Contamination-Free Manufacturing: Tool Component
Qualification, Verification and Correlation with Wafers
Samantha H. Tan, Ning Chen, Shi Liu and Kefei Wang
ChemTrace Corporation, 44050 Fremont Boulevard, Fremont, CA 94538 USA
Abstract. As part of the semiconductor industry “contamination-free manufacturing” effort, significant emphasis has been placed
on reducing potential sources of contamination from process equipment and process equipment components. Process tools
contain process chambers and components that are exposed to the process environment or process chemistry and in some cases
are in direct contact with production wafers. Any contamination from these sources must be controlled or eliminated in order to
maintain high process yields, device performance, and device reliability. This paper discusses new nondestructive analytical
methods for quantitative measurement of the cleanliness of metal, quartz, polysilicon and ceramic components that are used in
process equipment tools. The goal of these new procedures is to measure the effectiveness of cleaning procedures and to verify
whether a tool component part is sufficiently clean for installation and subsequent routine use in the manufacturing line. These
procedures provide a reliable “qualification method” for tool component certification and also provide a routine quality control
method for reliable operation of cleaning facilities. Cost advantages to wafer manufacturing include higher yields due to
improved process cleanliness and elimination of yield loss and downtime resulting from the installation of “bad” components in
process tools. We also discuss a representative example of wafer contamination having been linked to a specific process tool
component.
In order to address part of this need we have
developed new nondestructive test methods for chemical
sampling and analysis of quartz, polysilicon, metals (Al,
anodized Al and stainless steel) and ceramic surfaces.
These methods sample the contamination present on or in
the surface of the component part nondestructively. This
is also the area of the component most likely to contribute
to process contamination. Quantitative measurement and
up to sub-ppb detection limits are possible. In this paper
we discuss the analytical sampling methods and present
examples of results which are possible using these
procedures. We also discuss a representative example of
wafer contamination having been linked to a specific
process tool component that had not been cleaned
adequately.
INTRODUCTION
Constant efforts to improve the cleanliness of the
wafer fabrication process are essential to improve
manufacturing cost-effectiveness and to achieve yields
greater than 90% [1,2]. Process tool cleanliness is
important to the maintenance of a “contamination-free
manufacturing” environment and becomes more critical
with the continued device scaling ever present in the
progress of the semiconductor industry. Process tools
contain process chambers and components that are
exposed to the process environment or process chemistry
and in some cases are in direct contact with production
wafers. Inorganic and organic contamination incorporated
on or in the surfaces of these tool components has a
potentially serious negative impact. Any contamination
from these sources must be controlled or eliminated in
order to maintain high process performance and yields,
device performance, and device reliability. It is important
to establish cleanliness standards and measurement
methods for all of the material types used in the process
tool in this manner.
Traditional analytical techniques such as VPD/ICPMS, TXRF, XRF, SIMS, AES and XPS are commonly
used to detect inorganic and organic contamination on
wafer surfaces [3,4]. However, these techniques are
typically not suitable for the cleanliness measurement of
process tool components due to the large geometries and
topographies involved. New nondestructive test methods
must be developed to measure and verify the cleanliness
of 200mm and 300mm tool components.
METHOD FOR MEASUREMENT OF
TOOL PART CLEANLINESS
In summary, the measurement method consists of a
nondestructive wet chemical sampling techniques
followed by analysis using one of the following methods:
•
•
•
•
Inductively coupled plasma-mass spectrometry (ICPMS) for inorganics and metals
Ion chromatography (IC) for anions and cations
Gas chromatography mass spectrometry (GC-MS)
and thermal desorption GC-MS for organic residues
Total organic carbon (TOC) analysis for organic
residues
CP683, Characterization and Metrology for ULSI Technology: 2003 International Conference,
edited by D. G. Seiler, A. C. Diebold, T. J. Shaffner, R. McDonald, S. Zollner, R. P. Khosla, and E. M. Secula
© 2003 American Institute of Physics 0-7354-0152-7/03/$20.00
289
•
extraction and followed by multiple concentrated
HF/HNO3 depth etchings have been performed on exactly
the same area providing a depth profile of trace metal
surface concentration to a depth of approximately 100um.
Laser particle counter (LPC) for particles
Concentration (ppm)
The sampling technique is performed by the
controlled addition of small quantities of dilute acid, ultra
pure water (UPW) or organic solvent to the surface of the
component part. After the solution has reacted with the
surface or surface contamination dissolved the solution is
collected and analyzed.
As an example of this methodology we will discuss
the protocol for quartz tool component surface analysis.
In this case 1-2 mL of a dilute solution, that is
hydrofluoric and nitric acid based (HF and HNO3), is
applied to the tool part surface. This solution will cover
an area of approximately 50-100 cm2 and dissolves very
thin surface layer and extracts metals that may be present.
