Transcript ClareGrey

New NMR Approaches For Studying
Battery Materials and Fast Ionic
Conductors
Julien Breger, Meng Jiang, Young Joo Lee, Wonsub
Yoon, Namjum Kim, Francis Wang, John Palumbo
and Clare P. Grey
SUNY Stony Brook
Introduction
 Part I: Batteries
How do rechargeable batteries work?
What are some of the technological requirements for 21st century devices?
Where are some of the fundamental scientific breakthroughs required to
achieve these goals?
What “New” approaches can be used to investigate Li-ion battery
materials?
Introduction to NMR
Application to spinels and layered materials
 Part II: Ionic Conductors
Mechanisms for ionic conduction in solids
Devices that require rapid ionic conductivity
Studying motion by NMR
Applications to Aurivillius phases
Part I. Rechargeable Batteries:
Applications and Demands
 Portable (electronic) technologies have created an increasingly high
demand for batteries that
 last considerably longer and deliver power faster
 are non-toxic and may be readily recycled
 are cheaper and much lighter..
 Applications include:
 Electric and Hybrid Electric Vehicles (EVs and HEVs)
 Golf carts, wheel chairs and industrial vehicles
 Laptops
 Cell-phones
 digital cameras
 Camcorders, portable tools
 Artificial hearts (current battery lasts 4 hours!)
 memory backup, energy storage,
 generators, power for remote locations...
Different Applications Have Very
Different Power Requirements
Portable Electronics (Cell phones, laptops, PDA, digital
cameras)
Medical Devices
Low Power
(high energy)
Portable tools
Back-up power (UPS)
EVs and HEVs
Electric bikes/scooters
“Industrial” EV, forklifts,
golf carts
Power Storage for
Renewable Energy
High Power
Batteries come in two types:
 Primary: discharged once and discarded - lower energy/capacity
applications
 Secondary: rechargeable and can be used over again (e.g., Ni/Cad),
rocking chair batteries
-MnO2 (cathode)
Li (anode)
3V
Separator
(electrolyte)
Li/MnO2 bobbin cell
Chemistry is not reversible
Cathode: LiCoO2
Anode: Graphite LixC
Charge
Li+
Li+
Discharge
What is in a battery?
The Ni/Cd or NiCad Rechargeable Cell
Negative electrode
(anode)
Cd0 + 2OH--> Cd2+(OH)2 + 2e-
Electrolyte
(aqueous)
Cd
 OH- ions released at the
cathode travel to the anode
where they react with the
Cd metal.
I.e., electrolyte must allow
ionic but not electric conduction
NiOOH
Positive electrode
(cathode)
Ni3+OOH + H2O + e--> OH- + Ni2+(OH)2
OH1.35V Cell
e-
 Difference in the
couples (reduction potentials) of the two 1/2 cells determines the cell voltage
What is required to build
a rechargeable battery?
 Design a cell in which the chemistry is reversible over many cycles
e.g., the lead acid (car) battery:
(Pb0 + Pb4+O2 + H2SO4 -> 2Pb2+SO4 + 2H2O)
2V
Requirements:
 large differences in the potentials of the half cells (i.e., Pb4+/Pb2+ and
Pb2+/Pb0
=> High voltages
 Light materials
=> High energy densities
 An electrolyte that does not react with the anode or the cathode
(under as wide as possible temperature range)
 An electrolyte with high ionic conductivity
Allows rapid discharge
 Conducting anodes and cathodes
=> (and charge)
(Allows current to be removed)
I.e., high power densities
 Non-toxic materials
 Stable in charged and uncharged states
H2 evolution; unstable in
discharged state
Much higher voltages may be
achieved with Li-ion batteries
 In theory, a more than 2V gain in voltage can be achieved with a
Li+/Li anode
 Li is also v. light ---> higher energy densities
 Some electrochemistry:
E0 (V)
Li+(aq) + e- -> Li (s)
-3.04
PbSO4(s) + 2e- -> Pb(s) + SO42-(aq) -0.36
Cd(OH)2 + 2e- -> Cd(s)
-0.824
J. -M. Tarascon
Nature ‘01
[Cd2+(aq) + 2e- -> Cd(s)
-0.402]
2H+(aq) + 2e- -> H2
0
E.g., LiTiS2 ----> TiS2 + Li
Whittingham et al. ‘72
but….
The big advance in this field came with the
development of the SONY “Rocking-Chair”
battery in 1990
 LiCoO2, J. Goodenough (1980)
 2ndary host material, Murphy et al., and Scrosati et al. (‘78 and ‘80)
 Lithium shuttles backwards between two layered compounds
 Very high voltage (4 V; cf Ni/Cd @ 1.35 V)
Solid state chemistry at the cathode
end...
Positive Electrode
Polymer Binder
Co3+ is oxidized to Co4+ on Li
removal (deintercalation)
I.e., during charging
Intercalation
Oxide
Co
O
voltage vs. Li/Li+
Carbon black
Li
Ohzuku et al. J. Electrochem Soc. 140, 1862, ‘93
And at the anode end..
 Graphite anode forms an intercalation compound LixC
Charging
Tarascon Nature, ‘01
Energy density
SONY Cell
90 Wh/kg
Pb acid
30
Ni/Cd
30-35
Ni metal hydride 50
Cycle Life
Voltage
500-1000
4V
250-500
2V
300-700
1.3V
300-600
1.2V
Energy density (Whkg-1)
Why is more research needed?
Some disadvantages of the LiCoO2 cell
 Co is toxic and expensive
 Not sufficient Co globally to meet perceived
demands for rechargeables
 Only 0.5 of the Li can be removed. I.e., low capacity
 V. slow to charge and discharge (low power) - not suitable for E.V.s,
H.E.V.s or other high power applications
- Li
What is motivating all this work?
New markets for Li
Power
tools
Hybrid Vehicles
Bosch, Makita, Cooper, Milwaukee, Metabo all
have high end powertools with Li-ion batteries
spinels
Already happening
Currently all NiMH
Clearly effort to go to Li-ion (e.g., Toyota, Johnson
Controls, Samsung)
Conservative estimate of 3.5% HEV market penetration
in 2010 requires 1 Million KWh of battery capacity.
$1-3 Billion dollar market
Soon to come?
Power back-up
