1

JOINT DOE/EPAJET-REMPI

DIOXIN CEM TEST

FINAL REPORT

FOR JOINT DOE/EM, EPA, DLRSTUTTGART , AND SKY +

TEST PROGRAM

TO EVALUATE

JET-REMPI ANALYZER AS A

CONTINUOUS EMISSIONS MONITOR (CEM)FOR DIOXINS AND OTHER ORGANICS.

20. APRIL , 1997wr i t t en by :

N ina Be rgan F rench , Sky +

Br i an K . Gu l l e t t , U .S . EPA

Hara ld Ose r & H . -H . Gro thee r DLR S tu t t g a r t

Dav id Na t s chke , Acu r ex Env i ronmen t a l Co rp .

S K Y +

1

J O I N T D O E / E PA

J E T - R E M P I

D I O X I N C E M T E S T

F I N A L R E P O R TRESULTS OF JOINT PROGRAM TO TEST JET-REMPI

AS A CEM FOR DIOXINS

EXECUTIVE SUMMARY

A dioxin CEM offers two solutions to issues surrounding dioxin emissions: first, a method

to directly monitor compliance with emissions regulations, and second, a method to advance

prevention and control of dioxins before they are released.

A Jet-REMPI prototype developed by DLR Stuttgart, was tested from July to October,

1996 at the EPA National Risk Management Laboratory in Research Triangle Park, North

Carolina. The primary test objectives were to establish REMPI parameters (e.g. laser

frequency) and measure sensitivity of Jet-REMPI to actual dioxin congeners. These

experiments are not possible at the DLR laboratories due to safety regulations. The

experiments utilized a simple dioxin feeder which fed steady, known concentrations of vapor-

phase dioxin species into the Jet-REMPI analyzer.

Funding for this project came from the DOE/EM Mixed Waste Focus Area. In addition,

EPA provided infrastructure to conduct the test, including facilities, space, and programmatic

and technical guidance. DLR Stuttgart provided the Jet-REMPI hardware and labor costs for

DLR personnel.

During this test we measured zero-, mono-, di-, and tri- and tetra-chlorinated

dibenzodioxins. We believe these are the first data showing REMPI measurements of tri- and

tetra-chlorinated dibenzodioxins. The data are significant in both demonstration of detection

2

levels for these dioxins demonstrating that single-color (one wavelength) Jet-REMPI can

measure PCDD congeners up to tetra- levels of chlorination. The data indicate sensitivity of

30 ng/m3 (60 pptv) for 2,7 DCDD (di-chlorodibenzodioxin). We estimate that the detection

limit for a tetra-chlorinated dibenzodioxin will be approximately a factor of 6 higher for each

chlorine added, 62, or 36 times higher: approximately 1000 ng/m3.

A minimum detection limit of 30 ng/m3 is many orders of magnitude more sensitive than

other CEM techniques, but still four to seven orders of magnitude less sensitive than what is

ideally needed to directly measure the current proposed emissions standard of 0.2 ng/dscm

Toxic Equivalency (TEQ). If a CEM were to directly measure TEQ, we would ideally need

detection limits between 0.001 ng/m3 (for the most toxic congener (2,3,7,8 TCDD)) to 1.0

ng/m3 (for the least toxic octo- DD), depending on the Toxic Equivalency Factor (TEF) of

each of the 17 TEQ congeners.

No doubt, our ability to develop a dioxin CEM thus rests on our ability to improve

sensitivity. From this test, we identified several possible improvements, such as insuring

sample cooling in the expansion nozzle. The current nozzle may not have cooled adequately

because of the increased heating in the preceding sample line necessary to avoid condensation

of the heavier chlorinated compounds. The problem can be avoided by redesigning the inlet

valve assembly. Other hardware and software modifications can probably lead to a total of

two orders of magnitude improvement in Jet-REMPI detection limits.

The next step would be to design a sample pre-concentrator and a particle desorber /

separator. A sample preconcentrator can concentrate the sample as much as is needed to

make the measurement. A particle desorber, placed upstream of the sample concentrator, will

desorb dioxins from particles. The separator will separate out the particles, sending only the

vapor-phase dioxins to the analyzer.

With these enhancements, it may be possible to measure dioxins from a combustion-type

environment, in near real time, at sub-ng/m3 concentrations. To achieve this, we believe,

would require a several million-dollar development program. Indeed, simply developing the

REMPI capability for all 17 congeners in the TEQ (necessary for an “ideal” direct TEQ

3

monitor) would take thousands of hours in a research laboratory. Eventually, this can and

probably will be achieved, but in the meantime, there are more practical applications for a

dioxin CEM that can be accomplished in the near-tern for far less money.

Even with little or no sample pre-concentration, the next-generation REMPI system

capable of measuring vapor-phase 2,3,7,8 TCDD in low ng/m3 levels would be a remarkable

achievement. Such an instrument would be very useful in research laboratories studying dioxin

formation and control. These laboratory experiments for studying dioxin formation and

control usually operate at total dioxin concentrations around 100 to 1000 ng/m3 total.

Depending on how many of the 210 total congeners are present, detection limits around 5 to

10 ng/m3 for several congener of interest would be more than sufficient.

By combining data from carefully controlled experiments designed to isolate dioxin

formation mechanisms, we can begin to identify surrogates that indicate the presence or

imminent formation of dioxins. Hopefully these surrogates will be much easier to measure

than the full suite of 17 congeners in the TEQ, thereby allowing a much more usable

compliance instrument. In addition, this mechanistic information will lead to control strategies

to prevent dioxin formation in the first place.

4

1

ACKNOWLEDGEMENTS

The authors thank Dan Burns of Westinghouse Savannah River for providing contractual

support for this project. We also thank Dr. Steve Priebe from the DOE/EM Mixed Waste

Focus Area for technical and programmatic support. Finally, we thank the staff and

management at EPA National Risk Management Laboratory, and their support contractor,

Acurex Environmental, for superb test facilities and assistance in executing this test.

