Report for Congress: Review of the U.S. Army Proposal for Off-Site Treatment and Disposal of Caustic VX Hydrolysate From the Newport Chemical Agent Disposal Facility
Assessment of the Treatability of
Caustic VX Hydrolysate at the DuPont Secure Environmental Treatment
Facility
By
Carmagen Engineering, Inc.
in consultation with the
Centers for Disease Control and Prevention
November 3, 2004
SUMMARY
To completely ascertain the capability and effectiveness of the DuPont
Secure Environmental Treatment (SET) facility to treat caustic VX
hydrolysate (CVXH), the Centers for Disease Control and Prevention
(CDC) and Carmagen Engineering, Inc. (Carmagen), recognized that, in
addition to reviewing the DuPont treatability test results, the
Newport Chemical Agent Disposal Facility (NECDF) destruction process
and the analytical methodologies for CVXH clearance also had to be
assessed to ensure that the hydrolysate being shipped to the SET
facility will be adequately characterized and that VX and EA 2192
levels in the CVXH will meet Army clearance specifications. Please
note that in this report, the more technically accurate term CVXH
generally is used in place of Newport caustic hydrolysate or NCH.
These assessments were considered essential elements to ensure safe
SET facility operations. Therefore, the Carmagen Team (Team) focused
its review in three areas consisting of (1) process issues at NECDF,
(2) analytical methods, and (3) CVXH treatment at DuPont. The review
comprised several meetings with people from the Army, Chemical
Materials Agency, Parsons, and DuPont at which presentations were
made, followed by in-depth discussions. These meetings were followed
up by written questions and requests for additional documentation.
Documentation received in response to the Team’s questions and
requests for additional information was substantial.
The major findings from the three areas examined by the Team are shown
below. These findings are valid only for an 8% diisopropylcarbodiimide
(DIC)-stabilized VX hydrolysate. The current database is insufficient
to allow extrapolation to other VX loadings or stabilizers.
Process Issues (Chapter 2)
Only laboratory/bench-scale runs have been completed for the process,
and scale-up to the integrated full-size facility is based on
anticipated processing conditions. Recently, several safety studies
were completed that recommended changes in the design and operation of
the NECDF. The impact of the responses to these recommendations and
possible facility changes on the final process is unknown.
Finding 2.1. The database supports the efficacy of neutralizing
DIC-stabilized VX using sodium hydroxide at the 8% VX-loading rate.
Scale-up of the process from laboratory/bench scale to pilot scale
should be operationally feasible. However, because the NECDF will be a
pilot facility, changes must be anticipated in operating mode and
hydrolysate composition sent for off-site treatment.
Finding 2.2. VX loading (weight percent) and the specific
stabilizer (DIC; dicyclohexyldicarbodiimide [DCC]) employed
significantly impact the process, hydrolysate composition, analytical
methods validation, and possibly solids formation. Scale-up of the
process from 8% to 16% VX loading is of particular concern (because of
the similarity of the organic-phase volumes from 16% to 33% VX-loading
batches), the potentially high VX concentration in the resulting
organic layer, and the analytical problems identified with 33% VX
loading.
Finding 2.3. The impact is unknown of solids formation during
hydrolysis on operations (potential for blockage of the in-line static
mixer, control valves, and sampling system), VX analytic methods, and
off-site hydrolysate treatment. The transition from 8% to 16% VX
loading, as well as stabilizer change, is of concern and requires
additional detailed studies.
Analytical Methods (Chapter 3)
The purpose of the review and evaluation of the analytical methods was
to define the adequacy of the proposed NECDF analytical methods to
meet current programmatic requirements for detecting and quantifying
VX and EA 2192 in the CVXH.
Finding 3.1. The methods for analyzing VX and EA 2192 in 8% VX-loaded,
DIC-stabilized CVXH are adequate to detect and quantify at the
established clearance levels for VX and EA 2192 (non-detected with a
U.S. Environmental Protection Agency (EPA) method detection limit (MDL)
of ≤20 parts per billion [ppb] for VX and ≤1 part per million [ppm]
for EA 2192).
Finding 3.2. The use of the EPA’s method detection limit (MDL)
for clearance levels does not preclude analytical instrument detection
of low levels of VX and EA 2192 (generally <20 ppb VX and <1 ppm EA
2192) in the CVXH. The perception that the MDL clearance criteria
indicate absence of analytically detectable VX and EA 2192 could be
misleading. While CDC believes that utilizing the MDL approach would
not result in public health concerns, the Army needs to address
potential public misperceptions regarding the detection or
non-detection of VX in CVXH. A simpler reporting scheme (i.e.,
non-detected, detected at <20 ppb, or detected at >20 ppb) should be
considered.
Finding 3.3. The overall quality assurance (QA) and quality
control (QC) plan and procedures for the NECDF laboratory are well
designed and documented. However, NECDF laboratory personnel must
continue to implement the QA/QC plan by developing day-to-day
operational QC data to demonstrate that all analytical systems are
operational and under control before plant startup.
Caustic VX Hydrolysate Treatment (Chapter 4)
Once transported to the SET facility, CVXH will be further treated to
adjust the pH and remove the organic by-products by a series of
physicochemical and biologic processes. The DuPont treatability
studies were designed and executed to obtain scale-up parameters for
engineering design and regulatory compliance, rather than (except for
a few specific species) to assess fate, transport, and
biodegradability of environmental contaminates. The treatability
studies also investigated the capability of the SET facility to treat
alternating hydrolysate feeds from Aberdeen (sulfur mustard [HD]) and
Newport (VX).
Finding 4.1. The SET facility effectively treats the CVXH
generated from an 8% VX loading with DIC stabilizer (i.e., pH
adjustment, thiolamine destruction, conversion of ethyl
methylphosphonic acid to methyl phosphonic acid [MPA]), except for MPA,
for which only minimal reduction is demonstrated.
Finding 4.2. The SET facility treatment performance should be
unaffected when treatment of hydrolysate feeds from Aberdeen (HD) and
Newport (VX) is alternated.
Finding 4.3. The DuPont treatability studies have not yet
demonstrated the effective treatment of 16% VX-loaded CVXH, nor of 8%
VX-loaded CVXH with DCC or a mixture of DIC and DCC stabilizers.
Table of Contents
Summary
Table of Contents
1. Introduction
1.1. Background
1.2. Nature of the Caustic VX Hydrolysate
1.3. Clearance
1.4. Analytical Methods
1.5. Carmagen Engineering, Inc.
1.6. Report Outline
2. Process Issues
2.1. Introduction
2.2. Process Description
2.3. Process Chemistry
2.4. Findings
3. Analytical Methods
3.1. Introduction
3.2. Sampling Representativeness
3.3. Analysis of VX in Caustic VX Hydrolysate
3.3.1. Data Evaluation/Interpretation Criteria
3.3.2. Method Description and Documentation
3.4. Analysis of EA 2192 in Caustic VX Hydrolysate
3.4.1. Data Evaluation/Interpretation Criteria
3.4.2. Method Description and Documentation
3.5. Use of Analytical Data for Clearance
3.6. Quality Assurance and Quality Control Procedures
3.7. Findings
4. Caustic VX Hydrolysate Treatment
4.1. Introduction
4.2. Extent of Treatment
4.2.1. pH Adjustment
4.2.2. Hydrogen Peroxide Oxidation
4.2.3. PACT® Biotreatment
4.3. Environmental Persistence and Agent Loading Effects
4.4. Findings
5. Major Findings
6. References
1. Introduction
1.1 Background
The Newport Chemical Depot, Newport, Indiana, stockpile comprises the
chemical nerve agent O-ethyl S-[2- (diisopropylamino) ethyl] methyl
phosphonothiolate (VX) stored in bulk quantities (1269 tons in 1690
containers). VX contains phosphorus double-bonded to an oxygen atom
and single-bonded to a carbon atom. VX is stabilized with several
percent of either diisopropylcarbodiimide (DIC) or
dicyclohexyldicarbodiimide (DCC) to protect against decomposition.
Forty-six percent of the stockpile at Newport consists of VX
stabilized with DIC (potentially with small amounts of DCC stabilizer
as a contaminant), 16% stabilized with DCC, and 38% stabilized with
both DIC and DCC. VX is highly toxic and lethal in both liquid and
vapor forms. Because munitions containing agent and energetics are not
present at Newport, the process requirements for disposing of only ton
containers of agent are less demanding than the processing
requirements for the more complex stockpiles at most sites.
The Newport Chemical Agent Disposal Facility (NECDF) was designed and
is to be operated as a pilot-plant facility because the process has
been demonstrated only at a laboratory/bench scale. Production
operation will begin only after pilot-scale operations have been
completed, the data reviewed and assessed, and approval granted by the
State of Indiana and the federal government. Because pilot-plant
operations generally uncover unknown elements, the probability is high
of process modifications and change— including possible changes in the
analytical methods and procedures used to support plant operations and
hydrolysate clearance—during this piloting period.
