Developing a Reliable Fluorinated Greenhouse Gas (F-GHG)
 Destruction or Removal Efficiency (ORE) Measurement Method
 for Electronics Manufacturing: A Cooperative Evaluation with
                    NEC Electronics, Inc.
                         December 2008
 %
-------
Acknowledgements
The analytical measurements, data interpretation, and report preparations were funded by the
U.S. Environmental Protection Agency (EPA) under contract GS-10F-0124J to ICF International
and Air Products and Chemicals, Inc. EPA and the authors wish to express their appreciation
and thanks to NEC Electronics Inc., for their gracious support to this study by not only providing
their facilities but also their valuable assistance and advice. EPA looks forward to continued
collaborations with NEC Electronics Inc. and other electronics industry Partners in its
cooperative efforts to reduce greenhouse gas emissions and advance global climate protection.

-------
                                Table of Contents
                                                                                Page

Acknowledgements	3
1.0     Introduction	6
2.0     Experimental Setup	6
3.0     Data Analysis	9
       3.1    Determination of Scrubber Dilution	9
       3.2    Scrubber DRE Determinations	17
             3.2.1   TecHarmonic E-HTVS DRE	17
             3.2.2   Edwards TPU DRE	21
4.0     Conclusion	23

-------
                                    List of Tables
                                                                                   Page
Table I        Total equivalent flow measurements from E-HTVS determined using Xe and Kr	11
Table II       Total flow measurements from TPU determined using Xe and Kr	12
Table III       Ar concentration determined in TPU effluent during Ar flow testing	12
Table IV       Summary of the average total flow determinations for E-HTVS and TPU at NEC
              Electronics Inc	13
Table V       Total process flow determination for CVD Tool #1 using C2F6tool flows	15
Table VI       Total process flow determination for CVD Tool #2 using NF3 tool flows	16
Table VII      C2F6 DRE in E-HTVS during process  flow calibrations	18
Table VIII     Integrated F-GHG process and scrubber emissions for CVD Tool #1 (process) and E-HTVS
              during chamber cleans	21

-------
                                    List of Figures
                                                                                     Page
Figure 1     Actual sampling schematic used for E-HTVS and TPU at NEC Electronics Inc ................... 7
Figure 2     QMS response to Kr (top) and Xe during QMS calibration and E-HTVS spiking of Kr and
            Xe [[[ 8
Figure 3     Regression analysis of data obtained during calibration of QMS response to 86Kr and 132Xe in
            N2 [[[ 8
Figure 4     Calibration of FTIR response to C2F6 [[[ 9
Figure 5     Xe (top) and Kr (bottom) emissions from E-HTVS during spike calibration .................... 12
Figure 6     Kr (top) and Xe (bottom) emissions from TPU during chemical spiking of scrubbers ......... 14
Figure 7     Ar emission from TPU while Ar was being flowed through the process chamber on CVD
            Tool #2 [[[ 14
Figure 8     C2F6 emission profile for CVD Tool #1 during process flow calibration. C2F6 emissions from
            a 5 minute chamber clean are also shown [[[ 16
Figure 9     NF3 emission profile from CVD Tool #2 during NF3 process flow calibration ................... 17
Figure 10    Inlet and outlet C2F6 emission profiles for E-HTVS during process flow calibration .......... 18
Figure 11    FTIR spectrum of process emissions during C2F6/O2 chamber clean on CVD Tool #1 ......... 19
Figure 12    C2F6 and CF4 emissions from CVD Tool #1 (top) and E-HTVS scrubber during chamber

-------
1.0 Introduction

The purpose of this study was to accurately determine the Destruction or Removal Efficiency
(DRE) of two different commercially available Point Of Use (POU) abatement systems or
scrubbers for process emissions fluorinated greenhouse gases (F-GHGs) such as
perfluorocarbons (PFCs), hydrofluorocarbons (HFCs), sulfur hexafluoride (SF6), and nitrogen
trifluoride (NFS). A key component in accurately determining DRE was to determine the dilution
of process exhaust occurring in each scrubber.  This study used an experimental approach to
measure the dilution across the scrubber by injecting a chemical tracer that could not react in the
scrubber, or be produced as a by-product during scrubber operation. In this study Krypton,
Argon and Xeon were used as chemical spiking agents, as they met the requirements for this
application.