2

EXECUTIVE SUMMARY ......................................................................................................................... 1

ACKNOWLEDGEMENTS ........................................................................................................................ 1

CHAPTER 1. INTRODUCTION .............................................................................................................. 1

THE NEED FOR A DIOXIN CONTINUOUS EMISSIONS MONITOR....................................................................... 1

BACKGROUND: CHEMISTRY OF DIOXINS/FURANS ........................................................................................ 3

CHAPTER 2. JET-REMPI TECHNOLOGY........................................................................................... 5

DESCRIPTION OF REMPI TECHNIQUE ........................................................................................................... 5

DLR STUTTGART’S JET-REMPI PROTOTYPE.............................................................................................. 6

CHAPTER 3. DIOXIN CEM PERFORMANCE REQUIREMENTS.................................................... 11

CHAPTER 4. PROJECT TEAM AND TEST FACILITIES.................................................................. 20

PROJECT TEAM ........................................................................................................................................ 20

TEST HARDWARE DESCRIPTION................................................................................................................ 21

REMPI Analyzer System .................................................................................................................... 21

EPA dioxin feeder.............................................................................................................................. 22

CHAPTER 5. TEST OBJECTIVES......................................................................................................... 25

CHAPTER 6. TEST DESCRIPTIONS.................................................................................................... 27

PHASE I TEST SCHEDULE .......................................................................................................................... 29

CHAPTER 7. TEST RESULTS AND ANALYSIS ................................................................................. 30

INSTRUMENT SHAKE-DOWN ...................................................................................................................... 30

DIOXIN TESTING...................................................................................................................................... 30

CHAPTER 8. TECHNICAL ISSUES...................................................................................................... 41

PARTICLE MEASUREMENTS ...................................................................................................................... 41

SENSITIVITY ............................................................................................................................................ 41

SPECIES - WHAT TO DETECT? ................................................................................................................... 41

SYSTEM COST/COMPLEXITY...................................................................................................................... 43

3

CHAPTER 9. WHAT NEXT? ................................................................................................................. 44

CHAPTER 10. CONCLUSIONS ............................................................................................................. 45

REFERENCES.......................................................................................................................................... 47

1

CHAPTER 1 . INTRODUCTION

THE NEED FOR A DIOXIN CONTINUOUS EMISSIONS MONITOR

Current techniques to monitor emissions of polychlorinated dibenzodioxin (PCDD) and

polychlorinated dibenzofuran (PCDF) use sampling times in excess of hours, during which the

analytes are collected on adsorbing materials followed by sample extraction and preparation

for subsequent gas chromatography / mass spectrometry (GC/MS) analysis.1 These costly

and time demanding methods have drawbacks in that compliance measurements are made

only infrequently (perhaps once or twice per year). The consequences are over-designed air

pollution control systems and regulatory strategies that rely on indirect process monitoring

rather than direct monitoring and dioxin prevention strategies.

In light of these limitations, a continuous emission monitor (CEM) for PCDD and PCDF

offers four benefits to users:

1) Direct, rapid detection of PCDD and PCDF congeners, their indicators

(compounds measured in lieu of PCDD and PCDFs that indicate the

parallel presence of PCDDs and PCDFs), or their precursors (compounds

that have been shown to be chemical progenitors of PCDDs and PCDFs);

2) combustion system optimization through continuous, on-line monitoring

and process control;

3) a method to advance prevention of PCDD and PCDF formation rather

than rely on flue gas cleaning controls; and

4) assurance to stakeholders (permit writers, public, etc.) that the process is

operating safely.

The U.S. EPA Office of Solid Waste (OSW, which regulates hazardous waste treatment

processes) has identified continuous PCDD and PCDF monitoring as a research priority,

because PCDDs and PCDFs can drive risk assessments. 2 OSW recognizes that CEMs offer

continuous compliance assurance, compared to infrequent, extractive sampling. Their policy

2

provides the economic incentive for waste facilities to use CEMs by eliminating waste feed

characterization, compliance testing, and operating parameter monitoring for pollutants which

a facility uses CEMs. However, with little or no PCDD/PCDF CEM technologies currently

under development, DOE will not be able to implement this alternative mode of compliance .

A CEM for compliance purposes will likely require higher performance capabilities (especially

for sensitivity) than if the CEM were used for research purposes or as a method of

combustion optimization. These CEM sensitivity needs will be lessened by using a short

duration sample concentration method. However, as discussed later in this proposal,

sensitivity for a dioxin CEM is a primary issue, requiring measurements at concentrations two

to four orders of magnitude lower then ever achieved thus far.

As suggested by experts in this area,3 a dioxin CEM would be most useful first as a

research tool in laboratories studying PCDD and PCDF formation and control. As such, the

instrument would need to make rapid, accurate measurements of PCDDs and PCDFs but at

concentrations much higher than needed for a compliance CEM. This type of instrument

would greatly accelerate our understanding of PCDD and PCDF formation and the availability

of prevention and control techniques. Researchers have limited understanding of how

combustion processes affect PCDD and PCDF formation, largely due to their need to relate

time-integrated sampling data with dynamic formation mechanisms and combustor conditions

after an often multi-week analysis lag period. A real-time CEM would provide immediate

feedback on how variations in combustion operating parameters affect PCDD and PCDF

formation and/or destruction, thus allowing more accurate correlations and much more

comprehensive data analysis. An instrument capable of making these types of measurements

would also be valuable to DOE as new waste treatment processes are evaluated and readied

for permitting/ public acceptance.

As our understanding of PCDD and PCDF formation improves, it would be valuable to

build a database using emissions from actual waste treatment processes to correlate operating

conditions with PCDD and PCDF formation. Such a database could be used to devise

operating strategies to prevent formation of PCDD and PCDF. This database could also be

used to identify surrogates or indicators that can be monitored more easily and cheaper than

3

the PCDD and PCDF themselves, leading to less expensive, more widely-implemented

compliance and control strategies.

CEMs also provide data important for stakeholder's assurance that the combustion

processes are operating safely. Stakeholders such as public interest groups, permit writers, and

local citizens groups, can play a major role in permitting waste treatment facilities. Real-time

emissions data may accelerate their acceptance, saving time and money during the permitting

process.

