suementary inrmatin · 2014-11-20 · 1 eastchem, school of chemistry, university of st andrews,...

12
NATURE CHEMISTRY | www.nature.com/naturechemistry 1 SUPPLEMENTARY INFORMATION DOI: 10.1038/NCHEM.2101 Lee Johnson 1,6‡ , Chunmei Li 1,2‡ , Zheng Liu 1,6 , Yuhui Chen 1,6 , Stefan A. Freunberger 3 , Praveen C. Ashok 4 , Bavishna B. Praveen 4 , Kishan Dholakia 4 , Jean-Marie Tarascon 5 , Peter G. Bruce 6* 1 EastChem, School of Chemistry, University of St Andrews, North Haugh, St Andrews, Fife, KY16 9ST, UK 2 Laboratoire de Réactivité et Chimie des Solides — UMR CNRS 6007, 33 rue Saint-Leu, 80039 Amiens Cedex, France 3 Christian Doppler Laboratory for Lithium Batteries, and Institute for Chemistry and Technology of Materials, Graz University of Technology, Stremayrgasse 9, 8010 Graz, Austria 4 SUPA, School of Physics & Astronomy, University of St Andrews, North Haugh, St Andrews, Fife, KY16 9SS, UK 5 Collège de France, 11 place Marcelin Berthelot, 75231 Paris Cedex, France 6 Departments of Materials and Chemistry, University of Oxford, UK Table of Contents Materials and methods .......................................................................................................................................... 2 Stability of electrolyte solutions ............................................................................................................................. 3 Determination of donor number using NMR ......................................................................................................... 3 Figure S1 ................................................................................................................................................................. 4 Table S1 .................................................................................................................................................................. 4 Simulation of voltammograms of O 2 reduction obtained in DMSO and MeIm..................................................... 4 Figure S2 ................................................................................................................................................................. 5 Table S2 .................................................................................................................................................................. 6 Figure S3 ................................................................................................................................................................. 6 Shifts in the E 0 for O 2 /O 2 on replacing TBA + with Li + in MeIm and DMSO ............................................................ 6 UVvis analysis of O 2 disproportionation ............................................................................................................... 7 Figure S4 ................................................................................................................................................................. 7 Figure S5 ................................................................................................................................................................. 8 Figure S6 ................................................................................................................................................................. 8 Hydrodynamic voltammograms for O 2 reduction in the presence of Li + ................................................................ 8 Figure S7 ................................................................................................................................................................. 9 Estimating the free energy of equation (1) ............................................................................................................ 9 Calculation of capacity.......................................................................................................................................... 10 Influence of the solvent on the morphologies of Li 2 O 2 at high and low voltages ................................................ 10 References ............................................................................................................................................................ 12 and its consequences for LiO 2 batteries The role of LiO 2 solubility in O 2 reduction in aprotic solvents © 2014 Macmillan Publishers Limited. All rights reserved.

Upload: others

Post on 09-Jul-2020

1 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: SUEMENTARY INRMATIN · 2014-11-20 · 1 EastChem, School of Chemistry, University of St Andrews, North Haugh, St Andrews, Fife, KY16 9ST, UK 2 Laboratoire de Réactivité et Chimie

NATURE CHEMISTRY | www.nature.com/naturechemistry 1

SUPPLEMENTARY INFORMATIONDOI: 10.1038/NCHEM.2101

1    

A  unified  mechanism  of  O2  reduction  in  aprotic  Li+-­‐  electrolytes  and  its  consequences  for  Li-­‐O2  batteries  

Lee Johnson1,6‡, Chunmei Li1,2‡, Zheng Liu1,6, Yuhui Chen1,6, Stefan A. Freunberger3, Praveen C.

Ashok4, Bavishna B. Praveen4, Kishan Dholakia4, Jean-Marie Tarascon5, Peter G. Bruce6*

1 EastChem, School of Chemistry, University of St Andrews, North Haugh, St Andrews,

Fife, KY16 9ST, UK

2 Laboratoire de Réactivité et Chimie des Solides — UMR CNRS 6007, 33 rue Saint-Leu, 80039 Amiens Cedex,

France

3 Christian Doppler Laboratory for Lithium Batteries, and Institute for Chemistry and Technology of Materials,

Graz University of Technology, Stremayrgasse 9, 8010 Graz, Austria

4 SUPA, School of Physics & Astronomy, University of St Andrews, North Haugh, St Andrews, Fife, KY16 9SS, UK

5 Collège de France, 11 place Marcelin Berthelot, 75231 Paris Cedex, France

6 Departments of Materials and Chemistry, University of Oxford, UK

Table  of  Contents  

Materials  and  methods  ..........................................................................................................................................  2  Stability  of  electrolyte  solutions  .............................................................................................................................  3  Determination  of  donor  number  using  NMR  .........................................................................................................  3  Figure  S1  .................................................................................................................................................................  4  Table  S1  ..................................................................................................................................................................  4  Simulation  of  voltammograms  of  O2  reduction  obtained  in  DMSO  and  Me-­‐Im  .....................................................  4  Figure  S2  .................................................................................................................................................................  5  Table  S2  ..................................................................................................................................................................  6  Figure  S3  .................................................................................................................................................................  6  Shifts  in  the  E0  for  O2/O2

-­‐  on  replacing  TBA+  with  Li+  in  Me-­‐Im  and  DMSO  ............................................................  6  UV-­‐vis  analysis  of  O2

-­‐  disproportionation  ...............................................................................................................  7  Figure  S4  .................................................................................................................................................................  7  Figure  S5  .................................................................................................................................................................  8  Figure  S6  .................................................................................................................................................................  8  Hydrodynamic  voltammograms  for  O2  reduction  in  the  presence  of  Li

+  ................................................................  8  Figure  S7  .................................................................................................................................................................  9  Estimating  the  free  energy  of  equation  (1)  ............................................................................................................  9  Calculation  of  capacity..........................................................................................................................................  10  Influence  of  the  solvent  on  the  morphologies  of  Li2O2  at  high  and  low  voltages  ................................................  10  References  ............................................................................................................................................................  12  

1  

The role of LiO2 solubility in  O2  reduction  in  aprotic solvents  

and  its  consequences  for  Li–O2  batteries  Lee Johnson1,6‡, Chunmei Li1,2‡, Zheng Liu1,6, Yuhui Chen1,6, Stefan A. Freunberger3, Praveen C.

