United States j::::jj •:::::: :::::: ':::jjj:::::::::
Environmental Protection :::::: ::::: ::::" ::::"::::!!
Agency Sh Emission Redjction
Partnership for the Magnesium Industry
Characterization of
Cover Gas Emissions
from U.S. Magnesium
Die Casting
Office of Air and Radiation
EPA 430-R-04-004
www.epa.gov/highgwp/magnesium-sf6
May 2004
-------
Table of Contents
Page
Acknowledgements ES-1
Executive Summary ES-2
1.0 Introduction 1-1
2.0 FTIR Data and Hardware Quality Analysis/Quality Control Procedures 2-1
2.1 Data Analysis Procedures 2-1
2.1.1 The Spectrum Analysis Method 2-1
2.1.2 Creating the Spectrum Analysis Method 2-2
2.1.3 Reference Generation 2-5
2.2 Hardware Procedures 2-6
3.0 Extractive FTIR Sampling Systems 3-1
3.1 Hydrogen/Oxygen Analyzer 3-6
4.0 Test Results 4-1
4.1 HFC-134a Cover Gas with Nitrogen and Carbon Dioxide Diluents 4-1
4.2 Novec™ 612 Cover Gas with 85% CO2 and 15% Air Diluents 4-4
4.3 SF6 Cover Gas with Air Diluent 4-8
5.0 Conclusions 5-1
5.1 Cover Gas Test Observations 5-1
5.1.1 HFC-134a Cover Gas Testing with N2 and CO2 Diluents 5-1
5.1.2 Novec™ 612 Cover Gas Testing with CO2 and Air Diluents 5-1
5.1.3 SF6 Cover Gas Testing with Air Diluent 5-2
5.2 Cover Gas Degradation 5-2
5.3 Occupational Health and Safety 5-5
5.4 Global Warming Impact Discussion 5-6
5.5 Uncertainty Discussion 5-10
-------
List of Tables
1-1 Test Schedule at Palmyra and Hannibal Intermet Facilities 1-2
1-2 Magnesium Die Casting Machine Parameters 1-3
2-1 Analysis Method Parameters 2-3
4-1 Data Summary for Machine #32 Running HFC-134a with N2 Diluent 4-2
4-2 Data Summary for Machine #32 Running HFC-134a with CO2 Diluent 4-3
4-3 Data Summary for the Novec™ 612 Cover Gas on Machine #4, Sample Point #1
Casting Condition 4-6
4-4 Data Summary for the Novec™ 612 Cover Gas on Machine #4, Sample Point #1
Static Condition 4-6
4-5 Data Summary for the Novec™ 612 Cover Gas on Machine #4, Sample Point #2
Casting Condition 4-7
4-6 Data Summary for the Novec™ 612 Cover Gas on Machine #4, Sample Point #2
Static Condition 4-7
4-7 Data Summary for the SFe Cover Gas on Machine #32 4-9
4-8 Data Summary for the SFe Cover Gas on Machine #4, Sample Point #1 4-9
4-9 Data Summary for the SF6 Cover Gas on Machine #4, Sample Point #2 4-10
5-1 Machine #32 Percent Degradation for Cover Gas Testing 5-4
5-2 Machine #32 Percent Degradation for Cover Gas Testing 5-4
5-3 Comparison of 100-Year GWP Estimates from the Intergovernmental Panel on
Climate Change's Second (1996) and Third (2001) Assessment Reports 5-6
5-4 GHG Emission Comparison for Machine #32 Using HFC-134a and SF6 5-8
5-5 GHG Emission Comparison for Machine #4 Using Novec™ 612 and SF6 5-8
5-6 GHG (Weighted By Gas Flow Rate) Emission Comparison for Machine #32 Using
HFC-134aandSF6 5-9
5-7 GHG (Weighted By Gas Flow Rate) Emission Comparison for Machine #4 Using
Novec™ 612 and SF6 5-9
in
-------
List of Figures
1-1 Cold Chamber #32 at the Intermet, Palmyra Facility 1-4
1-2 Cold Chamber #4 at the Intermet, Hannibal Facility 1-5
2-1 Reference Generation Hardware Configuration 2-6
3-1 Sampling System Schematic 3-2
3-2 Sample Locations for Cold Chamber #32 at Palmyra 3-3
3-3 Sample Locations for Cold Chamber #4 at Hannibal 3-5
3-4 Nova 340WP Oxygen and Hydrogen Analyzer 3-6
IV
-------
Acknowledgements
This cooperative study is a product of the Australia/U.S. Climate Action Partnership, initiated in
June 2002. The analytical measurements and report preparation was funded by the U.S.
Environmental Protection Agency under contract GS-10F-0124J to ICF Consulting.
Measurements and analysis were conducted by URS Corporation. The authors wish to express
their appreciation and thanks to Intermet and their staff, specifically Al Dimmitt, for contributing
not only their facilities, but their valuable assistance and advice, to this measurement study. The
support of Australian Magnesium Corporation (AMC), the Cooperative Research Centre for Cast
Metals Manufacturing (CAST) and 3M™ for providing the cover gases, expertise and trial staff
for this study is also gratefully acknowledged.
ES-1
-------
Executive Summary
A study was conducted to evaluate alternative cover gases used in cold chamber die-
casting. Currently sulfur hexafluoride (SF6) is widely used for the protection of molten
magnesium but with an interest in reducing SFe emissions and global climate change impact, the
magnesium industry and EPA are evaluating alternative gases. This study examines the use of
SF6, AM-cover™ (supplied by Australian Magnesium Corporation (AMC)) and Novec™ 612
Magnesium Protection Fluid (supplied by 3M™) cover gases under casting and static operating
conditions at two Internet magnesium die casting facilities located in Palmyra and Hannibal,
Missouri. The AM-cover™ technology utilizes FIFC-134a to provide melt protection. Novec™
612 is a fluorinated ketone. Gas was extracted directly from the crucible headspace above the
molten magnesium to characterize degradation product formation resulting from the interaction
of the cover gases with the heated melt surface. The results reported are from measurements
taken inside the crucible headspace and should not to be mistaken for ambient air emissions data.
Tables ES-1 and ES-2 summarize details and results from the study. Measurements were
conducted for various sampling scenarios including different cover gases, cover gas mixtures,
cover gas flow rates, and die casting processes. The cover gas degradation estimates listed in
Tables ES-1 and ES-2 have been corrected for crucible dilution effects.
Table ES-1. Machine #32 Cover Gas Measurements and Observed Degradation
Table
4-1
4-1
4-2
4-2
4-2s
4-7
4-7
4-7
Cover Gas Mixture
Components
HFC-134a/N2
HFC-134a/N2
HFC-134a/C02
HFC-134a/C02
HFC-134a/C02
SFs/Air
SFs/Air
SFs/Air
Sample
Location
#3
#4
#4
#6
#6
#3
#6
#4
Cover Gas Mixture
Flow"
(1pm)
21
21
20
20
20
65
65
65
Cover Gas
Delivery
Cone.
(ppm)
4,000
4,000
4,000
4,000
4,000
19,000
19,000
19,000
Cover Gas
Measured
Cone."
(ppm)
83.9
68.2
337.5
206.2
268.6
12,078
12,277
11,930
Cover Gas
Degradation
97%
98%
89%
93%
92%
8%
6%
9%
aAs provided by Internet and AMC
ES-2
-------
Table ES-2. Machine #4 Cover Gas Measurements and Observed Degradation
Table
4-3
4-4s
4-5
4-6s
4-8
4-9
Cover Gas Mixture
Components
Novec612/CO2/Air
Novec612/CO2/Air
Novec612/CO2/Air
Novec612/CO2/Air
SF2 diluent are: CO2, CO, HF, C^f^ and COF2. Please see Table 2-1 for a listing of chemical
formulas and compound names. For machine #32 furnace measurements, HF and C2p6
concentrations were on the order of 100 to 200 parts per million (ppm) and 2 ppm, respectively.
COF2 concentrations were below the detectable limit of the FTIR instrument when using an N2
diluent, but increased to levels greater than 10 ppm with a CC>2 diluent. There was no marked
difference in the results obtained during part casting and static (i.e., during periods when no part
casting occurred) operating periods. The time series plots for these and other compounds are
illustrated in Appendix A. The plots illustrate that additional degradation products, such as
H2CO, CH4, C2H2, and C2H4 are formed with the addition of ambient air during the ingot loads.
Some compounds such as H2CO, NO, N2O and NO2 also had background levels inside the
headspace that sharply increased during ingot loading. Detection of C2H2 and C2H4 was
sporadic, with a few spikes occurring during ingot loading. Other than these spikes,
concentrations were close to or below detectable limits.
Novec™ 612 Cover Gas Testing with CO2 and Air Diluents
The primary degradation compounds measured while using Novec™ 612 as a cover gas
are: CO, COF2, C3F8, C2F6 and HF. Measurements were conducted during both active casting
and static conditions on die casting machine #4. HF was present at relatively constant levels
during both operating conditions; however, the concentrations were higher at sample point #2,
which is closer to the pump that loads molten magnesium into the die. For example, near the
ingot feed door (sample point #1), HF concentrations were on the order of 50 ppm, while on the
pump side, concentrations were on the order of 110 to 125 ppm. C2F6, CsFe and COF2
concentrations were on the order of 1 to 2 ppm, 5 to 10 ppm and 5 ppm, respectively during both
casting and static conditions. Additional degradation products detected included CH4, SiF4, and
ES-2
-------
H2CO. SiF4 was only detected during casting conditions, but at concentrations less than 1 ppm.
Similarly, low levels of NO (< 1 ppm) and NO2 (1 to 4 ppm) were only detected during casting.
These levels tended to increase during ingot loading periods.
SF6 Cover Gas Testing with Air Diluent
The primary degradation compounds measured while using SF6 as a cover gas are: HF
and SO2. HF concentrations were on the order of 10 to 30 ppm, while NO, N2O and NO2 levels
were on the order of 2 ppm; however, these levels increased during ingot loading. H2CO levels
remained close to the FTIR detection limit for all cold chamber measurements.
