Advanced Technology Center
1861 Lefthand Circle
Longmont, CO 80501
Phone:
(303)678-0700
FAX:
(303) 442-0711
Measurement
of Trace Oxygen
in Corrosive
Gases by GC-DID
Mark Raynor, Terry Gerhart, Jon Welch hans, Brad Grissom, Clark McGrew and
Virginia Houlding
Matheson Tri-Gas Inc., Advanced Technology Center, 1861 Lefthand Circle,
Longmont CO 80501
Venue:
Gas Workshop:
Date:
Time:
Place:
SEMICON West, 2001
Determination of Low Levels of O2 in Specialty Gases
Tuesday, July 17, 2001
1:30 pm -4:00 pm
San Francisco Marriott Hotel
55 4th Street
San Francisco, California
Measurement
of Trace Oxygen
in Corrosive
Gases by GC-DID
Mark Raynor, Terry Gerhart, Jon Welch hans, Brad Grissom, Clark McGrew and
Virginia Houlding
Matheson Tri-Gas Inc., Advanced Technology Center, 1861 Lefthand Circle,
Longmont CO 80501
The device yield and performance characteristics of many microelectronic products are
critically dependent on the presence of trace impurities in the semiconductor process
gases used in their manufacture. In particular, atmospheric impurities, such as oxygen,
in corrosive gases can alter certain semiconductor manufacturing processes. In plasma
etching of polysilicon films, for example, O2 has been reported to affect the CI2 discharge
properties and hence the etch rate of the process (1). A reliable analytical method is
therefore necessary to monitor and control the level of this important impurity.
Typically, oxygen in corrosive gases such as C12,HCI and HBr, is analyzed by gas
chromatography with discharge ionization detection (GC-DID). With this technique, the
corrosive matrix gas is prevented from entering the detector by using a pre-column that
preferentially retains the corrosive matrix peak, but allows the elution of atmospheric
impurities on to the analytical column. The pre-column is then back-flushed to vent the
corrosive gas matrix, while the impurities of interest on the analytical column are
separated and detected by the DID.
Unfortunately, difficulties are often encountered with oxygen impurity analysis. GC-DID
systems that show good performance for measurements made with standards in an inert
helium matrix, may not necessarily perform in the same way when exposed to corrosive
gas matrices. Furthermore, instruments that detect O2 at high concentrations may not
detect the impurity at ppb levels. The choice of pre-column is critical to the detection of
oxygen at ppb and ppm levels. Porous polymer GC columns that are commonly used for
analysis have been found to absorb trace oxygen in some corrosive gases. Column
passivation, involving a single injection of a high concentration oxygen gas standard is
commonly employed to prevent oxygen absorption. However, this is not a permanent
solution, as continued heating of the packing material or exposure to corrosive gas may
result in a gradual increase in oxygen absorption over time, affecting instrument
calibration. This paper focuses on the analysis of O2 impurity in corrosive gases. The
GC-DID technique, as well as factors influencing detection of O2 in HCI, HBr and CI2, will
be discussed and results obtained with different pre-columns will be presented.
H.H. Sawin, L.D. Baston, D. Gray, L. Tepermeister, M.T. Mocella and G.C. Zau,
Threshold levels and effects of feed gas impurities on plasma etching processes, J
Electrochem. Soc., 137, 3526 (1990).
-? ,
MEASUREMENT
CORROSIVE
OF TRACE OXYGEN
GASES BY GC-DID
IN
Mark Raynor, Terry Gerhart, Jon Welch hans, Brad
Grissom and Virginia Houlding
Advanced Technology Center, Matheson Tri-Gas, Inc.
Longmont, CO 80501
SEMICON West 2001, Gas Workshop, July 17, 2001
A
MATHESON
..TRI'GAS
Semi-GasOivision
Introduction
.Micro-electronic
product performance can be
effected by the presence of trace impurities in the
process gases used in manufacture
.Oxygen
is particular problem
such as HCI, HBr and CI2
in corrosive
gases
-In plasma etching of polysilicon films, O2 has
been reported to affect the CI2 discharge
properties and hence the etch rate (1)
.A
reliable analytical method is necessary
monitor and control O2 levels
(1) H.H. Sawin et al., J. Electrochem.
to
Soc.. 137. 3526 (1990)
A
~.
MATHESON
TRI'GAS
Semi.Ga,Di,,;,ion
.Galvanic
cell
-Very
sensitive
corrosive
.Mass
technology
but
not
compatible
with
gases
spectrometry
-50
ppb
on
MDL
in helium
matrix.
difficult
possible.