Ashok4, Bavishna B. Praveen4, Kishan Dholakia4, Jean-Marie Tarascon5, Peter G. Bruce6*

1 EastChem, School of Chemistry, University of St Andrews, North Haugh, St Andrews,

Fife, KY16 9ST, UK

2 Laboratoire de Réactivité et Chimie des Solides — UMR CNRS 6007, 33 rue Saint-Leu, 80039 Amiens Cedex,

France

3 Christian Doppler Laboratory for Lithium Batteries, and Institute for Chemistry and Technology of Materials,

Graz University of Technology, Stremayrgasse 9, 8010 Graz, Austria

4 SUPA, School of Physics & Astronomy, University of St Andrews, North Haugh, St Andrews, Fife, KY16 9SS, UK

5 Collège de France, 11 place Marcelin Berthelot, 75231 Paris Cedex, France

6 Departments of Materials and Chemistry, University of Oxford, UK

Table  of  Contents  

Materials  and  methods  ..........................................................................................................................................  2  Stability  of  electrolyte  solutions  .............................................................................................................................  3  Determination  of  donor  number  using  NMR  .........................................................................................................  3  Figure  S1  ................................................................................................................................................................ .  4  Table  S1  ..................................................................................................................................................................  4  Simulation  of  voltammograms  of  O2  reduction  obtained  in  DMSO  and  Me-­‐Im  .....................................................  4  Figure  S2  ................................................................................................................................................................ .  5  Table  S2  ..................................................................................................................................................................  6  Figure  S3  ................................................................................................................................................................ .  6  Shifts  in  the  E0  for  O2/O2

-­‐  on  replacing  TBA+  with  Li+  in  Me-­‐Im  and  DMSO  ............................................................  6  UV-­‐vis  analysis  of  O2

-­‐  disproportionation  ...............................................................................................................  7  Figure  S4  ................................................................................................................................................................ .  7  Figure  S5  ................................................................................................................................................................ .  8  Figure  S6  ................................................................................................................................................................ .  8  Hydrodynamic  voltammograms  for  O2  reduction  in  the  presence  of  Li

+  ................................................................  8  Figure  S7  ................................................................................................................................................................ .  9  Estimating  the  free  energy  of  equation  (1)  ............................................................................................................  9  Calculation  of  capacity..........................................................................................................................................  10  Influence  of  the  solvent  on  the  morphologies  of  Li2O2  at  high  and  low  voltages  ................................................  10  References  ............................................................................................................................................................  12  

1  

The role of LiO2 solubility in  O2  reduction  in  aprotic solvents  

and  its  consequences  for  Li–O2  batteries  Lee Johnson1,6‡, Chunmei Li1,2‡, Zheng Liu1,6, Yuhui Chen1,6, Stefan A. Freunberger3, Praveen C.

Ashok4, Bavishna B. Praveen4, Kishan Dholakia4, Jean-Marie Tarascon5, Peter G. Bruce6*

1 EastChem, School of Chemistry, University of St Andrews, North Haugh, St Andrews,

Fife, KY16 9ST, UK

2 Laboratoire de Réactivité et Chimie des Solides — UMR CNRS 6007, 33 rue Saint-Leu, 80039 Amiens Cedex,

France

3 Christian Doppler Laboratory for Lithium Batteries, and Institute for Chemistry and Technology of Materials,

Graz University of Technology, Stremayrgasse 9, 8010 Graz, Austria

4 SUPA, School of Physics & Astronomy, University of St Andrews, North Haugh, St Andrews, Fife, KY16 9SS, UK

5 Collège de France, 11 place Marcelin Berthelot, 75231 Paris Cedex, France

6 Departments of Materials and Chemistry, University of Oxford, UK

Table  of  Contents  

Materials  and  methods  ..........................................................................................................................................  2  Stability  of  electrolyte  solutions  .............................................................................................................................  3  Determination  of  donor  number  using  NMR  .........................................................................................................  3  Figure  S1  ................................................................................................................................................................ .  4  Table  S1  ..................................................................................................................................................................  4  Simulation  of  voltammograms  of  O2  reduction  obtained  in  DMSO  and  Me-­‐Im  .....................................................  4  Figure  S2  ................................................................................................................................................................ .  5  Table  S2  ..................................................................................................................................................................  6  Figure  S3  ................................................................................................................................................................ .  6  Shifts  in  the  E0  for  O2/O2

-­‐  on  replacing  TBA+  with  Li+  in  Me-­‐Im  and  DMSO  ............................................................  6  UV-­‐vis  analysis  of  O2

-­‐  disproportionation  ...............................................................................................................  7  Figure  S4  ................................................................................................................................................................ .  7  Figure  S5  ................................................................................................................................................................ .  8  Figure  S6  ................................................................................................................................................................ .  8  Hydrodynamic  voltammograms  for  O2  reduction  in  the  presence  of  Li

+  ................................................................  8  Figure  S7  ................................................................................................................................................................ .  9  Estimating  the  free  energy  of  equation  (1)  ............................................................................................................  9  Calculation  of  capacity..........................................................................................................................................  10  Influence  of  the  solvent  on  the  morphologies  of  Li2O2  at  high  and  low  voltages  ................................................  10  References  ............................................................................................................................................................  12  

© 2014 Macmillan Publishers Limited. All rights reserved.