This thin layer is collected by a pipette and injected into
an ICP/MS instrument for analysis. This analytical
procedure is very similar to the vapor phase dissolution
method (VPD) established for the measurement of trace
metals on silicon wafer surfaces. The sensitivity of the
method when applied to silicon wafers is 108 to 109
atoms/cm2 [3,4].
Quantitation of the contamination
present on quartz surfaces is made possible by measuring
the sample area and depth. The accuracy’s obtained are
approximately +/- 20% for quartz surfaces.
Ultrapure water is used to extract anions including
fluorides and chlorides. In this case, subsequent analysis
is by IC. Particles from the tool part surface are measured
after extraction into a solution and the particles in the
solution are measured using a modified liquid particle
counting technique. Surface organic contamination is
extracted using an organic solvent. The extract is
subsequently analyzed by GC-MS analysis. Volatile
organic compounds may be outgassed and collected by a
special outgassing apparatus and subsequently analyzed
using TD GC-MS.
These techniques have been found to be generally
useful and nondestructive for metals (e.g. Al, anodized Al,
10,000 Z1 Zone 2
1,000
Bulk Material
10
Al
1
1.00
Na
Mg
0.10
Ca
Cu
0.01
Fe
Ti
0.00
10
20 30 40
50
60 70
80 90
Depth (microns)
FIGURE 1. Dilute acidic extraction/ICP-MS depth profile
Concentration (ppm)
of a quartz ring manufactured by process A
10,000 Z1
1,000
Zone 2
Bulk Material
10
Al
1
1.00
Na
Mg
0.10
0.01
0.00
Ca
Cu
Fe
Ti
10
20 30 40
50
60 70
80 90
Depth (microns)
stainless steel), quartz, polysilicon and ceramic component
surfaces used in process equipment. In the case where the
FIGURE 2. Dilute acidic extraction/ICP-MS depth profile of
collection solution is nondestructive and the part is small
enough, total immersion in the sampling solution is also
possible and has been used in these measurements as well.
a quartz ring manufactured by process B
The depth profile information is important since
production of quartz components requires annealing steps
to relieve the stress in the quartz. Any surface
contamination on the quartz surface will be driven into
the bulk of the material during annealing. The depth
profile produced by the surface acidic analysis and
multiple depth etchings provide contamination
information from the surface to over 100 um in depth.
We have broken the regions of interest into three zones.
Zone 1 is closest to the surface and contamination in this
zone is introduced from machining, handling, cleaning,
and packaging operations. Sub-surface contamination in
Zone 2 reflects process contamination from machining,
VERIFICATION OF TOOL PART
CLEANLINESS
After manufacture and precision cleaning, tool part
cleanliness may be verified using the described
techniques. For example, trace metals present on the
surface of quartz rings produced by two different
manufacturing sources are presented in Figures 1 and 2.
For the case presented in Figs. 1 and 2, a dilute acidic
290
steel part A is seen to be heavily contaminated with
organics while part B has no detectable organic residues
with the exception of isopropyl alcohol used for cleaning.
The results also show that Part B contains a very high
sulfur content that is known to be detrimental to the wafer
process. Stainless steel part A is most likely a different
type of stainless steel.
polishing, bead blasting, welding, and annealing. The
contamination in Zone 2 typically originates from the
surface and is driven into the bulk material of the
component part. It is important to fabricate a critical
chamber part from high purity machining tool and
processes to prevent sub-surface contamination of the
chamber part. The results in Figures 1 and 2 clearly show
differences between two different fabrication processes.
Both processes used the same starting quartz material, as
illustrated by the concentration levels in the bulk of the
material.
Organic contamination residue on quartz parts can
be measured in a very similar way. In this case ultrapure
water is used to extract the contamination on the part
surface. Application and collection is performed in the
same manner as described above.
The organic
contamination dissolved by the ultrapure water is
measured using Total Organic Carbon Measurement
Instrument (TOC). The surfaces of the quartz bell jar,
pre- and post-clean, were analyzed using this method and
the results are presented in Table 1. These data illustrate
that organic surface residues have been reduced by about
a factor of 5 as a result of the cleaning step.