Measured Percent Degradation for Cover Gases
Tables ES-1 and ES-2 list the degradation estimates for all cover gases measured. The
degradation estimates, which are corrected for dilution effects (i.e., the effects of air ingression
into the crucible headspace), are calculated as the percent difference between the delivery
concentration and the measured concentration in the crucible headspace. For all of the tests,
average degradation estimates for F£FC-134a with N2 and CO2 diluents, and Novec™ 612 were
98 percent, 91 percent and 90 percent, respectively. The level of degradation did not vary
significantly between casting and static operating conditions. In comparison, degradation
estimates for SF6 were on the order of 10 percent for four tests; however, for one test (on cold
chamber machine #4 - sampling location #1) a percent degradation of close to 50 percent was
observed. The reason for this high level of cover gas degradation observation is unknown.
Potential Climate Impact
A key factor in evaluating substitute cover gas compounds is their composite global
warming potential (GWP) compared to SFe. For each cover gas compound and its associated
degradation products, a composite global warming impact estimate was developed using IPCC
third assessment report GWP factors1. The overall GWP-weighted gas emissions rate for each
test scenario (i.e., F£FC-134a/CO2 or N2, Novec™ 612/CO2/Air on both machines) was estimated
using the measured average concentrations of each gas, their molecular weights and the delivery
cover gas flow rates.
Based on this approach, results indicate that F£FC-134a and Novec™ 612 have a
greenhouse gas (GHG) emissions impact that is more than 95 percent lower than SFe. Although
cover gas degradation product gases such as CF4 and C2F6 contribute to the total GWP of the gas,
1 IPCC, Climate Change 2001: The Scientific Basis. Intergovernmental Panel on Climate Change, 2001, Cambridge
University Press. Cambridge, U.K.
ES-4
-------
their contribution is negligible when considering their relatively low concentrations, cover gas
degradation levels on the order of 90 percent, as well as the large GWP reduction that occurs
when switching from SF6 (GWP = 22,200) to HFC-134a (GWP = 1,300) or Novec™ 612 (a
specific GWP was not available; however, fluorinated ketones, of which Novec™ 612 is one, are
known to have GWP's close to I2'3'4).
Milbrath, D. 3M™ Novec1™ 612 Magnesium Protection Fluid: It's Development and Use in Full Scale Molten
Magnesium Processes, Proceedings of the 60th Annual International Magnesium Association Conference, May
2003, Stuttgart, Germany.
3 Taniguchi, N., et al. Atmospheric Chemistry ofC2F5C(O)CF(CF3)2: Photolysis and Reaction with Cl atoms, OH
radicals, and Ozone. J. Phys Chem. A., 107(15); 2674-2679.
4 ICF, Re-evaluation of a C-6 Oxyfluoro carbon (trade name Novec 1230) and References, Memo prepared by ICF
Consulting, Inc. for the Environmental Protection Agency, Global Programs Division, under EPA Contract Number
68-D-00-266. September 10, 2003.
ES-5
-------
1.0 Introduction
This report presents and interprets the results of a series of cover gas measurements on
two cold chamber molten magnesium die casting machines. Measurements were conducted by
URS Corporation (URS) at two Intermet production facilities located in Palmyra and Hannibal,
Missouri between September 29th and October 7th, 2003. Measurements were made in a
continuous and real-time fashion with an extractive-type Fourier Transform Infrared (FTIR)
spectroscopic system and an extractive-type oxygen (62) continuous emission monitor (CEM).
The focus of the study was to assess degradation products and emissions for three
different cover gases in cold chamber applications. The cover gases are used to prevent surface
burning of the molten metal during processing. The three cover gases evaluated during the study
were: 1) SF6; 2) AM-cover™ (supplied by Australian Magnesium Corporation (AMC)), which
uses HFC-134a; and 3) Novec™ 612 Magnesium Protection Fluid (supplied by 3M™). The
main objectives of this study were as follows:
• To determine the level degradation of the cover gas within the confines of process
crucibles containing molten magnesium.
• To determine the nature of the reaction products expected as the cover gas is degraded
during melt protection. Direct measurement by FTIR, and subsequent spectral analyses,
was employed to identify the gaseous fluorides, acids and perfluorocarbons that may
result from cover gas decomposition.
• To determine the greenhouse gas (GHG) emissions from the new cover gas technologies
and overall reduction in GHG emissions attributable to the use of HFC-134a and
Novec™ 612 instead of SF6.
The measurement schedule, sampling locations, and test conditions are summarized in
Table 1-1. Typical die casting process parameters are summarized in Table 1-2. The
measurements were conducted under these conditions during casting activity, as well as static
conditions (i.e., the machine was not casting). Figures 1-1 and 1-2 illustrate the cold-chamber
machines that were tested at the Palmyra and Hannibal facilities.
Testing was carried out using existing cover gas distribution and flow controls that were
not optimized for either of the alternate cover gases. Due to the high degree of reactivity of
HFC-134a and Novec™ 612 some modification of the gas delivery system would be required to
achieve the most effective and efficient application scenario possible.
1-1
-------
Table 1-1. Test Schedule at Palmyra and Hannibal Intermet Facilities
Date
1-Oct
1-Oct
1-Oct
1-Oct
1-Oct
3-Oct
3-Oct
3-Oct
Time
0932-1036
1036-1236
1618-1845
1845-1916
1916-2000
0959-1037
1114-1154
1200-1222
Die Casting
Machine
Cold #32
Cold #32
Cold #32
Cold #32
Cold #32
Cold #32
Cold #32
Cold #32
Cover Gas Mixture
Components
HFC-134a/N2
HFC-134a/N2
HFC-134a/CO2
HFC-134a/CO2
HFC-134a/C02
SF6/Air
SF6/Air
SF6/Air
Sample
Location
#3
#4
#4
#6
#4, #6
#3
#6
#4
Cover Gas Mixture Flow3
(1pm)
21
21
20
20
20
65
65
65
Cover Gas
Delivery
Cone. a
(ppm)
4,000
4,000
4,000
4,000
4,000
19,000
19,000
19,000
Test
Condition
Casting
Casting
Casting
Casting
Static
Casting
Casting
Casting
Ingot Type
AZ91D
AZ91D
AZ91D
AZ91D
AZ91D
AZ91D
AZ91D
AZ91D
FTIR Path
Length
5.1
5.1
5.1
5.1
5.1
20.1
20.1
20.1
6-Oct
6-Oct
6-Oct
7-Oct
7-Oct
7-Oct
1137-1652
1318-1358
1450-1654
1530-1655
0830-1025
1036-1227
Cold #4
Cold #4
Cold #4
Cold #4
Cold #4
Cold #4
Novec™ 612/CO2/Air
Novec™ 612/CO2/Air
Novec™612/CO2/Air
Novec™ 612/C02/Air
SF6/Air
SF6/Air
#1
#1
#2
#2
#2
#1
612/CO2=48.6,Air=7.3
612/CO2=48.6,Air=7.3
612/CO2=45.6,Air=6.8
612/C02=45.6,Air=6.8
37.5
37.5
126
126
126
126
5,000
5,000
Casting
Static
Casting
Static
Casting
Casting
AZ91D
AZ91D
AZ91D
AZ91D
AZ91D
AZ91D
5.1
5.1
5.1
5.1
5.1
5.1
aAs provided by Intermet, AMC and 3M™
1-2
-------
Table 1-2. Magnesium Die Casting Machine Parameters
Parameter
Facility
Furnace Temperature (°F)
Ingot Weight (Ibs)
Furnace Capacity (Ibs)
Ingot Type
Mg Casting Rate (seconds/part)
Mg Pump Type
Mg Shot Weight (Ibs)
Metal Throughput (Ibs/hr)
Product
Molten surface area (sq ft)
Ingot Loading
Cold Chamber #32
Palmyra
1,260-1285
25
3,000
AZ91D
43
Centrifugal
4.4
368
Valve covers
10.4
Automatic Feed
Cold Chamber #4
Hannibal
1,275
25
3,000
AZ91D
48
Electromagnetic Pump
5.4
405
Valve covers
10.4
Automatic Feed
1-3
-------
Figure 1-1. Cold Chamber #32 at the Intermet, Palmyra Facility
1-4
-------
Figure 1-2. Cold Chamber #4 at the Intermet, Hannibal Facility
1-5
-------
2.0 FTIR Data and Hardware Quality Analysis/Quality Control
Procedures
Almost every chemical compound absorbs infrared (IR) light to some degree in a
particular region of the mid-infrared spectrum. These absorption properties can be used to
identify and quantify chemical compounds in a complex mixture of gases. As stated by Beer's
Law, the magnitude of a compound's IR absorbance is directly proportional to the product of its
concentration in the mixture and the sample cell optical path length. This is otherwise known as
the compound's optical depth. The extractive FTIR instruments used by URS are able to achieve
parts-per-billion (ppb) detection levels because the optical path length within the measurement
cell is magnified many times by reflecting the IR beam between a series of mirrors before it
reaches the detector. The mirrors provide a fixed optical path length best suited to the gas
mixture being sampled. In this case, optical path lengths of 20.1 meters and 5.1 meters were
utilized.
The use of FTIR as an analytical tool requires extensive quality analysis/quality control
(QA/QC) procedures on both data analysis and hardware to ensure valid data results. FTIR is
often perceived as having poor results due to improper implementation of the proper procedures
and protocols. In accordance with ISO requirements, URS utilizes an extensive protocol (Radian
DCN#96-133-403-01) to maintain consistency in hardware setup and data analysis. The
following sections describe QA/QC procedures used for data analysis and hardware.
2.1 Data Analysis Procedures
2.1.1 The Spectrum Analysis Method
An infrared spectrum analysis is performed by matching the features of an observed
spectrum to those of reference gases of known concentration. If more than one feature is present
in the same region, then a linear combination of references is used to match the compound
feature. The standards are scaled to match the observed band intensities in the sample. This
scaling also matches the unknown concentrations. An infrared spectrum can be collected and
analyzed in approximately one second, but spectra are normally averaged over a one- or two-
minute integration period to produce adequate signal-to-noise limits and ppb detection levels.
The scaled references are added together to produce a composite that represents the best
match with the sample. A classical least squares mathematical function is used to match the
standards' absorption profiles with those of the observed spectrum in specified spectral analysis
2-1
-------
regions. The compounds of interest together with compounds expected to cause spectral
interference are included in the analysis region.