Page 2: SUEMENTARY INRMATIN · 2014-11-20 · 1 EastChem, School of Chemistry, University of St Andrews, North Haugh, St Andrews, Fife, KY16 9ST, UK 2 Laboratoire de Réactivité et Chimie

NATURE CHEMISTRY | www.nature.com/naturechemistry 2

SUPPLEMENTARY INFORMATIONDOI: 10.1038/NCHEM.2101

2    

Materials  and  methods  

Me-­‐Im  was  distilled  under  vacuum.  DMSO  was  distilled  under  vacuum  over  NaNH2.  CH3CN  and  DME  were  distilled  under  argon  over  CaH2.  All  solvents  were  further  dried  for  several  days  over  freshly  activated  molecular  sieves  (type  4  Å).  The  final  water  content  was  ≤  4  ppm  (determined   by   Karl   Fischer   titration).   Battery   grade   tetrabutylammonium   perchlorate  (TBAClO4)  and  battery  grade  lithium  perchlorate  (LiClO4)  were  dried  as  described  previously1.  High   purity   N5   O2   was   obtained   from   BOC   industrial   gasses   and   was   used   in   all  measurements.  All  materials  were  stored  in  an  argon  glove  box.    

Electrochemical  measurements  were  performed  using  a  VMP3  electrochemical  workstation  (Biologic)   and   a   multi-­‐necked,   air-­‐tight   glass   cell   within   a   glove   box.   Electrochemical  measurements  were   carried  out   at   room   temperature   and   IR   correction  was   used.   2  mm  diameter  polycrystalline  Au  disks  (BAS  Inc.)  were  employed  as  the  working  electrodes.  Prior  to  use  the  working  electrodes  were  polished  in  a  glove  box  with  0.05  μm  alumina  slurry  in  ethanol  and   rinsed  with  copious  amounts  of  ethanol   followed  by  drying  under  vacuum.  A  platinum   wire   served   as   the   counter   electrode.   Measurements   were   performed   using   a  reference   electrode   based   on   LiFePO4.   It  was   first   pre-­‐oxidized   (20%   of   total   capacity)   to  LixFePO4,   which   has   a   constant,   potential.   This   is   then   placed   in   0.1   M   LiClO4   in   DMSO  separated  from  the  bulk  electrolyte  with  a  porous  frit,  thus  providing  a  fixed  potential.  The  LixFePO4  reference  was  corrected  to  Li/Li+  by  subtracting  3.45  V.  Electrolytes  were  saturated  with  O2.   Rotating   ring/disk   electrode,   RRDE,  measurements  were   performed   using   a  MSR  rotator   and   a   removable   disk   rotating   ring/disk   electrode   containing   an   Au   disk   (5   mm  diameter)  and  a  glassy  carbon  ring  (Pine  instruments).  Measurements  were  performed  in  a  glove  box   in   a   round  bottom   flask   and  all   necks  other   than   the  neck   containing   the   shaft  were  sealed.    The  electrode  was  polished  as  described  above  between  measurements.  The  collection  efficiency  at  the  ring,  N,  in  each  experiment  was  determined  using  an  ideal  redox  couple.  For  the  polarization  measurements  in  Fig.  6,  discharge  at  a  planar  2  mm  Au  working  electrode  was  performed  at  a  constant  current  of  60  µA  cm-­‐2  in  O2  saturated  100  mM  LiClO4  solutions  of  the  various  aprotic  solvent.    

In   situ   surface   enhanced   Raman   spectroscopy   (SERS)   was   carried   out   in   an   airtight   3-­‐electrode   electrochemical   cell.   A   previously   electrochemically   roughened   Au   working  electrode   (see   reference   for   experimental   procedure)2   was   placed   behind   a   1   mm   thick  sapphire   window.   Raman   spectra   were   recorded   using   a   custom-­‐built   confocal   Raman  microscope  with  a  50x  0.45  NA  objective  (Nikon).  The  Raman  system  was  equipped  with  a  785   nm   diode   laser   (Sacher   Lasertechnik)   and   a   spectrometer   (Shamrock   SR-­‐303i,   Andor  Technology)  which  employed  a  400  lines/mm  grating,  blazed  at  850  nm  and  was  equipped  with  a  deep  depletion,  back-­‐illuminated  and  thermoelectrically  cooled  CCD  camera  (Newton,  Andor  Technology)  for  the  detection  of  Raman  signal.    

© 2014 Macmillan Publishers Limited. All rights reserved.

Page 3: SUEMENTARY INRMATIN · 2014-11-20 · 1 EastChem, School of Chemistry, University of St Andrews, North Haugh, St Andrews, Fife, KY16 9ST, UK 2 Laboratoire de Réactivité et Chimie

NATURE CHEMISTRY | www.nature.com/naturechemistry 3

SUPPLEMENTARY INFORMATIONDOI: 10.1038/NCHEM.2101

3    

Porous   composite   cathodes   for   the   study   of   solvent   dependent   Li2O2  morphologies  were  prepared  using  carbon  (Super  P,  Timcal  Ltd)  and  10  %w.t.  PTFE,  cast  onto  a  stainless  steel  mesh.  Discharge  of  the  cathodes  was  carried  out  at  constant  voltage  in  a  three  electrode  cell.  LixFePO4  was  again  used  as  the  reference  electrode.  The  morphology  of  the  pristine  carbon  electrode  and  discharged  samples  were  observed  by  FE-­‐SEM  using  a  JSM-­‐6700F  (JEOL).    

 

Stability  of  electrolyte  solutions  

Parasitic  side-­‐reactions  at  the  electrolyte/cathode  interface  in  Li-­‐O2  cells  have  been  reported  widely.   In   parallel,   many   authors   have   reported   successful   fundamental   electrochemical  studies   in   the   same   or   similar   electrolyte   solutions3,4.   Some   of   the   side-­‐reactions   are  attributable  to  decomposition  of  the  carbon  cathode,  used  in  most  studies  of  Li-­‐O2  cells,  but  not   in   the   fundamental   studies   reported   here,   which   use   gold.   Furthermore,   the   short  timescales   of   electrochemical   studies   (e.g.   CV)   compared   with   investigations   of   side-­‐reactions   on   discharge/charge   in   Li-­‐O2   cells,   means   that   the   quantity   of   side   reaction  products  is  much  less  in  the  electrochemical  studies.  This  is  reinforced  by  the  fact  that  the  electrochemical   data   presented   in   the   paper   are  well   fitted   by   O2   reduction  without   any  corrections   for   side-­‐reactions  and  no  evidence  of   side-­‐reaction  products  were  detected   in  SERS.        