(small and large)
Future market?
A Large Growth is Predicted for the Li-ion
Battery Technology - which will be driven
by improvements in performance
Ni-Cd
Ni-MH
inexpensive, high
power, low energy
expensive, high power,
average energy density, no
further improvement
expected
Market
Worldwide
battery
market
$1 Billion
$0.6 Billion
Li-ion
Still expensive, lower
power, high energy,
safety
$4 Billion
From The Cobalt
report (2005)
Alternative LiB Materials Under
Consideration: Voltage vs. Capacity
Cathodes
LiCoO2
Anodes
C
Li metal
 “Issues and challenges facing rechargeable lithium batteries”, J.-M.
Tarascon and M. Armand, Nature 414, 359-367 (2001)
Some Advances in the Anode Field:
Nanoparticles and Composites
 Metals and alloys show v. high capacities (e.g, Si = 4200 mAhg-1 approx. 10x
that of graphite), but suffer from extremely large volume expansions
 => use a composite (of nanoparticles/domains) to absorb stresses during
cycling (less problem with reactions with electrolytes at low voltage - fewer
safety issues)
 Field sparked off by the discovery of Tin-Based Amorphous Oxides (TCO):
Sn1.0B0.56P0.4Al0.4O3.6 (Kubota, Matsufuji, Maekawa, Miyasaka, Science, 276, 1997)
TCO : SnO + SnO2 --> Li2O + Sn ----> LixSn
Li2O
CoO + 2Li ---> Li2O + Co (740 mAhg-1)
S. Grugeon… J.-M. Tarascon, 2003
Other approaches extrusion/displacement reactions
 InSb + Li
 Cu2Sb + Li
--->
--->
Li3Sb + In --> LixIn
Li3Sb + In
M. M. Thackeray, J. Vaughey (1999)
J. Dahn
 Cu2.33V4O11
+ x Li ---->
LixCu2.33-y V4O11 + yCu
M. Morcrette, J. -M. Tarascon
(2001)
Reversibility?
What other cathode materials are
being investigated?
 Both Fe and Mn oxides have been studied extensively
as alternative cathode materials (cheap and nontoxic)
E.g.:
 3 dimensional structures such as:
-Manganese Spinels
LiMn2O4 --> MnO2+ Li
-Good electronic conductivity - high power
low capacity
LiFePO4 and related phosphates - low capacity
-Poor electronic conductivity
- good rate performance
(with C coating)
 New layered Materials such as Li(NiMn)0.5O2,
Li(Co1/3Mn1/3Ni1/3)O2 (Mn4+)
 High capacity --- poorer rate performance?
 Layered lithium vanadates (for static power supplies)
C
LiFePO4
Improving the materials performance requires a
fundamental understanding of how materials
function and what structural/electronic properties
limit battery performance
 Structures of the materials as they are cycled
Where are the Li+ intercalated into the structure?
 Electronic properties How do they change as Li+ is removed?
 Ionic conductivities
How do the Li+ ions move through the lattice?
 Effect of structure and electronic properties on voltage (e.g., Co3+ -> Co4+;
Fe2+->Fe3+)
Li+ in oct vs. tet. sites
 A whole variety of experimental and theoretical methods have been used to study
these systems including:
 Electrochemistry (voltages … infer structural changes)
 Diffraction methods (long range structure)
 XANES (local structure and oxidation state)
 Solid state NMR (local structure)
WHY NMR?
Solid State NMR can be used to obtain:
Local atomic structure + Local electronic structure
•In theory, we can use NMR to study these systems and determine:
where the Li+ ions are, their local coordination environments, and the Mn
electronic structure
...…..at each stage of charge/discharge
to obtain fundamental information about how the battery works
Mn(IV)
…and how the battery fails
(III)
Mn
Oxidn
states?
LiMn2O4
We perform lithium NMR, since the lithium
is directly involved in the electrochemical
process...
But what is nuclear magnetic
resonance (or NMR)?
 Require nuclei (e.g., 1H, 13C, 6Li, 7Li, 31P)
that have non-zero nuclear spins
 Spin-1/2 vs. quadrupolar (I>1/2)
 Nuclei behave like tiny bar magnets in a
magnet field
Bo
DE
Each nucleus has a characteristic energy
splitting DE, which depends on the magnetic
field strength B0 => element specific
Detect signal at frequency corresponding to
DE, with intensity proportional to no. of spins
=> quantitative
Basement @
Stony Brook
NMR of Solids
•NMR spectra of solids are broadened by the
anisotropic interactions (interactions whose
magnitude depends on the orientation wrt
field), e.g.,:
•Chemical shift anisotropy
•Dipolar coupling (Nuclei interact with each
other, in the same way that two bar magnets
interact)
•Quadrupolar interaction
•6Li (I = 1); 7Li (I = 3/2)
B0
I
•Much of the chemical information can be lost
•Cf liquids NMR where the tumbling of the
molecules removes these interactions
Liquid NMR, e.g.,
CH3CH2OH
1H
Solid State
q
rIS
S
HD = -D(3cosq - 1)IzSz
1H
D =  I  Sh/2pr3
NMR
Most of the anisotropic interactions may be
removed by Magic Angle Spinning: The bigger the
interaction, the faster we need to spin
Bo
Axial CSA
Powder pattern
nr
rt
(3cos2q - 1) = 0
Slow MAS
Frequency, 
nr
time
Faster MAS
Rotor period
= 1/nr
nr
NMR is also very
sensitive to motion
f (Hz)
Slow exchange
E.g., Two Site Exchange
kr(A-B)
A <----------------> B
k << f
Intermediate
regime
k same order as f
Fast exchange
k >> f
Timescale
Hz
kHz
T1r Lineshape
changes
Loss of
T2
satellite
transitions
MHz
T1
NMR can be used to:
• extract correlation times and
activation energies for motion
•determine which sublattices are
mobile
Many of the battery materials are
paramagnetic, which introduces additional
complications
d
electrons
Mn4+
Li2MnO3 (LiMn2O4)
Magnetic
momentdepends
on the no.
of unpaired
eno field
meff
Co3+
B0
Co4+
LiCoO2
Li1-xCoO2
NMR
ESR
T1e