Testing was conducted in a fully functional semiconductor manufacturing facility, owned and
operated by NEC Electronics Inc. in Roseville, CA. Two different process tools, referred to as
CVD Tool #1 and CVD Tool #2, each equipped with a POU scrubber, were tested. CVD Tool
#1 equipped with a TecHarmonic E-HTVS POU scrubber was tested, and CVD Tool #2
equipped with an Edwards TPU POU abatement device was also tested.

2.0 Experimental Setup

To carry out the objectives of this study it was necessary to monitor both process and scrubber
emissions simultaneously, and determine scrubber dilution using chemical spiking. Process and
scrubber emissions data were collected in parallel using Fourier Transform Infrared
Spectroscopy (FTIR). Data used to determine scrubber dilution were collected using Quadrupole
Mass Spectrometry (QMS).  A  schematic showing the experimental testing set up is shown in
Figure 1.

Two FTIRs were used to determine process and scrubber emissions. Both systems were MKS
2010 Multi Gas Analyzers equipped with liquid nitrogen cooled mercury cadmium telluride
(MCT) detectors. One FTIR was equipped with a 10 cm path length single pass gas cell, and
was used to sample process effluent. The other FTIR was equipped with a 5.6 m path length
multi pass gas cell,  and was used to sample scrubber effluent. Both FTIR were operated at
0.5cm"1 resolution.  Four scans were co-added for each data point yielding a sampling frequency
of 2.2 sec.

A Balzers QMS system was used to sample scrubber effluent during dilution determination. The
QMS was operated in Selective  Ion Monitoring (SIM) mode and a secondary electron multiplier
was used to enhance sensitivity. A 1 sec sampling frequency was used for each data point.  To
account for potential changes in QMS sensitivity, ion signals were normalized to the signal
obtained for the nitrogen fragment (N+), which is formed during electron impact ionization of
N2.

Sampling of effluent streams was done using metal bellows sampling pumps that were located
after the instruments. The sample flow rate was controlled using adjustable flow rate valves.
The sample line pressure for both FTIRs and the QMS were monitored using capacitance

-------
manometers.  A filter was installed in the sample line used for monitoring scrubber emissions to
ensure that particulate emissions from the scrubber would not coat the FTIR internal optics, or
the pressure reducing orifice used for the QMS.  Since the scrubber DRE determination and the
scrubber dilution determination were independent  events, it was possible to use the same sample
line for both operations.  This was accomplished by switching the instrument inlet sample fitting
from the FTIR to the QMS.

The QMS was calibrated to determine its response to Xe, Ar and Kr on site using a dynamic
dilution blending system.  Test atmospheres containing H2, He, Kr, Ar and Xe were created by
blending a calibration standard containing 1% of each species in N2 with N2 diluent. The QMS
responses  to 84Kr and 132Xe during calibration are shown in Figure 2. From regression analyses
of these data a calibration curve was generated and is shown in Figure 3. These data were used
to quantify Kr and Xe emissions from the scrubber.
                                                                   To Scrubbed Exhaust
 Effluent from Process
 Pump to Scrubber
                                                                      Sample Pump
Figure 1: Actual sampling schematic used for E-HTVS and TPU at NEC Electronics Inc.
Schematic shows the calibration and sampling capability for each instrument.

-------
                       Kr Emissions from E-HTVS during Flow Calibration
  |  4O
                                       QMS Scan
                      Xe Emissions from E-HTVS during Flow Calibrattions
                                      .00      1000      1200      1 400      1 600      1 8PO
                                       QMS Scan
Figure 2: QMS response to Kr (top) and Xe during QMS calibration and E-HTVS spiking
of Kr and Xe. (Note: Kr and Xe emissions in the figure titles denote the flow of tracer
materials and have nothing to do with process emissions.)
                              QMS Responser to Kr and Xe
    1 .OOE-O3 -


    9.00E-04 -


    8.00E-04 -


    7.OOE-O4 -


    6.OOE-O4 -


    5.00E-04 -


    4.00E-04 -


    3.OOE-O4 -


    2.OOE-O4 -


    1 .OOE-O4 -


    O.OOE+00 -
                                                                                   S6,
Figure 3: Regression analysis of data obtained during calibration of QMS response to  Kr
    1 ^7
and   Xe in NI. These data were used to determine Kr and Xe concentration during
chemical spiking of scrubbers.