 

Determination  of  donor  number  using  NMR  

As   the  DN  of  Me-­‐Im  was  unknown,   23Na  NMR  was  used   to  estimate   this   value.    Here  we  exploit  the  linear  trend  in  DN  against  23Na  NMR  shift  and  extract  the  DN  of  Me-­‐Im  from  this  trend,  Supplementary  Fig.  S15.  23Na  NMR  was  carried  out  in  a  10  mM  solution  of  NaTFSI  in  a  range   of   aprotic   solvents   (listed   in   Table   S1).   Using   these   data   the   DN   of   Me-­‐Im   was  estimated  to  be  47.  

 

© 2014 Macmillan Publishers Limited. All rights reserved.

Page 4: SUEMENTARY INRMATIN · 2014-11-20 · 1 EastChem, School of Chemistry, University of St Andrews, North Haugh, St Andrews, Fife, KY16 9ST, UK 2 Laboratoire de Réactivité et Chimie

NATURE CHEMISTRY | www.nature.com/naturechemistry 4

SUPPLEMENTARY INFORMATIONDOI: 10.1038/NCHEM.2101

4    

 

Figure  S1  |  Plot  of  23Na  NMR  shift  versus  solvent  DN  of  NaTFSI  dissolved  in  various  aprotic  solvents.  The  solvents  used  are  listed  in  Table  S1.  

 

Table  S1  |  Solvent  DN  and  the  corresponding  23Na  NMR  shift  in  each  solvent.    

  DN   23Na  NMR  shift  CH3CN   14   -­‐6.92  DME   20   -­‐5.89  DMSO   30   -­‐0.14  Pyridine   33   1.24  Me-­‐Im   47   8.84  

1,2-­‐diaminoethane   55   12.53      

 

Simulation  of  voltammograms  of  O2  reduction  obtained  in  DMSO  and  Me-­‐Im  

BASi   DigiSim   Simulation   Software   was   used   to   generate   fits   to   the   experimental  voltammograms  of  O2  reduction  in  DMSO  and  Me-­‐Im.  CVs  obtained  at  100  mV  s-­‐1  were  fit  using  a  single  E  step  and  CVs  obtained  at  5  mV  s-­‐1  were  fit  using  an  EC  mechanism,  where  the  C  step  is  irreversible.  The  resulting  fits  to  experimental  data  for  100  mM  TBA+  and  100  mM  Li+  are  shown  in  Supplementary  Fig.  S2  and  the  parameters  obtained  are  given  in  Table  S2.   As   E01   varies  with   the   concentration   of   LiClO4   (Fig.   2),   this   parameter  was   allowed   to  change  during  simulation.    

© 2014 Macmillan Publishers Limited. All rights reserved.

Page 5: SUEMENTARY INRMATIN · 2014-11-20 · 1 EastChem, School of Chemistry, University of St Andrews, North Haugh, St Andrews, Fife, KY16 9ST, UK 2 Laboratoire de Réactivité et Chimie

NATURE CHEMISTRY | www.nature.com/naturechemistry 5

SUPPLEMENTARY INFORMATIONDOI: 10.1038/NCHEM.2101

5    

Fits   to   experimental   CVs   obtained   at   5   mV   s-­‐1   were   carried   out   in   order   to   observe   the  disproportionation  of  O2

-­‐:    

2O2-­‐(sol)  +  2Li+(sol)  →  Li2O2(solid)  +  O2(sol)                     (S1)  

The   standard   heterogeneous   rate   constant,   k0,   and   the   O2   concentration,   C,   were  maintained  at  the  values  found  at  higher  scan  rates.  The  CVs  are  shown  in  Supplementary  Fig.   S3   and   the   first   order   rate   constants,   k,   for   LiO2   disproportionation,   equation   (S1),   in  Me-­‐Im  and  DMSO  with  100  mM  LiClO4  are  0.015  s-­‐1  and  0.030  s-­‐1,  respectively.    

 

 

Figure  S2  |  (circles)  CVs  and  (red  lines)  fits  for  O2  reduction  in  Me-­‐Im  and  DMSO  at  a  Au  electrode   in   100  mM   TBAClO4   and   100  mM   LiClO4.    CVs  obtained  at  100  mV   s-­‐1   and   fits  based  on  a  single  E  step.  The  resulting  variables  obtained  from  the  fits  are  given  in  Table  S2.  

 

 

 

 

© 2014 Macmillan Publishers Limited. All rights reserved.

Page 6: SUEMENTARY INRMATIN · 2014-11-20 · 1 EastChem, School of Chemistry, University of St Andrews, North Haugh, St Andrews, Fife, KY16 9ST, UK 2 Laboratoire de Réactivité et Chimie

NATURE CHEMISTRY | www.nature.com/naturechemistry 6

SUPPLEMENTARY INFORMATIONDOI: 10.1038/NCHEM.2101

6    

Table  S2  |  Variables  obtained   from   fits   to   the  CVs   shown   in   Supplementary  Fig.   S2.  D  –  diffusion   coefficient   and   k0   –standard   heterogeneous   electron   transfer   rate   constant   for  O2/O2

-­‐.  During   fitting,   the  O2   concentration,  C,  was   constant  at  1.34  mM  and   the  mean  α  used  were  0.48  and  0.44  for  DMSO  and  Me-­‐Im,  respectively.  