NMR timescale
Time averaged
value of magnetic
moment a c
In order to extract structural information, we need to understand
the effect of the unpaired electrons on the NMR spectra
The Fermi-contact shift results in large
shifts - these shifts contain both
structural and electronic information
 The Fermi Contact Shift:
measure of the unpaired electron spin density
transferred from the paramagnet to the NMR
nucleus
Hc = IzAs<Sz>
<Sz>  c
E.g., LiMn2O4 6Li MAS NMR
Fermi Contact Shift
Typical chemical
shift for diamagnetic
compound
520
Mn
t2g
Li
C.f. J-coupling and
magnetic coupling
H
N
1000
C
H
800
600
400
ppm
200
0
The sidebands are largely caused by the
dipolar coupling, which contains
geometric information
 The dipolar interaction: through space coupling to nearby nuclear or e- spins
Oh (2b)
rIS
<SZ>
**
*
**
Mn
*
***
**
* **
684
q
I
Li2MnO3
1461
B0
Li
755
734
Mn
Li0.5Zn0.5[Mn1.5Li0.5]O4
Mn
Oh (4a)
Li
2325
•Very different
sideband patterns are
observed, depending
on the arrangement
of Mn around the
central Li.
* ** * * ** *
*
** **
***
5000 4000 3000 2000 1000
ppm
0
Zn
-1000
NMR spectra are also sensitive to
electronic conductivity mechanism
520
• LiMn2O4 is a hopping semiconductor
• Li “sees” an average oxidation state
=> only one local environment
Li (8a Td)
Mn (16d Oh)
1000
C.f. metals - Knight shifts seen
800
600
400
ppm
200
0
But… how do we explain the different local
environments seen in LiMn2O4 as a function
of synthesis temperature?
520
Synthesis
Temperature
.. and obtain
chemical
information from
these systems
512
850 ºC
?
635
590
554
512
588
551
650 C
575
546
498
600 C
550 C
1000
630
Materials
synthesized at
lower
temperatures
show:
•better cathode
performance
•higher
average Mn
(IV) oxidation
states
800
600
400
ppm
200
0
Understanding the 6Li NMR Shifts:
Use of Model Mn III/IV compounds
MnO


LiMn3O4

II
I
(Mn3.5+)
LiMn2O4
(Mn3+) LiMnO2
Li5Mn4O9
Li7Mn5O12


Li2MnO3
(Mn4+)

ion
Insert
Mn3O4
l -Mn2O3
ction
Extra

 -MnO2
(Mn4+)