BACKGROUND: CHEMISTRY OF DIOXINS/FURANS

Dioxins and furans represent a class of chemical species composed of paired, multiple

chlorinated benzene rings joined by a third ring structure. In dioxins, the third ring contains

two oxygen atoms, while in furans, only one oxygen is present. One to eight chlorines can be

bound to the benzene ring structures, in varying arrangements. The chlorine number and

position account for 210 different dioxin/furan congeners. Compounds, or congeners, with

the same number of chlorine atoms are called homologues, and they all fall within a single

congener class (.e.g. tetra-chlorinated dioxins).

Dioxins and furans are produced in many combustion processes as a result of combustion

inefficiencies. A great deal of research during the past decade has focused on identifying the

pathways to formation - in hopes of understanding how also to prevent formation, and to

learn when dioxins pose actual risk to human health and the environment. Today, it is widely

believed that improper combustion is the primary contributor to dioxin and furan formation.

Sources of dioxins include waste treatment processes, industrial and residential wood burning,

smelters, coal and biomass combustion, metal recycling, iron sintering plants, and automobiles.

Dioxins and furans are also naturally-occurring. They are primarily believed to be

anthropogenically-derived (based on “old earth” analyses), although recent scientific results

suggest that biological composting my provide conditions for formation. Other potential

sources of combustion-derived dioxins and furans, e.g. diesel engines, have only received

limited attention. Advances in analytical and sampling techniques are now making these

4

sources more amenable to assessment.

5

CHAPTER 2. JET-REMPI TECHNOLOGY

DESCRIPTION OF REMPI TECHNIQUE

Resonce Enhanced Multiphoton Ionization (REMPI) is a highly sensitive, highly species-

selective, gas-phase analysis technique for combustion research.4 To achieve good wavelength

resolution, the gas sample has to be cooled by expansion through a nozzle. Adiabatic

expansion results in low sample temperatures which increase the electronic ground state

population. The enhanced population of the ground state gives an increase in sensitivity and

very sharp REMPI transitions. One or two lasers are used to ionize the cooled gas molecules

in a small volume by absorption of two or more photons, one of which is resonant with an

electronic transition in the target molecule (see Fig. 1).

hν1

hν1

a)

hν3

hν2

b)

Two photon ionization schemes

a) One colour two photon prozess

b) Two colour two photon prozessIP

0

Energy Ionization continuum

Ground state

Excited electronic states (S1)Molecular specific

Fig. 1: Two-photon ionization scheme

Aromatic molecules have conveniently accessible vibronic levels of the first excited singlet

6

states (S1 level), at energies just exceeding half of the molecular ionization energy. Thus, single

color, resonant two-photon ionization schemes (1+1 REMPI) can be used for these species.

An advantage of this approach is that "soft ionization" at relatively low laser intensities is

feasible. Typically an unfocused laser beam is used and there is minimal fragmentation of the

parent ion. In contrast, the three-photon (2+1 REMPI) approach used for molecules such as

alkyl chlorides requires the use of higher laser intensities which may result in significant ion

fragmentation and, thus, may compromise selectivity for some molecules.

In a REMPI system, a Time-of-Flight Mass Spectrometer (TOF-MS) analyzes the ions by

mass. This two-dimensional detection scheme by mass and wavelength provides high species

selectivity.

The potential for REMPI as a CEM for combustion emissions has been discussed in the

literature.5,6 However, typical sensitivities of conventional REMPI apparatuses are around the

ppb level, which is clearly insufficient for monitoring chlorinated aromatics in a waste

treatment process such as an incinerator. Thus, DLR Stuttgart has improved REMPI, and

coined the phrase Jet-REMPI to describe their system.

DLR STUTTGART’S JET-REMPI PROTOTYPE

Major improvements in sensitivity without loss in selectivity can be achieved with Jet-

REMPI. The DLR holds a European patent on several of these concepts,7 and a Japanese and

US patent are pending. The first improvement involves the location for ionization. In a

supersonic jet the temperature drop occurs only in an relatively narrow zone downstream of

the nozzle (i. e. the zone where the beam still forms a jet). If ionization is carried out further

downstream (i. e. in the molecular regime) as in conventional REMPI setups, then the

sensitivity drops due to a decrease in the beam density. Conversely, when ionizing right in the

transition zone between the jet regime and the molecular regime, the highest sensitivities in

conjunction with the lowest temperatures are obtained (see Fig. 2).

7

Jet

Molecularbeam

Valve

Laserbeam

x

Figure 2: Jet-REMPI principle

This simple and effective configuration cannot be achieved with conventional laser

ionization sources; specially designed ion extraction optics are needed. Further sensitivity

improvements have been accomplished by increasing the ionization volume (without loss in

resolution due to focusing techniques) and by minimizing collisions of charged particles with

the walls. To implement these improvements a skimmer is not used and a pulsed sample valve

is used to meet the vacuum requirements.

These improvements have been shown to increase REMPI sensitivity by a factor of 200

for dichlorotoluene to 100 ppt, with a signal to noise ratio of 2, and by a factor of 1000 for

naphthalene. Further sensitivity improvements are possible, for instance by increasing the

laser repetition rate.

8

Control andData Aquisition

Reflectron-TOF-MSIS

GasSample

COHERENTInfinity

Nd-YAG

LambdaOPPO

SHG + PS

Gate

Trigger

Trigger

RS232

Preamplifier

DSO1Gs/s

Signal250 µs

RS232

1 mJ2 ns0.15 cm

-1

Gated Valve

Figure 3. Jet-REMPI dioxin analysis prototype set-up. Setup of the mobile Jet-REMPIapparatus. IS = ion source, SHG = second harmonic generation, PS = prism separator,OPPO = optical parametric power oscillator, DSO = digital signal oscilloscope, RS232= serial interface type, and Gs/s = DSO sampling rate in gigasamples per second.