Electrolyte   105  x  D  /  cm2  s-­‐1   103  x  k0  /  cm  s-­‐1  Me-­‐Im  

   TBAClO4   9.49   2.72  LiClO4   10.4   2.75  

     DMSO      TBAClO4   8.6   2.34  

LiClO4   8.9   2.34    

 

 

Figure  S3  |  (circles)  CVs  and  (red  lines)  fits  for  O2  reduction  in  Me-­‐Im  and  DMSO  at  a  Au  electrode   in   100  mM   LiClO4.     CVs   obtained   at   5  mV   s-­‐1   and   the   fits   are   based   on   an   EC  mechanism.    

 

Shifts  in  the  E0  for  O2/O2-­‐  on  replacing  TBA+  with  Li+  in  Me-­‐Im  and  DMSO,  Fig  2  

The  shifts  in  the  standard  potential  for  O2/O2-­‐,  E01,  on  replacing  TBA+  with  Li+  in  Me-­‐Im  and  

DMSO,   extracted   from   the   CVs,   are   plotted   in   Fig.   2   in   the   main   paper.   For   Me-­‐Im   the  variation   is  small  and   linear,  whereas  for  DMSO  it   is   larger  and  curves  upwards,   indicating  that  the  potential  increases  more  rapidly  at  higher  Li+  concentrations.  Recognizing  that  the  main  difference  between  Me-­‐Im  and  DMSO  is  that  the  latter  has  a  lower  DN,  we  interpret  the   different   trends   as   indicating   that   in  Me-­‐Im   (high   DN),   O2

-­‐   interacts   relatively  weakly  with  Li+  cations,  e.g.  possibly  via  solvent  separated  ion  pairs  (Li+-­‐S-­‐O2

-­‐);  explaining  why  there  is   little  difference  in  the  interactions  between  O2

-­‐  and  TBA+  and  those  between  O2-­‐  and  Li+  

© 2014 Macmillan Publishers Limited. All rights reserved.

Page 7: SUEMENTARY INRMATIN · 2014-11-20 · 1 EastChem, School of Chemistry, University of St Andrews, North Haugh, St Andrews, Fife, KY16 9ST, UK 2 Laboratoire de Réactivité et Chimie

NATURE CHEMISTRY | www.nature.com/naturechemistry 7

SUPPLEMENTARY INFORMATIONDOI: 10.1038/NCHEM.2101

7    

(very  similar  CVs  from  TBA+  to  Li+,  Fig.  2).  Whereas  in  DMSO,  solvent  separated  ion  pairs  at  low  Li+  concentrations  increasingly  give  way  to  contact  ion  pairs  (Li+-­‐O2

-­‐)  or  larger  aggregates  (clusters)  as  the  concentration  increases,  resulting  in  a  greater  positive  shift  in  potential.  The  existence   of   ion   pairing   and   the   trend   towards   contact   ion   pairs   and   aggregates   with  increasing  salt  concentration  is  well  known  in  organic  electrolytes6,7.  

 

UV-­‐vis  analysis  of  O2-­‐  disproportionation  

In  order  to  determine  the  rate  of  O2-­‐  disproportionation,  equation  (S1),   in  aprotic  solvents  

containing   Li+,   the   reaction   was   followed   by   UV-­‐vis   spectroscopy   using   a   Lambda   35  spectrometer  (PerkinElmer  Instrument).  Excess  KO2  was  added  to  the  selected  solvent  and  stirred   overnight.   This   was   then   centrifuged   and   the   clear   KO2   saturated   solution   was  decanted  in  an  Ar  filled  glove  box.  2.7  ml  of  KO2  solution  was  added  to  a  sealed  cuvette.  0.3  ml  of  1  M  LiClO4  solution  was  quickly   injected  into  the  cuvette  to   initiate  the  reaction  and  precipitate  Li2O2;   the  absorbance,  A,  at  300  nm  wavelength  was  recorded  as  a   function  of  time.     The   analysis   was   performed   in   Me-­‐Im   and   DMSO.   However,   due   to   an   overlap  between   the   adsorption  peaks  of  O2

-­‐   and  Me-­‐Im  at   300  nm  a   rate   constant   could  not  be  obtained  in  this  solvent.  The  resulting  plot  of  ln(A)  against  time  is  shown  in  Supplementary  Fig.  S4,  consistent  with  a  first-­‐order  reaction.  Using  these  data  the  first  order  rate  constant  for  the  disproportionation  of  O2

-­‐  in  DMSO  was  calculated  to  be  0.07  s-­‐1.    

 

 

 

Figure   S4   |   UV-­‐vis   spectroscopy   analysis   of   O2-­‐   disproportionation   in   DMSO.   UV-­‐vis  

adsorption   by   O2-­‐   at   300   nm   recorded   using   a   saturated   solution   of   KO2   in   DMSO   after  

addition  of  100  mM  LiClO4.  Circles  –  experimental  points,  solid  line  –  linear  least  squares  fit.      

© 2014 Macmillan Publishers Limited. All rights reserved.

Page 8: SUEMENTARY INRMATIN · 2014-11-20 · 1 EastChem, School of Chemistry, University of St Andrews, North Haugh, St Andrews, Fife, KY16 9ST, UK 2 Laboratoire de Réactivité et Chimie

NATURE CHEMISTRY | www.nature.com/naturechemistry 8

SUPPLEMENTARY INFORMATIONDOI: 10.1038/NCHEM.2101

8    

 

Figure  S5  |  SERS  of  superoxide  during  O2  reduction.  In  DMSO  on  O2  reduction  the  peak  for  adsorbed  O2

-­‐  in  the  100  mM  TBA+  solution  is  unchanged  on  replacing  TBA+  by  Li+.  The  Raman  shifts  for  O2-­‐  and  LiO2  are  1110  cm-­‐1  and  1132  cm-­‐1,  respectively.      