The NECDF was designed to destroy VX using caustic hydrolysis in a hot
(194 degrees Fahrenheit [°F]) solution of sodium hydroxide. Initially
the plan was to further treat the resulting hydrolysate on-site by
Supercritical Water Oxidation (SCWO) and to ship the final SCWO
effluent (brine) to a Treatment, Storage, and Disposal Facility (TSDF).
Mechanical problems encountered in the SCWO engineering-scale test,
conducted in Corpus Christi, Texas, in 2000, led to initiation of
studies to directly ship the NECDF hydrolysate to an off-site
treatment facility as an alternative to on-site SCWO treatment. The
terrorist attacks of September 11, 2001, and continuing questions
about the feasibility of implementing SCWO on-site in any reasonable
timeframe supported the decision to adopt “Project Speedy
Neutralization.” This involves shipment of the neutralized product
(i.e., caustic hydrolysate) off-site for further treatment.
Detailed testing of the caustic hydrolysis process began with the
Alternative Technologies and Approaches program in the 1990s. The
“recipe” for NECDF agent destruction using sodium hydroxide was tested
on a laboratory scale, and an agent loading of 33% was chosen for the
program.
Confirmation of the efficiency of destruction of VX depends on the
analytical methods available to monitor for residual VX and EA 2192
levels in the resultant hydrolysate. During the past ten years changes
in analytical techniques and instrumentation, coupled with increased
personnel experience with these analyses, have lowered the detectable
concentration of VX to the low parts per billion (ppb) levels and the
detectable concentration of EA 2192 to the low tenths of parts per
million (ppm) levels for 8% VX loading hydrolysate. However, the
complexity and variability of the 33% VX loading hydrolysate continued
to complicate the VX analysis.
By October 2003, the Project Manager for Alternative Technologies and
Approaches had begun to investigate the use of reduced VX loading to
preserve resources and obviate the need to resolve differences in data
and data interpretation for the 33% VX-loading hydrolysate. The
program plans to begin operations at 8% VX loading of DIC-stabilized
agent and then, through a carefully monitored ramping-up process, move
to 16% VX-loading, DIC-stabilized agent.
1.2 Nature of the Caustic VX Hydrolysate
The hydrolysate that will result from the caustic hydrolysis of VX at
the NECDF comprises an aqueous phase and an organic phase. The organic
phase exists both as an emulsion with droplets distributed throughout
the continuous aqueous phase and as a visible organic layer that
floats on top of the continuous aqueous phase. The extent to which a
separate organic phase floats on the lower aqueous layer depends on
the VX loading. As the VX loading increases, the quantity of organic
phase available to form an organic layer (above that which forms a
stable [or metastable] emulsion) increases.
At 33% agent loading (weight percent), the organic layer was
significant (3%–5% by volume). The VX concentration in this organic
layer was approximately 20 times the concentration in the bulk
hydrolysate (>20 ppb), although disagreement exists within the program
about the validity of the measurements (Wojciechowski, 2003). For 16%
agent loading, the organic layer was 2–3 volume percent; for 8% agent
loading, the separate “organic layer” was only a sheen at the surface
of the hydrolysate. The “organic layer” has not been analyzed at 8%
and 16% agent loadings; only mixed (homogenized) samples were
analyzed. Obtaining samples of this organic layer for 8% agent loading
poses significant technical difficulties. Centrifugation of a
550-milliliter (mL) sample of 8% CVXH showed that the maximum organic
layer that could be “separated” was 0.45%–0.5%. These differences
demonstrate the significant impact of agent loading on hydrolysate
characteristics.
1.3 Clearance
Since its inception, a key tenet of the Army Chemical Demilitarization
program has been safety of the workers and public. Department of the
Army (DA) Pamphlet (PAM) 38561, entitled “Toxic Chemical Agent Safety
Standards,” defines the approach for verifying the thoroughness of the
neutralization process as using laboratory analyses to ensure that the
chemical agent is ≤20 ppb. This concentration is measurable and is a
quantifiable upper limit concentration in drinking water (20 ppb
criterion is for soldiers). However, the procedure and methodology to
verify the 20 ppb criterion in CVXH have been a challenge (see Section
3). As stated in the Low Level VX (LLVX) panel report (Science
Applications International Corporation [SAIC], 2003):
The panel is not aware of any document that clearly states the exact criteria for offsite shipment of VX hydrolysate from NECDF or any document that codifies the Army’s commitment to the public for offsite shipment.
The report, Generation and Clearance of Hydrolysate for Treatability Studies in Support of Newport Operations, states:
To clear the hydrolysate, the analytical results must be non-detect for VX with a method detection limit (MDL) of less than or equal to 20 parts per billion (ppb). Non-detect is defined as the absence of a signal in the VX retention time window for ion 128, or a signal with a signal-to-noise (S/N) ratio of less than or equal to 3, or a concentration below the calculated MDL.
These criteria are incompatible in that an analytical response for VX
could be classified as “analytically detected” by implementation of
the “analyte retention time/signal-to-noise (S/N) ratio equal to or
greater than 3” detection criteria, but reported as “nondetect” by the
“less than the established MDL” criterion (see Section 3).
1.4 Analytical Methods
Significant resources were expended for almost a decade to develop an
analytical method that could reliably and accurately measure VX
concentration in CVXH at lower and lower levels for a 33% VX agent
loading without success. The newer analytical methods demonstrated the
presence of detectable levels of VX in 33% DIC-stabilized CVXH and the
inability to demonstrate an MDL of ≤20 ppb. This unexpected result led
to an aggressive investigation of the causes and possible solutions
for addressing the issue to bring the plant into operation.
An independent assessment panel was convened in October 2003 to
evaluate the significance of the observation of “persistent” LLVX in
caustic hydrolysate at the 33% agent loading level and to determine
whether data were sufficient to confirm whether VX forms in CVXH (SAIC,
2003). Two conclusions of the panel were:
There are significant uncertainties in the Solid Phase Extraction (SPE)/gas chromatography-ion trap mass spectrometry (GC-ITMS) method that make it difficult or impossible to quantify LLVX.
It is not possible to determine the origin of the “persistent” LLVX in VX hydrolysate from the currently available data. The panel could not rule out formation of VX in VX hydrolysate or the hypothesis that has been advanced that there is a quasi steady state concentration of VX in VX hydrolysate due to a competition between agent destruction and formation. The current data from the analytical method did not enable the panel to determine if detectable VX was originating from VX hydrolysate (that is, either residual untreated VX or formation within the VX hydrolysate matrix) or was formed during the analytical procedure.
Consequently, efforts during the past year have been devoted to evaluating the effect of reduced VX loading on
VX caustic hydrolysis destruction,
VX reformation during long-term storage, and
VX formation after a reduction in pH accompanied by a concomitant formation of an organic layer.
This evaluation has paralleled the development, evaluation, and validation of analytical methodologies for measuring VX and EA 2192 in the 8% DIC-stabilized CVXH. At the time of this writing, methods for analyzing VX and EA 2192 in 8% DIC-stabilized CVXH and VX in 16% DIC-stabilized CVXH had been established in the NECDF laboratory, and the performance of these methods had been validated through various precision and accuracy studies. Implementation and validation of methods for ethyl methylphosphonic acid (EMPA), methyl phosphonic acid (MPA), and thiolamine in 8% DIC-stabilized hydrolysate are expected to be completed shortly. Similar work on other methods required for 16% DIC-stabilized hydrolysate and 8% DCC-stabilized hydrolysate were scheduled for completion later in 2004. Validated methods for anticipated processing conditions are essential to ensure that hydrolysate shipped off-site to a TSDF meets Army criteria.
1.5 Carmagen Engineering, Inc.
The Centers for Disease Control and Prevention (CDC) engaged Carmagen
Engineering, Inc. (Carmagen) to assemble a group of knowledgeable
experts (Team) to help evaluate the DuPont Technical Assessment on
U.S. Army Newport (Indiana) Project (March 2004). The Team consisted
of a former chairman of the National Research Council Stockpile
Committee, a retired assistant director for the CDC/NCEH Division of
Laboratory Sciences, a retired Program Manager for Chemical
Demilitarization, a professor at Stevens Institute of Technology, a
retired regional laboratory director for EPA, and a former
environmental health and safety manager/process design manager for
ARCO Chemical. Specifically, Carmagen was asked to evaluate the
“Treatability of Newport (Indiana) Caustic Hydrolysate” portion of the
DuPont report.
To ascertain the capability and effectiveness of the DuPont Secure
Environmental Treatment (SET) facility at Chambers Works (Deepwater,
New Jersey) to treat CVXH, the Team recognized that an assessment of
the NECDF destruction process and an examination of the analytical
methodologies to be used for CVXH clearance were required to ensure
that the hydrolysate being shipped to the SET facility will be
adequately characterized and that VX and EA 2192 levels in the CVXH
will meet Army clearance specifications. These assessments were
considered essential elements to ensure safe SET facility operations.