Li2Mn4O9 (Mn4+)
Li4Mn5O12 (Mn4+)
847
Lithium
“Hyperfine Shift
Scale”
1980
Litet[Li1/3Mn5/3]octO4
687
•use shift as a “fingerprint” of different
Li local environments
Li2Mn4O9
Mn3+, Mn4+
K2NiF4 cmpds
4000
2000
0
-2000
Mn3+ rock salts
ppm
Li in Mn4+ layers
Li in Li layers
Mn4+ layered materials
Mn3.5+ spinel
Tet Li
Oct Li
2500
2000
Mn4+ spinel
Tet Li
1500
1000
500
0
-500 ppm
The Fermi contact shift depends on the
coordination environment & nature of the
orbital overlap
Compound Site No. of Li-O-M
Bonds
Li
12
Li44Mn
Mn55OO1212 8a
16d
12
(16c)
(6)
(12)
Li22Mn
Mn42O
Li
O99
Li2MnO3
8a
4h
Li2MnO3
2b
2c
12
4
8
12
4
8
Bond Mn oxidn NMR Shift
Angle State
121.0
4
847
96.2±0.
1980
3
(171.9)
(89.6)
120.7
4
687
180
4
905/850
90
90
1817/1770
180
922/875
90
Mn
Li
16c
O
We can rationalize the shifts by using the same
approach used to analyze magnetic couplings
between spins:
“Goodenough
- Kanamori
Rules”
90º
Bo
t2g
Li
t2g
Spinel:
+ve shift: ~ 300 ppm
Li
122º
dz2
180º
-ve shift:
~ -100 ppm
Transfer to an empty dz2: spins align with e- in
t2g orbitals to maximize exchange interaction
12 x 122º
12 x ~ 64 ppm
DFT calculations can be used to help understand
the causes of these hyperfine shifts, by calculating
the unpaired e- spin density on Li
Layered
LiCr
LiCr Co
O 0.1Co1.9O2
Cr3+, d3; isoelectronic with Mn4+
1/8
LiCoO2
90°
interaction
180°Li
180 deg
interaction
interaction
7/8
8
550.0
475.0
400.0
325.0
250.0
175.0
100.0
25.00
6.250
0
-5.000
-10.00
-20.00
-50.00
-125.0
-200.0
-275.0
Co
Li
6
Li
Co
4
Li
0
-ve
spin density
= -ve shift
90 deg
interaction
90°
interaction
2
2
Dany Carlier
Gerd Ceder
Michel Menetrier
+ve
2
Co
10
Cr
Cr
t2g
4
Co
O
6
8
10
Plot spin densities of:
I.e., unpaired e- density
We can now use this hyperfine shift
scale to investigate Li local structure
in cathode materials
EXAMPLE 1: The effect of cation doping on the Spinel Structure
 E.g. I, Spinels: MnO2 (Mn4+) - + Li -> LiMn2O4 - + Li --> Li2Mn2O4 (Mn3+)
One source of capacity fade comes from the Jahn Teller Distortion that occurs in
the discharged state (for Mn oxidation states of 3.5 and less).
 Expansion and contraction of the unit cell can cause grains to lose contact with
each other and the carbon in cathode =>suggested to be responsible for severe
drop in capacity
Local
Macroscopic
Mn3+
 Potential solution: Cation doping is used to raise the average manganese oxidation
state of Mn at the end of discharge to above 3.5
Li - Excess Spinels
E.g., Li[Li0.05Mn1.95]O4 = Li[Li0.05Mn4+0.25Mn3.51.70]O4
5 “Mn3.5+” ions oxidized per Li+ dopant : Average oxidation state = Mn3.56+
Tet
Li
Oct Li
(spinel)
*
*
2500
2000
1500
639
674
2300
583
550
Increasing
oxidation
state
Mn3.5+
LiMn2O4
*
1000
500
ppm
0
*
-500
-1000
0
The Additional Resonances are due to Mn4+
sites near tet Li
Eg energy levels perturbed near the
defect sites  localization of e- holes
Mn3+ Mn4+ Mn3+ Mn4+
eg
t2g
Oct
Li
Mn
(IV)
Defect
639
583
550
Li Oct Site
1000
3.5+)
4+
Li(OMn
x
12-x(OMn )x
674
*
•Mn4+-rich regions
are created near
Oct Li
y
500
*
0
*
-500
-1000
520
Lithium-excess materials are also formed at
low temperatures, even for “stoichiometric”
compounds
MAS NMR
45 kHz MAS
Oct. Li
513
2300
*
*
*
*
*
Mn oxidn
state
increases
Li1+xMn2-xO4
+ Mn2O3
635
630
590
588
575
554
551
546
512
498
Tet. Li
7Li
850 C
LiMn2O4
650 C
600 C
550 C
1000 800 600 400 200
x40
ppm
6Li
0
MAS NMR
9 kHz MAS
4000
3000
2000
1000
0
PPM
Li NMR can be used to determine whether
Li+ substitutes on the oct. or tet. site of
the spinel cathode materials LiM’xMn2-xO4
*
*
LiZn0.1Mn1.9O4
2500 2000 1500 1000 500
ppm
407
*
283C
Li1.05Mn1.95O4
0
*
*
*
*
*
*
-500 -1000
*
397
718
649
573
502
*
Oh
1320
499
709
2300
Oh
Td
*
*
*
*
LiNi0.1Mn1.9O4
*
674
639
583
550504
2300
*
Oh
*
*
633
562
*
*
LiCr0.1Mn1.9O4
1390
746
501
Td
•Helps explain why Zn2+ systems don’t
work so well - Zn2+ in tet site blocks Li+
diffusion (J. -S. Kim.. M. M. Thackeray,
JES, ‘03)
•Concentration of “extra” peaks related to
oxidation state of dopant metal
*
*
*
*
2000 1500 1000 500
ppm
*
0 -500-1000
Following the
Electrochemical Process
Li1.05Mn1.95O4 at 4V
Swagelock-type Cell
End Bolt
4.6
4.4
Copper Plunger
4.2
Swagelock
Li Anode
Filter Paper
Polypropylene
Separator
LiMn2O4
Cathode
End Bolt
Potential (V)
Battery Cycler
Charging
4.0
Disharging
3.8
3.6
3.4
3.2
0
20
40
60
80
Capacity (mAh/g)
100
Capacity (mAhg-1)
120
140
99 %
Li1.05Mn1.95O4:
75 %
1st Charge
50 %
635
534
556
797
803
710
708
627
616
618
592
2300
Oct Li
485
25 %
Increasing Mn
oxidation state
 Li+ ions are removed
from local
environments
containing
progressively more
Mn4+ ions as the
charging proceeds
10 %
2500
2000
 Li+- ion mobility
increases - particularly
in 1st 1/2 of charge
637
2100
603 557
525
488
 Li remains in the oct.
site throughout
1500
1000
0%
500
ppm
0
-500
-1000
Loss of structure => Mobility
Li+ (+ e- ?) jump rates > 2 kHz
Slow exchange
k << f
25%
485
f (Hz)
Intermediate
regime
k same order as f
557
10%
Fast exchange
k >> f
1000
500
ppm
0
-500
Local Structure in Doped Spinels
