of 12
S1
Supporting
Information:
Ni
Mo
Nanopowders
for
Efficient
Electrochemical
Hydrogen
Evolution
James
R.
McKone,
Bryce
F.
Sadtler,
Caroline
A.
Werlang,
Nathan
S.
Lewis,
Harry
B.
Gray*
California
Institute
of
Technology,
Division
of
Chemistry
and
Chemical
Engineering,
1200
E.
California
Blvd.,
Pasadena,
CA
91125
S.1.
Materials
and
Methods
Preparation
of
powders
A
sample
synthetic
procedure
for
Ni
Mo
nanopowder
with
a
6/4
ratio
of
Ni/Mo
is
shown
below.
Other
ratios
were
also
synthesized
by
keeping
the
total
concentration
of
dissolved
metal
species
roughly
constant.
The
synthesis
was
adapted
from
several
previous
studies
involving
precipitation
of
Ni
Mo
oxides
or
deposition
of
active
Ni
Mo
electrocatalyst
on
metallic
substrates.
1
3
1.5
g
(5.2
mmol)
nickel
nitrate
hexahydrate
(Aldrich
97%)
and
0.6
g
(3.4
mmol
Mo
basis)
ammonium
heptamolybdate
(Aldrich
reagent
grade)
were
added
to
5
mL
deionized
(Nanopure,
>18
M
)
water,
to
which
was
added
2
mL
of
30%
ammonium
hydroxide
(Macron
Chemicals).
After
adding
the
ammonium
hydroxide,
the
green
solution
turned
a
deep
blue
(main
text
figure
1,
left)
color
and
the
molybdate
salt
dissolved
readily.
This
solution
was
added
at
once
to
45
mL
diethylene
glycol
(Aldrich
ReagentPlus)
at
room
temperature.
The
mixture
was
placed
in
a
100
mL
tall
form
beaker
and
heated
on
a
hotplate
set
to
350
°C
with
stirring
(500
rpm)
using
a
magnetic
stir
bar.
When
the
temperature
reached
approximately
70
°C,
a
green
solid
began
to
precipitate.
Upon
reaching
~110
°C,
the
reaction
had
become
blue
green
and
completely
opaque,
and
the
S2
remaining
water
in
the
mixture
began
to
boil.
At
this
time,
the
reaction
was
removed
from
the
hotplate,
allowed
to
cool
for
approximately
30
seconds,
and
then
transferred
into
4
15
mL
centrifugation
tubes.
While
still
hot,
the
suspension
was
centrifuged
at
3000
rpm
for
~10
minutes,
after
which
there
was
approximately
1
mL
1
of
green
solid
in
each
vial
along
with
a
pale
blue
supernatant.
For
each
fraction,
the
supernatant
was
discarded
and
the
green
solid
was
washed
with
5
mL
of
deionized
water.
Thorough
washing
was
aided
by
sonication
using
a
Ti
horn
sonicator
(Qsonica,
500W
model)
set
to
20%
power
to
fully
re
suspend
the
solid.
After
the
water
wash
and
a
second
centrifugation,
the
green
solid
appeared
to
decrease
somewhat
in
volume
and
the
supernatant
was
very
light
blue.
Subsequent
washing
with
water
gave
colorless
supernatant,
so
only
one
water
wash
was
employed.
After
a
subsequent
washing
with
acetone
and
centrifugation,
the
solid
was
suspended
a
final
time
in
a
minimum
of
methanol
and
poured
into
a
crystallization
dish.
The
dish
was
heated
on
a
hotplate
set
to
60
°C
for
several
hours,
yielding
a
pale
green
solid
(main
text
figure
1,
center).
The
green
solid
was
transferred
to
a
ceramic
crucible
and
heated
in
a
home
built
tube
furnace
(components
from
Omega
Engineering)
under
forming
gas
(5%
H
2
,
95%
N
2
,
500
sccm,
Airliquide),
first
at
200
°C
for
30
minutes,
and
then
at
450
°C
for
1
hour.