Figure 3 shows a sketch of the DLR Stuttgart Jet-REMPI analyzer.8,9,10,11, Here, REMPI

ionization is obtained by a laser beam directed in a horizontal plane orthogonal to the vertical

molecular beam axis and the horizontal axis of the mass spectrometer. The light of a

9

Nd:YAG (niodinium-yttrium, actinium, gallium) pumped laser (Coherent Inc., Infinity

40-100Ô, Santa Clara, CA), OPO system (Lambda Physic, Scanmate OPPOÔ, Goettingen,

Germany) is frequency-doubled to produce REMPI probe pulses in the 275-320 nm spectral

region to enable the target detection. An unfocused laser beam was used in this work, with a

cross section of 3 mm2, typical pulse energies of 0.3 mJ, and a duration of 2 ns. The samples

are jetted into the ionization chamber through a pulsed valve (General Valve Corporation,

Fairfield, NJ) that delivers 250 ms sample pulses at a repetition rate of 30 Hz

A linear Reflectron-Time-of-Flight Mass Spectrometer (TOF-MS) from Stefan Kaesdorf

(Munich, Germany) analyzes the ions by mass. This two-dimensional detection scheme by

mass and wavelength provides high species selectivity.

This prototype was tested at DLR with a variety of chlorinated organic compounds, albeit

not dioxins and furans due to laboratory safety constraints. The first field test was conducted

in May, 1996, at an incinerator in Karlsruhe, Germany. The prototype did not include a stack

interface or sample nozzle. The Jet-REMPI analyzer was connected to a laboratory-scale

municipal waste incinerator (about 3 m high) using a standard collection nozzle, by stack-

sampling personnel at Karlsruhe. During the tests, the incinerator was burning synthetic fuel

under well defined conditions. The Jet-REMPI prototype was used to monitor organic

compounds downstream of the air pollution control equipment. CO was monitored

simultaneously using a separate technique. No reference samples were taken, so the data

obtained was qualitative, not quantitative.

Data from the Karlsruhe test included time profiles for selected aromatics. In some cases

this data correlated easily to the CO profiles. In other cases, there are systematic mismatches,

probably caused by time dependent temperature effects in the filter, sampling line etc. It also

became clear that particulate matter in the ionization zone leads to an enormous background

signal. It is essential to prevent particles from entering the ion source, an important

consideration for future sampling interfaces to REMPI.

10

CHAPTER 3. DIOXIN CEM PERFORMANCE REQUIREMENTS

Measurement of PCDD and PCDF with REMPI requires consideration of several issues:

n Only PCDD and PCDF homologues which have at least all of the 2,3,7,

and 8 positions occupied by chorine atoms are important in determining

toxicity. There are 17 of these congeners from the tetrachlorinated

dibenzodioxin (TCDD) to the octachlorodibenzodioxin (OCDD). The

TCDD congener 2,3,7,8-TCDD and the pentachlorodibenzodioxin

(PeCDD) congener 1,2,3,7,8- PeCDD are especially important because of

their large contribution to the toxic equivalency, or TEQ (equivalence

factors of 1 and 0.5, respectively). We must estimate which specific

congeners should be measured, and to what level of sensitivity, to be of

value as a compliance CEM.

n The non-rigidity of PCDD and PCDF molecules, in particular the

“butterfly” vibration around the axis through the oxygen atom(s), may

demand very intense cooling to increase the electronic ground state

population.

n The work of Zimmermann et al.12 indicates that chlorination of aromatic

compounds increases the ionization energy and shifts the S1 state towards

lower levels. As a consequence, for the lower chlorinated dioxins the S1

state is more than one-half the ionization gap. However, for higher

chlorinated dioxins, the S1 state will be less than half the ionization gap,

which could require a two-color (two wavelength) REMPI system to excite.

Previous research13 indicates that the transition of the S1 state from greater

to less than one-half the ionization energy may occur near the

tetrachlorinated congeners. This is important because the TCDD isomers

contain the most toxic congener, 2,3,7,8-TCDD, and are therefore

11

significant toward TEQ determination. Therefore, the potential ability of

the relatively inexpensive one-color (1+1) REMPI method, described in

this work, to excite the TCDD to over one-half of the ionization gap

would drastically simplify detection of this important congener.

n Increased chlorination of the parent dioxin (or furan) molecule results in a

decrease in the lifetime of the first excited state (increased intersystem

crossing), leading to reduced REMPI sensitivities or precluding the single

wavelength method altogether

n Highly chlorinated PCDD and PCDFs have extremely low vapor pressure.

Since REMPI is a gas-phase measurement technique, all sample lines,

valves, and pumps must be heated to prevent condensation of the target

molecules.

This leads to three key questions:

1) Can TCDD isomers be measured with one-color (1+1) REMPI?

2) Does REMPI show promise for measuring PCDD and PCDFs, indicators,

or precursors with sufficient sensitivity and species selectivity to be useful

as a continuous emissions monitor?

3) What are the performance requirements for a PCDD and PCDF, indicator,

or precursor CEM?

This test program addresses the first two questions by measuring PCDD isomers with zero

to four chlorines.

The third question is much broader and requires a combination of research in mechanistic

studies and a prototype PCDD and PCDF CEM to collect data for identifying surrogates or

indicators that can be more easily measured. In addition, performance requirements for an

instrument can depend on the application. For a dioxin/furan CEM, the first, and probably

shorter-term application is in research laboratories to accelerate scientific understanding of

dioxin formation and control. Current laboratory research is extremely limited by current data

12

techniques which are expensive and have long turn-around times (extractive samples followed

by wet chemistry analyses). The amount of data that could be generated by a real-time

instrument is extraordinary and could be used to identify surrogates that (at higher

concentrations) are true indicators of dioxins’ presence, and easier and cheaper to measure. We

estimate performance requirements for this research-laboratory application as follows:

13

PARAMETER REQUIREMENT(S) COMMENT(S)

Detection limit 500 to 5,000 ng/dscm TEQ Concentrations typically used in

dioxin research experiments.

The lower the instrument’s detection,

the more useful it will be. Must relate

detection limit of individual congeners

to 100 to 1,000 ng/dscm TEQ.

