UNITED
STATES
ENVIRONMENTAL
PROTECTION
AGENCY
National
Vehicle
and
Fuel
Emissions
Laboratory
2565
Plymouth
Road
Ann
Arbor,
Michigan
48105­
2498
MEMORANDUM
FROM:
Todd
Sherwood,
Office
of
Transportation
and
Air
Quality
TO:
Environmental
Protection
Agency,
Air
Docket
A­
28­
2001,
E­
DOCKET
OAR­
2003­
0012
SUBJECT:
Telephone
Conversation
with
Jim
Roberts
of
Multimetco,
Inc.,
regarding
PGM
Recycling
DATE:
March
16,
2004
On
September
11,
2003,
I
spoke
with
Jim
Roberts,
President,
Multimetco,
Inc.,
regarding
the
platinum
group
metal
(
PGM)
recycling
industry.
According
to
their
website
(
www.
multimetco.
com),
Multimetco
is
one
of
the
largest
precious
metals
recyclers
of
spent
catalyst
and
automobile
catalytic
converters
in
the
United
States.
During
our
conversation,
Mr.
Roberts
informed
me
that
the
recycler
generally
keeps
roughly
10
percent
of
the
value
of
the
PGMs
recovered
from
a
spent
auto
catalyst.
This
10
percent
includes
losses
incurred
during
the
refining
process.
The
remaining
90
percent
of
the
value
is
returned
to
the
owner
of
the
auto
catalyst
(
i.
e.,
the
entity
that
has
hired
the
recycler
to
recover
the
PGMs
from
the
auto
catalyst).
In
other
words,
90
percent
of
the
value
of
the
PGMs
recovered
from
a
spent
auto
catalyst
is
returned
to
the
owner
of
those
PGMs.

Attachment:
PGMs
Recycled
from
Spent
Auto
Catalyst
at
Multimetco,
Inc.,
by
Jim
Roberts
1
PGMs
Recycled
from
Spent
Auto
Catalyst
at
Multimetco,
Inc.

Jim
Roberts
(
Multimetco,
Inc.)

History/
Background
In
1989
Multimetco
purchased
the
assets
of
the
Plasma
Smelting
Division
of
Texas
Gulf
Minerals
and
Metals.
In
the
early
80'
s
Texasgulf
had
worked
with
Tetronics
in
England
to
adapt
the
Tetronics
plasma
arc
torch
to
the
smelting
of
a
chromite
ore
fines
from
South
Africa
that
contained
PGMs.
During
the
development
phase
Texasgulf
sold
off
their
South
African
properties,
leaving
the
future
of
the
plasma­
smelting
project
in
jeopardy.
Astute
Texasgulf
engineers
quickly
recognized
that
their
plasma­
smelting
concept
could
be
applied
to
the
smelting
of
spent
auto
catalyst
for
the
recovery
of
the
contained
PGMs.

After
extensive
pilot
testing
at
Tetronics
Texasgulf
decided
to
build
their
smelter
in
Anniston,
Alabama,
located
almost
midway
between
Birmingham,
Alabama
and
Atlanta,
Georgia.
The
plasma
arc
furnace
began
smelting
late
in
1984­
the
point
in
time
that
Texasgulf
had
projected
significant
quantities
of
spent
autocat
would
begin
to
outturn
from
the
first
cars
fitted
with
catalytic
converters
(
1975
model
year).

The
scale­
up
from
pilot
to
commercial
operations
wasn't
without
operational
problems,
but
after
a
year
or
so
of
operations
Texasgulf
realized
that
an
even
larger
problem
was
commercial.
This
commercial
problem
had
at
least
two
aspects:
the
supply
of
scrap
auto
catalyst
was
clearly
below
their
projections,
and
players
in
the
scrap
metal
business
proved
to
be
tough
to
deal
with.
Their
original
assumptions
were
that
they
would
be
able
to
secure
sufficient
catalyst
by
dealing
with
4
or
5
collectors
who
would
buy
the
scrap
converters
from
salvage
yards
and
muffler
shops,
de­
can
the
catalyst,
and
sell
the
loose
catalyst
to
Texasgulf.
They
soon
realized
that
the
collectors
could
(
and
would)
withhold
supply
to
get
a
better
price
once
they
sensed
how
urgently
Texasgulf
needed
feed
for
the
smelter.
This
problem
was
compounded
by
the
fact
that
everyone
involved
in
the
business
had
over­
estimated
the
ramp­
up
in
availability
of
scrap
autocat.
Probably
the
two
most
important
miscalculations
were
that
many
cars
being
scrapped
that
had
originally
been
equipped
with
converters
no
longer
had
converters,
and
the
salvage
yard
operators
didn't
immediately
begin
to
remove
converters
that
were
on
the
cars,
resulting
in
many
cars
being
crushed
with
the
converters
still
on
the
car
(
this
may
still
be
the
case).