After
heating,
the
resulting
black
solid
was
cooled
completely
under
forming
gas
and
then
withdrawn
toward
the
end
of
the
tube.
Before
fully
withdrawing
the
crucible,
however,
a
small
quantity
of
isopropanol
(Aldrich)
was
transferred
into
the
crucible
so
as
to
wet
the
solid
and
prevent
pyrophoric
re
oxidation.
Warning:
if
this
step
was
not
taken,
on
withdrawing
the
solid
to
air
the
crucible
became
hot
to
the
touch
and
portions
of
the
solid
spontaneously
ignited
to
red
heat,
closely
resembling
wood
fire
embers.
Upon
cooling,
portions
of
the
solid
had
turned
brownish
green.
Allowing
the
solid
to
spontaneously
react
in
air
still
yielded
active
catalyst
material,
since
deposited
films
were
subsequently
reduced
again
before
testing
(vide
infra).
Wetting
with
isopropanol,
however,
inhibited
any
observable
spontaneous
oxidation.
S3
The
black
solid
was
suspended
in
a
minimum
of
isopropanol
and
ground
with
a
mortar
and
pestle
to
homogenize
large
agglomerates.
Then
the
resulting
viscous
suspension
was
transferred
to
a
glass
vial
and
sonicated
using
a
bath
sonicator
(Branson
1210)
for
at
least
30
minutes
to
yield
a
smooth
black
colloidal
“ink”
(main
text
figure
1,
right).
The
Ni
Mo
ink
remained
well
suspended
for
only
a
few
minutes,
but
could
be
easily
re
suspended
by
further
sonication,
vigorous
shaking,
or
agitation
in
a
vortex
mixer.
Preparation
of
nanopowder
coated
electrodes
For
preparation
of
electrodes,
Ni
Mo
nanopowder
inks
as
described
above
were
prepared
in
mass
concentrations
ranging
from
1
100
mg
mL
1
.
Mass
concentrations
were
verified
by
measuring
on
a
microbalance
the
dry
mass
left
behind
by
a
known
volume
after
the
solvent
had
fully
evaporated.
Known
quantities
were
then
transferred
by
pipette
to
the
surface
of
cleaned
Ti
foils
(Alfa
Aesar)
that
had
been
cut
into
5
10
mm
squares.
Coated
foils
were
then
annealed
under
forming
gas
at
400
450
°C
for
30
minutes,
and
cooled
to
room
temperature
under
forming
gas.
This
second
reduction
step
was
required
to
give
high
HER
activity,
and
appeared
to
greatly
improve
the
conductivity
of
the
catalyst
coating.
The
foils
were
fashioned
into
electrodes
by
contacting
to
a
copper
wire
using
silver
paint
(SPI
supplies).
The
wire
was
threaded
through
a
6
mm
diameter
glass
capillary
and
all
surfaces
except
the
active
electrode
area
were
coated
with
two
part
epoxy
(Hysol
9460).
Preparation
of
metallurgical
samples
Metallurgically
prepared
Ni
Mo
alloy
samples
in
the
form
of
cylindrical
rods,
with
Mo
contents
of
1%,
4%,
and
12%,
were
procured
from
Princeton
Scientific
Corp.
A
99.999%
pure
Ni
rod
was
also
procured
from
Alfa
Aesar.
The
rods
were
cut
into
coin
sized
samples
using
a
slow
speed
saw
equipped
with
a
low
concentration
boron
nitride
blade
(Buehler).
Samples
were
then
lapped
and
polished,
first
using
silicon
carbide
paper
on
a
rotating
wheel
(South
Bay
Technology)
and
finished
using
diamond
polishing
compound
S4
(Buehler)
on
an
automatic
polishing
apparatus
(Buehler
Minimet).
Electrodes
were
then
prepared
in
the
same
fashion
as
with
the
Ti
foil
substrates
for
the
nanopowder
samples.
The
final
polishing
step
made
use
of
50
nm
diamond
grit
combined
with
a
few
drops
of
Marble’s
reagent
(~10
wt%
CuSO
4
in
6M
HCl).