2.1.2 Creating the Spectrum Analysis Method
The spectrum analysis method used for the tests at the Internet facilities was developed
by URS by selecting the spectral regions and sub-regions that are least affected by primary IR
absorbers (H2O and CO2, in this case) while also producing the best detection limit possible for
the target compounds. Target compounds are initially determined prior to sampling based on
cover gas composition. However, many degradation product gases were found during data
analysis requiring many iterations of data processing and interpretation. Typically, an analysis
method is iteratively refined by using it to analyze a representative set of infrared spectra while
varying the method. The optimum method is indicated when both the 95 percent confidence
levels and the bias on the individual compounds are minimized. Table 2-1 lists the range of
references included in the analysis method used by the FTIR systems for the tests. Each
reference is described in terms of its optical depth (i.e., concentration times cell path length
(ppm-meters) range. For the Internet testing, new gas references were required for HFC-134a,
Novec™ 612 and CO2 which is discussed in a Section 2.3.
2-2
-------
Table 2-1. Analysis Method Parameters for
Major Contaminants and Spectroscopic Interferants
Chemical
Formula
H20
C02a
SF6
CH2FCF3
Novec™612a'b
SO2
CO
HF
COF2
C2H2
C2H4
C2F6
CF4
CH2F2
CHF3
CH4
OF2
H2CO
C3F8
C4F8
NO
N2O
NO2
Compound
water
carbon dioxide
sulfur hexafluoride
HFC-134a
Novec™612
suflur dioxide
carbon monoxide
hydrofluoric acid
carbonyl fluoride
acetylene
ethylene
hexafluoroethane
carbon tetrafluoride
dimethyl fluoride
methyl fluoride
methane
oxygen difluoride
formaldehyde
perfluoropropane
octafluorocyclobutane
nitric oxide
nitrous oxide
nitrogen dioxide
SF6
(ppm-meters)
30,600-2,748,900
1,360,368-
5,107,369
56-280
n/a
n/a
518-10,415
1,032-101,790
1-2,000
50-1,250
111-5,550
86-2,576
112-280
28-112
28-560
177-5,908
87-30,019
1,750-14,000
92-1,838
213-1,058
28-560
114-12,961
102-4,076
34-2,097
HFC-134a
(ppm-meters)
30,600-2,748,900
1,360,368-5, 107,369 (CO2)
70-2,1 10 (N2)
n/a
9,700-27,575
n/a
n/a
1,032-101,790
1-2,000
50-1,250
111-5,550
86-2,576
112-280
28-112
28-560
177-5,908
87-30,019
1,750-14,000
92-1,838
213-1,058
28-560
114-12,961
102-4,076
66-2,097
Novec™612
(ppm-meters)
30,600-2,748,900
70-2,110
n/a
n/a
21-991
n/a
1,032-101,790
1-2,000
50-1,250
111-5,550
86-2,576
112-280
28-112
28-560
177-5,908
87-30,019
1,750-14,000
92-1,838
213-1,058
28-560
114-12,961
102-4,076
66-2,097
""References generated at URS laboratory facilities in Austin, Texas. Note that two different optical depth ranges are
required to quantify CO2 during the HFC-134a tests. High optical depth was required when CO2 was used as the
diluent gas, and a low optical depth was required when N2 was used as the diluent gas.
Chemical formula was not provided by 3M™; however, the gas is known to be a fluorinated ketone
(C3F7C(O)C2F5).
After setting up the FTIR instruments on-site, signal-to-noise ratio (SNR) assessments
were performed. This was determined by measuring the noise equivalent absorbance (NEA) of
each FTIR system while sampling nitrogen. The NEA is derived by ratioing two consecutive
single beam spectra to produce a "zero" spectrum, then measuring the peak-to-peak absorbance
at a frequency region of interest. This represents the noise level of the instruments under field
conditions. By determining the concentration level for each contaminant that scales down its
analyzed spectral features to the NEA (representing a SNR of 1 or better), the compound's SNR-
limited minimum detection limit (MDL) can be estimated.
2-3
-------
Due to the complexity of the sample matrix, detection limits are calculated using two
different methods. The first method is a noise-based detection limit which involves data
collection during a "non-process" condition, or when the analyte of interest is not present in the
sample stream for a minimum of approximately 20 data points. The standard deviation over this
period is calculated and the noise-based MDL is determined. The second method is conducted
when the analyte of interest is present in the sample stream. A theoretical noise-based detection
limit is determined by comparing the ratio of absorbance intensities and optical depth for the
peak-to-peak noise to the ratio of the absorbance intensities and optical depth for the lowest
concentration reference used in the analytical method. The equation below shows the calculation
for theoretical noise-based detection limits represented as MDL or minimum detection limit.
Peak - to - Peak Noise Reference Absorbance
2 x MDL x Path Length Reference Concentration x Reference Path Length
Note that the MDL is multiplied by two as a conservative estimate. For all MDL
calculations, a peak-to-peak noise of 1 x 10"3 absorbance units is used. In some instances
compounds absorb infrared light in regions that are interfered with by higher concentration
compounds. An example would be CF4. CF4 is difficult to detect in a sample stream containing
high HFC-134a concentrations since its strongest absorbance peak is in a region that HFC-134a
absorbs infrared in. Therefore detection limits are affected and reported as such.
When spectroscopic interferences are taken into account for those contaminants that have
overlapping absorption features, an increase in their MDLs is expected; consequently, the second
method is used. To determine this MDL, a set of spectra was collected during "non-process"
periods on the FTIR system and the detection limits during these sets of data were calculated.
During this time, it is assumed that there are no process gases present and that any reported is a
mathematical anomaly created by interferences. Three times the standard deviation of the set of
data is a typical approximation of the method limited MDL. Since emissions are present in the
crucible head space even when the tool is idle, this calculation is computed over periods where
ambient air is running through the sampling cell. This method is used as an alternative to the
theoretical "noise based" detection limit to factor in the effects of interferences. The calculation
is a more conservative and practical calculation and therefore is used wherever possible.
2-4
-------
2.1.3 Reference Generation
Since the use of HFC-134a and Novec™ 612 within the magnesium industry is relatively
new, references were required to be generated for both gases. Additionally, high concentration
references for CC>2 were required since it was used as a diluent gas with Novec™ 612 and HFC-
134a at concentrations of 85 percent and greater. HFC-134a references were obtained and
formatted for utilization with MKS software (MKS was formerly, On-Line Technologies). For
the Novec™ 612 cover gas, a sample was obtained from 3M. A series of gas references for
Novec™ 612 had to be generated to accurately quantify sample data. A reference set was
generated such that it bracketed the expected concentration range from the field sample.
References were generated from certified gas standards made gravimetrically on NIST certified
scales with a solution of Novec™ 612. The standard was certified at ± 2 percent at 201 ppm
Novec™ 612 by HP Gas Products located in Baytown, Texas. The standard was diluted at 5
different levels with nitrogen and gas reference samples were measured in a static condition
using the following procedure.
1. Evacuate and fill the FTIR sample cell with nitrogen 5 times
2. Evacuate the cell using a roughing vacuum pump
3. Add ultra high purity (UHP) nitrogen to a cell pressure of 400-650 torr
4. Add Novec™ 612 gas standard while recording pressure differential
5. Fill cell to 750 torr with N2
6. Measure gas reference
With this approach the pressure was monitored at each reference step with a calibrated
Baratron pressure sensor made by MKS. By knowing the amount of standard added with respect
to pressure, the concentration can be calculated by the following equation.
_. ,, „ . Pressure of the Standard Added _, . _
Reference Concentration = x Bottle Concentration
Total Pressure
Each gas reference sample was saved and used to generate a 5-point calibration curve that
was then applied to the Novec™ 612 data. Additionally, a gas standard of 62.92 percent CC>2 was
obtained for the high concentration requirements and the procedure was repeated for CO2. A
total of 12 references were collected to generate a twelve-point calibration at high concentrations
of CC>2. Figure 2-1 is a schematic of the configuration used for generating the references.
2-5
-------
Pressure
Measurement
MKS
Baratron
Pressure
Measurement
Vacuum
Pump
Gas
Standard
Figure 2-1. Reference Generation Hardware Configuration
2.2 Hardware Procedures
A series of on-site calibration and system checks was performed on each FTIR and
sampling system prior to testing to ensure data of known quality. These checks consisted of the
following:
• Cell Leak Checks
This test checks the integrity of each cell by pulling a vacuum on it and then
monitoring the leak rate. The acceptance criteria for this test is a leak rate < 2
torr/minute. The FTIR sample cells on-site were verified to have a leak rate well
under 1 torr/minute prior to testing.
• Infrared Detector Linearity Checks
For best results, it must be assured that the infrared detector yields a linear response
throughout a reasonable absorbance range and all the frequencies in a set of test
2-6
-------
spectra. A software linearizer is used to continuously adjust the MCT detector preamp
signal in order to achieve the desired response. To optimize the linearizer, background
spectra are acquired with and without a polyethylene film card in the IR beam.
Comparison of the strongly absorbing polyethylene bands in the low, mid and high
frequency regions against a clean background enables the processor to appropriately
set the linearizer terms (offset, linear, quad, cubic and delay). This procedure was run
prior to the start of testing, and subsequent spectra were visually checked on a periodic
basis to confirm that linearity was maintained.
Noise Equivalent Absorbance (NEA) or Signal-to-Noise Ratio (SNR) Tests
This provides a measure of the system noise, that is the sensitivity of the instrument
for the specified spectral resolution (0.5 cm"1, in this case) and number of scans (256,
or 2 minute of signal averaging, in most cases). An NEA/SNR test was run upon set-
up, then re-checked before the second die-casting machine was sampled. The results
for both systems, which were used to assess the field detection limits, were as follows:
20.1m Path Length System 2010
Range = 1000-1100cm'1, RMS Noise=0.1783 milliAU, SNR=2435
Range = 2450-2550cnT1, RMS Noise=0.1836 milliAU, SNR=2366
Range = 4200-4300cm'1, RMS Noise=0.4385 milliAU, SNR=990
5.11m Path Length System 2030
Range = 900-1000cm'1, RMS Noise=0.1 milliAU, SNR=4200
Range = 2450-2550cnT1, RMS Noise=0.1 milliAU, SNR=3644
Range = 4100-4200cnT1, RMS Noise=0.37 milliAU, SNR=1167
Path Length
The sample cell used for these tests was geometrically fixed at 20.1 meters for one
FTIR system, and 5.1 meters for the other.