Random cation (Li+) doping on the octahedral
site will:
Lix [Li0.05Mn1.95]OhO4
1. Create Mn4+ ions nearby the dopant cations
2. Prevent Li+ removal from sites near
octahedral Li+
3. Create a 3D framework with a distribution of
charges from 1 to 4+
All these factors help prevent the formation of a
series of phases with long range lithium-ion
ordering during cycling
=> Helps improve the capacity retention?
Panasonic
Tet. Li
Oct. Li
Example II: Ni2+/Mn4+ Layered Cathodes,
Li[NixMn(2-x)/3Li(1-2x)/3]O2: A Combined XAS,
Diffraction, DFT and NMR Study
LiNiO2
Li[Li1/3Mn2/3]O2
x = 1/2
1/3
1/10
 These compounds may be viewed as
solid solutions between Li2Mn4+O3
(Li(Li0.33Mn0.67)O2) and Li(NiMn4+)0.5O2
Li(Ni1/2Mn1/2)O2
0
V
LiCoO2
LiMnO2
Ni, Mn
Li
Capacity:
mAhg-1
DATA From: Z. Lu, D. D. MacNeil, J. R. Dahn,
ESSL4, (2001) A191-A194.
Li(NiMn)0.5O2 (T. Ohzuku, J. Dahn)
Isostructural with LiCoO2
Oxidation/reduction process involves
multiple electrons (Ni2+ -> Ni4+ ???)
Approx. 200 mAhg-1 reversible capacity
can be obtained
 Similar capacities obtained
by other groups for:
 x = 1/2 (T. Ohzuka)
 Li excess materials
(C. Johnson and M. M.
Thackeray; Z. Lu and J.
Dahn)
20 cycles C/20
LiMn0.5Ni0.5O2
 Is the redox active metal
Ni2+?
 How do these systems
function?
 How does local structure
effect electrochemical
performance?
a. Local Structure:
Li[Li1/3Mn2/3]O2
•Li sites in Li and
Mn layers readily
resolved
Mn
Li
90
Li in lithium layers
Mn/Cr
Mn
y
*
z
MAS speed: 36 kHz
*
3000
*
2000
*
*
1000
0
-1000
PPM
Ni doped compounds show similar
spectra
(d ) no unpaired t e
Ni2+
8
2g
-
Li[Li1/9Mn5/9Ni3/9]O2
Li
Li2MnO3
shifts
90
Mn/Ni
Mn/Cr
Mn/Cr
Mn/Ni
90
Li in the Li layers
Li
y
180
z
Li in the Ni/Mn
layers near
Mn4+
4000
Mn/Cr
Mn:
6 5
z
x
2000
0
y
-2000
ppm
•Li2MnO3-type local environments still observed
Li
“Li[Ni0.5Mn0.5]O2”
•6Li NMR shows that there are still Li ions in the Ni/Mn layers, near Mn4+ even
tho’ these are not predicted based on the stoichiometry
Ni/Mn
Li2MnO3
shifts
Li
Li
Li+[Ni2+0.5Mn4+0.5]O2 -->
90
Mn/Cr
Mn/Ni
Li1-xNix[Mn0.5Ni0.5-xLix]O2
y
z
Li in the Ni/Mn
layers near
Mn4+
4000
2000
•This is consistent with
Ni2+
0
ppm
-2000
ppm
substitution in the Li layers (c.f. Dahn 2002)
Neutron Diffraction (ISIS, GEM) are
consistent with Ni2+/Li+ exchange
model
Bank 4
7Li(Ni
 Extra peaks: superstructure
0.5Mn0.5)O2
Rwp = 6.32%
006
104
10-2
110
10-8
009
101
peaks as seen for honeycomb ordering of Li2MnO3 (J.
Solid-State Chem. August
2005)
 Exchange of Ni and Li ions
between the TM and Li
layers: 8 ± 2 % of site
exchange: consistent with
NMR
Li
Li/Mn
layer in
Li2MnO3
Mn
776
Local structure as a function of
Ni content?
Mn/Cr
Mn/Ni
Li[Ni1/10Mn19/30Li8/30]O2
1560
90
1365
Li
Li in the Li
layers
567
Li2MnO3
shifts
x = 1/10
y
z
•Li in Ni/Mn layers is
predominantly near Mn4+
and not Ni2+
for all compositions
•What does this tell us
about the cation ordering in
the Ni/Mn layers?
Li[Ni1/3Mn5/9Li1/9]O2
x = 1/3
LiNi0.5Mn0.5O2
Li in the Ni/Mn
layers near Mn4+
4000
3000
2000
x = 1/2
1000
0
-1000
-2000
-3000
A Model for Cation Ordering in the
Ni/Mn layers
Ni content
Local environment
Probability
(I) Random
x = 1/10
Li(OMn)6
Li(OMn)5(ONi)
Li(OMn)5(OLi)
Li(OMn)4(ONi)2
Li(OMn)4(ONi)(OLi)
Li(OMn)4(OLi)2
Expt.
(II) Ni for Li
0.06
0.06
0.16
0.02
0.13
0.17
.74
.23
.0
.03
.0
.0
.73
.27
negligible
Average charge of cation in layer = 3+ (c.f. LiCoO2)
Li2MnO3: Li+ : 2Mn4+
Honeycomb ordering
minimizes Li - Li contacts
How can we minimize Ni2+ - Li+ contacts?
Mn
A Model for Cation Ordering in the
Ni/Mn layers
Ni content
Local environment
Probability
(I) Random
x = 1/10
Li(OMn)6
Li(OMn)5(ONi)
Li(OMn)5(OLi)
Li(OMn)4(ONi)2
Li(OMn)4(ONi)(OLi)
Li(OMn)4(OLi)2
(II) Ni for Li
0.06
0.06
0.16
0.02
0.13
0.17
.74
.23
.0
.03
.0
.0
Replace on 1 Mn4+
and 2 Li+ sites by
3 Ni2+
Mn
Expt.
Ni
.73
.27
negligible
A Model for Cation Ordering:
High Ni content
Ni content
Local environment
X = 1/3
Li(OMn)6
Li(OMn)5(ONi)
Li(OMn)4(ONi)2
 III Li/Ni avoidance
Li
Mn
Probability
Model II. Ni for Li
.34 - .6
.6 - .4
.05 - .0
Maximize
Li-Ni
avoidance
*
Ni
Expt.
.55
.45
negligible
A Model for Cation Ordering:
High Ni content: Ni = 1/2; 9 % Li in
Ni/Mn Layer
Ni content
Local environment
X = 1/2
Li(OMn)6
Li(OMn)5(ONi)
Li(OMn)4(ONi)2
Li(OMn)3(ONi)3
Li
Probability
Model II. Ni for Li
.18
.36
.30
0.13
Ni/Mn ?
ordering
III Li/Ni avoidance
Ni
C.f. Lu, Chen and Dahn, Chem Mat ‘03
Ni
Expt.
.35
.65
negligible
0
Using the total scattering to obtain local ordering: Pair
Distribution Analysis (PDF) of neutron diffraction data
Rwp =
6.32%
•Uses total scattering: Bragg and Diffuse
•Provides local structure
to give g(r) - radial
distribution function
Fourier Transform
Ni,Li-O:
2.07 Å
40
Li(Ni0.5Mn0.5)O2
Coherent scattering
lengths (fm):
Ni: 10.30; 62Ni: -8.70
7Li: -2.22; 6Li: 2.00
Mn: -3.75
7Li ZERONi
6Li
30
6Li-O
M-M
20
Use of different isotopes
allows different M-M and MO contacts to be separated
7Li
10
7Li-O
0
0
1
2
3
4
5
6
7
8
-10
-20
-30
-40
Mn-O: 1.95 Å
gC(r) =
r (Å)
 bi b j