 

Figure  S6  |  CVs  obtained  in  O2  saturated  Me-­‐Im,  DMSO  and  CH3CN  at  a  Au  electrode  in  0.5  M   LiClO4.   The   scan   rate  was   100  mV   s-­‐1.   The   2nd   reduction   peaks   have  moved   to   higher  potentials   with   higher   Li+   concentration.   The   CVs   collected   with   the   higher   voltage   limit  (black  curves)  show  that  O2

-­‐  is  still  formed  in  solution  even  at  these  high  Li+  concentrations  in  high  DN  solvents.    

 

Hydrodynamic  voltammograms  for  O2  reduction  in  the  presence  of  Li+  

RRDE  measurements  were  performed  at  a  polycrystalline  Au  disk  in  Me-­‐Im,  DMSO,  DME  and  CH3CN  and  the  resulting  polarization  curves   for  O2   reduction   in  0.1  M  LiClO4  solutions  are  shown  in  Fig.  4.  Here  we  scan  the  potential  of  the  disk  and  record  the  current  while  rotating  the  electrode  to  enhance  mass  transport.    The  ring  is  held  at  a  constant  potential  such  that  any   O2

-­‐   present   in   solution   is   oxidized   and   the   recorded   current   is   proportional   to   the  

amount  of  O2-­‐(sol)  formed  during  O2  reduction  at  the  disk.  As  the  2nd  reduction  process  forms  

solid  Li2O2  on  or  near  the  disk  surface   its  oxidation  cannot  be  observed  at   the  ring.   In   the  data  shown,  the  ring  current  is  corrected  to  account  for  the  ratio  of  product  formed  at  the  disk  to  that  collected  at  the  ring  using  an  ideal  one  electron  redox  mediator8.  The  black  lines  in  Fig.  4  show  the  current  recorded  at  the  disk,  iD  and  the  green  lines  correspond  to  the    

© 2014 Macmillan Publishers Limited. All rights reserved.

Page 9: SUEMENTARY INRMATIN · 2014-11-20 · 1 EastChem, School of Chemistry, University of St Andrews, North Haugh, St Andrews, Fife, KY16 9ST, UK 2 Laboratoire de Réactivité et Chimie

NATURE CHEMISTRY | www.nature.com/naturechemistry 9

SUPPLEMENTARY INFORMATIONDOI: 10.1038/NCHEM.2101

9    

 

Figure   S7   |   Polarization   curves   obtained   in   O2   saturated   Me-­‐Im,   DMSO,   and   CH3CN  containing  0.5  M  LiClO4.    The  RRDE  was  a  5  mm  diameter  Au  disk  with  a  GC  ring  and  the  rotation  rate  was  2000  RPM.  

 

current  detected  at  the  ring  iR  (presented  as  a  negative  current  for  easy  comparison  to  the  disk  current).  Finally,  the  red  lines  correspond  to  the  difference  (iD  -­‐  iR),  i.e.,  that  which  does  not   lead   to  O2

-­‐   in   solution;  we   assign   this   current   to   the   surface   reduction   to   form   Li2O2,  which  is  confirmed  by  in  situ  SERS.  The  RRDE  results  in  both  Me-­‐Im  and  DMSO  show  no  red  area  i.e.  (iD  –  iR)  at  high  potentials,  and  the  peak  current  was  dependent  on  the  rotation  rate  of  the  RRDE.  Both  of  these  results  are  consistent  with  the  reduction  of  O2  to  O2

-­‐  in  solution  and   not  with   formation   of   Li2O2   on   the   disk   surface.   At   lower   voltages,   the   current   ratio  between  the  ring  and  disk  began  to  drop  below  one  (red  area  grows)  and  at  slightly  more  negative  potentials  the  current  at  the  ring  began  to  decrease,  consistent  with  the  onset  of  reduction  of  O2

-­‐  to  Li2O2*  at  the  disk.  This  process  also  initiated  a  decrease  in  the  current  at  the  disk  due  to  the  Li2O2  blocking  the  electrode  surface.   It  should  be  noted  that  RRDE  is  a  steady   state   technique   and   that   this   decrease   in   current   is   not   due   to   any   diffusion  limitation.  In  the  lowest  DN  solvent,  CH3CN,  there  is  almost  no  ring  current  and  the  red  area  dominates  indicating  almost  no  O2

-­‐  in  solution.  The  disk  current  was  relatively  insensitive  to  rotation  rate  and  the  peak  current  was  considerably  lower  than  in  the  absence  of  Li+.  These  results  indicate  a  surface  confined  process  consistent  with  the  formation  of  a  Li2O2  film  on  the   electrode.   Finally,   in   DME   properties   of   both   the   high   and   low   DN   systems   were  observed.  The  magnitude  of  the  red  area  is  between  the  high  and  low  DN  solvents,  there  is  a   significant   ring   current   indicating   the   presence   of   O2

-­‐   in   solution,   similar   to   high   DN  solvents,  but  at  the  same  time  the  ring  current  never  equals  the  disk  current,  even  at  high  potentials.   In   addition,   the   current   was   noticeably   lower   than   in   the   absence   of   Li+,  confirming  a  surface  process  occurred  at  all  potentials.    

 

Estimating  the  free  energy  of  equation  (1)  

Equation  (2)  is  the  sum  of  the  two  redox  couples,  O2/O2-­‐  and  Li/Li+,  therefore,  summing  the  

two  E0   values   for   these   couples   and   recalling  ΔG0  =   -­‐nFE0,   gives   the  Gibbs   free  energy   for  

© 2014 Macmillan Publishers Limited. All rights reserved.