El sensitivity
to calibrate,
detector
dependant
drifts
.APIMS
-very
sensitive.
analysis,
requires
special
source
for
corrosive
gas
expensive
.Paramagnetic
susceptibility
-sensitivity
limited
to
ppm
levels
.GC-DID
-low
ppb
enter
sensitivity.
but
need
to ensure
matrix
does
not
detector
A.
MATHESON
..TRI.GAS
Semi-Ga, Division
GC-DID Requirements
.Discharge
ionization detector -Universal
response due
to the high ionization potential of Helium (19.8 eV). He
capable of ionizing all compounds except for Ne (21 eV)
.DID
requires
.Leak
gettered
high purity helium carrier gas
free connections
.Important
.Various
to keep corrosive
Separation
-Series
matrix from entering
DID
Strategies
by-pass
-Heartcut
-Fore-flush
-Back-flush
A
MATHESON
..:r:~r~f:t.n~"-,--
..
Sequence of Events in GC-DID Analysis
of Corrosive Gas Matrix
Pre-column
He
-+1
1
Scrubber
Analytical Column
-Ii
t==j
Matrix
He
.-1
with Back-flush
~
Impurities
-~III
~
-~
III~
He
Scrubber
.-
..-£::a.-
~DID
I ,
He
~
MATHESON
..TRI'GAS
Semi-Gils
Oivision
5
GC-DID
Flow
Schematic
Valve 1.
Injection/Back-flush
Pre-column
Valve 2 -Analytical
Column Selection
Valve 3.
DID/Bypass
A
~.
MATHESON
TRI-GAS
Sem;-G.. D;,,;,lon
~
Calibration
.Many
and Validation
GC calibrations
are performed
.Important
to perform the cali~ration
gas as that of the sample
in inert gas matrix
in the same matrix
-Retention
time changes ~ay result in incorrect
identification
of impurity
-Changes
in peak shape may affect quantification
-Reactions
that don't occur in inert matrix may occur
in corrosive gas matrix a~d affect detection limits
.With
the use of a dilution manifold, the corrosive matrix
can be spiked dynamically with known concentrations
of standard gas
.Standard
in corrosive
matrix used to calibrate
GC-DID
A
MATHESON
~ .TRJoGAS
Semi-GosDil/ision
7
GC-DID Sampling and Dilution
System
tielium Purge
Standard Gas in
~
[1-~.:-.~~~---1
I~
: i ReOOout :
:'::::::::::::::
i
i
:,
,:
:
,
:
i
.Valve
:
l
~~
,
HCI
HelIum
Purge
Rotameter
~
C¥J+1
ToScrub~r
A
MATHESON
..TRI-GAS
Sem;.tia, D;,,;,ion
8
"¥ ,
~
--,
Detection of Oxygen in HBr: Matrix Spiked with Gas Standard
WXX),
&XXXJ
..4(XXX) !
<:mX>
~
...200!)
-
1(XXX)0
2
0. C...'.nIr3Uon
Porapak R precolumn, Porapak Q
and Mol Sieve 13X analytical
columns in parallel
Based on five replicate injections at
each concentration
Linear response (R2 = 0.9999)
2..
5.0
7.5
Time (mln)
A
MATHESON
..TRI.GAS
Semi-Go.
Divi.;ol1
11
Absorption
of Trace
Oxygen
by Worous
I
Polymer
Column
c
0.
j
,.,
2.'
'0
"-(m'")
7.5
10 ppm standard in He prior
to atmospheric exposure
00
2.'
,..
7.'
00
25
5.0
7.5
n-Imln)
"-(mln)
10 ppm standard In He after
exposure to atmosphere
and reinstallation
10 ppm standard in He
using new or reconditioned
column
A
MATHESON
...TRI'GAS
Sem;G.. Div;,lon
12
GC-DID Response
to O2 in He After Exposure
to CI2
10 ppm Gas standard
containing H2, O2, N2, CO and
CH4 in Helium
Pre-column: Porapak R at 40°C
Analytical column: Molecular
Sieve 13X at 30°C
A
~
MATHESON
.TRI'GAS
Semi-Gas Oivis;on
13
Porous Polymer Pre-column
.O2 analysis possible using polrous polymer
in HCI and HBr matrices
.Reactivation
required
-Injection
.In
or column
passivation
of high concentration
pre-column
techniques
of O2
-Recondition
column
in dryl O2 at 200°C
-Continuous
addition
of O2 ~o carrier gas
CI2 matrix, low levels of O2 is absorbed
with porous polymer packing i
-Detector
matrix
.SiJica-based
response
for all analytes
I
pre-column
may be
and/or reacts
is reduced in CI2
investigated
A
..TRpGAS
MATHESON
Sem"G.. Division
14
~