x=0.5 for all the Li3+,Al-,Bx04 (A=P, V, As; B = Si, Ge, Ti) systems except for the Li3+,Pl -,Ge,04 system. For the Li3+,P1 -,Ge,04 system, the increase in lattice size by the substitution of a P5@ (0.17 A) site by the larger Ge4@ (0.390 A) contributes to the conductivity enhancement." 'I
The maximum bulk conductivity, ca. 2 x 10- S cm- at 300 K was obtained for x=O.75. Figure 3 presents the relationship between the lattice volume and electrical properties for the bulk component of the Li3.5A0.5B0.504 (A = P, As, V; B = Si, Ge, Ti) system. The activation energy for Lie migration decreased and the bulk conductivity at 300 K was enhanced with an increase in lattice volume. The Lie ions can more easily migrate with the enlargement of the tunnel size. The highest bulk conductivity in the oxide- based electrolyte was around 4 x 10- S cm- ' at room temperature for Li3.5V~.5Tio.504. This conductivity is still about 1/30 of that of the Li3N single crystal and Li2S-based glasses. It seems that the conductivity is enhanced by the utilization of the larger A' and B4@ ions. However, the electrical properties have not been reported for y11-Li3P04 type electrolyte with larger lattice parameters than the
104 c I V.Ti
10-6 0 t' . P,Si 10-7 1
330 335 340 345 350 355 360 v / A'
- - z 50 *
a is .
Fig. 3. Variation of conductivity at 300 K and activation energy with unit cell volume for Lis.sAo.sBo.s04 (A = P, V, As; B = Si, Ge, Ti). Conductivity 0, Activation energy 0.
parameters of Li3.5V0.5Ti0.504. This result shows that the tunnel size is too small for Lie to migrate in the y11-Li3P04- type structure.
&Alumina has a two-dimensional layer structure with a suitable distance between the layers for migration of the large Na' ion^."^"^] It is expected that the smaller Li' ion can more easily migrate through the layer in the p-alumina structure. On the contrary, the conductivity of Lie p- alumina (Li20. llA1203) is 1.3 x 10-4Scm-' at 298 K, which is about two orders of magnitude lower than that of Na@-P-al~mina.[~~] Lie ions do not migrate so easily through the wide p-alumina layers. Li' ions between the layers approach the oxygen in the layer because of a high Li'-O bonding energy.
The polycrystalline material Nal +,Zr2SiXP3 - ,OI2 (NASI- CON) designed by Goodenough et al. has a three- dimensional network structure ( R k ) which possesses a suitable tunnel size for the migration of the large Na@ ions.[159161 LiM 2(Po4)3, M(Iv) = Ge,[439441 TiJ45-531 Hf,[541 and Zr,[55-571 also forms a NASICON-type structure.[581 The NASICON-type structure is shown in Figure 4. Two polyhedra, M06 octa- and PO4 tetrahedra, are linked by their corners to form the [M2(P04)3le rigid skeleton and the Lie ions migrate through the tunnel three-dimensionally in the structure. Two different lithium-ion sites, Al and A2 exist in the NASICON-type structure. For LiM2(P04)3 (M=Ge,Ti,Hf,Zr), the A1 sites are fully occupied, but the A2 sites are completely vacant. In the case of the Lil +,Al,Ti2 -x(P04)3 system, a tetravalent Ti4@ ion was partially substituted by A13@ ions. The additional Lie ions occupy the A2 site with an increase in x for the Lil + ,A1,Ti2 - ,(Po& system.