-------
The FTIR was calibrated on-site using test atmospheres created through dynamic dilution
blending of a 1% C2p6 in N2 standard.  Several mixtures were created to calibrate the response of
the FTIR equipped with the 5.6m gas cell.  Figure 4 shows the calibration curve generated for
the CF3 symmetrical deformation of C2F6 (centered at 714 cm"1) as measured by the maximum
absorbance of the R branch for each calibration point.  The relative error of this calibration was
determined to be 3.1 %.
           Absorbance for R Branch of CF3 Symetrical Defromation (714 cm") as function of C2F6
                                    Concentration
    0.8 -i
                                            Intercept: 0.0173 ± 0.014
              200
                       400      600       800       1000
                          Concentration of Added C2F6 (ppmv)
                                                          1200
                                                                  1400
Figure 4: Calibration of FTIR response to CiFe.  Note 1: These data were generated on-site
at NEC Electronics Inc. using a 1% CiFe in NI standard. Note 2: The least squares line is
not shown in Figure, but the relevant parameters of regression line and statistics are
presented.

3.0 Data Analysis

3.1  Determination of Scrubber Dilution

One of the primary goals of this  study was to accurately determine the dilution that occurs when
gas emitted from the process chamber passes through the scrubber. Dilution can occur from
many sources including effluents from other chambers, combustion gases and by-products added
to and generated within the scrubber, vapors added as the gas stream passes through the water
scrubber portion of the system, in-board leaks, and back diffusion from main headers. The
method of determining dilution in this study was to use a purely experimental approach where  a
chemical was spiked into the gas stream entering the scrubber at a known flow rate, and
determined in the scrubber effluent stream.  From the determined concentration and the

-------
controlled flow rate added to the process exhaust duct, a total flow from the scrubber could be
calculated:

                           TF=  Sf/(CanX10-6)                     (1)

Where Sf represents the spike gas flow and is reported in liters, and C^ represents the analyte
concentration reported in ppmv.

The experiment conducted to determine total flow from the E-HTVS scrubber consisted of using
the calibration system shown in Figure 1 to add calibration gas into the process effluent through
the FTIR sample line where the process effluent was monitored. While calibration gas was being
added, the QMS was used to sample scrubber effluent. The flow of calibration gas was
controlled with a 0 - 5 slm MFC that was calibrated for nitrogen. Five flow rates were added to
the scrubber:  1, 2, 3,  4 and 5 slm. At 1% concentration, each of these flows corresponded to Kr
and Xe flows of 0.01, 0.02, 0.03, 0.04 and 0.05 slm, respectively.  Concentration profiles for Xe
and Kr determined from QMS data during this experiment are shown in Figure 5 for CVD Tool
#1, which was the tool used for testing the E-HTVS scrubber.

Applying Eq. 1 to the data shown in Figure 5 for the E-HTVS scrubber yielded the total
equivalent flows determined for both Xe and Kr, which are contained in Table I. The average
total flow for each species was:

                    Xe:  741 ±3                Kr: 765 ±3

Here the reported standard deviations are the best estimate of the standard deviations of the mean
for total flows contained in Table I.  Combining the data for Kr and Xe yielded a best estimate
of effective flow of 757 ± 2 slm, which is the variance weighted average of the mean.

The precision of these measurements were ± 0.3 %. The differences between the flow values
(3%) obtained for Kr  and Xe could be attributable to other factors affecting accuracy, such as the
accuracy of the Kr and Xe concentrations in the calibration standards.
                                       10

-------
Table I:  Total equivalent flow measurements from E-HTVS determined using Xe and Kr.
Xe/Kr Flow
(slm)
0.010
0.020
0.030
0.040
0.050
Xe Cone
(ppmv)
12.7 ±1.7
26.9 ± 2.0
40.1 ± 3.1
55.4 ± 4.5
67.6 ± 4.7
Total
Equivalent
Flow
(slm)
787 ± 122
744 ± 52
748 ± 58
722 ± 59
740 ± 51
Kr Cone
(ppmv)
14.4 ±1.3
28.1 ± 2.3
37.8 ± 2.7
50.2 ± 2.8
64.8 ± 2.4
Total
Equivalent
Flow
(slm)
694 ± 63
712 ± 58
794 ± 57
797 ± 44
772 ± 29
                      Xe Emissions from E-HTVS during Flow Calibrattions
  [  40
                                      QMS Scan
                      Kr Emissions from E-HTVS during Flow Calibration
             200      400      600
                                    .00     1000     1200      1 400     1 600     1 SJpO
                                      QMS Scan
Figure 5:  Xe (top) and Kr (bottom) emissions from E-HTVS during spike calibration.
Calibration data were generated using test atmospheres as described above.
                                      11