Therefore the Carmagen Team focused on three areas:
Process Issues (NECDF),
Analytical Methods, and
Caustic VX Hydrolysate Treatment (DuPont).
The review comprised several meetings
with people from the Army, Chemical Materials Agency, Parsons, and
DuPont at which presentations were made and discussed in depth. These
meetings were followed up by written questions and requests for
additional documentation. Documentation received in response to the
Team’s questions and requests for additional information was
substantial.
1.6 Report Outline
The report contains five chapters.
Introduction—Discusses the historical evolution of the NECDF project and the charge to and approach taken by the Carmagen Team.
Process Issues—Discusses the impact of VX loading on the process, i.e., nature and extent of the two-phase CVXH, VX partitioning to the organic layer, clearance quality assurance (QA) and quality control (QC), scale-up, and storage.
Analytical Methods—Reviews and evaluates the use and data quality objectives of VX and EA 2192 measurements, sampling procedures, validation of methods, and QC of the analytical processes.
Caustic VX Hydrolysate Treatment—Describes pH adjustment, oxidative pretreatment, PACT® biotreatment, and VX and EA 2192 destruction.
Major Findings—Presents major findings.
2. Process
Issues
2.1 Introduction
Although the primary purpose of this report is to examine issues
associated with the treatability of the hydrolysate produced by the
Newport facility, as noted in the Introduction, a discussion of
processing issues is important. The composition of hydrolysate sent
for treatment depends on the nature of the VX being hydrolyzed (i.e.,
agent loading, stabilizer), neutralization process, process operating
conditions, process effectiveness, and consistent process operation.
Confirmation of the composition of the hydrolysate (efficacy of
treatment) is related to the accuracy of the analytical methodologies
(see Chapter 3) and whether the sample(s) used for the analysis
represent the batch being processed. The satisfactory treatment of
each batch is determined on the basis of analysis of the hydrolysate
samples.
Only laboratory/bench-scale runs have been completed for the process,
and scale-up to the integrated full-size facility is based on the
anticipated processing conditions. At startup, NECDF intends to
operate the reactor at a VX loading of 8%, rather than the 33%
originally planned. This has process and operational consequences that
are discussed later in this chapter. The Army proposes that VX loading
will be increased to 16% as experience is gained with the process and
equipment, and when analytical methodologies and successful off-site
treatment capability demonstrated at the higher loading are validated.
The change from the proposed 33% VX loading to 8% VX loading will
increase substantially the total quantity of hydrolysate to be treated
and the length of time the Newport facility will operate.
2.2 Process Description
The process for VX neutralization at Newport uses batch processing
(Figure 2-1). Each batch consists of the following sequential steps:
The reactor is charged with caustic.
The reactor is heated to approximately 194ºF.
The reactor circulation loop is activated, and the agitator in the reactor is started.
Agent is added to the reactor using a feed line in the recirculation piping. The amount of agent added is determined by the VX loading target for a given batch. Two phases are present in the reaction mixture—an aqueous phase and an organic phase. The relative volumes of the two phases are determined by the VX loading.
VX and caustic are mixed by the agitator in the reactor and by the static mixer in the recirculation piping. The static mixer is designed to achieve an organic droplet size of approximately 10–30 microns (µm).
After the reaction has been circulated at temperature (194ºF) for a period of time sufficient to complete the hydrolysis reaction, the mixture is cooled and a sample taken from the recirculation line. If the sample meets the criteria for VX and EA 2192 destruction, the resulting mixture (the hydrolysate) is pumped from the reactor to storage. If the VX and EA 2192 destruction criteria are not met, then the mixture is reheated, and processing continues. This is repeated until the batch is successfully processed.
After the batch is processed, it will be transferred to intermediate storage, and then shipped off-site for final treatment.
Figure 2-1. Hydrolysate reactor
Texas A&M performed a safety study of the Newport facility using fault-tree techniques. One scenario examined was “Offsite Transfer of Hydrolysate Containing Excess VX Concentration.” The Executive Summary of this report stated:
Fault tree analysis techniques were applied to the VX project speedy neutralization (PSN) process and related process support systems in order to estimate the frequency that the cited hazard scenario can be expected to occur. The study results indicate that the best estimate for an annual frequency of this undesired event is 5 x 10-5.
The annual frequency is estimated at 1
in 20,000 chance for CVXH being outside of specification for VX (>20
ppb). The existing design does not detect contamination of acceptable
hydrolysate after the batch sampling procedures have been completed.
The amount of potential contamination is minor and not thought to
present a public health risk. This issue could be corrected by good
engineering practices such as physical isolation using piping blinds,
spool pieces or a double block and bleed valve configuration or by
development of a sampling method at the storage tank. CDC has alerted
Army representatives regarding this design issue as part of the normal
oversight activity. Any potential design changes to the facility and
schedule impacts need to be balanced by the national security risk
associated with extended storage of the VX.
At the time of this report, the recommendations in Safety Study
Reports by Texas A&M and other safety studies relating to the design
and operation of the Newport facility were still being evaluated for
implementation.
2.3 Process Chemistry
The process chemistry involved with VX neutralization is complex when
an extremely high destruction of VX is required. At the time this
report was written, investigations into the process chemistry are
still under way, and not all of the details of the main and side
reactions involved (e.g. solids formation) were fully understood. The
major variables that affect the chemistry include the agent loading
(i.e., the relative amount of agent per unit volume of caustic in the
reactor at the start of the batch) and the type of stabilizer present
in the agent being processed. (The stabilizers used to minimize the
decomposition of the VX during storage were DIC or DCC or DIC + DCC).
The main reaction by which VX is neutralized by caustic is well
understood and is pseudo-first order with respect to VX concentration.
However, the presence of two phases (organic and aqueous), the
presence of VX in the organic phase, the creation of EA 2192, and the
presence of stabilizers complicate the physical and chemical process.
If all other system parameters and the composition of initial caustic
solution remain constant, then the size, composition, and partitioning
of the reaction products between the aqueous and organic layers depend
on the VX loading. Mass transfer limitations become more pronounced as
the droplet size increases and the organic layer is formed. This will
affect the rate, as well as the pathways of the reactions, and may
produce different final products. In addition, some of the ton
containers are now known to contain gelled/solid material. How much of
this material will be removed with the VX and how much will remain in
the ton container is uncertain. The effect of any gelled/solid
material on the chemistry or operation of the neutralization reactor
mixing process and sampling system also is unknown.
The purpose of the agitator and the static mixer are to mix the phases
and to transform the organic phase into tiny droplets. The smaller the
droplet size, the faster the diffusion processes in leaching and
neutralizing the VX in the organic droplets. Therefore, VX is rapidly
destroyed at the start of the batch operation; then a slower,
diffusion-limited process follows as the VX in the organic phase
droplets is neutralized. Moreover, the size and chemical compositions
of the dispersed droplets and the organic layer will differ for the
various VX loadings and stabilizer types.
In a response to questions from CDC, the Army and its contractor
(Parsons) summarized these issues:
Because of the highly reactive nature of hot caustic, less than 0.1% of the VX added to the reactor during the FILL period accumulates in the reactor with virtually all of this residual VX removed during the first minute of REACT. Additional REACT time is needed to destroy residual VX that partitions into the organic phase during FILL, and to ensure that EA2192 is non-detect. ………. It is expected that the NECDF’s full-scale pilot reactor will provide the necessary mixing and droplet size to produce non-detectable levels of both VX and EA2192. This conclusion is based on laboratory-scale results and full-scale pilot plant calculation results provided herein. The actual reaction time required to obtain non-detectable levels of both VX and EA2192 will be determined during Controlled Start-up testing of the full-scale NECDF pilot plant. If the reaction time required to obtain non-detectable levels of both VX and EA2192 determined during Controlled Start-up differs from that which was predicted during laboratory-scale testing, the types and configuration of the elements within the static mixer and the volumetric flow rate through the recirculation line can be changed, as needed.
This response accurately describes the
process, neglecting the effect of any gelled/solid materials in the
feed to the reactor or generated within the reactor.
As previously noted, the reaction originally was designed to have used
a 33% agent loading in each batch. However, studies demonstrated that,
at 33% agent loading, a significant organic phase (3%–5% by volume)
formed during the reaction, and this organic layer separated from the
aqueous phase during storage and floated on top. Remaining
(un-neutralized) VX partitioned into this organic phase, and the VX
concentration in this organic phase was approximately 20 times the VX
concentration in the bulk hydrolysate (nominally <20 ppb). Therefore,
operation with 33% agent loading could have resulted in a
“significant” volume of organic phase with a “high” VX concentration
in storage tanks and during transportation. This was considered
unacceptable, and modifications to the process were proposed and
implemented.