Table 4 shows the porosities, the conductivities at 298 K, and activation energies for bulk and grain boundary
A&. Muter. 1996, 8, No. 2 0 VCH Verlagsgesellschaft mbH, 0-69469 Weinheim, 1996 0935-9648/96/0202-0131 $10.00+.25/0 131
G.-y. Adachi, N . Imanaka, H . AonolFast Li* Conducting Ceramic Electrolytes ADVANCED MATERIALS
Fig. 4. The NASICON-type structure of LiM2(P04), (M = tetravalent cation, A, =lithium ion site). AZ site is not shown in the left figure.
properties of the Lil +xAl,Ti2 -,(PO& system. For LiTi2(P04)3 (x=O), a highly dense ceramic cannot be obtained by sintering, even at high temperature, and its conductivity is low. However, high-density and high-Lie- conductivity samples were obtained by Ti-site substitution for AI3@. The maximum conductivity 7.0 x 10-4Scm-1 at 298 K for Li1.~Alo.3Ti1.~(P04)3, is comparable to that of the highest Lie conducting materials, such as single crystal Li3N and the Li2S-based glasses described above.
The Lie ion conductivity is the highest value reported for NASICON-type Lie conducting electrolytes. However, the most distinct difference from yII-Li3P04 type materials is that the activation energy for the bulk component is constant at around 28 kJmol-', irrespective of the substitutional amount.
The lithium insertion onto A2 sites by M3@ did not influence the Lie mobility in the bulk. On the other hand, the activation energy at the grain boundary decreased with increasing x for the Lil +xA1,Ti2-,(P04)3 system. The trend towards a reduction in activation energy is similar to that towards a reduction in the porosity. The conductivity enhancement on A13@ ion substitution in LiTi2(P04)3 is mainly attributed to the decrease in the activation energy at the grain boundary by the high densificati~n.[~~] The conductivity for NASICON-type polycrystalline electrolytes at around room temperature are mainly controlled by the grain boundary component. Thus, the control of the grain boundary condition is a key factor in obtaining highly conducting NASICON-type Li@ ionic electrolytes.
0 0.1 0.2 0.3 0.4 0.5
30 & . h 1 .- g
Fig. 5. The conductivity at 298 K and the porosity for NASICON-type solid electrolytes. LiGez(P04)3 + yLizO system A: LiTi2(P04)3 + yLi20 system 0; LiHf2(P04), + yLi20 system rn porosity (-).
Another effective way of obtaining a high density ceramic is to add lithium compounds such as Li20, Li3P04 or Li3B03 into LiM2(P04)3, M = Ge, Ti, or Hf.r58-601 Figure 5 shows the conductivity and the porosity for the LiM2(P04)3 + yLi20 systems. The conductivity was greatly enhanced by the addition of a lithium compound. The maximum conductivity, ca. 5 x S cm- ', for the M =Ti system is almost equal to that of the Lil +xAI,Ti2 - x(P04)3 system in Table 4. The lithium salt does not form a solid solution with LiTi2(P04)3 and functions only as a binder to yield high density ceramics. This high densification results in the conductivity enhancement.
Figure 6 shows the activation energy of the bulk and the grain boundary for the LiM2(PO& + yLi20 system. The activation energy of Lie migration at the grain boundary was also decreased by the addition of the Li20, but that at the bulk component was unchanged. The activation energies for the bulk component have been estimated to be ca. 37 kJmol-' for LiGe2(P04)3, 28 kJmol-' for LiTi2(P0,J3, and ca. 41 kJmol- for LiHf2(P04)3, respec- tively.[s81 The decrease in the activation energy with the LizO addition for the grain boundary is similar to the A13@- substituted Lil +xAlxTi2 - x(P04)3 systems1531 and also shows
Table 4. The porosities, the conductivities at 298 K, the activation energies for bulk and grain boundary of sintered pellets for the Li, +,AI,Ti~_,(POa)3 system
Sample Porosity (%) ~
Activation energy (kJ . mol-') bulk grain boundary
34.0 5.6 4.1 1.4 2.0
2.0 x lo6
7.0 1 0 - ~ 5.0 1 0 - ~ 2.5 1 0 - ~
6.5 x 29 28 28 28 29
46 34 35 35 34
0 VCH Verlagsgesellschaft mbH, 0-69469 Weinheim, 1996 0935-9648/96/0202-0132 $ l O . O O + . Z S / O Adv. Mater. 1996,8, No. 2
G.-y. Adachi, N . Imanaka, H . AonolFast Li' Conducting Ceramic Electrolytes
7 45 - c
40- x 2 B - 8 :
35 .* I .* I 9 30-
' M=Hf (bulk)
A A A A M=Ti (bulk)
1 2 0 ' I I I
0.0 0.1 0.2 0.3 0.4 0.5 Y
Fig. 6. The activation energies of the bulk and the grain boundary components of NASICON-type solid electrolytes. LiGeZ(PO& + yLizO system: bulk A , grain boundary A; LiTiz(P04)3 + yLizO system: bulk 0, grain boundary 0; LiHfZ(P04), t yLizO system: bulk m, grain boundary 0.