4
This
chemo
mechanical
polishing
step
led
to
very
smooth
surfaces
with
visible
contrast
in
the
scanning
electron
microscope
that
we
attribute
to
polycrystalline
grain
boundaries
(Figure
S1).
Without
including
etchant
in
the
final
polishing
step,
electrochemistry
results
did
not
yield
a
clear
relationship
between
Mo
content
and
electrochemical
activity.
We
believe
this
is
because
polishing
damage
can
disrupt
the
surface
chemical
composition
of
metal
alloys,
resulting
in
surface
composition
that
does
not
reflect
the
bulk
alloy
composition.
4
Figure
S1.
Scanning
electron
micrograph
of
polished
Ni
Mo
sample
exhibiting
contrast
due
to
grain
boundaries.
Electrochemical
methods
All
electrochemistry
data
were
collected
on
a
Gamry
Reference
600
potentiostat,
which
is
capable
of
potential
control,
current
control,
and
impedance
analysis.
Electrolytes
for
hydrogen
evolution
experiments
were
either
0.5
M
H
2
SO
4
,
1
M
NaOH,
or
2
M
KOH
solution,
as
noted
in
the
main
text.
Polarization
data
were
collected
at
slow
scan
rates
(2
S5
10
mV
sec
1
)
with
solution
agitation
by
fast
stirring
using
a
magnetic
stir
bar,
and
uncompensated
resistance
was
corrected
using
the
current
interrupt
method.
5
Stability
data
were
collected
under
galvanostatic
conditions
without
solution
agitation,
and
were
not
corrected
for
uncompensated
resistance.
Research
grade
H
2
(g)
(Airliquide)
was
bubbled
at
1
atm
through
the
working
compartment
during
all
experiments,
so
as
to
ensure
a
well
defined
thermodynamic
potential
for
the
HER.
Reference
electrodes
for
acid
experiments
were
either
saturated
calomel
(SCE)
or
mercury/mercury
sulfate
(MSE).
Also
for
acid
experiments,
the
counter
electrode
was
a
home
built
Ru/Ir
oxide
(from
pyrolysis
of
Ru
and
Ir
chloride
salts,
Aldrich)
electrode
supported
on
Ti
mesh
(Alfa
Aesar),
6
contained
behind
a
Nafion
®
(Fuelcellstore.com)
separator.
For
alkaline
experiments,
reference
electrodes
were
either
SCE
or
mercury/mercury
oxide
(Hg/HgO)
filled
with
1M
NaOH
solution,
and
the
counter
electrode
was
a
large
area
Ni
mesh
contained
in
the
same
compartment
as
the
working
electrode.
During
alkaline
experiments,
few
bubbles
were
observed
on
the
Ni
counter
electrode,
implying
that
the
main
reaction
at
the
counter
electrode
was
oxidation
of
the
Ni
electrode
to
Ni
oxide
(insoluble
under
alkaline
conditions)
and/or
hydrogen
oxidation.
In
all
cases,
the
reversible
hydrogen
electrode
(RHE)
potential
was
determined
explicitly
in
the
electrolyte
of
interest
by
saturating
the
solution
with
H
2
(g)
at
1
atm
and
measuring
the
open
circuit
potential
of
a
freshly
cleaned
Pt
electrode.
Generally
these
calibration
experiments
were
carried
out
after
testing
non
noble
metal
electrodes
so
as
to
avoid
Pt
contamination.
Microscopy
and
elemental
analysis
Scanning
electron
microscopy
(SEM)
was
performed
using
a
Zeiss
model
1550
field
emission
scanning
electron
microscope,
equipped
with
an
Oxford
X
Max
SDD
X
ray
Energy
Dispersive
Spectrometer
(EDS)
system.
Further
structural
characterization
was
performed
using
an
FEI
Tecnai
F
20
transmission
electron
microscope
(TEM)
operated
at
200
kV
and
equipped
with
an
EDAX
energy
S6
dispersive
x
ray
detector
and
a
Gatan
Image
Filter
(GIF).