Multimetco's
owners
had
been
in
the
scrap
metal
business
for
several
generations,
first
in
Belgium
(
Metallo
Chimique),
and
subsequently
in
the
U.
S.
(
Chemetco),
operating
secondary
smelters
recovering
copper,
lead,
and
tin.
Having
had
supply
problems
of
their
own,
Chemetco
and
Metallo
had
backward
integrated
into
scrap
collection.
By
1989
Chemetco
had
established
approximately
40
warehouse
locations
in
major
cities
throughout
the
U.
S.,
and
had
already
begun
to
purchase
scrap
converters
on
a
small
scale.
The
belief
was
that
the
Chemetco
warehouse
network
was
just
what
the
2
Texasgulf
smelter
was
lacking,
so
Multimetco
was
formed
in
July
1989
to
operate
the
Anniston
plasma
smelter.

In
our
early
days
Multimetco
was
supplied
its
feed
of
spent
autocat
from
the
Chemetco
purchasing
network.
In
1993,
due
to
extremely
tough
times
in
the
copper
business,
the
owners
decided
to
split
off
the
copper,
lead­
tin,
and
PGM
operations
into
stand­
alone
businesses.
Since
that
time
Multimetco
has
operated
it's
own
purchasing
network.

Commercial
Operations
Multimetco
operates
purchasing
facilities
in
the
following
cities:
Boston,
Buffalo,
Pittsburgh,
Tampa,
Miami,
Memphis,
Dallas,
Houston,
San
Antonio,
Los
Angeles,
Chicago,
and
Anniston.
Ten
of
these
twelve
facilities
also
operate
trucks,
providing
"
picked­
up"
purchasing.
The
trucks
range
in
size
from
2­
ton
stake­
sides
up
to
tractor­
trailer
rigs.
Although
Multimetco's
owners
had
extensive
experience
in
the
scrap
metals
side
of
the
business,
auto
converters
provided
several
new
twists.
The
first
was
the
fact
that
buying
converters
meant
buying
by
the
piece,
not
by
the
pound.
The
computer
inventory
system
for
other
scrap
metals
had
to
be
revamped
to
accommodate
this
oddity.
The
second
was
the
practice
of
buying
on
a
"
picked­
up"
basis,
often
paying
in
cash.
All
other
scrap
had
been
brought
to
the
warehouses,
bought
by
the
pound,
and
paid
by
check.

Probably
the
most
difficult
commercial
aspect
of
this
new
business
was
(
and
still
is)
training
buyers
how
to
grade
and
sort
converters.
In
our
universe
of
scrapped
automobiles
there
are
at
least
thousands
of
different
sized
and
shaped
converters,
and
maybe
that
many
different
catalyst
formulations.
The
converters
we
buy
every
day
cover
the
time
frame
from
the
first
clean­
air
regulations
in
1975,
which
covered
only
carbon
monoxide
and
hydrocarbon
emissions,
up
to
the
current
stricter
regulations
that
include
nitrous
oxides.
Obviously
many
different
catalyst
formulations
and
converter
combinations
have
been
used
to
comply
with
the
ever­
tightening
standards.

The
task
is
to
develop
a
databank/
history
that
allows
a
rough
classification
of
each
converter
into
one
of
several
broad
categories,
primarily
based
on
the
value
of
the
contained
PGMs.
Our
purchasing
categories
currently
are
as
follows:

Pellet
converters
Double
plug
pellet
(
3­
way)

Small
monolith
Large
monolith
Large
GM
monolith
Small
foreign
Large
foreign
Pre­
converters
and
after­
markets
3
As
the
descriptions
imply,
size
does
matter
when
it
comes
to
converters­
large
converters
contain
3.5
to
4
pounds
of
catalyst,
often
in
two
"
bricks",
while
small
converters
average
closer
to
2
pounds,
usually
a
single
"
brick".
But
there
are
"
larges"
that
are
bought
as
"
smalls",
and
vice­
versa,
because
of
the
loadings
(
PGM
content)
of
the
catalyst.
All
of
us
that
buy
catalyst
"
in
the
can"
have
to
continually
update
our
information
about
the
metal
content
of
various
converters,
and
we
can
all
look
forward
to
even
more
radical
changes
as
we
begin
to
encounter
the
Pd­
Rh
catalysts
that
have
been
favored
by
car
manufacturers
in
the
last
several
years.
I
should
also
point
out
that
converters
are
not
always
"
full";
off­
spec
engine
conditions
can
cause
the
converter
to
overheat,
cracking
or
melting
the
monolith
brick,
which
will
result
in
loss
of
catalyst.
As
a
consequence,
at
the
point
of
purchase
each
converter
must
be
inspected
to
insure
that
the
catalyst
is
still
intact.
Often
this
is
not
possible
until
attached
tailpipe
is
cropped
off.