Additional investigation showed that operation at 16% agent loading
reduced the organic layer to approximately 2–3 volume percent. At 8%
agent loading, the organic layer was only a sheen on the surface of
the hydrolysate (approximately 0.5% by volume determined by
centrifuging the sample). The VX concentrations in the organic phase
for 8% and 16% agent loadings had not been determined at the time this
report was written.
Significant changes in organic liquid loading occur between 8% VX
loading and 16% VX loading (approximately a 1:5 volume ratio at a
minimum) and between the 16% and 33% VX loading (approximately 1:1.5
ratio). The physical and/or chemical processes involved and the
reason(s) for such a significant increase in organic loading between
8% and 16% VX loading have been the subject of some investigation, but
no conclusion has been reached.
Laboratory studies have demonstrated that the reaction times required
to complete neutralization vary with agent loading and stabilizer.
With DIC-stabilized agent (approximately 46% of the Newport
stockpile), the reaction times are 2.5–4 hours for 8% loading and 4–6
hours for 16% loading. With DCC-stabilized agent, the reaction time is
10–14 hours for 8% loading. The reason(s) for the apparent additional
processing time required by DCC-stabilized agent is (are) not fully
understood. The amount of stabilizer in each ton container also can
vary significantly. Therefore, what is valid for 8% VX loading
stabilized with DIC may not be valid for 16% VX loading case and other
stabilizers. The data do not warrant generalizations that apply to all
VX loadings and stabilizers.
In addition, laboratory studies have determined that solids are
generated during the neutralization process. These solids have been
variously described as a sticky gel and as a more coherent material.
The amount of solids, their composition, and the amount of VX, (if
any) these solids contain have not been determined.
The presence of solids in the hydrolysate within the reactor may be
problematic in the full-scale unit and impact plant operations.
Concern has been expressed that the solids may precipitate onto the
surfaces of the agitator in the reactor and result in an imbalance
that could cause mechanical failure of this item. A more likely source
of concern may be the potential blockage of the in-line static mixer
or deterioration of the performance of control valves, particularly
the three-port valve that controls the introduction of chemical agent
to the reactor and the transfer of the hydrolysate to the storage
tanks. The in-line mixer is constructed deliberately with small flow
paths (10–30 µm) to break up the organic phase into small droplets.
Any solids formation could result in blockage, with the potential for
reduced production rates and the need to remove the in-line mixer for
cleaning. Solids also can also be deposited on the surfaces of the
internal parts of the three-port valve, impacting valve closure and
enabling leakage of agent, thereby contaminating previously sampled
and acceptable hydrolysate batches as they are transferred from the
reactor to the storage tanks. Another possibility is that modification
of the process equipment to incorporate an upstream filter may be
required. Furthermore, the solids may negatively impact the sampling
system and the analytic measurements and treatment of the hydrolysate.
Appendix K of the documentation, provided in response to CDC Question
1, discusses solids formation. The “Conclusions” section of this
document states
Formation of solids in 8 weight % hydrolysate have (SIC) the potential to impact process throughput due to reactor hardware plugging in the pumps or static mixer. Preventative maintenance needs to be scheduled as experience determines.
Difficulties have been encountered clearing the hydrolysate with gelatinous material. When hydrolysate fails to clear, more processing is required. Detailed analysis of the gelatinous material may lead to procedures that could expedite clearance.
Further testing is underway to characterize the observed solids and identify whether stabilizer type (DCC vs. DIC) or VX loading causes changes in solid volume or content.
At [the Chemical Agent Munitions Disposal System (CAMDS)], twenty five batches of DCC hydrolysate and one batch containing DIC hydrolysate were processed without process failure due to these solids. (Note—Whether a static mixer with very small passages [such as at Newport] was installed at CAMDS) is not known)
In the subsystem hazard analysis of the process, the following finding (Failure Mode and Effects Analysis [FMEA] Item 01-04-134) was noted:
Over-or Under-Reaction Creates Gelatinous Matrix in Neutralization Reactor Containing VX
Several mis-operations and reaction inconsistencies can result in the creation of a gelatinous matrix in the neutralization reactor (1- and 2-L401). It might not be possible to completely prevent this occurrence. A study is being performed to identify ways to dissolve or solubilize
any gelatinous matrix that might form. Additional information or data from the study could determine methods to prevent the polymer formation and ways to mitigate such a formation if it occurs. This evaluation addresses FMEA Item 01-04-134.
Whether this finding in the safety studies documenting issues
associated with solid/gel formation in the reaction system has been
addressed at the time this report was completed is not known.
Except for solids formation and its possible effects, the scale-up of
the reactor from laboratory to full-scale operation should succeed.
Adequate heating and cooling have been provided for the reactor
system, the equipment is simple in design and the batch will be run
until the analytic methods demonstrate that VX and EA 2192 have been
adequately destroyed. However, the effect of gelled/solid material in
the ton containers passed into the reactor does not appear to have
been examined in detail. Therefore, no conclusion can be reached about
the effects of such material on the neutralization reaction, the
destruction efficiency, and the operation of the reaction system.
2.4 Findings
Scale-up of the process for 8% VX loading from laboratory-scale data should be operationally feasible. The database supports the efficacy of neutralizing 8% VX (stabilized with DIC) using sodium hydroxide. However, the Newport facility will be a pilot operation when it starts operation, and changes must be anticipated in operating mode and hydrolysate composition sent for off-site treatment.
VX loading and the specific stabilizer employed significantly impacts the process, hydrolysate composition, analytical methodology, and possibly solids formation. Scale-up of the process from 8% to 16% VX loading is of particular concern (because of the similarity of the organic-phase volumes between 16% and 33% VX loading batches) and the analytical problems identified with 33% VX loading.
The effects of solids formed during the hydrolysis reaction in the process on the hydrolysate and on the efficacy of treatment at a TSDF are unknown. The solids may contain VX. The impact of solids formation on the operation of the reaction system and, in particular, the potential for blockage of the in-line static mixer and other components (including the sampling system) is unknown. In addition, the presence of solids may impact the VX analytics, as well as the off-site hydrolysate treatment process.
At the time this report was written, all the findings from safety studies had not been fully addressed. In particular, findings relating to possible solids formation in the reactor and the required process modifications to provide additional assurance that no off-specification CVXH is shipped from the Newport facility may affect the CVXH composition shipped off-site.
3. Analytical Methods
3.1 Introduction
The purpose of this review and evaluation is to define the adequacy of
the proposed methods for the analysis of VX and EA 2192 in the CVXH to
meet the programmatic requirements of the NECDF. The scope of this
review is limited to laboratory analyses of hydrolysate from the
neutralization of DIC-stabilized VX at the 8% VX-loading level.
Adequate analytical data were not available to evaluate analyses of
hydrolysate related to other VX-loading levels or stabilizers.
3.2 Sampling Representativeness
We recognize that the validity of the clearance process depends on the
sample taken and delivered to the laboratory for analysis; the sample
must truly represent the total hydrolysate process batch. To evaluate
the sample procurement process, all available documents describing the
design and operation of the equipment and the sampling procedures were
reviewed. We also had detailed discussions with NECDF personnel.
NECDF personnel believe the sampling will be highly representative on
the basis of the mixing capability of the reactor, the design and
operation of the sampling equipment, and the detailed protocols that
have been established. On the basis of our understanding of reviewed
information, we agree—as long as solids formation does not block the
sampling points. The planned sampling program should provide
representative samples for CVXH batches to the laboratory for
analyses.
QA/QC procedures are in place to ensure and document adequate training
of personnel, performance of sampling equipment, availability and
quality of supplies, proper and complete recordkeeping, establishment
and maintenance of chain of custody, and the safety of plant and
laboratory personnel.
Maintaining representativeness of the analytical sample during
transfer of the 5-mL analytical portion from the plant batch sample
will be challenging because of the potential for separation of an
organic layer. The laboratory method for VX analysis in CVXH calls for
the analyst to “verify hydrolysate is as homogeneous as possible”
during the subsampling process. This process can be highly subject to
analyst technique error and will require careful QC.
3.3 Analysis of VX in Caustic VX Hydrolysate
3.3.1 Data Evaluation/Interpretation Criteria
Instrument or qualitative detection as defined in Laboratory Field
Instruction (LAFI)-A-30-053:
Consider VX present in the sample if the following criteria are met:
Retention time of analyte peak within +/- 0.1 minute of average standard VX retention time.
The m/z 128 ion, the m/z 139 ion, and the m/z 167 ion maximize within 0.05 minute of each other.
The m/z 139 and 167 ions may not be present at concentrations <1 microgram per milliliter (µg/mL) in the sample.
The m/z 128 ion response must be at least three times the background noise level, i.e., S/N ratio 3 or greater.
Quantitative criterion as defined by the Army is as follows:
MDL, calculated according to U.S. Environmental Protection Agency (EPA) procedure published in the Code of Federal Regulations (40 CFR, Part 136, Appendix B) <20 ppb.