Estimate sensitivity around 5 ng/m3

for each congener of interest would

be more than adequate.

Species identified 1,2,3,4, TCDD, di-chlorodioxin,

and others to be determined.

1,2,3,4 TCDD is safe to use in

laboratory and chemically

representative of toxic congeners of

interest.

Even a limited set of species will be

useful if related to TEQ.

Phase gas phase acceptable Gas/particulate phase highly

desirable, but could make do with just

gas phase in beginning.

Gas Matrix Dioxins in standard combustion

gases with other

organics/inorganics.

Useful even without full matrix

detection capabilities.

Maximum

temperature

to be determined

Moisture dry to saturated The more tolerant to moisture, the

more useful the instrument. Moisture

usually fairly constant with time.

Could calibrate at regular intervals.

Accuracy to be determined, +/- 100% would

be probably still satisfactory.

Response Time Seconds to hours

• Table 1. Estimated performance goals for laboratory-based dioxin/furan CEM.

14

The second long term application for a dioxin CEM is to be installed on a stack, either

permanently or temporarily (as in a mobile instrument), to monitor dioxins for regulatory

compliance, improved system performance, and increased stakeholder acceptance.

Performance requirements are similar to those shown in Table 1, except the detection limits

are driven by regulatory requirements on emission limits (0.1 ng/dscm), considerably below

concentrations tested in research laboratories studying formation and control. Table 2

summarizes estimated performance goals for a field-based CEM. Response times

aresufficiently long, however, that the sample can be concentrated to achieve these

sensitivities.

PARAMETER REQUIREMENT(S) COMMENT(S)

Detection limit 0.1ng/dscm TEQ (approx. 70 ppq

based on M.W. of 2,3,7,8 TCDD),

or approx. 10 ng/dscm TOTAL.

For direct TEQ measurement,

estimate need sensitivity between

0.001 ng/m3 and 1.0 ng/m3

depending on Toxic Equivalency

Factor (TEF) of each congener.

Can use sample pre-concentrator, or

identify surrogates to measure at

higher concentrations.

Species identified To be determined. May need to

detect class of congeners, or subset

of congeners included in TEQ.

A limited set of species, or a class of

congeners will be useful if it can be

related to TEQ.

Phase must be able to account for gas-

phase and particle-bound species.

Will require sample interface to

collect and digest/vaporize particles.

Gas Matrix Dioxins in standard combustion

gases with other

organics/inorganics.

Must define matrix detection

capabilities.

Maximum

temperature

to be determined

Moisture Must be able to measure in all

environments from dry to saturated.

Moisture usually fairly constant with

time. Could calibrate at regular

intervals.

15

PARAMETER REQUIREMENT(S) COMMENT(S)

Accuracy Probably +/- 50% to 100% relative

accuracy compared to EPA

Reference Method (M23)

Reliability Probably >95%.

Cost to be determined. $400K not realistic. $200K may be

OK for limited applications,

depending on maintenance and

operating costs.

Ease of operability Must be a black box requiring only

trained technicians.

Table 2. Estimated performance goals for field-based dioxin/furan CEM.

In the future, another application for this type of CEM is process control to prevent

dioxin formation. This would require monitoring specific surrogates at relatively high

concentrations, with short response times.

16

PRACTICAL VS. SCIENTIFIC INSTRUMENT

A practical CEM and a scientific CEM will be designed to the following criteria, although

those denoted by (p) will be of higher importance for a practical instrument where as criteria

which are marked (s) will have priority in the scientific case.

Low price($150K -

$200K)

(p)

Compactness (p)

Tolerant against

vibrations

(p)

Tolerant against

temperature changes

(p)

Easy to handle (p)

Simplicity of

ionization scheme

(p)

Sensitivity (s)

Many compounds

accessible

(s)

This list makes Nd:YAG pumped Ti:Sa lasers an obvious choice for a practical instrument.

Conversely, an OPO laser has a larger wavelength range and is more suitable for a scientific

instrument. For the highly chlorinated aromatics a two colour technique is required for

REMPI transitions. To achieve this two laser sources are necessary, one fixed and one

tuneable. For the fixed source waste light from the OPO pump laser can be used. Thus the

OPO system will yield a scientific tool which is suitable for both one and two colour REMPI.

Determining which compounds need to be measured is less easy. For scientific

applications, as complete a data set as possible seems desirable. However during one

17

experiment only a limited number of species profiles can be monitored because each

compound requires a discrete resonance wavelength for measurement.

Another important task is the measurement of Toxic Equivalency (TE). Only dioxins and

furans with 4 or 5 Cl atoms make a significant contribution to the TE. This is significant

because REMPI sensitivities decrease sharply with increasing degree of chlorination. The

experiments described in this report demonstrate that tetra-chlorinated dioxins are measurable

with the simple one colour 1 + 1 REMPI. Thus the critical issue which remains for the on-

line-determination of TE’s are the penta- chlorinated dioxins and furans. It is known that in

incineration the pentachlorofurans, in particular, can represent a large contribution to the TE.

Consequently, it should be a goal of future Jet-REMPI Dioxin CEM programs to demonstrate

ionization for penta-chlorinated dioxins and furans.

For a practical instrument two types of compounds must be measurable; (i) precursors and

(ii) Toxic Equivalency (TE) surrogates.

(i) Precursor concentrations yield the most important information for process control. Tri-

chlorophenols are regarded as the most viable precursors to dioxin/furan formation.

Relevant isomeric constitutents of this class and of other clases of chlorophenols have yet

to be established. Identification of these species will be an important task in future Dioxin

CEM studies. After the most relevant precursors have been identified, we must design a

practical technique to efficiently ionize these compounds.

(ii) TE Surrogates: Ideally, one could identify surrogate compounds related to TE’s to monitor

in place of highly chlorinated dioxins/furans (> five or six chlorines, difficult for single-

color REMPI). Currently, in the limited studies addressing dioxin surrogates,

chlorobenzenes are most often used, even though they correlate rather poorly with TE’s.

Obviously it would be to correlate surrogates and TE based on a chemical kinetic model.