that the conductivity is mainly controlled by the grain boundary condition.
The tunnel size for the structure is easily adjusted by changing the substituting tetravalent cation M4@ (Ge or Hf) in LiTi2(P04)3. The intra-grain Li@ migration is directly affected by the change of the lattice size caused by the ion substitution. Figure 7 presents the relationship between the activation energy for Lie ion migration for the bulk component and the cell volume for LiMxTi2 - x(P04)3.[581 The cell volume for LiMXTi2-,(P04)3 changes with M4e
0 A A
1150 1200 1250 1300 1350 1400 1450 1500
Cell Volume / 8,
Fig. 7. Relationship between the activation energy for Li@ intra-grain migration and the cell volume (hexagonal unit) for LiM,TiZ_ .(PO&-bulk (M=Ge 0, Hf A , and x=O 0).
(M = Ge and Hf) substitution. The minimum activation energy (27-28 kJmol-') is obtained for the samples with cell volumes around 13 10 A3, a value which is almost equal to that of LiTiz(P04)3.
The activation energy appreciably increases when the cell volume becomes smaller or larger than that of LiTi2(P04)3. This clearly indicates that the LiTi2(P04)3 structure has the most suitable lattice size for Lie migration. A smaller tunnel size makes the carrier Lie migration more difficult and a larger size leads the mobile Lie ions to approach the oxygens in the polyhedra from the center of the Lie ion site. A similar phenomenon is observed for the above mentioned LIB-,!!?- alumina system composed of a two-dimensional layered structure.[421 The conductivities for typical NASICON-type solid electrolytes are summarized in Table 5. The Ti-based materials have the maximum conductivity of 5- 7 x S cm- among the NASICON-type Li' ionic conductors. However, it must be pointed out that the Ti4@ ion in the LiTi2(P04)3-based electrolytes is easily reduced to Ti3@ by Li metal used as an anode.1611
As far as the application in lithium batteries is concerned, the other LiGe2(PO&-based or LiHf2(PO4)3-based systems having conductivities of 1-3 x 10-4Scm-1 would be more appropriate candidates among the identified NASICON-type oxide-based Lie conductors. The details of the structure and the electrical properties of the oxide-based Li' conducting ceramics have been reported in a previous publication.[621
Perovskite-type AB03 materials come in several forms, such as A@B5@O3, A2@B4@03, and A3eB3@03. In 1984, Latie et al. reported that the perovskite-type LixLali302/3 - x- Nbl -xTi,03 (El: A-site vacancies) is a Lie conducting material (ca. I O - ~ S ~ ~ - ' at room temperature for x = 0.05).[631 Since then, the Li3xLa2/3 - x 0 1 / 3 - zxTi03 system which exhibits high bulk conductivity was investi- gated.r64-681 In this system, A-sites are made up of Lie ions, La3' ions, and vacancies. The high conductivity for these materials comes from the presence of vacancies on A-sites. The lithium ions freely migrate through A-site vacancies of the perovskite lattice. The total conductivity (reciprocal of bulk resistivity plus grain boundary effects) and the bulk conductivity are as high as 7 x 10-SScm-l and 1 x ~ o - ~ ~ c m - ' at room temperature for x=0.1O-O.12, respectively. The conductivity is mainly controlled by that of the grain boundary.