3.3.2 Method Description and Documentation
LAFI-A-30-053 provides a comprehensive, step-by-step description of
the method for analyzing VX in CVXH. The method is based on multiple
hexane extractions of the hydrolysate, followed with solid-phase
extraction techniques for initial fractionation of the extract, then
final separation and detection of the VX using gas chromatography (GC)
coupled with ion-trap (IT) mass-spectrometry/mass-spectrometry (MS/MS)
techniques. The use of high-resolution capillary GC coupled with the
dual-phased MS/MS IT techniques gives this method extremely high
selectivity and sensitivity for VX in the hydrolysate. Stated in
layman’s terms, the method can detect and quantify VX in the highly
complex CVXH mixture at <20 ppb with a high level of confidence
against both false positives and false negatives.
The laboratory QC procedures defined in LAFI-A-30-053 and in Section
11.2 of the NECDF Laboratory Quality Control Plan, Revision 2, are
consistent with procedures and requirements published in EPA SW-846.
Implementation of these procedures should provide the QC data needed
to define the overall validity of the analytical results.
Evaluation of MDL data for 8%VX-loaded, DIC-stabilized hydrolysate
shows that, with this type of hydrolysate, the NECDF laboratory can
consistently generate MDL values below the 20-ppb criterion. In a
study to characterize batch-to-batch variation, the NECDF laboratory
generated three MDL values for each of two batches of hydrolysate. The
six MDL values ranged from 6 to 17 ppb, with a mean of 11 ppb, with no
appreciable differences between the two hydrolysates.
In summary, the current method for analyzing VX in CVXH is adequate to
detect and quantify VX well below the established clearance level of
20 ppb. The GC/IT/MS/MS technique provide a method with extremely high
analyte selectivity and sensitivity. The method consistently shows an
instrument detection limit below the 5–10 ppb range.
3.4 Analysis of EA 2192 in Caustic VX Hydrolysate
3.4.1 Data Evaluation/Interpretation Criteria
Instrument or qualitative detection as defined in LAFI-A-30-030:
Consider EA 2192 present in the sample if the following criteria are met:
Retention time of analyte peak is within +/- 1.0 minute of the average retention time of the standard EA 2192 during instrument calibration.
The m/z 162 ion is present with a 128/162 ion ratio of 0.3.
At EA 2192 concentrations <1 mg/mL the 128/162 ion ratio may not equal 0.3, but m/z 162 ion must be present.
The m/z 128 ion response must have a minimum S/N ratio of 3.
Quantitative criterion as defined by the Army:
MDL, calculated according to EPA procedure published in 40 CRF, Part 136, Appendix B, <1 ppm.
3.4.2 Method Description and Documentation
LAFI-A-30-030 provides a comprehensive, step-by-step description of
the method for analyzing EA 2192 in CVXH. The method consists of a
simple 1:25 dilution of the CVXH sample, followed by analyte
separation using liquid chromatography (LC) techniques, with final
detection and quantification using dual-phase IT/MS/MS. The use of
LC/IT/MS/MS techniques results in a highly sensitive, extremely
selective analysis of EA 2192 in the CVXH.
Laboratory QA/QC procedures defined in LAFI-A-30-030 and the NECDF
Laboratory Quality Control Plan are consistent with those published in
EPA SW-846. Analytical data characterizing the performance of this
method are limited. MDL data show values of 0.23 ppm and 0.09 ppm;
both well below the clearance level of 1 ppm. Precision and accuracy
data show overall very good precision of the method with analyte
recoveries ranging from 82% to 95%.
In summary, the current method for analyzing EA 2192 in CVXH is
adequate to detect and quantify EA 2192 in laboratory-generated
hydrolysate well below the established clearance level of 1 ppm. Data
also indicate that the qualitative (analytical presence) instrument
detection limit of the method is consistently <0.1 ppm.
3.5 Use of Analytical Data for Clearance
The Army has stated its intended use of VX and EA 2192 analytical data
in the clearance of CVXH for off-site shipment, as follows:
Since its inception, a key tenet of the Army Chemical Militarization program has been the safety of the workers and the public. Department of the Army (DA) Pamphlet (PAM) 385-61, entitled “Toxic Chemical Agent Safety Standards,” defines the approach for verifying the thoroughness of the neutralization process as using laboratory analysis to assure that the chemical agent is at a level less than or equal to 20 ppb. This level has been deemed protective of soldiers and Department of Defense personnel. The Project Manager for Alternative Technologies and Approaches (PMATA) elected to use the standard EPA method detection limit (MDL) as the means for determining whether the detection limit specified in the DA PAM has been met. Thus, the requirement for successful neutralization of VX is that the hydrolysate must be non-detect for VX with an MDL of 20 ppb or less.
The Army also has stated that EA 2192
must be “non-detect with an MDL of 1 ppm or less.”
As discussed in Sections 3.3 and 3.4, we believe that NECDF methods
LAFI-A-30-053 for VX in CVXH and LAFI-A-30-030 for EA 2192 in CVXH can
provide valid qualitative and quantitative data for detecting and
quantifying VX and EA 2192, respectively, in the concentration ranges
needed for programmatic clearance of the hydrolysate material for
off-site shipment. NECDF’s intended practice for measuring and
reporting “non-detects” is potentially misleading. Specifically, we
are concerned with the Army’s plan to classify and report analytical
results above the instrument detection level, but below the
established MDL, as “non-detects.” While CDC believes that utilizing
the MDL approach would not result in public health concerns, the Army
needs to address potential public misperceptions regarding the
detection or non-detection of VX in CVXH. A simpler reporting scheme
(i.e., non-detected, detected at <20 ppb, or detected at >20 ppb)
should be considered.
The Army’s clearance criteria of “non-detect with an MDL less than an
established concentration level” combines two related, but different,
analytical chemistry concepts. First, “instrument or analytical
detection” is a qualitative-based “yes or no” criterion. Second, MDL
is a statistically calculated, quantitative criterion.
The first criterion, “detection,” addresses two questions: (a) Was an
instrument response observed at the expected retention time of the
analyte? and (b) If so, was the level of that response greater than
three times the background noise (S/N ratio >3)? If the answers to
both of these questions are “yes,” then according to instructions in
LAFI-A-30-053 and LAFI-A-30-030, the analyte (either VX or EA 2192) is
considered “present” or “detected.” If the answer to either question
is “no,” then the result of the analysis is a “non-detect.”
The second criterion, MDL, addresses the level of confidence in the
quantitative value calculated from the observed instrument response
using an established calibration curve for the instrument. EPA’s
definition of an MDL, calculated according to the published procedures
in 40 CFR, Part 136, Appendix B, is the minimum concentration of a
substance that can be measured and reported with 99% confidence that
the analyte concentration is greater than zero. This is a highly
conservative criterion designed to all but completely eliminate
false-positive results. Failure to meet the quantitative-based MDL
criterion does not negate the analytical “presence” established by the
“detection” criterion.
Our issue is that the Army, through its current use of the EPA MDL
concept, could improperly classify analytical data as “non-detects”
when, in fact, the data have been determined analytically as
“detects.” Although EPA-prescribed uses of the MDL concept may be
appropriate for many applications in regulatory monitoring, in this
public health-driven application, it is open to criticism when
low-level instrument detects are discarded.
We are not suggesting that using the MDL concept and reporting
“analytical detects” as “non-detects” will compromise the process of
clearing the CVXH concentration at 20 ppb for VX and 1 ppm for EA
2192. Rather the issue is improper classification of analytical
results. Usually no issue would involve MDL, if the MDL was used only
to help determine a quantitation level at which a reliable number can
be provided to help make an action decision. In this case, the Army
used “detection,” not a quantitative level, as its primary clearance
criterion. We stated in sections 3.3 and 3.4 that the current NECDF
methods can support a clearance process on the basis of quantifiable
measurements. The Army could report analytical results as “less then,”
rather than as “detects” and “nondetects,” which would more accurately
represent the analytical data.
3.6 Quality Assurance and Quality Control Procedures
The Laboratory Quality Control Plan clearly defines the comprehensive
laboratory QA/QC procedures and techniques. This document defines the
procedures for: preparation and verification of analytical standards;
the certification, maintenance, and calibration of analytical
instruments; the certification of methods and personnel; and the QC
procedures, techniques, and samples used to define the operational
status of the analytical processes and the basic validity of the
analytical data. The overall QA/QC plan and procedures are well
designed and documented.
3.7 Findings
The planned sampling program should provide representative samples for CVXH batches.
The current method for analyzing VX in CVXH (LAFI-A-30-053) is adequate to detect and quantify VX in laboratory-generated, 8% VX-loaded, DIC-stabilized hydrolysate well below the established clearance level of 20 ppb.
The current method for analyzing EA 2192 in CVXH (LAFI-A-30-030) is adequate to detect and quantify EA 2192 in laboratory-generated, 8% VX-loaded, DIC-stabilized hydrolysate well below the established clearance level of 1 ppm.