This should be another goal of future Dioxin CEM research. In the absence of such

models, surrogates should be chosen that chemically resemble the constituents of the TE

as closely as possible. The surrogates should also exist in relatively high concentrations and

18

should be easely measured. Dioxins and furans with low chlorination might fulfill these

requirements.

19

CHAPTER 4. PROJECT TEAM AND TEST FACILITIES

PROJECT TEAM

This project was conducted jointly by the U. S. EPA National Risk Management Research

Laboratory (NRMRL) and DLR Stuttgart Institut fur Physikalische Chemie der Verbrennung,

with Acurex personnel under contract to EPA. Sky +, a U.S.-based small business specializing

in CEM technology development, managed the project. Key personnel were as follows:

Name Affiliation Project Role

Dr. Brian Gullett EPA - NRMRL Principal InvestigatorDan Burns DOE/Westinghouse

Savannah RiverContract Coordinator for DOE

Dr. Steve Priebe DOE/Mixed Waste FocusArea, INEL

Technical representative fromsponsoring organization

Cliff Brown DOE/Mixed Waste FocusArea, ORNL

Progrmmatic representative fromsponsoring organization

Dr. Harald Oser DLR Stuttgart Jet-REMPI prinicipal investigatorDr. Horst-HenningGrotheer

DLR Stuttgart Jet-REMPI project manager

David Natschke Acurex Environmental, Inc. Lead EPA technical/laboratorysupport personnel

Kevin Bruce Acurex Environmental, Inc. Acurex project managerDr. Nina French Sky + Project manager

DLR Stuttgart, a German non-profit research laboratory widely recognized for laser

diagnostic capabilities, is the only research group known to have built and tested a Jet-REMPI

prototype on effluent from a hazardous waste incinerator. They were keenly interested in this

test program because DLR is not permitted to use dioxins in their research laboratories,

limiting their ability to complete proof-of-principle experiments. Such experiments are crucial

to identify resonance frequencies for dioxin molecules and to verify accuracy and detection

limits.

20

EPA National Risk Management Research Laboratory (NRMRL) in Research Triangle

Park, North Carolina provided facilities and expertise to conduct controlled experiments with

known quantities of dioxins. EPA NRMRL also has expertise in dioxin chemistry, formation,

and control - knowledge crucial to adapt the Jet-REMPI technique into a useful dioxin/furan

CEM.

Funding for this project came from the DOE/EM Mixed Waste Focus Area, which paid

for EPA contract personnel to conduct the tests, some hardware for the dioxin feeder, travel

costs for two of the DLR personnel, hardware shipping costs from Germany, reference

method sampling and analysis, and project management and reporting costs. EPA provided

much of the infrastructure to conduct the test, including facilities, space, and programmatic

and technical leadership. DLR Stuttgart provided the Jet-REMPI hardware and all labor costs

for Drs. Harald Oser and Horst-Henning Grotheer.

TEST HARDWARE DESCRIPTION

REMPI ANALYZER SYSTEM

The REMPI analyzer prototype consists of a probe, laser, mass spectrometer, pump,

electronics, and mounting racks. It is described in Chapter 2 of this report. Note: a preferred

laser configuration would be to combine a Coherent Nd:YAG with a Coherent OPO or

Continuum Nd:YAG with a Continuum Sun Light OPO. Cost of laser is about $200K. Cost

of TOF mass spectrometer is about $40K. Pump equipment and electronics add

approximately another $100K to cost of system as tested. The mounting racks consist of

moveable racks approximately 2m long x 0.9m wide x 1.5m high. The total system weight is

approximately 1000kg.

EPA DIOXIN FEEDER

To test the ability of Jet-REMPI to measure dioxin species, a temperature-sensitive

permeation vial system14,15 was developed to deliver gaseous species to the Jet-REMPI system.

The key to this system is vapor permeation from a capillary independent of gas flow. The

21

permeation vial was housed within a glass spiral mounted in the oven of a Hewlett Packard

model 5890 gas chromatograph (GC) which acted as a constant temperature (± 1oC) bath.

Nitrogen was used to carry the volatilized test compound from the permeation vial to the inlet

of the Jet-REMPI analytical system. The mass volatilized was determined by mass loss of the

permeation vial and coupled with a mass flow controller allowed calculation of concentrations.

The following equation describes the permeation rate:

where

r = rate of diffusion (ng/min)

T = temperature of vapor (K)

Do = diffusion coefficient at STP (cm2/sec)

M = molecular weight (g/mole)

A = cross sectional area of the capillary (cm2)

L = length of diffusion path (cm)

P = atmospheric pressure (mm Hg)

ρ = vapor pressure of chemical at temperature T (mm Hg)

As the above equation indicates, the permeation rate is dependent upon the temperature,

dimensions of the capillary, and vapor pressure of the particular compound. It is independent

of the gas flow rate.

The permeation vial was constructed from a glass weighing vial which was modified to

incorporate the capillary into its cap. For this work the capillary was 1 cm long with a 2 mm

I.D. bore. The permeation vial was housed in a glass spiral through which the carrier gas

flowed. The spiral and permeation vial were mounted in the oven of a Hewlett Packard model

5890 gas chromatograph (GC) which acted as a constant temperature bath. Testing showed

that the GC provided better than 1 °C temperature control. Nitrogen was used to carry the

volatilized test compound from the permeation vial to the inlet of the Jet-REMPI analytical

system. A Tylan mass flow controller was used to control the nitrogen flow rate, which was

r TD MAL

PP

=−

1 90 1040. * log

ρ

22

necessary for calculating the test compound concentration.