-------
The same approach was used for determining flow from the TPU.  Figure 6 shows Xe and Kr
emissions from the TPU during spiking of the scrubber. The same Xe and Kr flow rates of 0.01,
0.02, 0.03, 0.04, and 0.05 slm were used. Table II contains the total flow determined for each
species at each spike flow rate. The average total flow for each species was:
                        Xe: 751 ±3
Kr:  838 ±2
Here the reported standard deviations are the best estimate of the standard deviations of the mean
for total flows contained in Table II. Combining the data for Kr and Xe yielded a best estimate
of effective flow of 815 ± 2 slm, which is the variance weighted average of the mean.

The precision of these measurements were ± 0.3 %. The difference between the flow values
observed for Kr and Xe on the TPU (11%) were greater than those observed on the E-HTVS
(3%). An additional flow determination method was deployed.

An additional experiment using Ar was conducted on the TPU. The experiment consisted of
flowing Ar from the CVD Tool #2 through the scrubber and determining the Ar concentration
emitted from the scrubber. Figure 7 shows the Ar emission profile for 4 process flows: 2.0, 1.5,
1.0 and 0.5 slm. As indicated in Figure 7, Ar background concentration was relatively high due
to the natural atmospheric abundance of 0.94 %. This background was subtracted from the
measured Ar concentration during the Ar flows to yield a concentration that could be attributable
to the added Ar. Table III contains the data from this experiment.  The data in Table III was
used to determine the TPU flow by dividing the Ar process flow (in si) by the differential Ar
concentration (AAr). The weighted average total flow determined during this experiment was
826 ± 2 slm. This value is in close agreement (± 1.5%) with the reported effective dilution for
the Kr spike test (838 slm).
Table II: Total flow measurements from TPU determined using Xe and Kr.
Xe/Kr Flow
(slm)
0.010
0.020
0.030
0.040
0.050
Xe Cone
(ppmv)
12.3 ± 1.3
27.2 ± 2.2
42.2 ± 2.7
51.6 ±4.1
63.6 ± 5.0
Total
Equivalent
Flow
(slm)
813 ± 86
735 ± 60
711 ±45
775 ± 62
786 ± 62
KrConc
(ppmv)
12.7 ± 0.9
22.8 ± 2.5
35.0 ± 1.5
47.6 ± 1.6
59.9 ± 1.9
Total
Equivalent
Flow
(slm)
787 ± 56
877 ± 95
857± 37
840 ± 28
835± 26

Table III: Ar concentration determined in TPU effluent during Ar flow testing. AAr
concentration is the measured concentration minus the baseline.
Ar Process Flow
(slm)
Ave. Ar Cone.
(ppmv)
A Ar Cone
(ppmv)
Total TPU Flow
(slm)
                                       12

-------
Baseline
(0)
2.0
1.5
1.0
0.5
Baseline
(0)
8685 ± 22
11,098 ±154
10,302 ± 33
9996 ± 27
9321 ± 18
8655 ± 18
-
2413 ±156
1617 ±40
1311 ±35
636 ± 28
-
-
828 ± 53
927 ± 23
762 ± 20
786 ± 35
-
A summary comparing total flows determined for both scrubbers is contained in Table IV.
Results for Kr and Xe on the E-HTVS scrubber were within 3%, while on the TPU system the
difference was 11%.  In both cases the Kr based determination was higher than Xe
determination. The dilution determined using Ar on the TPU was found to be within 1.5 % of
the value determined using Kr.  Both values were > 10 % higher than the value determined using
Xe. Considering that using Ar supplied by the process tool is independent of the calibration
standard accuracy for the spiking process, this result suggests that inaccuracies in the Xe
determination may more likely have occurred. Given the relative close agreement in results for
all flow determinations, this was not further investigated.  However, the Xe discrepancy warrants
further investigation in subsequent evaluation of the methodology.