1


(
r

r
)
r

ij   4
r i j  b 2



p r0
9
REVERSE MONTE CARLO
SIMULATIONS:
•Generate giant cell
•Randomly swap atoms in ab planes
•If swap improves fit of PDF, swap is allowed
Random
R-3m space group 12*12*2 cluster
Before RMC
After RMC
%Ni-Ni pairs
16.1% (418 pairs)
11.8% (306 pairs)
%Ni-Mn pairs
42.4% (1099 pairs)
50.2% (1301 pairs)
%Mn-Mn pairs
24.2% (626 pairs)
20.7% (536 pairs)
%Li-Mn pairs
9.1% (235 pairs)
8.8% (227 pairs)
%Li-Ni pairs
7.8% (201 pairs)
8.1% (209 pairs)
%Li-Li pairs
0.50% (13 pairs)
0.5% (13 pairs)
Total
100% (2592 pairs)
100% (2592 pairs)
Correlation
cNiMn*
-0.04
-0.23 (± 0.04)
Correlation cLiNi*
-0.09
-0.16 (± 0.1)
Correlation
cLiMn*
-0.06
-0.06 (± 0.1)
RMC
After RMC
Ni/Mn Ordering ?
Honey comb
NMR; Yoon et al.
TEM; Meng et al
ESSL 2003/2004
Flower: A. van der Ven,
G. Ceder, Electrochem.
Commun. 2004
Li
Mn
Ni
Mn ring
Ni ring
Ni
Test the PDF Data Against
Different Structural Models
Ni-Ni
1st coordination shell (%)
2nd coordination shell (%)
Ordering
schemes
Ordering
schemes
10*10*2
10*10*2
Order- Order- Rand
ed
ed
-om
flower Honey
-comb
After
RMC
Order- Ordere Randed
d
om
flower Honeycomb
After
RMC
16.67
13.75
16.4
12.0
16.67
20.50
16.9
21.1
50.00
47.50
42.0
50.5
33.33
25.00
41.1
33.5
16.67
18.75
23.2
19.1
33.33
37.50
23.7
27.2
pairs
Ni-Mn
pairs
Mn-Mn
pairs
Flower: Carlier, A. van der Ven and Ceder
Can we combine all the experimental
data to provide more constraints all
the on ordering?
EXAFS - Mn near 1-2 Li and 4-5 Mn/Ni (Yoon et al.)
PDF - Ni near < 3 Ni; Ni near 3 Mn
Li
III Li/Ni
avoidance
IV Constrain NiNi contacts
Ni
Ni
NMR
Simple honey comb implies
A. Too few Li(OMn6) sites
B. Too many Li(OMn4)(ONi)2 sites
Disordered flower
motifs emerge..
•Large coulombic energy associated with Li+ /Mn4+
“honey-comb ordering” drives Li substitution in
Ni/Mn layer and Ni substitution in Li layer
•difficult to make a pure layered material directly
How does Li ordering and Li/Ni exchange affect
electrochemical performance?
b. Electrochemistry:
The Redox-active metal?
Battery
cycler
+
/ V vs. Li/Li
Potential
Voltage Profile for
LiNi0.5Mn0.5O2
5
4
3
2
1
0
100
200
300
Capacity / mAhg
-1
400
In situ XAS experiments show that
the Ni is oxidized on charging
1.8
1.2
K-edge
Li2MnO3
1.5
Normalized Intensity (a. u.)
Normalized Intensity (a. u.)
(a)
Mn
pristine
scan 3
scan 6
scan 9
scan 12
Li2MnO3
0.6
Charging
K-edge
LiNiO2
0.9
pristine
scan 3
scan 6
scan 9
scan 12
scan 14
LiNiO2
0.6
0.3
0.0
0.0
6.535
1.2
Ni
(b)
6.540
6.545
6.550
6.555
Energy / KeV
6.560
6.565
6.570
8.33
8.34
8.35
8.36
Energy / KeV
With X. Yang, and J. McBreen, BNL
Which Li sites are involved in the
electrochemical processes?
•Li NMR results show that Li is
removed from the Li layers and
the Ni/Mn layers
Li0.4Ni05XMn
05O2
= 0.6
CHARGE
Li0.7Ni05XMn
= 0.3
05O2
Mn/Ni/Li
Mn/Cr
Li0.9NiX05=Mn
0.105O2
Li
Mn/Ni
180
Mn/Ni
Mn/Cr
layer
594
Li
711
90
Li Layer
z
y
Sites in the transition metal
layers are always thought to
remain stuck in the layers
1486
1370
x
4000
2000
LiNi0.5XMn
0.5O2
= 0.0
0
ppm
-2000
-4000
NMR of Li[Li1/9Ni1/3Mn5/9]O2 shows that the Li in the
T.M. layers is removed on charging, but it returns on
discharging
CHARGE
x=0.1
231 mAh/g
Li is not inert
x=0.4
154 mAh/g
End of 1st cycle, discharge to 2.5 v
L[Ni1/3Li1/9Mn5/9]O2 powder
after first cycle between 4.8 and 2.5 V
77 mAh/g
x=0.7
Li in
Ni/Mn
layers
Pristine vs. end of
1st charge cycle
1498
1324
590
733
x=0.9
24 mAh/g
x=1
pristine
4000
4000
2000
0
-2000
-4000
3000
2000
1000
0
ppm
-1000
-2000
-3000
Calculations* were used to understand
the mechanism for Li extraction out of
the Ni/Mn layers
How and why does this happen?
Oxygen
Li Layer
Metal Layer
Li in metal Layer
*J. Reed, A. van der Ven