Spectrometer Frequency and Resolution Checks
A real-time check of frequency position and resolution was performed prior to and
directly after each round of testing by monitoring a specific water absorption line
(present in ambient air). The position of this line must not deviate more than + 0.005
cm"1 from the reference value over the course of each test. Likewise, the linewidth
(directly related to instrument resolution) of this line must not deviate more than +
0.05 cm"1 from the reference value over the course of each test.
2-7
-------
Spectral Background
A spectral background is essentially a "blank spectrum" in that it does not contain any
of the target compounds present in the sample. It was created by purging the cell with
ultra-high-purity (UHP) nitrogen while collecting a spectrum. This spectrum was then
used by the analytical software to ratio against each sample spectrum to produce an
absorbance spectrum for quantitative analysis. A new spectral background was
generated prior to each sampling run.
Sample Cell Exchange Rate
Given sampling flow rates on the order of 2 liters/min through either cell during the
majority of monitoring tests, a complete sample exchange takes place every 6 seconds
for the 5.1 meter cell, and 15 seconds for the 20.1 meter cell. Since spectral signal
averaging was conducted over 30 second and 1 minute intervals, each collected
spectrum represented an integrated average over multiple sample cell exchanges.
FTIR Measurement Error
As with all analytical devices, extractive FTIR measurements are known to have a
given error associated with them. Steps were taken throughout the measurement
process to minimize sampling error. Sampling error is dependant on many factors
including interferences contained in the sample stream, and optical depth of references
that are applied. Errors were minimized by applying a series of references at various
optical depths to account for any nonlinearities or dynamic concentrations in the
sample matrix. Spectra were also manually inspected for qualitative and quantitative
validation. As a result of these efforts, it is believed that the measurements taken in
this study have a level of uncertainty that is on the order of 10 percent.
2-8
-------
3.0 Extractive FTIR Sampling Systems
Two extractive-type FTIR systems were used for the testing conducted at Internet. MKS
FTIR spectrometers and sample cells were used to speciate and quantify the gaseous compounds
at each die-casting crucible. In general, the system components include a inconel sample probe
(3/8" OD), a heated PFA-grade Teflon extraction line, the on-line FTIR spectrometer interfaced
to a heated, nickel-coated sample cell, a sample pump, and rotameter. Given this configuration,
real-time monitoring consisted of pulling a gas stream continuously from the sample probe
through the sampling system into the heated FTIR sample cell. Sample flow was maintained at
approximately 6.5 ft3/hr by a diaphragm pump connected to the outlet of the FTIR cell. The
rotameter at the sample cell exhaust was used to monitor the system sample flow. A schematic is
shown in Figure 3-1.
Inside each FTIR cell, a set of optically matched gold-plated mirrors reflects an infrared
beam through the sample gas multiple times. As the beam passes through the sample, the
molecules in the sample absorb some of its energy. After exiting the cell, the infrared beam is
directed to a liquid-nitrogen cooled mercury/cadmium/telluride (MCT) detector, a
photoconductive device that produces an electrical voltage proportional to the amount of infrared
light that strikes it. The strength of the absorption at particular frequencies is a measure of the
compound's concentration. The total distance traveled by the infrared beam inside the cell is the
cell path length, and is an important variable used in determining sample concentrations. For
this project, cell path lengths were fixed at 20.1 and 5.1 meters.
The FTIR sample cell and extraction lines were maintained at a temperature of 150°C (to
prevent any condensation losses and preclude the formation of HF mists). Cell pressures were
continuously recorded during measurement periods using a pressure sensor calibrated over the
0-900 torr range. Instrumental resolutions were set to 0.5 cm"1 and signal averaging was
performed over 30 second and one-minute periods.
As shown in Table 1-2, a total of two cold chamber die-casting machines were tested at
the Palmyra and Hannibal facilities. Figure 3-2 illustrates the sample locations on the crucible
lid for cold chamber #32 at the Palmyra facility. While 6 sample points were available for
extracting gas, only points #3, 4 and 6 were used. Points 3 and 4 are normally used as cover gas
injection points; when sampling from these points, the cover gas was equilibrated at the same
flow rate over the remaining four injection points. Sampling from point 6 had no impact on
cover gas distribution because it is the location of a thermocouple port.
5-1
-------
Sample Locations
Roughing Pump
Manual Valve
Rotameter
Cold Chamber #32 Crucible Lid
Exhaust —-
Heated
Extraction
Line
Figure 3-1. Sampling System Schematic
3-2
-------
Cold Chamber #32 Crucible Lid
#1
#6
#2
\
\
#5
#4
Figure 3-2. Sample Locations for Cold Chamber #32 at Palmyra
#3
\
3-3
-------
Figure 3-3 shows the sample locations for cold chamber #4 at Hannibal. Two sampling
points were used, with one near the ingot feed door (sampling point #1) and the other near the
pump feed to the die (sampling point #2). Samples at both points were obtained through a
stainless steel tube inserted into the headspace along the side of the thermocouples.
Consequently, unlike machine #32, the cover gas distribution regime remained constant. For
both machine #4 and #32, magnesium ingots were fed to the crucible automatically
approximately every 3 to 5 minutes. Please note that the sampling regime (in terms of sampling
points and elevation above the melt surface) differed between the two cold chambered machines.
As a result, comparison between machines should not be considered to be under strictly identical
conditions.
5-4
-------
Cold Chamber #4 Crucible Lid
#2
\\
Figure 3-3. Sample Locations for Cold Chamber #4 at Hannibal
3-5
-------
3.1 Hydrogen/Oxygen Analyzer
During the testing
period, a Nova Model
340WP portable analyzer
was used on the HFC-
134aandNovec™612
cover gas tests for the
continuous measurement
of oxygen concentrations.
The instrument is shown
in Figure 3-4. This
instrument uses an
electrochemical sensor to
measure oxygen. A
chemical reaction occurs when the sensor is exposed to oxygen, resulting in a millivolt output
proportional to the oxygen concentration in the sample gas. This small voltage is used to display
the measured oxygen concentration on the instrument's front panel meter. The analyzer was
placed inline with the exhaust of an FTIR sample pump and readings were recorded into field
logbooks. The sample system was closed up to that point, eliminating the introduction of
ambient air to the sample.
Figure 3-4. Nova 340WP Oxygen and Hydrogen Analyzer
-------
4.0 Test Results
This section presents all the test data and is broken into three main sections divided by
cover gas type: SFe, HFC-134a and Novec™ 612. Table 1-1 shows the test schedule, flow rates
and concentrations used during testing. Data collected during sampling downtime (i.e., probe
taken out while switching sample points) is excluded from the calculations. Additionally, there
are some instances where data values were below detection limits; consequently, "BDL" is
reported as the value. In order to calculate average values over a period where data points are
below detection limits, a value of "MDL", the method detection limit, divided by two is used in
the average calculation. Normally a range for average values is reported using zero for the low
range and the detection limit for the high range. However, the data tables are simplified by using
an "average value" of the detection limit divided by two to calculate averages. This only occurs
during instances where data values are both below and above calculated detection limits.
Detection limits are calculated as 3 times the standard deviation when the analyte is not present
in the sample streams. When the analyte is present, a method based detection limit is determined
by using the noise-based equation defined in Section 2.2.
The data in this section are presented in summary Tables 4-1 through 4-9. Appendix A
presents charts detailing data trends and process activities, such as ingot loading, for all the
compounds that were above detection limits.
4.1 HFC-134a Cover Gas with Nitrogen and Carbon Dioxide Diluents
Table 4-1 presents results for data collected at sample locations 3 and 4 on cold chamber
#32 (see Figure 3-2) using a nitrogen diluent. The data from the two sample locations were
relatively similar, with HFC-134a and C$6 concentrations on the order of 80 ppm and 1.5 ppm,
respectively. HF concentrations were considerably lower at location #4 (<1 ppm) compared to
location #3 (60 ppm), which may be an indication of increased ambient air infiltration occurring
at the crucible lid near location #3. Degradation product gases at detectable parts per million
levels include: CO2, CO, HF, COF2, C2F6, NO, N2O, NO2 and CH4. NO, N2O, NO2; levels
sharply increased during the dressing, which may indicate that their formation is a function of
ambient air dilution. This is possibly confirmed by slight concentration increases during ingot
loading. Gases that were below detectable limits include: OF2, H2CO, C3F8, C4F8, C2H2, CF4,
CH2F2, and CHF3.
The next test conducted on cold chamber #32 was with HFC-134a cover gas but with
CO2 instead of nitrogen diluent. Table 4-2 summarizes the data.