Table IV: Summary of the average total flow determinations for E-HTVS and TPU at
NEC Electronics Inc. All flows reported as slm.
POU System
E-HTVS
TPU
Xe Spike
741 ±3
751 ±3
Kr Spike
765 ±3
838 ±2
Ar On-Tool
N/A
826 ±2
                  Kr Concentration Determiend during TPU Spike Flow Calibration
                                       13

-------
                  XeConcentration Determiend during TPU Spike Flow Calibration
                                         300
                                       QMS Sscan
Figure 6:  Kr (top) and Xe (bottom) emissions from TPU during chemical spiking of
scrubbers.
                        Ar Emission from TPU during NFS DRE Testing
Figure 7:  Ar emission from TPU while Ar was being flowed through the process chamber
on CVD Tool #2. Baseline Ar concentration was close to that expected in air.

Once the total flow from the scrubbers has been determined it was then possible to determine the
dilution of process effluent occurring across the scrubbers. This calculation requires measuring
the dilution that occurs as gases from the process chamber are pumped out of the chamber and
fore line and sent into the corrosive scrubbed exhaust. The experiment to measure the dilution of
CVD Tool #1 and CVD Tool #2 chamber effluent consisted of flowing C2F6 into the CVD Tool
#1 chamber and NFs into the CVD Tool #2 chamber, with the radio frequency  (RF) power in the
chambers  turned off, at several flow rates.  The determined C2p6  and NF3 concentrations in the
chamber effluent could be used to calculate the total process flow  entering the  scrubber from the
following  equation:
                                       14

-------
                           TPF= PGf/(CPGX10-6)
                                                                   (2)
Here the total process flow (Tpp) is determined from the ratio of the process gas flow (PGf) in
slm divided by the measured concentration (CPG) in ppmv.

Process flow data for the E-HTVS system were obtained by flowing C2p6 through the CVD Tool
#1 chamber at several flows with the RF power turned off. Figure 8 shows the C$6 emission
profile for 5 different flows. From these data the total process flow was determined and is
contained in Table V.  The average total process flow was 83.58 ± 0.03 slm.  Combining this
value with the total scrubber flow yielded the effective dilution of influent gas across the E-
HTVS:
                  Dilution Factor = (757 ± 2)7(83.58 ± 0.03) = 9.05± 0.02

This value was used for all subsequent DRE calculations. This value represents the effective
dilution that was determined for the E-HTVS.  These data do not necessarily imply a total flow
of 757 slm was flowing through the abatement device, but do imply that the chemical spiking
agents were diluted to an equivalent flow of 757 slm. The source of this dilution was not
determined, but could be attributable to factors such as in-board leaks in the duct work between
the scrubber outlet and the sample location or mixing at the sample location from the main
header of the scrubbed corrosive exhaust system. Having an accurate determination of this
effective  dilution is critical in determining an accurate DRE for the scrubber.
Table V: Total process flow determination for CVD Tool #1 using
                                                                    tool flows.
C2F6 Flow
(slm)
1.0
0.8
0.6
0.4
0.2
Ave. C2F6 Concentration
(ppmv)
11,714±61
9461 ± 50
7232 ± 43
4889 ±32
2500 ±26
Total Process Flow
(slm)
85.4 ±0.4
84.6 ±0.4
83.0 ±0.5
81. 8 ±0.5
80.0 ±0.8
                                       15

-------
                  C2F6 Emissions from E-HTVS during Process Flow Calibrations
                                 300      400
                                        FTIR Scan
Figure 8:  CiFe emission profile for CVD Tool #1 during process flow calibration.
emissions from a 5 minute chamber clean are also shown.
Total process flow into the TPU was determined by flowing NFs from the chamber with RF
power off. Process emissions from six flows are shown in Figure 9. From these data the total
process flow was determined to be 15.50 ± 0.01 slm. Table VI contains NF3 concentrations
determined for each flow.  The total process flow and total TPU exhaust flow were used to
calculate the dilution occurring in the TPU:

                  Dilution Factor = (831 ± 1)7(15.50 ± 0.01 ) = 53.6 ± 0.1
Having determined the total process flow for each tool and scrubber allows determination of the
dilution occurring across the scrubber.  These data were used to determine the scrubber F-GHG
ORE.
Table VI: Total process flow determination for CVD Tool #2 using NFa tool flows
NF3 Flow
(slm)
0.50
1.00
1.50
0.20
0.25
0.30
Ave. NFs Concentration
(ppmv)
31,735 ±157
67,886 ± 974
93,964 ± 685
13,098 ±83
16,147 ±50
19,020 ±41
Total Process Flow
(slm)
15.8±0.1
14.7 ±0.1
16.0 ±0.1
15.3 ±0.1
15. 5 ±0.1
15. 8 ±0.1
                                       16

-------
                  NFS Emissions from AMAT HDP during Process Flow Calibration
Figure 9:  NFs emission profile from CVD Tool #2 during NFs process flow calibration.
3.2 Scrubber DRE Determinations
Determination of the E-HTVS scrubber performance was performed using two different testing
conditions.  The first was to measure the scrubber effluent of C$6 during the total process flow
calibrations and the second was to measure the scrubber effluent during C2P6/O2 chamber cleans.
Two chamber cleans were monitored. The first was a short 5-minute clean of a chamber that was
relatively clean (little tungsten accumulation in the chamber).  The second clean was longer and
was performed after an accumulation of approximately (ca.) 20 jim of film that had built up in
the chamber.  This clean used an optical end-point mechanism to determine the length of the
clean. The process recipe for the chamber clean was to flow 1.2 slm C$6 and 1.2 slm O2.
Determination of the TPU performance could only be done using NFs flows from the tool gas
delivery system, as the tool chosen for this project was not operational at the time of testing.
Several NFs flows were sent through the process chamber to determine NFs DRE.  However, no
process wafers could be run and the system could not be evaluated under manufacturing
conditions.
3.2. 1 TecHarmonic E-HTVS DRE
The E-HTVS DRE for C$6 was determined during process flow calibrations. Figure 10 shows
C$e emissions from process and scrubber during the flow calibration. Using the average
measured C$6 concentration for both inlet and outlet samples from each step and applying the
dilution correction yields the C$6 DRE for the E-HTVS scrubber during process flow
determination. These averages for each C$6 flow are contained in Table VII. The associated
DRE values for each C$6 flow are also contained in Table VII.
                                       17

-------
Table VII: C2F6 DRE in E-HTVS during process flow calibrations. Dilution adjustment
was based total measured flows into and out of the scrubber.
C2F6 Flow
(slm)
1.0
0.8
0.6
0.4
0.2
C2F6 Cone. In
Process
Effluent
(ppmv)
11,714±61
9461 ± 50
7232 ± 43
4889 ±32
2500 ±26
C2F6 Cone. In
E-HTVS
Effluent
(ppmv)
1288 ± 10
1064 ± 10
736 ± 10
505 ±8
250 ±4
Dilution
Adjusted C2F6
Concentration
(9.05X in ppmv)
11,656
9629
6661
4570
2263
DRE
(%)
0%
-2%
8%
7%
9%
From the data contained in Table VII above, a 95% Confidence Interval (CI) is determined
to be 5.0 ± 4.4 % indicating that all values in Table VII are statistically zero at the 95 % CI.
                  C2F6 Emissions from E-HTVS during Process Flow Calibrations
1 .0 slm





	 i
f>*»"^
^ O.8 slm
V*ww**««*
•
t . 	 ^
\
^ O.4 slm
*




' ^•«a^«|!'m-


T
I
              5O     1OO     15O     2OO     25O     3OO     35O     4OO     45O     5OO

                                       FTIR Scan
                   C2R6 Emissions from E-HTVS during R rocess Rlow Calibration
                                                             400     450     500
Figure 10:  Inlet and outlet C2Fe emission profiles for E-HTVS during process flow
calibration.
                                      18

-------
The E-HTVS DRE for C^s was also determined during chamber cleans.  Emissions from the
chamber clean included significant levels of tetrafluoromethane (CF4), which was formed as by-
product. Other chamber clean by-products included carbonyl fluoride (COF2), tungsten
hexafluoride (WF6) and carbon monoxide (CO) as identified in the FTIR spectrum shown in
Figure 11. Emission profiles of CF4 and C^s for two chamber cleans are shown in Figure 12.
Emission profiles of CF4 and C2p6 from the E-HTVS are also shown in Figure 12.  These data
were numerically integrated over time to yield emission volumes. Comparison of the CF4 and
C2F6 emission volume entering the scrubber with that emitted from the scrubber can be used to
directly calculate DRE for each compound.