and Gerd Ceder
Spontaneous migration
into Li layer
Oxygen
Li Layer
Metal Layer
Extended Cycling or Cycling to High Voltage: Li
does not return to the transition metal layer
pristine
Li Layer
after 20 cycles
To 4.6V
Mn/
Ni
layer
*
*
2000
1000
ppm
0
-1000
To 5.3 V
What exactly is going on?
4.6V
pristine
5.3
V
Ni, Mn, Li
Li, Ni
Charge to high V
Ni, Mn
Li
Conclusions
NMR spectra are very sensitive to local structure and
electronic structure.
Nexelion
SONY’s new LiIB:
•The spectra can be rationalized by considering the overlap
between the Li, O and Mn (paramagnet) orbitals
•This has been used to spinels to:
•Locate Li on oct vs. tet sites
•Examine effect of doping on electronic structure
•Follow the electrochemical process
• and in Li[NixLi1/3-2x/3Mn2/3-x/3]O2 to:
LiCoO2 + CoNiMn
layered oxide hybrid
cathode
+ C/Sn hybrid anode
33% more capacity
•Determine occupancies for Li in the transition metal layers
•Derive a model for cation ordering in the transition metal layers
•Follow the changes that occur on cycling the battery
•(The Lis in the T.M. layers can participate in the
electrochemical process; --> mechanism for increasing
cation disorder following multiple charge cycles)
Ni
Part II:
NMR Studies of
Oxygen Ion Motion
Why is Anionic Mobility
Important?
The SOFC O2/H2 Cell
Porous anode
Doped ZrO2
Electrolyte
Porous
Cathode
-for stationary power
O2 + 2e- -> 2O2*Requires high temperatures (1000 oC
H2 ->2H++2e-
for thick electrolytes)
O2-
Interconnection
Porous Support Tube
Air electrode
electrolyte
H2
H2 O
Fuel electrode
e-
O2 (air)
Air Flow
Fuel Flow
Other applications that require
anionic conduction include..
O2 Sensors
•
Lambda O2 sensor for O2
detection in cars
• (regulate gas/O2 ratios)
• -again uses YSZ
• -high temp (500 ºC) required
for rapid response
O2/N2 Separations
Problems with SOFCs
 High operating temperatures
 Expensive - particularly due to materials
required for high temp. operation
(Ceramic vs. stainless)
 Stability of materials at high
temperatures
 Degradation of seals
Solutions ----> Improve catalyst in
electrodes (particularly on cathode side)
Improve ionic conductivity of electrolyte
Increase surface area of 3-phase
boundary
Use thinner electrolytes
cathode
O2
O2
O2
<- e-
LSM
O2-
O2-
O2-
Electrolyte (YSZ)
Improving Ionic
Conductivity
 Investigate known alternative electrolytes
LSGM - alternative to YSZ for “low temperature” (700 ºC) operations
Sr2+ and Mg2+ doped LaGaO3
La (Sr) =A
Ga
(Mg)
YSZ
=B
O=X
J. Goodenough
 Improve our understanding of mechanisms for ionic conduction - better
electrolytes?
How Do Oxygen Anions
Move in Solids?
Vacancy mechanism
Interstitial Mechanism
Major
mech. for
O2-
Vacancies/interstitials are created in a solid by cation doping:
e.g. Y3+ - doped ZrO2
YxZr1-xO2-x/2
c.f., Y3+ - doped CaF2
YxCa1-xF2+x
x/2
interstitial F-
So what’s the problem?
 Large activation barriers, due to the 2- charge on the anion
=> need high T.
 Since the vacancy is produced by adding a lower valent
cation….
e.g., Y3+-ZrO2
2-
3+
2-
3+
4+
4+
…vacancy-dopant clustering
can occur at low temperatures
Understanding the conduction
process: Why NMR?
 Many anionic conductors show considerable disorder on the anion sub-lattice.
- diffraction studies do not necessarily provide an accurate model for the local
structure.
 However… the nature of the defects and local order controls the mechanisms of
conduction in these materials.
 If we can observe the different oxygen sites in the structure by NMR, then we
can determine which oxygen anions are responsible for the anionic conductivity.
3+
24+
The BIMEVOX Family (Bi4V2-xMxO10-x/2) - some of
the highest oxide conductors discovered to date