Abundance
40000
00
S8
IPA
A
30 0
35000
00
30000
00
Ethylhexanol
25000
00
20000
00
DOP
12 7
2-Ethyl-hexanal
15000
00
29 2
10 9
10000
00
36 1
23 01 26 7
A
5000
00
0
SURFACE CONCENTRATION
( x 1014 atoms/cm2 )
TOC
DL
Pre-Cleaned
300mm Bell Jar
Post-Cleaned
300mm Bell Jar
5.0
703
141
B
Time-
5.0
0
10.0
0
15.0
0
20.0
0
25.0
0
30.0
0
35.0
0
Figure 3. TD GC-MS analysis of organic contamination
products from two stainless steel process tool component parts
CORRELATION OF TOOL PART
CLEANLINESS WITH WAFER
CLEANLINESS
TABLE 1. Total organic carbon (TOC) concentrations
from an intact 300mm-quartz bell jar
We will now give an example where surface
contamination was detected on process monitor wafer
surfaces and traced back to a process tool component part
using these methodologies. In this case, magnesium and
copper were observed on process monitor test wafers at a
concentration of greater than 1x1010 atoms/cm2. The
likely source of this contamination was traced to ceramic
“lift pins” which are commonly used in process tools to
raise the wafer up off of a wafer chuck for transfer to the
next process operation. The lift pins were removed from
the tool and analyzed for surface metal contamination
using the dilute acidic extraction/ICP-MS method.
Analysis results for pre-cleaned (contaminated pin as
received), post-cleaned (pin cleaned and analyzed), and
new pin (packaged new pin) are shown in Table 2. High
levels of Mg and Cu are observed on the pre-cleaned pin.
Magnesium is observed on all the pins at about the same
concentration level of 1.5x1015 atoms/cm2 and is most
likely a binder additive that is added to the bulk ceramic
Specific components present in organic residues on
part surfaces can be identified by direct outgassing of the
organic residues in a TD GC-MS analysis instrument
providing the parts fit within the analysis chamber. A
large stainless steel collection chamber has been built for
analysis of organic residues on larger parts. In this case
the parts are heated at 200 oC for 30 mins while the
chamber is purged with helium (He) gas. The volatile
organic residues desorbed from the part surface into the
He gas are then trapped by passage of the gas through an
organic absorbent called Tenax in a glass tube. The
Tenax tube is subsequently analyzed for organics by TD
GC-MS instrument. As heating in the analysis chamber
occurs the trapped organic impurities are released and
analyzed. The results obtained for two large stainless
steel process tool component parts were analyzed by this
method the results are presented in Figure 3. Stainless
291
during manufacturing. Copper showed a dramatic
decrease from 6.6x1014 atoms/cm2 to a detection limit
level of 2x1011 atoms/cm2 after cleaning. It was found that
Cu is present on unused or “new” pins at the significant
level of 2.1x1013 atoms/cm2. This contamination is
present on the pins in the as-received condition from the
manufacturer. The next step in identifying the root cause
of the contamination being introduced on wafer surfaces
was the finding that “new pins” were being installed
during process tool preventative maintenance procedures
without further cleaning.
Particulate contamination measurement results performed
in this manner for polysilicon tool component parts are
presented in Fig. 4.
5. In this case, two parts from different
manufacturers with different cleaning process were
examined. Particles were extracted from the surface of
the parts by full immersion in an ultrapure water bath
while being exposed to ultrasonic energy agitation.
Measurements were made by a laser particle counter as a
function of immersion time. Particle density calculations
were normalized using the calculated surface area of the
part.
SURFACECONCENTRATION( x 1010 atoms/cm2)
Element
Aluminum
Calcium
Chromium
Copper
Iron
Magnesium
Manganese
Molybdenum
Nickel
Potassium
Sodium
Titanium
Zinc
DL
(Al)
(Ca)
Cr)
Cu)
(Fe)
(Mg)
(Mn)
(Mo)
(Ni)
(K)
(Na)
(Ti)
(Zn)
100
100
50
20
50
100
10
3
20
100
100
50
30
Pre-Clean
4,000,000
310,000
65,000
66,000
3,900,000
110,000
31,000
20,000
2,300,000
84,000
210,000
7,400
100,000
Post-Clean
310,000
110,000
<50
20
9,800
130,000
750
12
120
4,800
40,000
2,600
1,200
SURFACE CONCENTRATION ( 1010 atoms/cm2)
New
Element
350,000
91,000
<50
2,100
2,900
180,000
280
<3
100
10,000
110,000
740
110,000
Barium
Boron
Calcium
Cobalt
Copper
Iron
Lithium
Magnesium
Manganese
Nickel
Potassium
Sodium
Strontium
Tin
Titanium
Tungsten
Zinc
Zirconium
TABLE 2. Trace metal analysis of ceramic lift pin surfaces,
pre- and post-clean, and “new.”