Plant
Operations
The
buyers
at
the
warehouses
box
and
accumulate
converters
until
they
have
a
trailer
load,
approximately
3000
pieces,
at
which
point
they
ship
to
the
plant
in
Anniston.
We
operate
6
hydraulic
guillotine
shears
to
open
the
converters
and
extract
the
catalyst.
The
capacity
of
these
6
shears
is
in
excess
of
3000
converters
in
an
8­
hour
shift.
The
409
stainless
steel
shells
are
then
sold
for
recycling.

The
pellet
catalyst
must
be
screened
to
remove
any
trash
or
oversize,
and
passed
over,
a
magnetic
separator
prior
to
sampling.
As
the
pellets
flow
by
gravity
from
an
overhead
hopper
a
5%
"
cut"
is
taken
by
a
pelican
sampler,
and
then
another
5%
sample
is
split
off,
resulting
in
a
0.25%
sample
by
weight.
A
quick
moisture
determination
is
made
in
a
microwave
oven,
and
if
required,
the
pellets
are
then
dried
in
a
fluid­
bed
dryer.

The
monolith
catalyst
consists
of
partial
and
whole
bricks
up
to
5"
x
4"
x
6"
long.
This
catalyst
is
fed
by
a
vibrating
feeder
onto
a
conveyor
belt,
passes
under
a
magnetic
separator
to
remove
large
magnetics
(
such
as
pieces
of
the
converter
shell),
and
into
a
hammermill.
It
is
crushed
to
minus
8
mesh,
conveyed
by
bucket
elevator
to
a
feed
hopper,
and
then
fed
by
gravity
into
bins
of
about
1500
lb
capacity.
As
the
bins
are
filled
the
catalyst
passes
through
a
Retsch'
sampler,
and
as
with
the
pellets,
two
successive
5%
samples
are
taken.
Moistures
are
determined
by
microwave
"
on
the
spot".

Although
the
plant
equipment
was
designed
to
process
auto
catalyst,
we
have
over
the
past
5
years
steadily
increased
our
intake
of
non­
autocat
feed.
We
have
been
forced
into
learning
how
to
process
other
types
of
feed
because
of
the
competitiveness
of
the
scrap
autocat
business
in
the
U.
S.
As
we
improved
our
furnace
throughput
we
found
that
attempts
to
significantly
increase
our
autocat
intake
were
counter­
productive.
To
get
more
material
we
had
to
take
it
away
from
somebody
else
with
higher
prices,
and
that
"
somebody
else"
invariably
would
protect
"
their"
turf
and
bid
yet
a
higher
price
for
the
converters.
This
lose­
lose
game
proved
to
be
no
fun
to
play,
down
right
painful
actually.

We
presently
supplement
our
autocat
feed
with
an
additional
50%
of
other
PGM­
containing
catalysts
and
residues.
Non­
autocat
feed
materials
include:

­
petroleum
catalysts
(
non­
soluble)
4
­
insols
from
leaching
processes
­
chemical
catalysts
­"
coked"
petroleum
catalysts
­
fines
from
CCR
units
­
refractory
rubble
from
fiberglass
manufacturers
­
abator
catalysts
from
stationary
sources
­
slags
and
refractories
from
other
melting/
smelting
operations
These
non­
autocat
materials
are
processed
(
including
crushing
when
required)
and
sampled
similar
to
the
autocat.
Witnessing
by
outside
representatives
is
common.

After
the
autocat
and
non­
autocat
feed
materials
have
been
sampled
and
assayed
they
are
ready
for
the
smelting
operation.
Constituents
that
must
be
analyzed
before
smelting
include:

­
PGM
content
­%
moisture
or
%
LOI
­
carbon
­
A1203,
Si02,
Zr0
­
iron,
or
other
reducible
metals
The
materials
are
blended
to
produce
a
slag
with
an
acceptable
liquidous
with
a
minimum
of
fluxing,
and
a
metal
with
7­
10%
PGM
content.
Moisture
and
carbon
must
be
controlled;
water
decomposes
in
the
plasma
arc,
consuming
power
and
carbon,
and
an
excess
of
carbon
results
in
a
slag
that
is
too
electrically
conductive,
the
result
of
which
is
insufficient
resistive
heating.