The mass volatilized was determined by mass loss of the permeation vial. A Mettler AT20

balance (2 µg resolution) was used to determine this mass loss. The concentration of dioxin

presented to the Jet-REMPI instrument was calculated from this mass loss divided by the total

nitrogen volume over the course of the experiment:

where:

[PCDD] = vapor phase concentration of dioxin

w1 = permeation vial weight prior to experiment (g)

w2 = permeation vial weight after experiment (g)

Q = nitrogen flow rate (L/min) as determined by the mass flow controller

t = length of experiment (min)

Process gas flow rates and injectant concentrations were varied to affect different gas

phase concentrations of TCDD. After exiting the re-evaporation chamber, process gases pass

through the REMPI system and on to a modified Method 23 sampling system. At the

conclusion of each experiment, the system was cooled and component parts analyzed for

residual TCDD by solvent rinsing. The Method 23 gas sampling and solvent rinse analyses

were used to determine the duration-averaged TCDD concentration. Complete delivery of

the TCDD was expected based on previous work.

The reference method samples were analyzed by GC/MS using the isotope dilution

technique. Isotopically labeled homologues of 1,2,3,4-TCDD were used to quantify the native

1,2,3,4-TCDD present and assess method recovery. Prior to overnight toluene solvent

extraction, the samples were spiked with a known concentration of 13C12 1,2,3,4-TCDD.

Following extraction, the samples were concentrated and subjected to a series of

chromatographic cleanup columns (acid/base silica, alumina, and activated carbon) to remove

analytical interferants. The samples were then concentrated to a known value and spiked with

[ ]( )

PCDDw w

Q t=

−∗

1 2

23

the 13C6 1,2,3,4-TCDD recovery standard and analyzed by GC/MS.

The GC/MS system was calibrated by determining the response of a fixed mass of 13C12

1,2,3,4-TCDD relative to the response of the varied mass of unlabelled 1,2,3,4-TCDD. The

resulting relative response factors were used to quantify the unlabelled TCDD present in each

sample based on the known mass of 13C12 1,2,3,4-TCDD present in each sample. Similarly,

response factors developed between the 13C12 1,2,3,4-TCDD internal standard and the 13C6

1,2,3,4-TCDD recovery standard were used to quantify the method performance by

determining 13C12 1,2,3,4-TCDD recovery.

24

CHAPTER 5. TEST OBJECTIVES

The goal of Phase I was to test and evaluate Jet-REMPI as a potential continuous

emissions monitoring technique. Our evaluation was based on performance criteria listed in

Table 1 for Jet-REMPI as a research-laboratory based CEM to understand dioxin formation

and control. Research-laboratory applications are most realistic for near-term Jet-REMPI,

although we will also evaluate the potential of Jet-REMPI to meet performance criteria

specified in Table 2 for application to stacks as a compliance tool.

Specific objectives of this Phase I Test & Evaluation Program were to:

1. Determine if tetra-chlorinated dioxins can be measured with 1+1 (single color) Jet-

REMPI.

2. Identify resonance wavelengths for di-, tri-, and tetra-chlorinated dioxins.

3. Generate calibration curves (instrument response versus known concentration) of Jet-

REMPI for several concentrations of a di- chlorinated dioxin congener.

4. Experimentally determine minimum detection limit of Jet-REMPI for a di- chlorinated

dioxin congener.

5. Demonstrate transient measurement capabilities.

6. Conceptually evaluate means to enhance detection and other critical performance

requirements such as cost, species selectivity, and particle measurements.

Other important evaluations for future test programs would be:

1. Evaluate potential interferences from combustion gases by testing on EPA IRF

combustor

25

2. Repeat some or all of Phase I tests with additional dioxin congeners.

3. Compare Jet-REMPI performance features to other organic CEM technologies (e.g.

GC/TOFMS)

4. Verify performance using EPA Multi-Fuels combustor

5. Field Demonstration - verify performance on industrial-scale Municipal Waste

Combustor in conjunction with already-scheduled Reference Method data collection.

26

CHAPTER 6. TEST DESCRIPTIONS

The DLR Jet-REMPI analyzer prototype was tested from July to October, 1996 at EPA

National Risk Management Laboratory in Research Triangle Park, North Carolina. The

primary test objective was to make measurements on actual dioxin molecules, which was not

possible at the DLR laboratories due to internal safety regulations. The Jet-REMPI analyzer

was operated with a simple dioxin feeder, parts of which have previously been used to study

dioxin formation and control. Costs for these tests were minimal, since most of the

equipment and REMPI staff did not require additional funding. EPA provided much of the

infrastructure to conduct the test.

The tests were conducted with di-, tri-, and tetra- chlorinated dioxin congeners. The

precision of Jet-REMPI was measured using two methods: first, the EPA analysis reference

method (Method 8280), and second, a gravimetric analysis of the feed chamber.

Measurements from M8280 would normally be the standard against which relative accuracy

would be compared. However, M8280 is very expensive, and the Phase I test budget did not

allow more than 10 M8280 measurements. Therefore, to accumulate enough data for

statistically-significant relative accuracy calculations, we used a gravimetric analysis of the feed

chamber. The accuracy of the gravimetric analysis was compared to M8280 during Test 4.

M23 was not strictly followed, since there was no particulate matter. As an alternative, we

used an XAD module and no heated filter. The heater was monitored closely to avoid thermal

breakdown of the TCDD. The appropriate heater temperature was determined by a TCDD

mass balance -- measuring the TCDD input vs. Method 23 results + QTC rinse analysis.

At estimated low ppt REMPI sensitivities, TCDD concentrations were so low that M23

samples were run for many hours in order to achieve required detection limits.

27

PHASE I TEST SCHEDULE

ID Task Name

1 Equipment Shakedown

2 Contingency for Equipment

3 Test Plan Complete

4 Testing

5 First Dioxin Test Complete

6 Results presentation

7 Final Report

8 Final Report Complete

9

8/5/96 8:00 AM

8/28/96 8:00 AM

9/26/96 8:00 AM

7/8 7/15 7/22 7/29 8/5 8/12 8/19 8/26 9/2 9/9 9/16 9/23 9/30 10/7July August September

28

CHAPTER 7. TEST RESULTS AND ANALYSIS

INSTRUMENT SHAKE-DOWN

After a one-week delay in U.S. Customs the Jet-REMPI hardware arrived at the EPA test

facility in Research Triangle Park. We then underwent 3 weeks of diagnostics and costmetic

surgery to repair minor damage suffered during shipment, and a laser software malfunction.