The use of EPA’s MDL for clearance levels does not preclude analytical instrument detection of low levels of VX and EA 2192 (generally <20 ppb VX and <1 ppm EA 2192) in the CVXH. The perception that the clearance criteria (defined as “non-detected” with a MDL of ≤20 ppb VX or ≤1 ppm EA 2192) indicate absence of analytically detectable VX and/or EA 2192 could be misleading. While CDC believes that utilizing the MDL approach would not result in public health concerns, the Army needs to address potential public misperceptions regarding the detection or non-detection of VX in CVXH. A simpler reporting scheme (i.e., non-detected, detected at <20 ppb, or detected at >20 ppb) should be considered.
The overall QA/QC plan and procedures are well designed, and documented. NECDF laboratory personnel must generate day-to-day operational QC data to demonstrate that all analytical systems are operational and under control before plant startup according to written plans and procedures.
4. Caustic VX
Hydrolysate Treatment
4.1 Introduction
The CVXH is the liquor obtained from the alkaline hydrolysis of the
chemical agent VX at elevated temperatures. The details of the
processes that generate the CVXH at the Newport facility are described
earlier in this report. Once transported to the DuPont SET facility,
CVXH will be further treated to remove the organic by-products by a
series of physicochemical and biologic processes. The exact
composition and phase characteristics of the CVXH received at the SET
plant will depend on the stabilizer type and VX loading used in the
NECDF process batch. The major parameters and characteristics of 8% VX–
loaded, DIC-stabilized hydrolysate (which is the main focus of this
report), as received by DuPont, are given in the Table 4.1 for two
separate CVXH samples.
pH |
TOC, mg/L |
COD mg/L |
TN mg/L |
EMPA mg/L |
MPA mg/L |
Thiolamine mg/L |
>13 |
33,852 |
61,000 |
6,739 |
39,135 |
2,789 |
11,200 |
13.1 |
44,147 |
4,334 |
35,937 |
2,826 |
42,900 |
total organic carbon (TOC), total nitrogen (TN), milligrams per liter (mg/L)
Table 4.1 Characteristics of caustic VX hydrolysate generated from 8% VX loading with DIC stabilizer
The DuPont treatability studies were designed and executed to obtain
scale-up parameters for engineering design and regulatory compliance.
Because of their relatively high concentrations in the CVXH, only
thiolamine, EMPA, and MPA were analyzed or monitored within the
treatment train or in the process effluent. Trace contaminates, such
as VX and EA 2192, were not monitored during the studies. (Note:
Because of the high 1500- to 2000-fold dilution factor in the DuPont
SET process, monitoring of these compounds may not be analytically
possible.)
The pH adjustment and neutralization of the CVXH is the first step of
the pretreatment process before introduction of the waste to the
biologic treatment system. CVXH neutralization is followed by peroxide
treatment to destroy odorous substances. The most recent
biotreatability studies, the final step in the treatment train, use
two-stage PACT®-activated sludge systems that are operated under
conditions emulating the actual plant flow rate and hydraulic
retention time. In addition to CVXH, the reactors received mustard
(HD) hydrolysate from the Aberdeen operations because an alternating
treatment scheme may be implemented at the DuPont SET facility.
The studies described in the two DuPont treatability reports (March 3,
2004, and July 19, 2004) were performed with different types of
hydrolysates. The inconsistencies in the samples used to conduct the
treatability studies make evaluation of the entire treatment process
on the same basis and extrapolation of the treatability studies to
pilot-plant performance challenging. For example, the pH adjustment
and neutralization experiments reported in the Basic Data Summary
Report (July 19, 2004) were conducted using 16% VX–loaded, DIC-stabilized
CVXH (actual), but the biotreatability studies were performed with 8%
VX-loaded, DIC-stabilized CVXH (actual). Although 20% sulfuric acid
was used in the pH-treatment experiments, DuPont proposes to use 5%
acid in the full-scale process. The heat of reaction for acidification
was measured for 8% and 16% VX-loaded CVXH (reformulated)1 with DCC
stabilizer, not DIC, which is the focus of our investigation. In
summary, the studies reported in the Technical Assessment and the
Basic Data Summary Report suffer from inconsistencies with respect to
the type of CVXH used in each test. The experimental findings do not
support the assumption that the CVXH has identical physical and
chemical properties regardless of the VX loading and stabilizer type.
The volume of the organic layer formed, which differs for 16% VX-loaded
CVXH and 8% VX-loaded CVXH, clearly indicates that the system
chemistry differs depending on how much VX is added to the caustic
solution. Moreover, the volume of the organic layer formed during
hydrolysis is not directly proportional to the VX loading. Therefore,
linear extrapolations of the experimental results obtained in the
preliminary treatment studies should not be used to predict
performance at higher agent loadings, and equating the 8% VX-loaded
7000-gallons per day (gpd) CVXH with 16% VX-loaded 3500-gpd CVXH
(Table 5, Basic Data Summary Report) for design and modeling purposes
should be avoided.
Because the Army’s stated objective is to begin operations with 8% VX-loaded,
DIC-stabilized CVXH, the assessment of the DuPont treatability studies
focused mainly on treatment of the CVXH at this condition.
Occasionally, however, other data and material reported by Parsons on
the VX alkaline hydrolysis treatment are cited to support the main
findings of this assessment. Data are insufficient to assess treatment
of CVXH at other VX loadings and for other stabilizers. In the
following sections, the hydrolysate acidification process, the
peroxide oxidation, and the biologic treatment studies are evaluated
and the major findings presented.
4.2 Extent of Treatment
4.2.1 pH Adjustment
The CVXH acidification experiments were conducted with actual CVXH
(16% VX-loaded, DIC-stabilized) titrated with 20% sulfuric acid to a
final pH of 4–6. The titration curve obtained from the actual CVXH was
compared with the aqueous layer from a centrifuged sample after
separation of the organic layer. The heat of reaction also was
computed, but for 8% and 16% VX-loaded, DCC-stabilized (reformulated)
CVXH. The results of these experiments demonstrated that
The organic layer is destroyed. pH adjustment produces a homogeneous amber yellow clear solution.
The process generates 3.07 calories per gram (cal/g) during the titration of 8% VX-loaded, DCC-stabilized (reformulated) CVXH, producing a temperature increase of 6.4 ºC. This energy is expected to dissipate through heat losses during plant operation, and cooling and heat exchanger installation will be unnecessary.
Removal of the organic layer lowers the buffering capacity of the mixture (hydronium ions appear to be consumed during destruction of the organic layers).
The process increases the volume of the CVXH waste by about 30%. If 5% sulfuric acid is used, as DuPont proposes to avoid cooling the reaction mixture, the volume increase will be close to 100%, further diluting the sample by a factor of 2. The effect of the 5% sulfuric acid on the organic treatment is unknown; the available reports did not present data using 5% sulfuric acid.
In response to the May 25, 2004,
clarification questions (Responses to CDC Clarification Questions,
Final, 17 June, 2004), Parsons indicates that pH adjustment does not
destroy the organic layer. DuPont’s 3 March 2004 report, “Treatability
of Newport (Indiana) Caustic Hydrolysate” (Reich et al), confirmed
that the adjustment of pH without additional treatment measures
aggravates the odor of hydrolysate. Furthermore, the uncharged form of
thiolamine is poorly-soluble and results in the formation of a large
organic layer, on the order of 10% by volume. This organic layer is
presumed to have a low flashpoint, which would add risk to the
shipping process.
However, DuPont and its treatability study as presented in the Basic
Data Summary Report, states
The sample was observed to change from a yellowish cloudy color
to a slightly amber clear color once a single phase was formed
which occurred around pH 6.0. Once a single phase formed, there
was no longer any organic material coating the glass.
Addition of a strong acid to the CVXH
profoundly affects the physical and chemical stability of the organic
droplets dispersed in the hydrolysis liquor and the dissipation of the
organic layer. Attachment 1, “Characterization of Droplets Resulting
from NECDF Static Mixers,” of the Parsons report (July 22, 2004)
states that the average size of the colloidal droplets ranges from 5
to 10 µm, with specific gravity of about 0.87 and strong negative
charges. This charge most likely keeps the droplets suspended,
preventing efficient collisions and subsequent aggregation. The
electrophoresis experiments to determine the particle surface charge
were performed with 16% VX- loaded, DIC-stabilized CVXH (actual). No
experimental data are presented in the Parsons white paper on the
properties of the droplets formed in the hydrolysate from the 8% VX-loaded,
DIC-stabilized CVXH. The Parsons reports documented, and experimental
observations by DuPont verified, that the volume of the organic layer
and the size distribution and dispersion of the droplets in the final
CVXH depends on the VX loading. The higher the loading rate the larger
the resulting organic layer volume. However a direct proportional
relation does not appear to exist (i.e., doubling the VX loading does
not increase the volume of the organic layer by a factor of two).