Page 10: SUEMENTARY INRMATIN · 2014-11-20 · 1 EastChem, School of Chemistry, University of St Andrews, North Haugh, St Andrews, Fife, KY16 9ST, UK 2 Laboratoire de Réactivité et Chimie

NATURE CHEMISTRY | www.nature.com/naturechemistry 10

SUPPLEMENTARY INFORMATIONDOI: 10.1038/NCHEM.2101

10    

equation  (2).  The  E0  values  for  the  O2/O2-­‐  couple  were  measured  in  each  of  the  four  solvents  

in   the  presence  of   the  TBA+  salt   (i.e.TBAClO4  solutions).  The  E0  values   for   the  Li/Li+  couple  were   measured   in   each   of   the   four   solvents   in   the   presence   of   ClO4

-­‐   ions   (i.e.   in   LiClO4  solutions).  The   reference  electrode  used   in  each  case   is  described   in   the  Methods  section  and   is  the  same  as  that  used  throughout  this  work.  LixFePO4  was  used   instead  of  Li  metal,  because  Li   is  not  stable  in  every  solvent.  3.45  V  was  subtracted  to  give  the  value  for  Li/Li+.  The  free  energy  of  equation  (3)  was  obtained  from  the  pressure  of  O2(g)   in  the  head  space  above  the  solution  and  the  concentration  of  O2(sol)   in  the  solvent,  which  are  in  equilibrium.  Combining  equations  (2)  and  (3)  with  (4),  for  which  the  free  energy  is  known  (-­‐221  kJ  mol-­‐1)9,  gives  an  estimate   for   the   free  energy  of  equation   (1).  The  approximations   inherent   in   the  estimation  of  ΔG0  for  equation  (1)  and  why,  despite  such  approximations,  they  still  predict  the   correct   trend   from   soluble   LiO2   in   high   DN   to   surface   LiO2*   in   low   DN   solvents,   is  discussed  in  the  paper.      

The  ΔG0    for  DME  is  25  kJ  mol-­‐1  so  the  equilibrium,  equation  (1),  lies  towards  surface  species,  however   a   concentration   of   0.4   mM   for   O2

-­‐   is   predicted   from   ΔG0   =   -­‐RTln[Li+(sol)][O2-­‐(sol)]),    

Li+  0.1  M.    

 

Calculation  of  capacity    

We   consider   an   electrode   based   on   carbon   composed   of   hexagonally   close   packed  cylindrical  pores,  which  we  optimize   for  deposition  by  a   surface  and  separately  a   solution  mechanism.  An  optimized  cathode  structure  for  a  surface  mechanism  (low  DN  solvents)  has  the  highest  possible  surface  area  commensurate  with  pores  of  sufficient  size  to  permit  mass  transport  in  the  electrolyte  within  the  pores.  The  maximum  thickness  of  a  Li2O2  layer  formed  on   the   surface   has   been   reported   as   ∼7   nm10.   Taking   a   pore   diameter   of   50   nm   as   a  minimum  requirement  for  mass  transport  and  a  wall  thickness  of  3  nm,  this  corresponds  to  a   BET   surface   area   of   170   m2   g-­‐1.   The   resulting   capacity   for   a   7   nm   Li2O2   film   is    2500  mAh  g-­‐1(C).  An  electrode  optimized  for  the  solution  growth  of  Li2O2  (high  DN  solvent)  requires   the   largest   possible   pore   volume.   Taking   a   pore   diameter   of   500   nm   for   the  cylindrical   pores   and  assuming  80  %  of   the  porous   volume  can  be   filled  with   Li2O2  during  discharge   (the   remaining   20  %   contains   the   electrolyte),   we   obtain   a   specific   capacity   of  8500  mAh  g-­‐1(C),  which  is  more  than  a  threefold  increase  in  capacity.  

 

Influence  of  the  solvent  on  the  morphologies  of  Li2O2  at  high  and  low  voltages  

For  each  of  the  solvents  studied  here  the  morphology  of  Li2O2  growth  has  been  examined  as  a  function  of  potential,  Fig.  7.  For  such  studies  a  cathode  with  a  relatively  large  surface  area  and  porosity  was  required,  so  that  both  solution  and  surface  pathways  can  contribute  and  to  allow  sufficient  Li2O2  to  form  so  that  the  morphology  of  Li2O2  may  be  readily  imaged  by  

© 2014 Macmillan Publishers Limited. All rights reserved.

Page 11: SUEMENTARY INRMATIN · 2014-11-20 · 1 EastChem, School of Chemistry, University of St Andrews, North Haugh, St Andrews, Fife, KY16 9ST, UK 2 Laboratoire de Réactivité et Chimie

NATURE CHEMISTRY | www.nature.com/naturechemistry 11

SUPPLEMENTARY INFORMATIONDOI: 10.1038/NCHEM.2101

11    

SEM.  Recognizing  that  the  vast  majority  of  morphological  studies  have  employed  carbon,  we  used  electrodes  formed  from  a  composite  of  Super  P  carbon  and  PTFE  (see  Methods  in  SI),  which   fulfill   the  criteria  of  a   relatively   large  surface  area  and  porosity.  For   the  highest  DN  solvent   (Me-­‐Im),   and  at  high  potentials  ≥  2.5  V   (high  V   in   Fig.  7  b),   large  agglomerates  of  Li2O2  particles  (∼200-­‐300  nm)  form  between   the  carbon  particles.  At   low  potentials  ∼2.2  V  (low  V   in  Fig.  7   c),   i.e.   corresponding   to   the   second   reduction  peak   in   the  CVs   for  Me-­‐Im,  there   is   little   evidence   of   Li2O2   particles.   These  morphological   observations   are   in   accord  with  the  mechanism  we  have  proposed,  as  well  as  with  the  SERS  and  electrochemical  data,  i.e.  particles  at  high  V  and  a  film  on  the  electrode  at  low  V  (see  e.g.  SERS,  Fig.  3,  for  Li2O2  film  growth  on  the  electrode  surface).  A  similar  behavior  is  observed  in  DMSO.  We  do  not  know  the  detailed  mechanism  of  how  dissolved  LiO2,  forms  a  Li2O2  precipitate  in  high  DN  solvents  at   high   voltages.   Presumably   there   is   desolvation,   formation   of   an   activated   complex,  disproportionation  and  nucleation  and  growth.  Computational  studies  have  given  valuable  insight   into   this   process   in   the   gas   phase,   although   of   course   conditions   in   solution   are  somewhat  different11.  Some  of  the  O2