To convert measured concentrations into volumes, the following equation was used:
                           VEM =
Where the total emission volume (VEM) is the summation of each FTIR data point where the
concentration of analyte Q is determined during time interval At and multiplied by the total flow
(Tf). The summation of the entire emission profile provides an emission volume for a given
analyte during the process. During this study, these calculations were performed using standard
spreadsheet software (Microsoft Excel). Use of this technique reinforces the importance of
accurately determining the total process and scrubber flows as described in the sections above.
 Arbitrary Y / Wavenumber (cm-1)

 File#2 =NE3_1343
                                                          Paged Z-Zoom CURSOR

                                                       9/25/2007 4:36 PM Res=None
Figure 11: FTIR spectrum of process emissions during
Tool #1.
                                                             chamber clean on CVD
                                       19

-------
                   C2F6 and CF4 Emissions from LPW1O8 During Chamber Clean
                                   11OO
                                 FTIR Scan
                   C2F6 and CF4 Emissions from E-HTVS during Chamber Cleans
                             9OO    1OOO
                                 FTIR Scan
                                        1 1OO   12OO    13OO   14OO
Figure 12:  C2F6 and CF4 emissions from CVD Tool #1 (top) and E-HTVS scrubber during
chamber cleans. The first clean was timed for 5 minutes. The second clean was for a
tungsten accumulation of ca. 20um and was run until end point.

Integration of the emission profiles shown in Figure 12 yielded emission volumes contained in
Table VIII for both process and scrubber emissions.  These integrated emission volumes were
compared directly to determine F-GHG DRE during the chamber cleans, which are also included
in Table VIII. The results are close in value to those presented for C$6 during the process flow
calibrations shown above.

The results for CF4 yielded an average DRE of 9.5 % with a 95 % CI of ±  9.8 %. Use of these
two data points yields DRE's in the range of -0.3 - 19.3 %, which includes zero. Therefore, we
cannot reject the hypothesis that the DRE equals zero.  Given that CF4 is more difficult to abate
than C2F6 due to higher bond energy of C-F compared to C-C, a zero DRE would be expected
and the data indicates statistically that this is the case. The actual deviation from a zero DRE as
                                       20

-------
determined in this study can be attributable to the propagation of errors associated with all the
factors used in calculating the values contained in Table VIII.
Table VIII: Integrated F-GHG process and scrubber emissions for CVD Tool #1 (process)
and E-HTVS during chamber cleans. Details on the cleans provided in text

CiFe Process Emission (si)
C2F6 E-HTVS Emission (si)
C2F6 DRE
CF4 Process Emission (si)
CF4 E-HTVS Emission (si)
CF4DRE
Clean #1
3.914
3.723
5%
1.011
0.952
6%
Clean #2
8.266
8.802
-5%
1.730
1.494
13%
The data contained in Table VIII can also be used to determine the process DRE for C2F6 and the
process emission factor for CF4.  The process DRE is defined as the ratio of C^s emitted from
the process to the total C2F6 added to the process. For clean #1 the total C2F6 added to the
process was 6.0 si based on a 1.2 slm process flow and 5 minute chamber clean, thus the C2F6
process DRE was 35 % for this clean. For clean #2, which reached end point at 11.1 minutes,
the C2Fe process DRE was 38% based on a total process flow of 13.3 si.
The CF4 emission factor is defined as the ratio of CF4 volume emitted from the process to the
C2p6 volume added to the process. For cleans #1 and #2 these values were 17 and 13%,
respectively. The slight differences between cleans 1 and 2 were attributed to clean #2 being
done when the chamber had more tungsten film accumulation in the chamber.

All of the DRE values reported for C2F6 and CF4 indicate that the E-HTVS system does not abate
these gases under the operating conditions tested.   The result obtained for CF4 is statistically
equivalent to zero at the 95 % CI.