Visual observations by Parsons personnel of the formation of the
organic layer estimated that the layer thickness remains unchanged for
up to 4 months. However, no kinetic information is provided about the
rate of formation of the organic layer.
Given that the organic droplets carry an overall negative charge,
addition of hydronium ions should compress the electrical double layer
that typically exists in the boundary of the organic-aqueous interface
and allow the attraction forces to take over. Because this is not
observed, i.e., addition of sulfuric acid does not appear to enhance
flocculation or layer formation and separation, we can conclude that
either the solubility of the organic phase is higher or its components
become chemically unstable and decompose at lower pH or both. The
disappearance of the organic phase during pH adjustment supports this.
The exact composition of the organic layer is not known, but the
response of the whole (as received) CVXH to the addition of sulfuric
acid suggests that it imparts alkalinity to the sample, probably
because of weak organophosphorous acids and carbonates in the process
water. More sulfuric acid (about 30 grams [g]) is required to reduce
the pH of the whole CVXH sample than the aqueous layer to a pH of 8
(Figure 1 of the Basic Data Summary Report). However, the two
titration curves intersect at a pH of 7 indicating that the same
amount of acid is needed to bring the solutions to this endpoint. From
that point, further addition of small amounts of acid brings about a
steep pH drop in the aqueous layer but has little effect on the whole
CVXH (as received), until about 380 g acid (x-axis of Figure 4-1),
where pH drops substantially. This behavior is consistent with a
chemically reactive solution. The organics exert a hydronium ion
demand in excess of the amount required to neutralize the base. The
organic layer appears to react with the hydronium ions participating
in a chemical reaction rather than to be simple acid-base equilibrium
chemistry. Moreover, the observation that this step modifies the
odorous intensity of the mixture provides additional evidence that the
organic components undergo significant chemical changes during pH
adjustment.
Figure 4-1 Caustic CVXH titration curves provided by DuPont in the Basic Data Summary Report.
4.2.2 Hydrogen Peroxide Oxidation
Once the pH of the hydrolysate is adjusted to a pH of 4–6, the mixture
is treated with 10% peroxide to control objectionable odors emanating
from the CVXH caused mainly by the volatilization of thiolamine.
Peroxide and the free radicals formed by its addition to the reaction
mixture attack the organics present in the hydrolysis liquor and
initiate thiolamine destruction. Again, these studies were conducted
with 16% VX- loaded, DCC-stabilized CVXH (actual or reformulated).
Thiolamine is destroyed quickly by the peroxide, with most of the
compound depleted within the first minute of reaction (Figure 4-2).
After 20 minutes, the concentration drops below the detection limit of
5 ppm. The degradation products of thiolamine are presented in the
Technical Assessment Report (March 3, 2004). Four compounds were
identified as possible thiolamine degradation products: acetic acid,
diisopropyl amine, urea, and 2-diisopropylaminoethyl ethyl disulfide.
Acetic acid and urea are readily biodegradable compounds and are
expected to break down in the two-stage PACT® bioreactors. However,
the biodegradability of isopropyl amine and the
2-diisopropylaminoethyl ethyl disulfide is not documented in the
Technical Assessment Report or the Basic Data Summary Report; only
qualitative references (page 49 of the Technical Assessment Report)
state that samples analyzed from the effluent of one of the
bioreactors had no detectable amounts of thiolamine or any of its
oxidation products. No other information is provided that confirms the
biodegradation of these two by-products. EMPA and MPA remain
unaffected by the peroxide process.
Figure 4-2 Destruction of thiolamine by hydrogen peroxide oxidation.
The oxidation step is an exothermic process releasing approximately 14
cal/g of heat. This value was obtained from a reformulated 16% VX–loaded,
DCC-stabilized, CVXH that was first treated with 20% sulfuric acid to
a pH of 6.4, then subjected to 20% weight equivalent of 10% hydrogen
peroxide solution. Gas-generation measurements conducted in 2-liter
flasks showed that the amount of gas generated during the peroxide
oxidation is negligible. The lack of gas evolution suggests that the
degradation of thiolamine is incomplete; in other words, the compound
is not mineralized to the simple innocuous carbon dioxide and water.
4.2.3 PACT® Biotreatment
Two sets of biodegradation experiments were conducted using one- and
two-stage PACT® bioreactors. The first treatability study was
performed with CVXH; in the second, both CVXH and HD hydrolysate from
Aberdeen were tested to determine the effect of alternating the
bioreactor feeds on the performance of the biologic system.
Co-processing will be necessary when both types of hydrolysates will
be sent for treatment to DuPont’s SET facility. The objectives and the
criteria of both studies were stated in the Basic Data Summary Report:
To confirm that the anticipated rates of CVXH can be processed successfully through the SET [wastewater treatment plant (WWTP)], enhancing the database provided by the original treatability study;
To assure that the CVXH can be processed at appropriate rates while HD hydrolysate from Aberdeen is being managed at the WWTP using a plan to either alternately campaign each hydrolysate or process the pretreated hydrolysates simultaneously;
To ascertain the degree of improvement in treatment that can be anticipated with a two stage PACT® system.
As for the earlier Treatability Study there were three general criteria for judging the treatment of CVXH to be successful:
Ability to maintain satisfactory control of wastewater and sludge odors.
Ability to maintain control of SET WWTP operations (e.g., effective dissolved organic carbon [DOC] removal, manageable foaming, pH control, solids management, etc.)
Ability to assure permit compliance
(e.g., effluent BOD5 [5-day biochemical
oxygen demand], BOD5 percent removal, effluent TSS, effluent NH3-N and
WET). In addition the fate of EMPA, MPA and thiolamine were monitored.
As mentioned before, the studies were
designed to provide information about system performance in terms of
regulatory compliance and to obtain design parameters for scaleup.
To ensure adequate treatment, two PACT® bioreactors were operated in
series. This biologic system, in addition to the microbial
degradation, was dosed with activated carbon, which in general
enhances the treatment capacity by removing recalcitrant compounds
that are resistant to biodegradation. Six reactors were set up to
evaluate various treatment scenarios using 8% VX-loaded, DIC-stabilized
CVXH and the HD hydrolysate. The flow rate and retention time in the
bioreactors were set to simulate actual plant conditions treating
7000-gpd CVXH and 15,000- and 25,000-gpd HD hydrolysate. A large
dilution of the hydrolysate, to the order of approximately 2000 times,
occurred at introduction of the pretreated CVXH to the biologic PACT®
system. Appropriate controls were used throughout the study, and all
pertinent system parameters were monitored to assess system
performance. However, the fate of individual compounds as they pass
through the bioreactors is not as well documented. Only EMPA and MPA
were monitored in the pilot-plant effluent.
The data presented in figures 7, 8, 9, and 10 and tables 8 and 9 of
the Basic Data Summary Report indicate that, after a short acclimation
period, the removal efficiency, as measured by DOC and BOD reduction,
stabilizes to an average of about 85%–90% in all reactors. Even during
the acclimation period, the removal does not drop below 75%. This
high-removal efficiency also is observed in the alternating Aberdeen/DIC
CVXH influents, indicating that the biologic system is not affected by
these input changes. The 7000–gpd, 8% VX-loaded CVXH is equated to
3500–gpd, 16% VX-loaded CVXH (Table 5). However no evidence suggests
that this is a valid approach. See Section 4.3 for a discussion of the
potential differences on the composition and general chemistry of the
8% and 16% VX-loaded CVXH.
The Technical Assessment and Basic Data Summary reports clearly
document the conversion of EMPA to MPA. Both compounds remain
unaffected by the pH reduction, and conversion during peroxide
treatment appears to be limited. Biologic treatment by the two-stage
PACT® process converts essentially all of the EMPA to MPA but appears
not to affect the MPA decomposition. Data are sufficient to support
this conclusion. The slight decrease in MPA effluent concentration
most likely results from partitioning in the organic sludge.
DuPont’s Technical Assessment and Basic Data Summary reports contain
no information about the fate of VX or EA 2192 during treatment of the
CVXH in the DuPont SET facility. The presence of these two compounds
in the plant effluent in trace amounts cannot be excluded.
4.3 Environmental Persistence and Agent Loading Effects
The major hydrolysis products of VX are well characterized, and the
reaction rate and pathways depend strongly on solution pH and
temperature (Figures 4-3 and 4-4). With solubility of approximately 30
grams per liter (g/L), VX is considered to be highly mobile in the
environment and can persist for days or even weeks in slightly acidic
waters. Other VX hydrolysis products in the CVXH include EMPA, which
has a half life in soils of about 8 days, with MPA being the major
transformation product.
Figure 4-3 pH dependence of apparent rate constant for VX hydrolysis
Figure 4-4 Temperature dependence of apparent rate constant for VX hydrolysis at a pH of 7.7.