-­‐  near  the  electrode  surface  will  disproportionate  and  form  Li2O2  likely  using  the  electrode  or  existing  Li2O2  surfaces  as  a  convenient  nucleation  site.  Most   will   diffuse   away   but   in   practical,   porous,   electrodes,   will   precipitate   in   the   pores,  because   the   rate   of   disproportionation   is   quite   fast.   The   morphology   in   CH3CN   is   very  different  from  high  DN  solvents  at  high  voltage.  There  is  no  evidence  of  Li2O2  particles  at  any  potential,  Fig.  7  h  and  i,  but  SERS  (Fig.  3)  and  the  electrochemistry  confirms  Li2O2  is  present  at  high  and  low  potentials.  In  the  case  of  DME,  Li2O2  particles  are  observed  in  the  pores  at  high   potentials.  Our   electrochemical   analysis   and   SERS,   Fig.   3,   indicates   the   simultaneous  formation   of   a   Li2O2   film   at   these   potentials.   Below   this   potential   no   Li2O2   particles   are  evident  and  SERS  (Fig.  3)  confirms  a  Li2O2  film  on  the  electrode  surface.  

The   variation   in   Li2O2   morphology   with   solvent   type   is   in   excellent   agreement   with   the  change  in  the  pathway  of  O2  reduction  from  one  dominated  by  solution  soluble  LiO2  in  high  DN  solvents,  (giving  rise  to  Li2O2  particles  in  the  pores  of  the  electrode  at  high  potentials)  to  the   surface   mechanism   in   low   DN   solvents   (Li2O2   film   on   the   electrode   surface).   These  results   are   also   in   good   accord  with   the   polarization  measurements   shown   in   Fig.   6.   The  intermediate  nature  of  the  DN  in  DME,  results  in  both  solution  and  surface  mechanisms  of  oxygen   reduction   occurring   at   the   same   potential,   the   result   being   the   simultaneous  formation   of   particles   that   fill   the   pores   in   the   electrode   as  well   as   a   film   that   coats   the  surface.    

 

 

 

 

© 2014 Macmillan Publishers Limited. All rights reserved.

Page 12: SUEMENTARY INRMATIN · 2014-11-20 · 1 EastChem, School of Chemistry, University of St Andrews, North Haugh, St Andrews, Fife, KY16 9ST, UK 2 Laboratoire de Réactivité et Chimie

NATURE CHEMISTRY | www.nature.com/naturechemistry 12

SUPPLEMENTARY INFORMATIONDOI: 10.1038/NCHEM.2101

12    

References  

1.   Peng  Z,  Freunberger  SA,  Hardwick  LJ,  Chen  Y,  Giordani  V,  Barde  F,  et  al.  Oxygen  reactions  in  a  non-­‐aqueous  Li+  electrolyte.  Angew  Chem,  Int  2011,  50(28):  6351-­‐6355.  

2.   Gao  P,  Gosztola  D,  Leung  LWH,  Weaver  MJ.  Surface-­‐Enhanced  Raman-­‐Scattering  at  gold  electrodes  -­‐  Dependence  on  Electrochemical  Pretreatment  Conditions  and  Comparisons  with  Silver.  J  Electroanal  Chem  1987,  233(1-­‐2):  211-­‐222.  

3.   Trahan  MJ,  Mukerjee  S,  Plichta  EJ,  Hendrickson  MA,  Abraham  KM.  Studies  of  Li-­‐Air  Cells  Utilizing  Dimethyl  Sulfoxide-­‐Based  Electrolyte.  J  Electrochem  Soc  2013,  160(2):  A259-­‐A267.  

4.   Lu  Y-­‐C,  Gasteiger  HA,  Shao-­‐Horn  Y.  Catalytic  Activity  Trends  of  Oxygen  Reduction  Reaction  for  Nonaqueous  Li-­‐Air  Batteries.  J  Am  Chem  Soc  2011,  133(47):  19048-­‐19051.  

5.   Erlich  RH,  Popov  AI.  Spectroscopic  studies  of  ionic  solvation.  X.  Study  of  the  solvation  of  sodium  ions  in  nonaqueous  solvents  by  sodium-­‐23  nuclear  magnetic  resonance.  J  Am  Chem  Soc  1971,  93(22):  5620-­‐5623.  

6.   Das  U,  Lau  KC,  Redfern  PC,  Curtiss  LA.  Structure  and  Stability  of  Lithium  Superoxide  Clusters  and  Relevance  to  Li-­‐O-­‐2  Batteries.  J  Phys  Chem  Lett  2014,  5(5):  813-­‐819.  

7.   Izutso  K.  Electrochemistry  in  Nonaqueous  Solutions.  WILEY-­‐VCH  Verlag  GmbH:  Weinheim,  2002.  

8.   Bard  AJ,  Faulkner  LR.  Electrochemical  Methods:  Fundamentals  and  Applications,  2nd  edn.  John  Wiley  &  Sons,  2001.  

9.   Snow  RH.  Thermodynamic  Evaluation  of  the  Possibility  of  Lithium  Superoxide  Production  Technical  Report;  1965.  

10.   Viswanathan  V,  Thygesen  KS,  Hummelshoj  JS,  Norskov  JK,  Girishkumar  G,  McCloskey  BD,  et  al.  Electrical  conductivity  in  Li2O2  and  its  role  in  determining  capacity  limitations  in  non-­‐aqueous  Li-­‐O2  batteries.  J  Chem  Phys  2011,  135(21):  214704.  

11.   Bryantsev  VS,  Blanco  M,  Faglioni  F.  Stability  of  Lithium  Superoxide  LiO2  in  the  Gas  Phase:  Computational  Study  of  Dimerization  and  Disproportionation  Reactions.  J  Phys  Chem  A  2010,  114(31):  8165-­‐8169.  

 

© 2014 Macmillan Publishers Limited. All rights reserved.