The
plasma
furnace
incorporates
extensive
water
cooling
of
the
castable
alumina
lining.
The
thickness
of
the
lining
has
been
minimized
consistent
with
the
wear
pattern
and
freeze
line
of
slag
on
the
lining.
A
complete
second
furnace
with
interchangeable
parts
is
always
available
to
minimize
reline
time.
The
spare
furnace
is
lined
and
thoroughly
dried,
only
requiring
plumbing
and
electrical
connections
before
use.
The
used
lining
is
crushed
and
recycled
through
the
furnace
after
each
reline
for
recovery
of
any
absorbed
PGMs.

After
blending
and
fluxing,
the
material
is
gravity
fed
through
the
roof
of
the
furnace,
falling
through
the
plasma
arc
into
the
bath.
Feed
is
continuous,
as
is
slag
discharge,
by
way
of
a
water­
cooled
copper
underflow
spout.
In
order
to
operate
at
high
throughput
rates
(
2,400
pounds,
per
hour)
it
was
necessary
to
develop
equipment
to
handle
the
slag
with
minimum
labor
requirements.
A
water­
cooled
slag
conveyor
for
solidifying
and
conveying
the
slag
was
developed
(
and
patented);
it
produces
a
slag
"
pancake"
more
amenable
to
crushing.
The
slag
is
crushed
and
magnetically
separated
to
recover
any
prills
of
collector
metal
that
are
carried
over
in
the
slag.
5
The
collector
metal
(
iron­
based)
accumulates
in
a
graphite
crucible
located
in
the
base
of
the
furnace.
The
metal
is
tapped
after
smelting
a
pre­
determined
quantity
of
feed
(
typically
~
80,000
lbs).
The
metal
is
ground,
screened,
blended,
and
sampled
before
shipping.

Sample
Prep
&
Analysis
After
a
bulk
sample
of
the
material
is
obtained,
the
representative
sample
is
then
ground
and
split
for
laboratory
analysis.
A
typical
sample
is
ground
to
at
least
60
mesh
(
smaller
particle
size
may
be
required
for
some
materials)
and
then
subdivided
with
a
rotary
type
splitter
until
the
appropriate
sample
weight
is
obtained.
These
splits
then
enter
the
lab
where
analysis
for
precious
metals,
oxide
(
matrix)
constituents,
carbon
and
moisture
content
are
determined.

Oxide
concentrations
are
determined
with
a
Spectrace
X­
ray
Energy
Dispersive
Fluorescence
Spectrometer.
The
system
offers
an
evacuated
sample
chamber
and
a
He
purged
detector,
which
aids
in
light
element
determination
and
improved
stability.
Major
elements
of
interest
to
our
processes
(
feed)
are
Al,
Si,
Ca,
Fe,
Zr,
and
Ce.
Estimates
of
PGMs
are
also
determined.
Sample
analysis
is
quite
simple
and
requires
loading
the
powder
into
sample
cups
and
running
on
the
spectrometer.
Where
better
precision
is
needed,
samples
may
be
pressed
into
pellets.

Precious
metal
analyses
(
Pt,
Pd
and
Rh)
are
performed
on
a
Thermo
Jarrell
Ash
Iris
Advantage
ICP.
This
instrument
features
a
Charge
Injection
Device
(
CID)
detector.
The
CID
detector
simultaneously
reads
multiple
orders
off
all
wavelengths
giving
the
entire
wavelength
spectra.
This
results
in
reading
multiple
lines
and
background
data
for
each
element
of
interest
in
every
sample,
a
distinct
advantage
when
isolating
interferents.
After
classical
fire
assay
techniques
to
remove
matrix
oxides
and
concentrate
the
PGMs,
samples
are
digested
in
aqua
regia
and
PGMs
are
analyzed
on
the
ICP.
All
samples
are
assayed
in
triplicate
with
strict
QA/
QC
procedures
to
assure
accuracy.

Another
new
addition
to
our
laboratory
is
a
Thermo
Jarrell
Ash
DC­
Arc.
This
instrument
also
has
the
CID
detector
and
similar
software
and
design
characteristics
as
the
ICP
Currently
this
instrument
is
being
used
for
internal
process
control
(
slag
analysis).
It
offers
extremely
sensitive
determination
without
the
need
for
fire
assay
procedures.
Samples
are
mixed
with
a
conductive
host
(
carbon),
and
packed
into
the
electrodes.
The
samples
are
then
sputtered
into
the
gas
phase
with
the
arc.
The
high­
energy
arc
causes
elemental
excitation
and
spectral
emission,
resulting
in
a
 
direct
read'
of
any
element
present
in
the
sample.
This
has
resulted
in
an
enormous
time
saving
and
reduction
in
the
use
of
laboratory
chemicals.