4-1
-------
Table 4-1. Data Summary for Machine #32 Running HFC-134a with Nitrogen
Diluent
H2O
(%)
02
(%)
C02
(ppm)
HFC 134a
(ppm)
HF
(ppm)
COF2
(ppm)
CO
(ppm)
C2H4
(ppm)
C2F6
(ppm)
CH4
(ppm)
NO
(ppm)
NO2
(ppm)
N2O
(ppm)
Point #3
Min
Max
Avg
Std Dev
0.35
2.19
0.65
0.40
na
na
na
na
1,526.79
11,317.12
4,591.43
2,646.50
22.73
148.41
83.85
29.24
BDL
274.65
64.44
72.70
BDL
3.65
0.60
0.66
179.12
846.21
424.71
129.67
BDL
BDL
BDL
na
0.71
1.59
1.10
0.18
BDL
25.13
6.98
4.16
6.99
22.55
14.85
2.13
BDL
1.16
0.27
0.21
4.73
12.81
7.94
1.84
Point #4
Min
Max
Avg
Std Dev
MDL
0.33
0.90
0.41
0.06
0.04
6.80
7.60
7.20
na
0.10
424.83
22,448.88
4,645.96
4,025.25
89.38
8.34
429.71
68.16
61.31
1.40
BDL
2.05
0.56
0.39
0.73
BDL
1.03
0.38
0.11
0.70
2.66
1,086.21
383.49
173.52
0.88
BDL
36.05
0.61
2.44
0.63
BDL
10.05
2.26
1.54
0.10
BDL
102.96
7.77
11.59
0.76
BDL
55.86
9.62
6.96
2.59
BDL
8.56
0.68
1.11
0.29
BDL
41.06
10.30
5.69
0.32
BDL = below detectable limit
MDL = method detection limit
na = not applicable
nc = not calculated
4-2
-------
Table 4-2. Data Summary for Machine #32 Running HFC-134a with CO2 Diluent
H2O
(%)
02
(%)
CO2
(%)
HFC 134a
(ppm)
HF
(ppm)
COF2
(ppm)
CO
(ppm)
CH4
(ppm)
C2F6
(ppm)
NO
(ppm)
NO2
(ppm)
N2O
(ppm)
Point #4
Min
Max
Average
Std Dev
0.28
3.50
0.45
0.52
6.00
7.60
6.78
0.69
48.15
91.62
79.94
7.05
166.32
602.16
337.47
86.93
0.10
262.45
197.50
61.24
BDL
71.71
37.73
12.74
2,187.62
11,476.85
5,643.24
1,703.74
BDL
4.85
1.89
0.70
1.44
4.35
2.80
0.62
7.33
27.36
17.13
3.77
7.07
23.98
15.23
3.88
13.63
34.11
23.49
2.89
Point #6
Min
Max
Average
Std Dev
0.25
0.30
0.29
0.01
na
na
na
na
72.36
78.66
75.01
1.39
147.58
270.23
206.22
31.93
214.40
420.31
283.95
58.65
4.47
13.09
8.05
2.01
1,724.29
3,071.40
2,267.02
300.93
1.38
2.90
2.26
0.33
1.47
2.35
1.83
0.22
26.79
37.99
32.07
2.93
2.86
5.60
3.89
0.65
18.04
25.98
21.85
1.60
Point #6 Static
Min
Max
Average
Std Dev
MDL
0.26
0.29
0.27
0.01
0.04
na
na
na
na
0.10
76.15
82.42
79.64
1.62
0.01
212.63
356.55
268.55
30.88
1.40
190.96
239.84
211.36
11.34
0.22
3.16
9.91
6.11
1.41
0.70
1,640.54
2,273.98
1,975.95
149.36
0.88
1.96
3.92
2.91
0.35
0.35
1.36
1.91
1.64
0.17
0.10
26.23
33.50
29.21
1.69
0.58
2.47
3.78
3.12
0.28
0.30
18.78
27.40
23.80
1.76
0.03
BDL = below detectable limit
MDL = method detection limit
na = not applicable
nc = not calculated
Table 4-2 includes data for sample locations #4 and #6 in addition to a sampling period
where the die casting machine was in a static condition (not casting). Results from the HFC-
134a/CC>2 cover gas mixture were similar to those observed during use of the HFC-134a/N2
cover gas mixture. The primary difference was the absence of C2H2 formation during the HFC-
134a/CC>2 test. HFC-134a presence in the crucible headspace was at higher concentrations
during the CO2 diluent test (200 to 300 ppm) compared to N2 (60 to 80 ppm). Additionally,
another variation was that the NO, N2O and NC>2 levels were more significant during the HFC-
134a/CC>2 test (on the order of 15 to 30 ppm, as opposed to less than 10 ppm). The data obtained
from sample location #4 and location #6 did not vary significantly, which may result from them
being spaced close together. However, location #4 had much higher concentrations for COF2
and CO. The additional source for CO may be from burning natural gas at the pump feed line
into the die, or as a byproduct of CO2 reacting with Mg.
4-3
-------
4.2 Novec™ 612 Cover Gas with 85% CO2 and 15% Air Diluents
A series of tests were run with Novec™ 612 used as a cover gas. Tests were run at two
locations, sample point #1 near the ingot load door and sample point #2 near the magnesium
pump. Tables 4-3 and 4-4 present data from sample location #1. Table 4-3 shows results during
casting conditions, while results presented in Table 4-4 were taken while the machine was in a
static condition. Results indicate that most of the Novec™ 612 cover gas (supplied at 126 ppm)
is consumed, with headspace concentrations on the order of 7 to 8 ppm. The level of Novec™
612 decomposition does not change significantly when comparing active casting (average 7
ppm) versus static conditions (average 8 ppm). The compounds detected include: CO, NO,
NO2, HF, COF2, CH4, C2p6 and CsFg. NO and NO2 were not detectable during the static testing.
As illustrated in Table 4-3, additional compounds, SiF4 and H2CO, were detected at
sample location #1. While H^CO formation maybe expected as a symptom of the presence of
hydrocarbons under intense heat, SiF4 formation was not expected since the source for silicon is
unknown. One possibility is that source of silicon may be from the insulation that coats the oven
that is used to warm ingots before loading into the crucible.
Tables 4-5 and 4-6 present data from the Novec™ 612 at sampling location #2. As at
sample location #1, data were collected with the machine in both a casting and static condition.
Novec™ 612 measurements were relatively consistent with those observed on the ingot side.
Sampling at location #2 also yielded similar compounds: CO, NO2, FTP, COF2, CH4, C2F6 and
CsFg. As at sample location #1, SiF4 and H2CO were detected during casting conditions. NO
was not detected at sample location #2, and is only detected on the ingot door side, which may be
indicative of degradation product formation caused by increased ambient air dilution during
ingot loading. This is possibly confirmed by the observation that no NO2 was detected during
the static periods.
By comparing all four tables for the Novec™ 612 cover gas a few observations can be
made. Both casting and static conditions produced C2F6 and CsFg, at concentrations on the order
of 1 to 2 ppm and 5 to 10 ppm, respectively. It was also noted that the formation of C2F6 and
CsFe were not affected by the machine being in a static (non-casting) condition. However, the
static condition on the pump side location (sample point #2) seemed to produce higher average
levels of COF2 (25 ppm as opposed to <10 ppm). It was also observed that HF levels were
higher on the pump side (sample location #2) compared to the ingot loading side (sample
location #1) for both casting and static conditions (i.e., 110 to 126 ppm compared to 50 ppm)
even though O2 levels remained relatively constant throughout the test conditions. Higher HF
4-4
-------
concentrations may be due to the fact that the pump side cover gas protective layer is not
penetrated during the addition of ingots. Consequently, the protective layer on the pump side
remains intact relative to the ingot loading side, and consequently, any excess Novec™ 612
present undergoes thermal degradation instead of reaction with the melt surface. As an additional
point of interest, in prior Novec™ 612 measurement trials5, the presence of perfluoroisobutylene
(PFIB) as a possible by-product of cover gas degradation was noted; however, in these
measurements, no PFIB was detected.
5 Mibrath, D., "Development of 3M™ Novec™ 612 Magnesium Protection Fluid as a Substitute for SF6 Over
Molten Magnesium," International Conference on SF6 and the Environment: Emission Reduction Technologies,
November 21-22, 2002, San Diego, CA.
4-5
-------
Table 4-3. Data Summary for the Novec™ 612 Cover Gas on Machine #4, Sample Point #1, Casting Condition
Minimum
Maximum
Average
Std Dev
MDL
H2O
(%)
BDL
0.54
0.15
0.12
0.05
02
(%)
7.50
7.60
7.55
0.07
0.10
CO2
(%)
55.43
98.20
87.02
9.56
0.11
Novec
(ppm)
3.68
10.99
6.85
1.47
0.02
HF
(ppm)
36.51
78.67
54.39
8.50
1.37
COF2
(ppm)
1.18
8.55
4.92
1.50
0.13
CO
(ppm)
939.94
24,340.95
10,847.66
6,695.05
0.45
CH4
(ppm)
1.11
39.94
7.04
6.17
0.76
C2F6
(ppm)
0.61
1.64
1.04
0.33
0.02
NO
(ppm)
BDL
3.62
1.47
0.69
0.55
NO2
(ppm)
1.20
10.74
4.05
2.19
0.66
H2CO
(ppm)
0.35
9.68
1.30
1.79
0.35
C3F8
(ppm)
3.19
11.00
6.46
1.76
0.46
SiF4
(ppm)
0.05
0.06
0.05
0.02
0.05
BDL = below detectable limit
MDL = method detection limit
nc = not calculated
Table 4-4. Data Summary for the Novec™ 612 Cover Gas on machine #4, Sample Point #1, Static Condition
Minimum
Maximum
Average
Std Dev
H2O
(%)
BDL
BDL
BDL
nc
02
(%)
7.70
7.70
7.70
0.00
CO2
(%)
94.61
98.59
96.85
0.90
Novec
(ppm)
6.04
10.68
8.12
0.93
HF
(ppm)
46.63
57.90
52.30
2.90
COF2
(ppm)
5.13
9.13
7.39
0.82
CO
(ppm)
2,822.24
5,995.77
3,932.86
765.93
CH4
(ppm)
0.67
1.30
0.92
0.15
C2F6
(ppm)
1.34
1.61
1.47
0.08
C3F8
(ppm)
4.82
6.16
5.47
0.33
4-6
-------
Table 4-5. Data Summary for the Novec™ 612 Cover Gas on Machine #4, Sample Point #2, Processing Condition
Minimum
Maximum
Average
Std Dev
MDL
H2O
(%)
0.09
0.28
0.18
0.04
0.05
Novec
(ppm)
7.86
15.63
12.23
1.66
0.02
CO2
(%)
73.94
94.18
86.10
3.79
0.11
CH,
(ppm)
0.79
38.80
5.02
4.18
0.76
CO
(ppm)
413.07
5,682.59
2,791.35
1,593.56
0.45
COF2
(ppm)
5.53
14.36
9.47
1.74
0.13
NO2
(ppm)
BDL
2.81
0.82
0.50
0.66
HF
(ppm)
BDL
151.56
112.93
34.37
1.37
C2F6
(ppm)
0.39
1.18
0.80
0.16
0.06
SiF4
(ppm)
BDL
1.81
BDL
0.28
0.61
C3F8
(ppm)
5.43
14.42
9.97
1.99
0.46
H2CO
(ppm)
BDL
5.01
BDL
0.37
1.29
BDL = below detectable limit
MDL = method detection limit
nc = not calculated
Table 4-6. Data Summary for the Novec™ 612 Cover Gas on Machine #4, Sample Point #2, Static Condition
Minimum
Maximum
Average
Std Dev
H2O
(%)
0.03
0.22
0.04
0.03
02
(%)
7.80
7.90
7.83
0.06
CO2
(%)
89.41
95.61
92.32
1.25
HF
(ppm)
80.50
177.54
126.21
12.38
CH4
(ppm)
0.92
11.77
8.45
1.85
Novec
(ppm)
5.17
12.79
10.54
1.36
C3F8
(ppm)
6.79
10.67
9.75
0.57
COF2
(ppm)
15.59
30.35
26.51
2.77
CO
(ppm)
4,520.21
11,817.11
8,108.58
1,768.96
C2F6
(ppm)
1.41
1.95
1.78
0.08
4-7
-------
4.3 SF6 Cover Gas with Air Diluent
In order to evaluate the replacement cover gases (HFC-134a and Novec™ 612), baseline
testing using SF6 was conducted on both machines at the Palmyra and Hannibal facilities. Table
4-7 thru 4-9 shows the data for cold chamber machines #32 and #4.