Max. temp.
reached by
standard
commercial
probes
 -Bi4V1.7Ti0.3O10.85

a
Bi4V2O11
Conductivity of selected oxide materials, after Kendell et al. Chem
Mat, ‘96, and Yan and Greenblatt, SSI ‘95.
The 17O chemical shift is very sensitive to the local
environment, so it should be possible to determine
mobilities for different sites…. e.g., Y2(SnyTi1-y)2O7
OTi2Y2
Y2Ti2O7
OY4
•Determine the Sn/Ti
ratio
17O
I = 5/2
17O
OSnTiY2
I = 5/2
Y2(Sn0.4Ti0.6)2O7
OSn2Y2
Y2Sn2O7
Y2Sn2O7
NMR Studies of Motion: Well established for
systems where motion approaches the NMR
(chemical shift) timescale near R.T.: E.g., fluorides
Temp.
(ºC)
R.t.
Two Site Exchange
a-PbF2
kr(1-2)
F(1 )----------------> F(2)
<--------------
kr(2-1)
120
tc =
2x10-5s
160
2x10-4s
200
Alpha-PbF2
250
2x10-3s
The conductivity may be dramatically enhanced by
substituting the M6+ cations by M5+ cations, either
stoichiometrically or by doping
17O
MAS NMR
8.4 T
Bi2O22+ layers
VO5
e.g., aBi4V2O11
= Bi2VO5.5 0.5
0.5 “octaheda”
c.f., Bi217O3
 = 195 ppm
615
(S. Yang et al. JACS 1989)
830
WO6
Nb5+-doped Bi2WO6
s = 10
With R. -N. Vannier
Lille
vs. 10-3 Scm-1
at 500 °C (Nb)
-5
Baux et al. SSI 1996
1000
500
0
-500
PPM
17O
17O
enrichment is straightforward for
moderate conductors
Bi2O22+
layers
MAS NMR
8.4 T
s = 10
WO6
•Heat in
600 °C
Bi2W0.9Nb0.1O5.95
-5
vs. 10-3 Scm-1 at 500 °C
(Nb)
Baux et al. SSI 1996
17O
2
gas at
Bi2WO6
X5
1000
800
600
400
200
0
-200
PPM
Since we can resolve the different sites by NMR
we can then determine by NMR which oxide ions
are responsible for conductivity
Two site exchange
Bi2O22+ layers
250oC
225oC
200oC
Coalescence => 1/tc = 7000 s-1
•Motion occurs via anion
vacancy jumps between
the axial and equatorial
WO6 oxygen atoms
150oC
RT
500
400
300
200
PPM
•Oxide ions in the Bi2O22+
layers are rigid
Increasing the doping level increases the
mobility, consistent with an anion vacancy
model
t = 2 x 104 s
5%
Nb5+
T
t = 3 x 105 s
in contrast to
earlier literature
reports..
10%
Nb5+
250oC
200oC
150oC
100oC
RT
500
400
PPM
500
400
PPM
The NMR results are consistent with new
conductivity data (on the same samples)
1000/T (K-1)
0
-1
-1
log s (S cm )
-2
-3
0.5
1
1.5
2
2.5
o Bi2WO6
D Bi2W0.95Nb0.05O6
•Bi2W0.9Nb0.1O6
-4
-5
-6
-7
-8
NMR measurements
Probing much slower motion:
2-Dimensional NMR methods
Preparation
p/2
• Probe motion
that occurs
during the
“mixing time”
Detection
p/2
Evolution
t1
Mixing
p/2
t2
B
A
A-B
A
F1
B-A
F2
B
2-D NMR methods can be used to detect the
slower jumps between the Bi2O22+ and WO6 layers
Room temperature
tm=50ms
Bi2O2 layer
W-O-Bi
W-O-W
175oC
tm=5ms
225oC,
tm=5ms
tc for W-O ax and eq
exchange ~ 10 ms,
250oC
while tc for
tm=5ms W-O/ Bi-O exchange
~ 10 ms
Variable Temperature 17O MAS NMR spectra of a
Bi4V2O11can again be used to determine directly
which sites are mobile
260
17O
MAS NMR
700
250oC
220oC
600 MHz
data
V-O-V
180oC
*
V-O-Bi
140oC
900
700
100oC
PPM
•Both V-O oxygen
sites are involved in
the motion.
*
615
30oC
2000
830 *
*
1000
900
0
-1000
700
PPM
•The resonances at
615ppm and 830ppm
gradually merge
 oxygen motion in
the vanadium oxide
layer increases as the
temperature
increased
PPM
Now 51V MAS NMR may be used as a 2nd probe of
structure (& motion) in this more complex material
Abrahams, I. J. Mat. Chem. 1998. Hardcastle, F. D. JSSC 1991. Delmaire, F. PCCP 2000
1:1
Idealized Structure of the
high temperature  phase
51V
MAS NMR
40
kHz
-440
-510
5-coord tet V:II
(distorted)
Structure refinements*:
-430
: 2 tet, 2 penta coord V sites
a: 3 tet, 1 penta coord V sites
* tet
*
sym.
*
*
15
kHz
*
*R.N. Vannier, G. Mairesse et al.
0
-200
-400
-600
-800
PPM
17O/51V
TRAPDOR NMR of Bi4V2O11confirms that the
resonances at 610 and 820 ppm are near vanadium.
Use to detect motion between the same sites
17O/51V
TRAPDOR at 10kHz spinning
17O
17
O
265
180o
90o
t = ntR
No irradiation
610
820
51V
51V
51V
irradiation
for 800msec
(rf = 100 kHz)
Difference
2000
1000
0
-1000 PPM
•The TRAPDOR
effect decrease
substantially at 180
ºC as these sites
become mobile
tc ~ 1 ms
t
Motion is seen at room temperature for the fastest
anionic conductor Bi4V1.7Ti0.3O10.85!
17O; 18
kHz
Oxygen in the Bi-O layer
17O; 9
260
kHz
V-O layer
670
250oC
0oC
180oC
-100oC
•No TRAPDOR
effect until -100 oC
100oC
252
-140oC
677
30oC
1000
0
PPM
1000
0
PPM
LaGaO3: 71Ga and 17O Can be Used
to Probe Local Structure
71Ga
NMR
17O
NMR:
No evidence for O mobility
Rt
100C
Pnma
150C
R3c
200C
RT
250C
80
200
C
70
60
50
40
30
20PPM
250
200
150
100
17O
(and 71Ga) are quadrupolar nuclei;
Large QCCs imply anisotropic
environments for O
mI
-5/2
Vzz
0+2Q’
0
-3/2
0
0+Q’
-1/2
0 Central
Transition
+1/2
0
0
Central
Transition
Central
Transition
0-Q’
+3/2
0
0-2Q’
+5/2
HZ
HQ(1)
HQ(2)
2nd order
quadrupolar
lineshape extract QCC
(3.1 MHz
eta close to 0)
Cation Doping Increases O
Mobility, as Seen by 17O NMR
250C
10% Sr
200C
10%Mg2+
Collapse of 2nd
order lineshape
=> hops
between sites
Vzz
150C
250C
100C
Mg-O-Ga
Ambient
350 300 250 200 150 100 50
rt
PPM
300
250
200
150
Summary
•Local probes such as NMR and PDF analysis can be used to examine both local and
long-range motion in perovskite-like materials
•We have now developed a number of approaches that allow us to follow oxide ion
mobility in solids at moderate temperatures
1. Normal VT methods (for more mobile systems), for hops between different sites
2. Two dimensional NMR experiments allow us to study much slower motions
(up to correlation times of approx 100 ms)
•These methods yield detailed information concerning which and how oxide anions
move in solids
This approach has been used, e.g, in Nb5+-doped Bi2WO6 to show that:
•motion involves jumps between the two sites in the WO62- layers
•Motion in 3D occurs with a much longer correlation time
•Methods are being applied to materials with lower conductivity such as doped
LaGaO3
Acknowledgements
CEMS
NSF, DOE (Office of
FreedomCAR) and Gillette
Nicolas Dupre (CNRS, Nantes)
Wonsub Younkee
Stony Brook
Paik
Young Joo Lee (LG Yoon
(BNL)
Chemicals)
Namjum Kim
BNL
(Stanford)
Xing Yang
James McBreen
Lille
Rose-Noelle Vannier
John Palumbo
Julien Breger
Meng Jiang
MIT
Gerd Ceder
Dany Carlier (Bordeaux); Anton van der Ven
(U. Michigan); John Gorman; John Reed;
Shirley Meng; Kisuk Kang
Yang Shao-Horn