For
TEM
analysis,
a
small
quantity
of
Ni
Mo
nanopowder
was
suspended
in
90%
isopropanol,
10%
acetylacetone
(acac,
Aldrich),
which
increased
the
stability
of
the
colloid.
The
suspension
was
drop
cast
onto
a
lacey
carbon/Cu
grid.
The
SEM
and
TEM
images
were
subjected
to
a
minimum
of
post
processing,
being
limited
to
normalization
of
pixel
intensity
maxima
uniformly
across
the
image
so
as
to
produce
optimal
contrast.
S.2.
Relationship
between
Mo
Loading
and
HER
Activity
in
Powders
We
observed
consistent
relationships
between
the
relative
quantity
of
Mo
included
in
the
initial
precipitation
reaction,
the
Mo
loading
subsequently
observed
in
the
catalyst
powders,
and
the
catalytic
activities
of
the
powders
under
alkaline
conditions.
Syntheses
were
attempted
with
Mo
loading
ranging
from
0%
to
100%,
but
the
actual
Mo
content
in
the
reduced
powders
only
ranged
from
0%
to
~60%
mole
fraction
(Figure
S2).
Interestingly,
an
ammonium
nickel
molybdate
reported
previously
by
Levin,
et
al.
exhibits
varying
Mo
content
corresponding
to
a
formula
of
(NH
4
)H
2x
Ni
3
x
O(OH)(MoO
4
)
2
,
with
0
<
x
<
1.5.
2
Thus
the
upper
limit
of
Mo
content
for
the
synthesis
of
this
compound
was
57%,
in
close
agreement
with
the
upper
range
observed
in
our
powders.
We
were
also
able
to
obtain
powder
samples
with
higher
Nickel
content
by
including
excess
Ni
relative
to
the
stoichiometry
limit
implied
by
the
aforementioned
formula.
When
excess
Mo
was
used
for
the
precipitation
reaction,
the
total
yield
of
powder
decreased
to
the
point
where
a
“pure
Mo”
synthesis
yielded
no
precipitation
at
all.
Instead
pure
Mo
oxide
nanopowder
was
procured
commercially
(Alfa
Aesar)
and
reduced
under
the
same
conditions
for
comparison
to
Ni
and
Ni
Mo
nanopowders.
S7
Figure
S2.
Plot
depicting
the
relationship
between
Ni/Mo
ratio
in
the
precursor
solution
(top
row,
above
horizontal
line),
the
resulting
Ni/Mo
ratio
in
the
solid
catalyst
as
measured
by
EDS
(below
line,
x
axis),
and
the
overpotential
required
to
pass
10
mA
cm
2
in
1
M
NaOH
solution
using
a
catalyst
mass
loading
of
~0.5
mg
cm
2
(below
line,
y
axis).
Lines
connect
data
points
for
a
given
sample
before
and
after
reaction
to
form
the
active
catalyst.
Based
on
the
factors
above,
we
propose
the
following
mechanism
for
the
generation
of
Ni
Mo
oxide
powders
in
the
observed
Ni/Mo
ratios.
Upon
heating
in
diethylene
glycol,
Ni
nitrate
reacts
with
ammonium
molybdate
to
give
the
mixed
nickel
molybdate
as
reported
by
Levin,
et
al.
up
to
the
maximum
Mo
content
that
can
be
incorporated
into
the
crystal
structure
for
solutions
containing
excess
molybdenum.
In
the
case
where
Mo
is
the
limiting
reagent,
all
Mo
in
the
starting
mixture
is
incorporated
into
the
resulting
precipitate.
If
Ni
is
in
excess
(e.g.
Ni/Mo
>
1.5),
some
or
all
of
the
nickel
nitrate
precipitates
as
nickel
hydroxide
upon
heating,
likely
due
to
loss
of
ammonia
from
solution,
destabilizing
the
nickel
hexammine
complex.
Conversely,
if
Mo
is
in
excess
(i.e.