Average SF6 concentrations in the cold chamber #32 were 1.2 percent. Compounds
measured within the crucible headspace included: CO2, CO, NO, N2O, NO2, CFLi, HF, SO2 and
H2CO, and details are shown in Table 4-7. An anomaly in the data was observed while sampling
at the location #3 on cold chamber #32. Immediately after connecting the sample probe to
location #3, hydrochloric acid (HC1) was detected. A maximum concentration of 7 ppm was
measured shortly after connecting the probe; however, after this spike, HC1 levels decayed below
the detection limit for the remainder of the test. This behavior is independent of the process
activity and therefore may be attributed to chlorine being stripped from the Teflon core in the
extraction line.
SF6 cover gas testing continued on cold chamber #4 at the Hannibal facility. SF6
measurements within the crucible headspace were between 0.2 and 0.33 percent. Data tables for
the SF6 testing on both the ingot loading side (sample location #1) and pump side (sample
location #2) are shown in Tables 4-8 and Tables 4-9. The tables indicate some uneven cover gas
distribution within the same crucible headspace. For example, SFe readings averaged at 2,036
ppm (ingot loading side) versus 3,347 ppm (pump side). The detected compounds, CO2, CO,
CH4, NO, N2O, NO2, HF, SO2 and H2CO were all slightly higher at the sample location #1. This
may be associated with the intrusion of ambient air into the headspace during ingot loading.
4-8
-------
Table 4-7. Data Summary for the SF6 Cover Gas on Machine #32
H2O
(%)
C02
(ppm)
CO
(ppm)
CH4
(ppm)
N20
(ppm)
NO
(ppm)
N02
(ppm)
SF6
(ppm)
HF
(ppm)
S02
(ppm)
H2CO
(ppm)
HCL
(ppm)
Point #3
Minimum
Maximum
Average
Std Dev
BDL
0.38
0.08
0.05
975.05
2,211.27
1,290.12
239.86
2.70
15.53
5.47
3.02
BDL
1.34
0.41
0.19
3.09
7.81
6.95
0.83
1.65
11.64
9.52
1.84
BDL
BDL
BDL
na
6,669.69
12,706.61
12,077.99
1,063.66
BDL
161.93
109.08
40.77
125.30
356.30
288.95
48.04
0.63
1.20
1.05
0.12
BDL
7.05
1.34
1.11
Point #6
Minimum
Maximum
Average
Std Dev
BDL
0.11
0.04
0.01
BDL
1,067.19
411.69
172.46
2.17
24.88
4.35
3.65
0.67
3.92
1.12
0.39
1.15
4.06
2.32
0.93
BDL
7.64
1.76
1.65
BDL
15.93
4.36
4.20
10,587.04
12,831.12
12,276.78
393.41
BDL
0.30
0.12
0.04
93.80
321.69
153.84
41.00
0.77
1.12
0.96
0.07
BDL
BDL
BDL
na
Point #4
Minimum
Maximum
Average
Std Dev
MDL
BDL
0.08
0.05
0.01
0.08
BDL
708.19
331.54
150.56
342.48
2.83
17.37
5.13
3.18
1.53
0.64
2.45
0.95
0.31
0.36
2.46
3.53
2.75
0.23
0.12
1.70
4.65
3.12
0.53
0.82
BDL
BDL
BDL
na
3.35
11,474.55
12,154.06
11,929.89
154.89
863.88
0.07
0.19
0.13
0.03
0.05
177.07
222.24
199.50
9.19
10.31
BDL
1.02
0.91
0.05
0.57
BDL
BDL
BDL
na
0.24
BDL = below detectable limit
MDL = method detection limit
na = not applicable
nc = not calculated
Table 4-8. Data Summary for SF6 Cover Gas with Air Diluent on Machine #4, Sample Point #1
Minimum
Maximum
Average
Std Dev
MDL
H20
(%)
0.35
1.04
0.68
0.15
0.05
02
(%)
19.90
20.00
19.95
0.06
0.1
CH4
(ppm)
BDL
32.73
2.16
3.36
0.76
CO2
(ppm)
206.74
1,848.61
448.76
252.26
49.56
CO
(ppm)
0.73
269.97
30.17
38.74
0.45
NO
(ppm)
BDL
8.37
1.97
1.27
3.19
N2O
(ppm)
2.22
14.30
5.09
1.91
0.5
NO2
(ppm)
4.11
34.87
14.13
5.52
0.66
HF
(ppm)
12.98
56.00
31.68
6.27
1.37
SF6
(ppm)
1,072.94
3,051.00
2,036.70
447.53
58.16
SO2
(ppm)
45.81
335.11
138.12
61.83
7.0
H2CO
(ppm)
BDL
5.93
0.40
0.53
1.29
4-9
-------
Table 4-9. Data Summary for SF6 Cover Gas with Air Diluent on Machine #4, Sample Point #2
Minimum
Maximum
Average
Std Dev
MDL
H20
(%)
0.16
0.53
0.26
0.08
0.05
02
(%)
20.50
20.70
20.55
0.10
0.1
CH4
(ppm)
BDL
2.56
0.56
0.24
0.76
C02
(ppm)
158.86
662.98
247.48
81.15
49.56
CO
(ppm)
1.42
71.67
6.83
10.13
0.45
NO
(ppm)
BDL
3.70
BDL
0.51
3.19
N20
(ppm)
1.56
3.92
2.33
0.41
0.5
N02
(ppm)
1.59
10.58
3.67
1.08
0.66
HF
(ppm)
BDL
25.41
14.07
6.39
1.37
SF6
(ppm)
2,059.83
3,899.50
3,347.14
386.22
58.16
S02
(ppm)
62.62
187.21
99.83
23.99
7.0
BDL = below detectable limit
MDL = method detection limit
nc = not calculated
4-10
-------
5.0 Conclusions
5.1 Cover Gas Test Observations
5.1.1 HFC-134a Cover Gas Testing with N2 and CO2 Diluents
The primary compounds detected when running HFC-134a and a N2 or CO2 diluent are:
CO2, CO, HF, C2F6, and COF2. The time series plots in Appendix A show that concentrations of
these compounds decreased during ingot loading, which indicates that the gases are originating
from inside the crucible. For machine #32, FTP and C2Fe concentrations were on the order of 100
to 200 parts per million (ppm) and 2 ppm, respectively. COF2 concentrations were below the
detectable limit of the FTIR instrument when using an N2 diluent, but increased to levels greater
than 10 ppm with a CO2 diluent. There was no marked difference in the results obtained during
cold-chamber casting and static (i.e., during periods when no melt casting occurred) periods. The
plots in Appendix A also illustrate that additional degradation products, such as H2CO, CH4,
C2H2, and C2H4 are formed with the addition of ambient air during the ingot loads. Compounds
including H2CO, NO, N2O and NO2 also had background levels inside the headspace that sharply
increased during ingot loading. Detection of C2H2 and C2H4 was sporadic with a few spikes that
occur during ingot loading. Other than these spikes, concentrations were close to or below
detectable limits. Another trend was that the CO, HF, NO, N2O and NO2 and COF2
concentrations were higher for the FCFC-134a/CO2 testing than F£FC-134a/N2.
5.1.2 Novec™ 612 Cover Gas Testing with CO2 and Air Diluents
The primary compounds detected when using Novec™ 612 as a cover gas are: CO,
COF2, CsFg, C2Fe and FTP. Measurements were conducted during both casting and static
conditions on a cold-chamber die-casting machine. FTP was present at relatively constant levels
during both conditions; however, the concentrations increase to higher levels during ingot
loading. CH4, SiF4, and H2CO were observed to increase during ingot loading. SiF4 and CH4
were not detected during the static tests, which may indicate that silicon is entering with the
ingot. H2CO was present without the addition of ingots. Low levels of NO (ingot side only) and
NO2 were detected during casting but were not present during the static conditions. These levels
also increased during ingot loading. Since the Novec™ 612 feed concentration was low
compared to FtFC-134a and SF6, the average concentrations detected within the crucible
headspace was accordingly much lower as well. The highest average concentration of Novec™
612 was 12 ppm versus average concentrations for FtFC-134a (ranging from 70 to 2,760 ppm) or
SF6 (ranging 2,040 to 12,280 ppm).
5-1
-------
5.1.3 SF6 Cover Gas Testing with Air Diluent
The primary compounds detected when using SFe are: HF and SC>2. For both cold-
chamber measurements, HF concentrations were on the order of 10-30 ppm. Low levels of NO,
N2O and NO2 (i.e., on the order of 2 ppm) were detected. Additional compounds detected
included H^CO. H^CO levels remained close to the FTIR detection limit.
5.2 Cover Gas Degradation
One of the main objectives with this cover gas study is to determine the level of
degradation. Degradation estimates are calculated as:
_. _^ , . Delivery Concentration - Measured Concentration
Percent Degradation =
Delivery Concentration
Tables 5-1 and 5-2 provide a summary of all the tests and calculated level of degradation.
The above equation assumes that the crucible headspace is a well-sealed environment preventing
infiltration of dilution air. However, it is not due to ingot loading and seal leaks in the crucible
lids. Consequently, in order to correct for ambient air dilution, a dilution factor is calculated by
measuring the concentration of specific compounds inside the crucible headspace.6
For the HFC-134a testing, dilution factors were estimated using delivery and measured
CO2 data. CO2 data is used wherever possible since the data is taken by FTIR on a continuous
basis. Using CO2 data is not valid when CO2 is formed as a by-product of the cover gas process,
such as occurs when using an N2 diluent. It is assumed that dilution levels experienced during
the CO2 test condition will be consistent with dilution levels present during the N2 diluent tests
since measurements were conducted on the same die casting machine and at similar cover gas
flow rates; consequently, dilution factors derived from the CO2 data are applied to the HFC-
134a/N2 data.