3.2.2. Edwards TPUDRE

As stated above the CVD Tool #2 was not operational.  This prevented evaluating the
performance of the TPU during wafer processing.  It was possible to flow NF3 from the process
chamber through the TPU.  The five flow rates used to collect the data in Figure 9 (data collected
at the inlet to the TPU) yielded undetectable levels of NF3 in the TPU effluent. Figure 13 shows
FTIR spectra obtained at the inlet and outlet of the TPU while 1.5 slm NF3 was flowing from the
process chamber.  These data can be used to report a TPU DRE of > 99.9 % based on an
estimated FTIR detection limit of 0.5 ppmv, an average inlet concentration of 93,964 ± 685
ppmv and a dilution factor of 53.6 X:
                                       21

-------
             Diluted NF3:
             Undetected NF3:
             DRE
93,964/53.6 = 1750 ppmv fully diluted
0.5/1750 = 0.00029
(1.000 - 0.00029)xlOO = 0.99971 = 99.971%
The TPU was proven to be effective at abating any NF3 potentially emitted from the process
under the conditions with which it was operated. The TPU thermocouple temperature gauge
reported an operating temperature of ca. 820° C. An experiment was conducted to determine if
the NF3 DRE would be reduced if the TPU operating temperature was reduced. Temperature
reduction was done by reducing the natural gas flow by ca. 10 and 20 %, which yielded
operating conditions of 665° C and 480° C, respectively.  Under these operating conditions NF3
was not detected in the TPU effluent suggesting that it may be possible to operate the TPU at
lower temperatures if its sole purpose is to abate NF3.  One observed consequence of operating at
lower temperatures was an increase in CO and CH4 emissions from the TPU relative to those
observed at 820° C.  Figure 14 shows CO and CFLt increased emissions while the temperature
was reduced.

Another by-product observed during NF3 abatement was nitric oxide (NO).  Figure 15 shows NO
emission profile during testing. NO emissions are correlated with NF3 flows from the tool.
                 IHJtxhaust

        W*^^***1     w""wv ^"^^Hfww^it^^
                Process 6
-------
4.0 Conclusion

The purpose of this study was to accurately determine the DRE of F-GHGs in two commercially
available POU abatement systems used in semiconductor manufacturing. A critical component
of an accurate DRE was to determine the dilution of process effluent occurring in the scrubbers.
Chemical spiking of inert gases was used to determine the dilution.  Kr and Xe were used for the
TecHarmonic E-HTVS scrubber. It was determined that both Xe and Kr yielded statistically
equivalent effective dilution for the E-HTVS.  Kr, Ar and Xe were used to determine the
effective dilution occurring in the Edwards TPU scrubber.  All three chemicals yielded similar
results.  The effective dilution for each system was combined with the process emission data to
determine the DRE of each scrubber for F-GHGs C2F6 and CF4 on the E-HTVS and for NF3 on
the  TPU.
The DRE determined for C$6 and CF4 on the E-HTVS were very low.  The results indicate a
statistically zero DRE was obtained for both C2F6 and CF4 on the E-HTVS.

The TPU was determined to have a very high DRE, > 99.9%, for NF3 when accounting for the
dilution occurring in the scrubber. The data also demonstrated that the TPU can be operated at
lower temperatures while maintaining a high DRE for NF3.  While other process and operating
parameters may need to be considered before reducing the temperature, it could be possible to
operate at lower natural gas flows and maintain a high NF3 DRE.

The process DRE for C2F6 during tungsten chamber clean was determined to be 35 - 38 %,
which was in close agreement with the 2006 IPCC reported value of 40 %. A CF4 emission
factor of 13 - 17 % was determined, which was also in close agreement to the 2006 IPCC
reported value of 20 %.
                                      23

-------
                    CH4 Emissions from TPU During NFS Breakthrough Testing
                                              CH4 Flow Decreased to 1 8

                                             TPU cooled to 478
                                          CH4 Decreased to 2O
                                CH4 Flow at 22 slm TPU at
                     CO Emissions from TPU During NFS Breakthrough Study
                                                                             sdbo
Figure 14: Relative CUt and CO emissions from TPU during reduced temperature
operation. The temperature was reduced by lowering the CH4 flow into the system.
                                        24

-------
                    NO Emissions from TPU During NFS Breaktrhough Test
Figure 15: Nitric oxide emission profile during NFs flows from CVD Tool #2 chamber
through the TPU. NO emissions are correlated with NFs flow rate
                                     25

-------