Figure 8 shows the relationship between the electrical properties and the lattice parameters for the perovskite-type
Table 5. The total conductivities at room temperature for the NASICON-type Lim conducting ceramics.
Sample Conductivity Ref. ( S . cm-')
Adv. Mater. 1996,8, No. 2 8 VCH Verlagsgesellschafl mbH. 0-65469 Weinheim, 1556 0535-9648/56/0202-0133 $ 10.00+.25/0 133
G.-y. Adachi, N . Zmanaka, H . AonolFast Li' Conducting Ceramic Electrolytes ADVANCED MATLRIALS
. l o 4
rn . b 10'
1 0 6
30 3.80 3.82 3.84 3.86 3.88 3.90
a / A Fig. 8. Plot of the conductivity at 400 K and activation energy for Lie ion migration versus perovskite cell parameter for Li0.34La0.51Ti02.94, Lio.3J'r0.56- Ti03.01, Lio.~~N&.~~Ti03.00, and Li, 38Smo.52Ti02.97. Conductivity 0; Acti- vation energy 0.
lithium conductors.[681 The atomic ratio for all the perov- skite-type materials in the figure was analyzed, since elemental Li partially vaporized during sintering at high temperature. The bulk conductivity increases and the activation energy for Li' ion conduction decreases with an increase in lattice parameter. The lattice size was increased by the substitution of the larger lanthanide ion. The minimum activation energy, 35 kJmol-' is obtained for Li0.34La0.51Ti02.94, which is still high compared with 27- 28 kJ mol- ' of the NASICON-type LiTiz(PO&-based ceramics. The conductivity will be enhanced by increasing the lattice size. This result clearly shows that a tunnel size is still narrow for Li' migration in the perovskite material similar to that for yII-Li3P04-type materials.
The vitrification of the polycrystalline electrolyte has also been attempted to eliminate the less conducting grain boundary layer in the polycrystal. The glassy material has larger spaces for the ion migration than the crystalline material. The enhancement of the conductivity by vitrifi- cation has been reported for some polycrystalline elec- trolytes. For example, the conductivity of glassy Li3.6P0.4Si0.604 is 5 x Scm-' at 298 K, which is 10- times higher than that of the polycrystalline form.[691 In these samples, the conductivity was mainly enhanced by an enlargement of the tunnel size for the ion migration and by the elimination of the grain boundary,
Kanehori reported a thin film solid-state rechargeable battery of Li/glassy Li3.6P0.4Si0.604/TiS2.[691 The amorphous solid electrolyte film was sputtered on the cathode TiS2 film. Then Li anode film was evaporated on the electrolyte. Total cell thickness is ca. 30 pm. The short circuit current was 1.3 mAcmW2 and the good recharge- able characteristics were obtained at a current density of
3-16 pAcm-'. The glassy samples of NASICON-type materials, LiTi2(P04)3 and Lil,3A10.3Til.,(P04)3 can be also obtained by rapid quenching["] and by an explosion method.[711 However, the conductivities of the materials are appreciably lower than those of the polycrystalline electrolytes. The vitreous electrolyte does not possess a suitable tunnel size for Li' migration, since the rigid [Ti2@'04)3]e skeleton bulk which is appropriate for Li' conduction, is destroyed by the vitrification. The vitrifying treatment is not appropriate in the case of these three- dimensional skeleton-type Li' ionic conductors. For the oxide-based glasses, the maximum Li' ionic conductivity is lower than 10-'Scm-' at 298 K.