For the Novec™ 612 cover gas tests, O2 values were used to determine dilution inside the
crucible. Steady state values for 02 concentrations during the Novec™ 612 testing were
averaged to determine the dilution factor. 02 readings were used for the dilution calculations
because O2 was the most suitable of the gases in the sample matrix. The level of oxygen in
ambient air was measured at a constant 20 percent by volume. It has the lowest chance of
6 Please refer to Section 5.5 for a discussion regarding the uncertainty associated with this methodology.
5-2
-------
interaction with other gases, therefore minimizing any bias from reaction products. A sample
calculation for a 7 percent C>2 reading is shown below.
_
Dilution Factor = - = 0.65
20%
The above calculation uses 20 percent by volume Ch for ambient air. For the Novec™
612 testing, 7.3 liters per minute (1pm) of air was delivered with 41.3 1pm of CC>2 for a diluent
gas. In this case, the C>2 reading must be corrected to account for the incoming O2 which is
determined as:
[73 i
: x 20% = 4.0%
41.3 + 7.3
The equation above assumes the C>2 concentration in the air of the delivery gas is 20
percent. Therefore the dilution factor for Novec™ 612 testing is calculated as:
/ A f)0/
Novec™ 612 Dilution Factor = — = 0.8
20%
5-3
-------
Table 5-1. Machine #32 Percent Degradation for Cover Gas Testing
Table
4-1
4-1
4-2
4-2
4-2s
4-7
4-7
4-7
Date
1-Oct
1-Oct
1-Oct
1-Oct
1-Oct
3-Oct
3-Oct
3-Oct
Time
0932-1036
1036-1236
1618-1845
1845-1916
1916-2000
0959-1037
1114-1154
1200-1222
Die
Casting
Machine
Cold #32
Cold #32
Cold #32
Cold #32
Cold #32
Cold #32
Cold #32
Cold #32
Cover Gas
Mixture
Components
HFC-134a/N2
HFC-134a/N2
HFC-134a/CO2
HFC-134a/CO2
HFC-134a/CO2
SFg/Air
SFg/Air
SFg/Air
Sample
Location
#3
#4
#4
#6
#6
#3
#6
#4
Cover Gas Mixture Flow "
(1pm)
21
21
20
20
20
65
65
65
Cover Gas
Delivery
Cone. "
(ppm)
4,000
4,000
4,000
4,000
4,000
19,000
19,000
19,000
Cover Gas
Measured
Cone.
(ppm)
84
68
338
206
269
12,078
12,277
11,930
Dilution
Facto rb
0.79 (+0.06)
0.79 (+0.06)
0.80 (+0.07)
0.75 (+0.01)
0.80 (+0.02)
0.69 (+0.17)
0.69 (+0.17)
0.69 (+0.17)
Cover Gas
Degradation
97%
98%
89%
93%
92%
8%
6%
9%
Test
Condition
Casting
Casting
Casting
Casting
Static
Casting
Casting
Casting
aAs provided by Internet and AMC
bRange represents the standard deviation for measured concentrations of indicator compounds.
Table 5-2. Machine #4 Percent Degradation for Cover Gas Testing
Table
4-3
4-4s
4-5
4-6s
4-8
4-9
Date
6-Oct
6-Oct
6-Oct
7-Oct
7-Oct
7-Oct
Time
1359-1525
1530-1600
1451-1655
1600-1654
1036-1227
0830-1025
Die
Casting
Machine
Cold #4
Cold #4
Cold #4
Cold #4
Cold #4
Cold #4
Cover Gas Mixture
Components
Novec 612/CO2/Air
Novec 612/CO2/Air
Novec 612/CO2/Air
Novec 612/CO2/Air
SF6/Air
SF6/Air
Sample
Location
#1
#1
#2
#2
#1
#2
Cover Gas Mixture Flow "
(1pm)
Novec/CO2=48.6, Air=7.3
Novec/CO2=48.6, Air=7.3
Novec/CO2=45.6, Air=6.8
Novec/CO2=45.6, Air=6.8
37.5
37.5
Cover Gas
Delivery
Cone. "
(ppm)
126
126
126
126
5,000
5,000
Cover Gas
Measured
Cone.
(ppm)
7
8
10
12
2,037
3,347
Dilution
Facto rb
0.77 (+0.07)
0.77 (+0.07)
0.76 (+0.06)
0.76 (+0.06)
0.77 (+0.07)
0.77 (+0.07)
Cover Gas
Degradation
93%
92%
89%
87%
47%
13%
Test
Condition
Casting
Static
Casting
Static
Casting
Casting
aAs provided by Internet and 3M™
bRange represents the standard deviation for measured concentrations of indicator compounds.
5-4
-------
For the SF6 cover gas tests, air was used as a diluent gas and therefore 62 readings were
20 percent inside and outside of the crucible. As a result, CFLt was chosen as an alternative to C>2
to calculate the dilution factor since its presence is constant in ambient air. Results illustrate that
CH4 values were stable within the crucible headspace and the impact of ingot loading is clearly
shown in time series plots. Dilution factors were determined by comparing CFLt values
measured in ambient air to readings taken inside the crucible headspace. As with 62, steady state
values for CH4 are used to determine dilution so that values are not biased higher during ingot
loading. During testing at the Hannibal facility, ambient air concentrations for the SFe testing
never reached steady state and the readings were deemed unreliable. Therefore, it was assumed
that the dilution for the Novec 612™ testing on the same machine is approximately the same as
the dilution for the SF6 testing.
Average percent degradation for HFC-134a with N2 and CC>2 diluents, and Novec 612™
cover gases were 98, 91, and 90 percent, respectively. SF6 degradation estimates on the cold
chamber tests were on the order of 10 percent; however, during one test an unexpectedly high
value of 47 percent was observed. The reason for this occurrence is unknown.
5.3 Occupational Health and Safety
Each of the cover gases evaluated in this study can produce by-products that may be of
concern from an occupational exposure standpoint. For example, formaldehyde (H2CO),
carbonyl fluoride (COF2), hydrofluoric acid (HF), and carbon monoxide (CO), have very low 8-
hour time-weighted average exposure limits of 0.75, 2, 3, and 25 ppm, respectively7. While gas
concentrations presented in this report are in some cases significantly higher than these OSHA
levels, it is important to note that these concentrations are not reflective of actual occupational
exposure conditions, in that they have been measured within the enclosed crucible headspace,
and not ambient air within operator "breathing" zones. To confirm this observation, during some
measurement activities at the Palmyra facility, FTIR samples were taken at distances appropriate
to operator "breathing" zones. Even though maximum crucible headspace concentrations for
H2CO and COF2 were above occupational standards, ambient measurements taken outside the
crucible were below the FTIR detectable limits of 300 ppb. While this test is by no means a true
industrial hygiene analysis, the results illustrate that given the high level of ventilation present at
these facilities, the crucible head space gases are contained to such an extent that significant
concentrations were not found in ambient air close to the crucible lid. However, further
occupational exposure monitoring would be required to confirm these observations.
OSHA Permissible Exposure Limits (PELs).
5-5
-------
5.4 Global Climate Change Impact Discussion
One of the benefits of using HFC-134a and Novec™ 612 as cover gases within
magnesium production and processing is that their contribution to global climate change is
significantly lower when compared to SFe. This is evident when comparing their estimated
global warming potentials (GWP). Table 5-3 presents GWP's of several compounds detected
during this study.
Table 5-3. Comparison of 100-Year GWP Estimates from the
Intergovernmental Panel on Climate Change's (IPCC's)
Second (1996) and Third (2001) Assessment Reports
Gas
Methane
Nitrous Oxide
HFC-134a
Perfluoromethane (CF4)
Perfluoroethane (C2F6)
Perfluoropropane (C3F8)
Sulfur Hexafluoride (SF6)
1996 IPCC GWP
21
310
1,300
6,500
9,200
7,000
23,900
2001 IPCC GWP
23
296
1,300
5,700
11,900
8,600
22,200
Sources:
IPCC (1996), Climate Change 1995: The Scientific of Climate Change. Intergovernmental Panel on Climate Change,
Cambridge University Press. Cambridge, U.K.
IPCC (2001), Climate Change 2001: The Scientific Basis. Intergovernmental Panel on Climate Change, Cambridge
University Press. Cambridge, U.K.
The crucible headspace contains a large variety of compounds, but only those with GWP
values were used to estimate the overall GWP impact of switching to alternate cover gases (i.e.,
HFC-134a and Novec™ 612) from SFe. This calculation was achieved by multiplying the
average concentrations (parts per million by volume) for each of the component cover gases and
applicable degradation products with their respective GWP factors (obtained from the Third
Assessment Report of the Intergovernmental Panel on Climate Change) to obtain a normalized
CC>2 GWP value. The total CC>2 GWP equivalent of the gases measured in the crucible headspace
for F£FC-134a and Novec™ 612 testing scenarios were summed and compared to the
corresponding SF6 condition (or baseline). Tables 5-4 and 5-5 show that when comparing the
composite, or overall GHG emissions, the alternate cover gases have a much lower impact. An
obvious source for this reduction can be found in a direct comparison of cover gas GWPs. SFe
has a GWP of 22,200, while HFC-134a's GWP is 1,300. Novec™ 612's GWP has not been
supplied by 3M, but is likely to be extremely low (i.e., Novec™ 612 is a fluorinated ketone,
5-6
-------
which is assumed to have a GWP on the order of I8). In addition to having lower GWPs, the
alternate cover gas compounds have much higher decomposition (on the order of 90 percent)
within the crucible headspace compared to SF6 (on the order of 10 percent). While the
decomposition of HFC-134a and Novec™ 612 does produce degradation products with GWPs,
their impact is minimal due to the very low concentrations generated. Compared against the
test, switching to HFC-134a and Novec™ 612 produces a reduction in overall global warming
impact of gases inside the crucible headspace on the order of 99 percent.9
The above comparison does not include the specific flow rates for each cover gas. In
order to provide a more comprehensive analysis of composite GWP, an additional comparison
was conducted. Using the ideal gas law, the molecular weights of each gas and the delivery flow
rate of the cover gas was used to estimate the composite emission rate in grams per hour (g/hr).