As discussed in Section 4.2.3, the
treatability studies with the 8% VX-loaded CVXH demonstrates
conversion of EMPA to MPA in the activated sludge bioreactors. MPA is
stable in the environment because it is resistant to hydrolysis,
photolysis, and thermal decomposition. It is also soluble in water and
has a low coefficient for sorption onto soil particles. Therefore, it
can migrate easily in the soil and groundwater (Munro et al., 1999).
Another major by-product of the hydrolysis of VX at neutral and high
pH values, is EA 2192 (S-(2-diisopropylaminoethyl)methyl
phosphonothioic acid), an environmentally persistent highly toxic
compound with infinite water solubility.
Some of the hydrolysis products, namely EA 2192, EMPA and MPA, are
stable at neutral pH; whether these, or other byproducts that are not
identified or exist at low concentrations, can react and form stable
VX molecules is questionable. This is a concern because the CVXH is
adjusted to a pH below 6 in preparation for the oxidation and biologic
treatment. Parsons attempted to partially address this concern by
studying the CVXH over a 5-hour period at a pH of 10 or 71 days at a
pH of 14. These conditions, however, do not represent the low (<6) pH
range in the system after pH adjustment. Neutral pH is a worst-case
scenario because of the stability of the by-products at those
conditions and the possibility of recombining to reform VX.
Thermodynamic analyses also should have been performed to assess the
tendency of the pH-adjusted CVXH to move toward VX reformation.
Because experimental data are not presented, the questions regarding
possible VX reformation remain unanswered.
4.4 Findings
The 8% VX-loaded, DIC-stabilized CVXH is treated by pH adjustment to a pH <6 to eliminate the two-phase mixture, followed by hydrogen peroxide oxidation to destroy the odor-causing thiolamine, and finally biologic treatment to convert most of the EMPA to MPA.
The DuPont SET facility effectively treats the CVXH generated from an 8% VX loading with DIC stabilizer, except for MPA, for which only minimal reduction is demonstrated.
Alternating feeds from Aberdeen HD hydrolysate and CVXH did not affect the performance of the DuPont bench-scale reactor.
The effects of the SET facility on the destruction of any trace quantities of VX and EA 2192 in the CVXH are unknown. In addition, the fate of diisopropyl amine and 2-diisopropylaminoethyl ethyl disulfide through the SET plant is not well documented.
The possibility of VX reformulation at
acidic (<6) pH conditions (after pH
adjustment) in the Dupont SET treatment process has not been
adequately investigated and remains unresolved.
Effective treatment of 16% VX-loaded CVXH and 8% VX-loaded CVXH with DCC or DIC/DCC stabilizers were not demonstrated in the DuPont studies.
5. Major Findings
NECDF was designed to destroy VX using caustic hydrolysis in a hot
solution of sodium hydroxide. Initially the plan was to further treat
the resulting waste on-site by SCWO and to ship the SCWO effluent to a
TSDF. After the terrorist attacks of September 11, 2001, the plan was
modified to eliminate on-site SCWO treatment and ship the resulting
hydrolysate directly off-site for treatment at a TSDF. Critical to
this modified plan was the development and validation of analytical
methods to clear the hydrolysate for shipment. The stringent Army
clearance levels for VX and EA 2192 proved challenging to the
analysts. The original plan to operate at 33% VX loading was
abandoned, and the program plans to begin operations at 8% VX loading
and move to 16% VX loading.
This programmatic change has necessitated an intensive effort to
develop the analytical methods needed to assess process performance
and suitability of the hydrolysate for off-site shipping, process
modification to ensure adequate mixing and VX droplet size, and search
for a TSDF capable of treating the hydrolysate. The current plans are
for NECDF to ship the CVXH to the DuPont SET facility in Deepwater,
New Jersey.
CDC engaged Carmagen Engineering, Inc., to assemble a team of experts
(Team) to assist in the evaluation of the DuPont SET facility’s
treatment of the CVXH. The Team recognized that an assessment of the
NECDF destruction process and an examination of the analytical methods
to be used for CVXH clearance were required to ensure that the
hydrolysate being shipped to SET will be adequately characterized and
that VX and EA 2192 levels in the CVXH meets Army specifications.
The Team addresses its findings in chapters 2–4 of the report. The
reader is encouraged to review all of the findings, as well as the
supporting documentation in each chapter. The major findings follow.
Process Issues (Chapter 2)
Finding 2.1. The database
supports the efficacy of neutralizing DIC-stabilized VX using sodium
hydroxide at the 8% VX-loading rate. Scale-up of the process from
laboratory/bench scale to pilot scale should be operationally
feasible. However, because the NECDF will be a pilot facility, changes
must be anticipated in operating mode and hydrolysate composition sent
for off-site treatment.
Finding 2.2. VX loading (weight percent) and the specific
stabilizer (DIC, DCC) employed significantly impact the process,
hydrolysate composition, analytical methods validation, and possibly
solids formation. Scale-up of the process from 8% to 16% VX loading is
of particular concern (because of the similarity of the organic-phase
volumes from 16% to 33% VX-loading batches), the potentially high VX
concentration in the resulting organic layer, and the analytical
problems identified with 33% VX loading.
Finding 2.3. The impact is unknown of solids formation during
the hydrolysis process on operations (potential for blockage of the
in-line static mixer, control valves, and sampling system), VX
analytic methods, and off-site hydrolysate treatment. The transition
from 8% to 16% VX loading, as well as stabilizer change, is of concern
and requires additional detailed studies.
Analytical Methods (Chapter 3)
Finding 3.1. The methods for
analyzing VX and EA 2192 in 8% VX-loaded, DIC-stabilized CVXH are
adequate to detect and quantify at the established clearance levels
for VX (20 ppb) and EA 2192 (1 ppm).
Finding 3.2. The use of EPA’s MDL for clearance levels does not
preclude analytical instrument detection of low-level VX and EA 2192
(generally <20 ppb VX and <1 ppm EA 2192) in the CVXH. The perception
that the MDL clearance criteria indicate absence of analytically
detectable VX and EA 2192 could be misleading. While CDC believes that
utilizing the MDL approach would not result in public health concerns,
the Army needs to address potential public misperceptions regarding
the detection or non-detection of VX in CVXH. A simpler reporting
scheme (i.e., non-detected, detected at <20 ppb, or detected at >20
ppb) should be considered.
Finding 3.3. The overall QA/QC plan and procedures for the
NECDF laboratory are well designed and documented. However, NECDF
laboratory personnel should continue implementing the QA/QC plan by
developing day-to-day operational QC data to demonstrate that all
analytical systems are operational and under control before plant
startup.
Caustic VX Hydrolysate Treatment (Chapter 4)
Finding 4.1. The SET facility
effectively treats the CVXH generated from an 8% VX loading with DIC
stabilizer (i.e., pH adjustment, thiolamine destruction, conversion of
EMPA to MPA), except for MPA, for which only minimal reduction is
demonstrated.
Finding 4.2. The SET facility treatment performance should be
unaffected when treatment of hydrolysate feeds from Aberdeen (HD) and
Newport (VX) are alternated.
Finding 4.3. The DuPont treatability studies have not yet
demonstrated the effective treatment of 16% VX-loaded CVXH, nor of 8%
VX-loaded CVXH with DCC or DIC + DCC stabilizers.
6. References
DuPont. DuPont Technical Assessment on U.S. Army Newport (Indiana)
Project. 3 March, 2004.
DuPont. Basic Data Summary Report for Newport Caustic Hydrolysate. 19
July, 2004.
Munro, N.B., Talmage, S.S., Griffin, G.D., Waters, L.C., Watson, A.P.,
King, J.F., and Hauschild, V. The sources, fate, and toxicity of
chemical warfare agent degradation products. Environ Health Perspect.
1999;107(12):933–974.
Parsons. CVXH Analytical Testing Results (Response to CDC Request for
Information: Item No. 1). 15 May 2004.
Parsons. Sampling Protocols (Response to CDC Request for Information:
Item No. 2). 14 May 2004.
Parsons. Analytical Methods for VX in Hydrolysate (Response to CDC
Request for Information: Item No. 4). 14 May 2004.
Parsons. Newport, Indiana Process Variability Impact on Secondary
Treatment (Response to CDC Request for Information: Item No. 6). 14
May 2004.
Parsons. Responses to EPA Questions to DuPont [16 questions]. 14 June
2004.
Parsons. Responses to Additional CDC and EPA Questions [9 EPA
questions, 19 CDC questions]. 17 June 2004.
Parsons. Responses to CDC 29 June, 2004 Questions [11 requests and 5
questions]. 22 July 2004.
SAIC. Assessment of Low-Level VX in CVXH LLVX Independent Assessment
Panel. October 2003.
Texas A&M University System. Quantitative Subsystem Hazard Analysis of
Potential for Off-Site Transfer of Hydrolysate Containing VX Above the
20 ppb Method Detection Limit. Draft Report. August 2004.
Wojciechowski P, Mokos J. Analysis for VX in Hydrolysate at NECDF.
Presentation to CDC, 27 March 2003.
Previous | Table of Contents | Next