4. Conclusions and Outlook
In order to produce a significant electric current in a battery, high ionic mobility in the electrolyte is essential. In practice, the battery performance mainly depends on the ionic species which is migrating. Lithium ions are the most promising because of their high energy density and very negative electrochemical potential. The lithium rechargeable batteries on the market utilize organic solvents which function as the high conductivity electrolyte. There are still some problems, such as leakage of the organic electrolyte and ignition of the electrolyte, or its decomposition products, at higher temperatures. High Lie-ion conducting solid electrolytes based on inorganic materials are therefore expected to lead to the most promising advances for the all- solid-type, lithium rechargeable battery.
For non-oxide-based ceramic electrolytes, many high Lie conducting materials, such as the Li1-Al203 composite system, Li3N-based single crystals, and LizS-based glassy materials, have been developed. Among these, the most suitable materials are found in the glassy Li2S-SiS2-Li3P04 system. These materials have high conductivity with high stability when in contact with lithium anode metal. The prospects for this electrolyte are indeed excellent for practical applications, provided that problems related to the hygroscopic nature of the material can be solved by appropriate sealing treatments.
Amongst the oxide solid electrolytes, such as the NASICON-type materials, the yII-Li3P04 type and the perovskite type materials, the NASICON-type exhibit the highest conductivity and the lowest activation energy for Li' migration, without any hygroscopic properties. Among the various NASICON-type materials, the LiTi2(P0&- based electrolytes have the most suitable lattice for Lie migration and exhibit the highest conductivity amongst the oxides.
When the material is applied in the polycrystalline form, the control of the grain boundaries is the most significant factor when controlling the conductivity. The grain boundary component of conductivity for L i T i ~ ( p 0 ~ ) ~ can be enhanced by Ti4e site replacement by M3@ ion, or by the addition of a
134 0 VCH Veriagsgeseiischaii mbH, 0-69469 Weinheim. 1996 0935-9648/96/0202-0134 $ 10.00+.25/0 Adv. Mater. 1996, 8, No. 2
G.-y. Adachi, N . Immaka, H. AonolFast Lie Conducting Ceramic Electrolytes
lithium compound resulting in an increase of sinterability. The highest conductivity in the NASICON-type Lie conductors is 7.0 x S cm- for Lil,3Alo.3Til,7(P04)~, which is almost equal to that of the Li2S-SiS2-Li3PO4 system. This material is another promising candidate as a constituent of the lithium rechargeable battery, if the reduction of constituent Ti4e by Li metal in the battery can be suppressed by improvement of the anode material or by controlling the layer between the anode and the electrolyte.
At present, LiGe2(P04)3-based or LiHf2(P04)3-based materials are the most appropriate candidates among the oxide materials. The conductivity of both systems is about one fifth that of the Li2S-SiS2-Li3P04 glass. However, these oxide materials have the remarkable advantage of a high stability toward humid air compared with the Li2S-based glasses.
The market for lithium-based rechargeable batteries is expected to grow rapidly for various kind of applications because of their excellent performance characteristics. For the purpose of obtaining a high current density, the preparation of the electrolyte as a thin film is one suitable approach. Hence, progress in thin film preparation tech- niques is one key factor towards the successful realization of the all-solid, lithium rechargeable battery.
Received July 18, 1995 Final version: October 13, 1995
[I] C. Tubandt, Z . Anorg. Allg. Chem. 1921, 115, 105.  C. Tubandt, H. Reinhold, Z . Electrochem. 1923, 29, 313.  L. W.Strock, Z . Phys. Chem. 1935, B31, 132. 141 B. Reuter, K. Hardel, Naturwissenschaften 1961, 48, 161.  B. B. Owens, A. G. Argue, Science 1%7,157,308.  T. Takahashi, S. Ikeda, 0. Yamamoto, J . Electrochem. SOC. 1972,119,477.  M. Tatsumisago, Y. Shinkuma, T.Minami, Nuture 1991, 354, 217.  D. 0. Raleigh, J . Appl. Phys. 1970, 41, 1876.  T. Takahashi, 0. Yamamoto, S. Yamada, S. Hayashi, J . Electrochem.