This equation can be described as follows:
Emission Rate — = ppm x MW x 1pm x -=-138.6 liters/mole x 1061
1 1 \ /
^ hour ) hour
ppm = measured average concentration in parts per million
MW = molecular weight in grams per mole
Ipm = gas flow in liters per minute
These values were summed to provide a composite GWP value that was weighted by the
cover gas flow rate. The overall GWP values were then compared against the corresponding
values for SFe. Based on this approach, both HFC-134a and Novec™ 612 were observed to have
a global warming impact greater than 99 percent lower than SF6.10 Details of the flow-weighted
GWP impacts are presented in Tables 5-6 and 5-7.
Milbrath, D. 3A4™ Novec™ 612 Magnesium Protection Fluid: It's Development and Use in Full Scale Molten
Magnesium Processes, Proceedings of the 60th Annual International Magnesium Association Conference, May
2003, Stuttgart, Germany.
9 Please refer to Section 5.5 for a discussion regarding the uncertainty associated with this methodology.
10 Ibid.
5-7
-------
Table 5-4. GHG Emission Comparison for Machine #32 Using HFC-134a and SF6
Table
4-1
4-1
4-2
4-2
4-2a
4-7
4-7
4-7
Die
Casting
Machine
Cold #32
Cold #32
Cold #32
Cold #32
Cold #32
Cold #32
Cold #32
Cold #32
Cover Gas Mixture
Components
HFC-134a/N2
HFC-134a/N2
HFC-134a/CO2
HFC-134a/CO2
HFC-134a/CO2
SF6/Air
SF6/Air
SF6/Air
Sample
Location
#3
#4
#4
#6
#6
#3
#6
#4
GWP
Weighted
CO2
4,591
4,646
799,398
750,129
796,386
1,290
412
332
GWP
Weighted
HFC-134a
109,000
88,603
438,713
268,088
349,120
0
0
0
GWP
Weighted
SF6
0
0
0
0
0
268,131,458
272,544,500
264,843,543
GWP
Weighted
CH4
161
179
43
52
67
9
26
22
GWP
Weighted
N2O
2,349
3,050
6,953
6,468
7,044
2,059
686
813
GWP
Weighted
C2F6
13,138
26,904
33,266
21,796
19,509
0
0
0
GWP
Weighted
C3F8
0
0
0
0
0
0
0
0
Normalized
CO2
Equivalent
129,239
123,382
1,278,373
1,046,533
1,172,126
268,134,816
272,545,623
264,844,710
Average by
cover gas
126,310b
1,165,677C
268,508,383d
Chg
from
SF6
>99%
>99%
-
Vindicates static test (i.e., die casting machine not casting)
bAverage composite GWP for HFC-134a/N2 cold-chamber tests (Table 4-1)
'Average composite GWP for HFC-134a/CO2 cold-chamber tests (Table 4-2)
dSF6 composite GWP baseline estimate for comparison with HFC-134a/CO2 andN2 cold-chamber tests (Table 4-7)
Table 5-5. GHG Emission Comparison for Machine #4 Using Novec™ 612 and SF6
Table
4-3
4-4a
4-5
4-6a
4-8
4-9
Die
Casting
Machine
Cold #4
Cold #4
Cold #4
Cold #4
Cold #4
Cold #4
Cover Gas Mixture
Components
Novec612/CO2/Air
Novec612/CO2/Air
Novec612/CO2/Air
Novec612/CO2/Air
SF6/Air
SF6/Air
Sample
Location
#1
#1
#2
#2
#1
#2
GWP
Weighted
CO2
870,240
968,471
860,991
933,429
449
247
GWP
Weighted
HFC-134a
0
0
0
0
0
0
GWP
Weighted
SF6
0
0
0
0
45,214,808
74,306,474
GWP
Weighted
CH4
162
21
115
159
50
13
GWP
Weighted
N2O
0
0
0
0
1,506
689
GWP
Weighted
C2F6
12,410
12,667
9,548
21,641
0
0
GWP
Weighted
C3F8
55,582
47,043
85,722
83,838
0
0
Normalized
CO2
Equivalent
938,394
1,033,063
956,376
1,028,438
45,216,813
74,307,423
Average by
cover gas
989,068b
59,762,118°
Chg
from
SF6
98%
-
Indicates static test (i.e., die casting machine not casting)
bAverage composite GWP for Novec™ 612/CO2/Air cold-chamber tests (Tables 4-3, 4-4, 4-5 and 4-6)
0 SF6 composite GWP baseline estimate for comparison with Novec™ 612/CO2/Air cold-chamber tests (Tables 4-8 and 4-9)
5-S
-------
Table 5-6. GHG (Weighted By Gas Flow Rate) Emission Comparison for Machine #32 Using HFC-134a and SF6
Table
4-1
4-1
4-2
4-2
4-2a
4-7
4-7
4-7
Die
Casting
Machine
Cold #32
Cold #32
Cold #32
Cold #32
Cold #32
Cold #32
Cold #32
Cold #32
Cover Gas Mixture
Components
HFC-134a/N2
HFC-134a/N2
HFC-134a/CO2
HFC-134a/CO2
HFC-134a/CO2
SFg/Air
SFg/Air
SFg/Air
Sample
Location
#3
#4
#4
#6
#6
#3
#6
#4
GWP
Weighted
CO2 (pfhr\
1
1
1,094
1,026
1,090
6
2
1
GWP
Weighted
HFC-134a
(g/hr)
363
295
1,392
850
1,107
0
0
0
GWP
Weighted SF6
(g/hr)
0
0
0
0
0
3,956,311
4,021,426
3,907,798
GWP
Weighted
CH4 (p/hr\
0
0
0
0
0
0
0
0
GWP
Weighted
N2O fa/hri
3
4
10
9
10
9
3
4
GWP
Weighted
C2F6 (g/hr)
59
121
143
94
84
0
0
0
GWP
Weighted
C3FS (g/hr)
0
0
0
0
0
0
0
0
Normalized
CO2
Equivalent
(g/hr)
432
427
2,638
1,979
2,290
3,956,326
4,021,431
3,907,803
Average by
cover gas
430b
2,302C
3,961,853d
Chg
from
SF6
>99%
>99%
-
Indicates static test (i.e., die casting machine not casting)
bAverage composite GWP for HFC-134a/N2 cold-chamber tests (Table 4-1)
"Average composite GWP for HFC-134a/CO2 cold-chamber tests (Table 4-2)
dSF6 composite GWP baseline estimate for comparison with HFC-134a/CO2 andN2 cold-chamber tests (Table 4-7)
Table 5-7. GHG (Weighted By Gas Flow Rate) Emission Comparison for Machine #32 Using Novec™ 612 and SF6
Table
4-3
4-4a
4-5
4-6a
4-8
4-9
Die
Casting
Machine
Cold #4
Cold #4
Cold #4
Cold #4
Cold #4
Cold #4
Cover Gas Mixture
Components
Novec 612/CO2/Air
Novec 612/CO2/Air
Novec 612/CO2/Air
Novec 612/CO2/Air
SFg/Air
SFg/Air
Sample
Location
#1
#1
#2
#2
#1
#2
GWP
Weighted
CO2 fo/hrl
2,893
3,220
2,686
2,880
1
1
GWP
Weighted
HFC-134a
(g/hr)
0
0
0
0
0
0
GWP
Weighted SF6
(g/hr)
0
0
0
0
384,894
632,539
GWP
Weighted
CH4 Mhr\
0
0
0
0
0
0
GWP
Weighted
N2O fa/hri
0
0
0
0
4
2
GWP
Weighted
C2F6 (g/hr)
129
183
93
207
0
0
GWP
Weighted
C3FS (g/hr)
790
668
1,143
1,117
0
0
Normalized
C02
Equivalent
(g/hr)
3,813
4,071
3,922
4,205
384,899
632,541
Average by
cover gas
4,003b
508,720C
Chg
from
SF6
99%
-
Indicates static test (i.e., die casting machine not casting)
bAverage composite GWP for Novec™ 612/CO2/Air cold-chamber tests (Tables 4-3, 4-4, 4-5 and 4-6)
CSF,5 composite GWP baseline estimate for comparison with Novec™ 612/CO2/Air cold-chamber tests (Tables 4-8 and 4-9)
5-9
-------
5.5 Uncertainty Discussion
The results of this measurement study should not be interpreted to represent an absolute
analysis of HFC-134a, Novec™ 612, and SFe cover gas degradation. While this study does
present a relatively accurate measurement analysis and approximate comparison of GHG
emissions, there are several areas of uncertainty inherent with this methodology. These areas of
uncertainty include FTIR error, dilution correction methods, and analytical and operational
variation across the two machines evaluated.
Measurements taken by the FTIR are subject to variability inherent with any highly
complex analytical device. While all prudent steps were taken to minimize this contributor to
uncertainty (see Section 3.2 (page 3-8)), some small degree of error is unavoidable.
The different cover gas feed matrices utilized throughout the study created difficulties in
defining a good indicator to estimate cover gas dilution within the crucible headspace. Three
indicator compounds were utilized to estimate dilution, CC>2, CH4, and 62. Although all these
gases may be produced from several sources and consumed in reactions with other components
in the headspace, they provided the best available approach under the circumstances (e.g., a
foreign indicator compound could not be added to the cover gas, since it could disrupt the cover
gas feed mix, and make its application to the melt surface unrepresentative of ideal "test"
conditions). However, considering the potential sources for indicator interference, and the
subsequent uncertainty associated with the development of dilution factors, the values (i.e., cover
gas destruction levels and GHG emissions) presented in this report should be considered as a
"best estimate" only, and not an absolute value.
Even though an effort was made to conduct the measurement study on machines as
identical as possible; there are some variations to consider when interpreting these results. For
example, cold chamber machine #32 uses much higher SFe concentrations (19,000 vs. 5000
ppmv SF6) and flow rates (65 vs. 38 1pm) than machine #4; consequently, it requires more than 4
times the cover gas for adequate melt protection. Additionally, the sampling ports used on both
machines were slightly different, which may impact cover gas distribution regimes. Since the
crucible headspace is a dynamic reaction space, it is not known how the sampling differences
between cold chamber machine #32 and #4 are reflected in the data. Consequently, due to the
differences in the analytical sampling and the actual furnaces used, when reviewing the results of
this analysis it is important that comparisons only be made between SF6 and the replacement
cover gas compounds for each respective machine, and not between HFC-134a and Novec™
612.
5-10
-------
&EPA
United States
Environmental Protection
Agency
Office of Air and Radiation (6202J)
Washington, DC 20460
EPA 430-R-04-004
May 2004
5-10
------- |