Ni/Mo
<
0.75),
some
Mo
stays
in
solution
as
soluble
molybdate
or
perhaps
an
oxo
molybdenum
glycolate
complex.
Catalytic
activities
for
Ni
Mo
powders
under
alkaline
conditions
correlated
only
weakly
with
Mo
content
(Figure
S2).
For
Mo
loading
of
0%
<
[Mo]
<
50%
the
activity
was
markedly
improved
over
100%
Ni
or
100%
Mo.
In
particular,
[Mo]
values
around
20
40%
S8
appeared
to
give
maximal
HER
activity.
This
is
consistent
with
previous
results
from
Brown
and
Mahmood,
who
saw
significant
enhancement
in
HER
activity
under
alkaline
conditions
for
[Mo]
>
10%
and
no
clear
further
increase
in
activity
for
10%
<
[Mo]
<
40%.
1
S.3.
Estimation
of
Turnover
Frequencies
A
spreadsheet
in
Excel
format
(.xlsx)
has
been
provided
as
a
separate
electronic
document
containing
the
formulas
and
values
used
to
estimate
turnover
frequencies
for
Ni
Mo
nanopowder
catalysts
as
well
as
planar
metallurgical
Ni
and
Ni
Mo
samples.
For
the
estimation
of
surface
area
for
nanoparticle
catalysts,
calculations
have
been
included
for
catalysts
consisting
of
pure
Ni
and
pure
Mo
as
well
as
the
60/40
ratio
in
order
to
illustrate
the
relative
insensitivity
of
the
calculation
to
the
relative
proportion
of
each
metal.
S.4.
Additional
Electron
Microscopy
Data
Additional
data
from
electron
microscopic
analysis,
including
SEM
and
TEM
micrographs
and
electron
diffraction
patterns,
are
included
on
the
following
pages
(figures
S3
S5).
S9
Figure
S3.
Scanning
electron
micrographs
of
Ni
Mo
nanopowder
with
progressively
increasing
magnification
from
top
to
bottom.
Note
on
the
bottom
micrograph
the
texturing
of
particle
surfaces,
implying
porosity
at
the
sub
100
nm
scale.
S10
Figure
S4.
Transmission
electron
micrographs
(top
left,
right)
and
electron
diffraction
image
(bottom)
of
Ni
Mo
oxide
powder
prior
to
reduction
to
Ni
Mo
nanopowder.
The
diffraction
pattern
indicates
a
low
degree
of
crystalline
order
for
this
precursor
material.
S11
Figure
S5.
Transmission
electron
micrographs
(top
left,
right)
and
electron
diffraction
image
(bottom)
of
Ni
Mo
nanopowder.
The
left
micrograph
is
representative
of
the
difficulty
in
obtaining
clear
micrographs
of
agglomerates
due
to
their
thickness
preventing
clear
contrast.
The
diffraction
pattern
indicates
a
significantly
higher
degree
of
crystalline
order
compared
to
the
Ni
Mo
oxide
precursor.
S12
SUPPORTING
REFERENCES
(1)
Brown,
D.
E.;
Mahmood,
M.
N.
US
Patent
4,358,475
1982
.
(2)
Levin,
D.;
Soled,
S.
L.;
Ying,
J.
Y.
Inorg.
Chem.
1996
,
35
,
4191.
(3)
Schmitt,
P.;
Brem,
N.;
Schunk,
S.;
Feldmann,
C.
Adv.
Funct.
Mater.
2011
,
21
,
3037.
(4)
Vander
Voort,
G.
F.
Metallography,
Principles
and
Practice
;
McGraw
Hill:
New
York,
1984.
(5)
Oelßner,
W.;
Berthold,
F.;
Guth,
U.
Mater.
Corros.
2006
,
57
,
455.
(6)
Angelinetta,
C.;
Trasatti,
S.;
Atanasoska,
L.
D.;
Minevski,
Z.
S.;
Anatasoski,
R.
T.
Mater.
Chem.
Phys.
1989
,
22
,
231.