Soc. 1979,126, 1654. [lo] C. Wagner, Naturwissenschaften 1943,31, 265. [Ill R. D. Shannon, Actu. Crystullogr. 1976, A32, 751. [I21 W. J. Fleming, J. Electrochem. SOC. 1977, 124, 21. [I31 N. Weber, J. T. Kummer, Proc. Annu. Power Sources Conf. 1%7,21, 37. [I41 M. S. Whittingham, R. A. Huggins, J. Chem. Phys. 1971,54,414. [I51 J. B. Goodenough, H. Y.-P. Hong, J. A. Kafalas, Muter. Res. Bull. 1976,
[I61 H. Y.-P. Hong, Muter. Res. Bull. 1976, 11, 173.  M . Watanabe, M. Kono, E. Hayashi, S. Mori, N. Ogata, Polym. Prepr.
[IS] C. A. Angell, C. Liu, E. Sanchez, Nuture 1993, 362, 137. [I91 C. C. Liang, J . Efectrochem. SOC. 1973, 120, 1289.  K. Fester, W. D. Helgeson, B. B. Owens, P. M. Skarstad, Solid State
 C . R. Schlaikjer, C. C. Liang, J. Electrochem. SOC. 1971, 118, 1447.  F. W. Poulsen, N. H. Andersen, B. Kindl, J. Schoonman, Solid State
 C . C. Liang, L. H. Barnette, J. Electrochem. SOC. 1976, 123, 453.  C. C. Liang, A. V. Joshi, N. E. Hamilton, J. Appl. Electrochem. 1978,8,
 U. v. Alpen, A. Rabenau, G. H. Talat, Appl. Phys. Lett. 1977, 30, 621.  B. A. Boukamp, R. A. Huggins, Mater. Res. Bull. 1978, 13, 23.  J. R. Rea, D. L. Foster, Mater. Res. Bull. 1979, 14, 841.
Jpn. 1993, 42, 2818.
Ionrcs 1983, 9&10, 107.
Ionics 1983, 9&10, 119.
 H. Obayashi, A. Gotoh, R. Nagai, Muter. Res. Bull. 1981, 16, 581.  H. Obayashi, R. Nagai, A. Gotoh, S . Mochizuki, T. Kudo, Mater. Res.
 R. Mercier, J . P. Malugani, B. Fahys, A. Saida, Solidstate Ionics 1981,5,
 H. Wada, M. Menetrier, A. Levasseur, P. Hagenmuller, Mater. Res. Bull.
 J. H. Kennedy, Y. Yang, J . Electrochem. SOC. 1986,133, 2437.  J. H. Kennedy, Z. Zhang, Solid State Ionics 1988, 28-30, 726.  S. Kondo, K. Takada, Y. Yamamoto, Solid State Ionics 1992, 53, 1183.  N. Aotani, K. Iwamoto, K. Takada, S . Kondo, Solid State Ionics 1994,
 K. Takada, N. Aotani, K. Iwamoto, S. Kondo, SolidStute lonics, 1995, 79,
 A. R. West, Z . Kristallogr. 1975, 141, 422.  Y. -W. Hu. I. D. Raistrick, R. A. Huggins, Muter. Res. Bull. 1976, 11,1227.  H. Y. -P. Hong, Muter. Res. Bull. 1978, 13, 117.  A. R. Rodger, J. Kuwano, A. R. West, Solid State Ionics, 1985, I S , 185.  P. G. Bruce, A. R. West, J. Electrochem. SOC. 1983, 130, 662.  M. S. Whittingham, R. A. Huggins, Nutl. Bur. Stand. (NBS) Spec. Publ.