Ceramic A
(Ba)
(B )
(Ca)
(Co)
(Cu)
(Fe)
(Li)
(Mg)
(Mn)
(Ni)
(K )
(Na)
(Sr)
(Sn)
(Ti)
(W)
(Zn)
(Zr)
Ceramic B
3,000
120,000
1,100,000
86
610
16,000
3,300
77,000
310
1,700
180,000
530,000
880
280
1,200,000
610
2,000
770
910
1,700
91,000
12
2,100
2,900
120
180,000
280
100
10,000
110,000
280
14
740
<5
110,000
6,000
Ceramic C
2.2
630
660
2.4
17
380
<3
340
7.5
3.4
44
350
1.7
1.1
110
<0.5
140
40
TABLE 3. Dilute acidic extraction ICP/MS analysis results
from the surface of ceramic from three different manufacturers
identified above as Ceramic A, B and C.
COMPONENT PART MATERIAL
SELECTION AND CONTAMINATION
10000
Normalized Particle
2
Counts per cm
In efforts to identify root cause contamination sources
in the above study, ceramic pins from three different
manufacturers were analyzed using the dilute acidic
extraction/ICP-MS method. The results are presented in
Table 3. Pins from the three different manufacturers are
identified as Ceramics A, B and C. Calcium, Na, K and
Mg are common impurities found in ceramic as they are
found in the binders commonly used in the manufacturing
of ceramics. Ceramic from manufacturers A and B
contain these metals plus other impurities such as B, Fe,
Zn, Cu, Ni and Zr. Ceramic C pins have very low metal
contamination as they are manufactured from a ceramic
that is not made with the traditional binder.
Particles on chamber parts can also be measured by
liquid extraction techniques aided by ultrasonic agitation.
Ultrasonic energy causes cavitation at the liquid-solid
interface promoting the loosening of surface particles.
1000
100
10
1
0
20
40
60
80
100 120 140 160 180
Extraction Time (min)
FIGURE
Figure 5.4. Baselining of polysilicon parts for 0.2-micron
particles
The results show that particle counts decrease with
increasing extraction times, until a baseline is reached.
292
Part cleaning methods were evaluated by direct
measurement of surface particle count density and
correlation to subsequent wafer level particle counts when
the component part was installed and used in the process
tool under normal process conditions. Polysilicon
component tool parts with particle counts of less than 20
/cm2 presented no particle problems to the wafer during
chamber start-up. In contrast, parts exhibiting >100
particle counts/cm2 showed high particle counts on the
wafer during start-up required chamber conditioning to
reduce particle counts to acceptable levels. These
measurement methods were subsequently used to evaluate
the effectiveness of cleaning procedures on particulate
contamination levels. The results of these studies
demonstrated that it was possible to minimize tool
qualification and process startup time by the optimization
of cleaning procedures through the use of these
measurement methodologies.
2.
3.
4.
CONCLUSIONS
New nondestructive methods for the measurement of
process tool component parts have been successfully
developed and applied.
These test methods are
applicable for parts made from a variety of materials
including quartz, polysilicon, metals (Al, anodized Al and
stainless steel) and ceramics. These methods sample the
contamination present on or in the surface of the
component part, the area of the component most likely to
contribute to process contamination. It has been shown
that these methods can be used to evaluate the
effectiveness of cleaning procedures and to verify
whether a tool component part is sufficiently clean for
installation and subsequent routine use in the
manufacturing line. These procedures provide a reliable
“qualification method” for tool component certification
and provide a routine quality control method for reliable
operation of cleaning facilities. Use of these methods in
an analytical problem-solving mode enables the linkage
of wafer level contamination to specific process tool
component sources. Cost advantages that occur as a
result of the application of these methods in the
semiconductor and other industries include higher yields
due to improved process cleanliness and elimination of
yield loss and downtime resulting from the installation of
“bad” components in process tools.
REFERENCES
1.
David Jensen,
”National Technology Roadmap for
Semiconductors”, Contamination-Free Manufacturing for
Semiconductors and Other Precision Products, edited by
293
Robert R. Donovan, Marcel Dekker, Inc., New York, pp.725.
International Technology Roadmap for Semiconductors,
Semiconductor Industry Association, pp. 155-157, (2000).
Novel Methods for Trace Metal Analysis in Process
Chemicals and DI Water and on Silicon Surfaces, P,
("Kim") Gupta, S. H. Tan, Z. Pourmotamed, F. Cristobal,
N. Oshiro and B. McDonald, Proceedings of Contamination
Control and Defect Reduction in Semiconductor
Manufacturing III, R. Novak, R. Reedy, T. Ito, and D. N.
Schmidt, editors, Electrochemical Society Proceedings PV
94-3, 200 (1994).
Samantha H. Tan, “Application of Vapor Phase
Decomposition Techniques (VPD/AAS and ICP-MS) for
Trace Element Analysis in Oxide Coatings on Silicon,”
Nuclear Instruments and Methods in Physics Research,
B99, pp. 458-461 (1995).