 Shi-chun Li, Jian-yi Cai, Zu-xiang Lin, Solid State Ionics 1988, 28-30,
 H. Aono, E. Sugimoto, Y. Sadaoka, N. Imanaka, G. Adachi, Bull. Chem.
 L. Shi-chun, L. Zu-Xiang, Solid Stute Ionics 1983, 9/10, 835.  L. Zu-xiang, Y. Hui-jun, L. Shi-chun, T. Shun-Bao, Solid Stare Ionics
 L. Zu-xiang, Y. Hui-jun, L. Shi-chun, T. Shun-Bao, Solid State Ionics
 M. A. Subramanian, R. Subramanian, A. Crearfield, Solid State Ionics
 S . Hamdoune, D. Tranqui, Solid State Ionics 1986, 18/19, 587.  D. Tranqui, S. Hamdoune, J. L. Soubeyroux, E. Prince, J. Solid State
 H. Aono, E. Sugimoto, Y. Sadaoka, N. Imanaka, G. Adachi,
 H. Aono, E. Sugimoto, Y. Sadaoka, N. Imanaka, G. Adachi,
 H. Aono, E. Sugimoto, Y. Sadaoka, N. Imanaka, G. Adachi, Chem. Lett.
 H. Aono, E. Sugimoto, Y. Sadaoka; N. Imanaka, G. Adachi, Solid State
 B. E. Taylor, A. D. English, T. Berzins, Mater. Res. Bull. 1977, 12, 171.  D. Petit, P. Colomban, G. Collin, J. P. Boilot, Muter. Res. Bull. 1986,21,
 L. Hagman, P. Kierkegaard, Actu Chem. Scand. 1968, 22, 1822.  H. Aono, E. Sugimoto, Y. Sadaoka, N. Imanaka, G. Adachi,
 H. Aono, E. Sugimoto, Y. Sadaoka, N. Imanaka, G. Adachi, Solid-State
 H. Aono, E. Sugimoto, Y. Sadaoka, N. Imanaka, G. Adachi, Chem. Lett.
 H. Aono, E. Sugimoto, Y. Sadaoka, N. Imanaka, G. Adachi, Chem. Lett.
 H. Aono, N. Imanaka, G. Adachi, Ace. Chem. Res. 1994,27,265.  L. Latie, G. Villeneure, D. Conte, G. L. Flem, J . SolidSture Chem. 1984,
I641 A. G. Belous, G. N. Novitskaya, S. V. Polyanetskaya, Yu. I. Gornikov,
 Y. Inaguma, C. Liquan, M. Itoh, T. Nakamura, Solid State Commun.
 H. Kawai, J. Kuwano, J . Electrochem. SOC. 1994, 141, L78.  Y. Inaguma, L. Chan, M. Itoh, T. Nakamura, Solidstate Ionics 1994, 70/
 M. Itoh, Y. Inaguma, W. -H. Jung, L. Chen, T. Nakamura, Solid State
 K. Kanehori, K. Matsumoto, K. Miyauchi, T. Kudo, SoZid State Ionics
 N. Machida, K. Fujii, T. Minami, Chem. Lett. 1991, 367.  N. Imanaka, T.Shimizu, G. Adachi, Solid State Ionics 1993, 62, 167.
Bull. 1981, 16, 587.
1983, 18, 189.
(US) 364, 139.
Soc. Jpn. 1992, 65,2200.
1986, 18/19, 549.
1988, 31, 91.
1986, 18\199 562.
Chem. 1988, 72,309.
J . Electrochem. Soc. 1989, 136, 590.
J. Electrochem. SOC. 1990, 137, 1023.
Ionics 1993, 62, 309.
J. Electrochem. SOC. 1993, 140, 1827.
Ionics 1991, 47, 257.
fzv. Akud. Nauk SSSR, Neorg. M a w . 1987, 23,470.
1993, 86, 689.
Ionics 1994, 70171, 203.
1983, 9/10, 1445.
Adv. Muter. 1996, 8, No. 2 0 VCH Verlagsgesellschaft mbH, 0-69469 Weinheim, 1996 0935-9648/96\0202-0l35 $ 10.00+.25/0 135