DIOXIN TESTING

The first PCDD congener tested was an unchlorinated dibenzodioxin (UCDD). This initial

compound was chosen because, for every chlorine atom present in the congener, the

sensitivity decreases by a factor of approximately 6.16 The decrease in sensitivity is due to

intersystem crossing [i.e., a reduced lifetime of the excited state (S1)], and can be overcome

only by using different ionization schemes.17 Although the measurement of UCDD thus

appears facile due to the relatively high sensitivity, it plays an important role in testing Jet-

REMPI for the detection of PCDD and PCDFs in general. Dibenzodioxins have a so-called

“butterfly” vibration around the symmetric line through the oxygen (O) atoms.18 The force

constant for this bending vibration is very low which results in a very narrow spacing between

the energy levels. If the molecules are insufficently cooled, both the ground state will be

populated and the closely adjacent, vibrationally excited states will be occupied leading to more

complicated or even “clogged” spectra. Such a complication may be suspected in the case of

Jet-REMPI with its short distance between the nozzle and skimmer.

Fig. 4 shows that this is not the case. Figure 4 shows that the cooling method of DLR Jet-

REMPI is sufficient to provide a well-resolved spectrum for UCDD with line widths (full

width, half maximum, FWHM) of typically less than 0.1 nm and clearly separated lines. In

other words, the cooling achieved with Jet-REMPI is sufficient to avoid spectral clogging due

to high population of excited states. The different peaks in the wavelength-dependent spectra

represent the excitation into different vibrational energy levels of the excited electronic state.

29

295.0 295.5 296.0 296.5 297.0

0

2

4

6

8

10

Ion signal [a. u.]

Wavelength [nm]

12C spectra13C spectra x 10

Figure 4: Wavelength dependent REMPI spectra for UCDD. Masses 184 and 185 are monitored

Further issues concerning UCDD deserve to be mentioned. Figure 5 shows the mass

spectrum at a resonance wavelength of 295.82 nm. In the spectrum the contribution due to

the 12C and 13C isotope parent signals is clear. However, no fragmentation is observed, which

is typical for our work. In Figure 2 the wavelength dependent spectra are plotted for these

parent peaks. One clearly sees that the 13C signals are shifted towards higher energies which is

expected for the vibrations of a heavier compound. This discrimination between 12C and 13C

isotopes is another indication of our high selectivity. In GC/MS 13C congeners can be

separated by mass but not by retention time since the retention time is virtually identical for12C and 13C compounds. In the case of Jet-REMPI one can use both the mass and the

ionization wavelength to discriminate between isomers.

30

0 50 100 150 2000

20

40

60

80

100

Ion signal [a. u.]

Mass [amu]

Fig. 5: Mass spectrum for UCDD at a wavelength of 295.82 nm.

After the experiments with UCDD the next step was to investigate 2,7-DCDD. The

wavelength spectra is shown in Figure 6. There are a lot more transitions than for UCDD, but

each of the transitions is well resolved. According to the literature19,20 the greater complexity in

the 2,7-DCDD spectrum is due to the fewer degrees of symmetry in this molecule in

comparison with UCDD. The Jet-REMPI mass spectra shown in Figure 7 again show no

fragmentation due to the soft ionization technique.

31

Fig. 6: Wavelength dependent Jet-REMPI spectra for 2,7-DCDD. Mass 252 is monitored.

Fig. 7: Mass spectrum for 2,7-DCDD at a wavelength of 305.6 nm.

32

A series of measurements at different concentrations were performed to estimate a calibration

curve and the sensitivity limits for 2,7-DCDD. The 2,7-DCDD concentration was varied by

altering the temperature of the dibenzodioxin feeder. The actual concentration at each

temperature was determined by operating the permeation system for several hours and then

using gravimetric analysis. The results are shown in Figure 8. These data, when extrapolated to

a signal/noise ratio (S/N) of 3/1, yield a detection limit of 30 ng/dscm at a pulse energy of 0.3

mJ. Actually, we could show that increasing the pulse energy to 1 mJ without fragmentation is

possible. In this case we would get a detection limit of 9 ng/dscm which translates into 75 ppt

by weight or 60 pptv for DCDD. This is the first on-line detection limit reported for a DCDD

isomer.

0.001

0.01

0.1

1

10

100

1000

1 102 104 106 108

Concentration (ng/dscm)

ion signal (a.u.)

Fig. 8: Jet-REMPI measurements of 2,7-DCDD, showing minimum detectability ofapproximately 30 ng/dscm.

Our work found that efforts must also be undertaken to ensure that the inlet valve can be

sufficiently heated to prevent condensation of dibenzodioxins during on-line measurements.

Such heating influences the beam cooling mechanism and requires careful optimization of the

33

inlet system.

1,7,8-trichlorinated dibenzodioxin (1,7,8-TrCDD)

The wavelength spectra for 1,7,8-TrCDD is much more complicated than that for 2,7-

DCDD, as shown in Figure 9. In addition the prototype instrument may not be providing

sufficient cooling (to be discussed later). The mass spectra is again fragment free. However,

there are more parent peaks due to the larger number of possibilities in the 35Cl/37Cl and

12C/13C distribution as shown in Figure 10.

300 304 308 312 316

0,0

0,1

0,2

0,3

0,4

0,5

Ion

sign

al [a

. u.]

Wavelength [nm]

Figure 9: Wavelength dependent REMPI spectra for 1,7,8-TrCDD. Mass 286 is monitored

34

50 100 150 200 250 3000

20

40

60

80

100Io

n si

gnal

[a. u

.]

Mass [amu]

Figure 10: Mass spectrum for 1,7,8-TrCDD at a wavelength of 304.9 nm

Tetrachlorinated dibenzodioxins (TCDD)

Two problems must be addressed, which may turn out to be limiting factors for reliable

measurement of tetra- and higher chlorinated dioxins. From the vapor-pressure data for these

species21 it is obvious that the sampling line and the inlet valve must be heated very carefully

to avoid condensation. As the degree of chlorination increases, it becomes more important to

ensure that all parts are well heated. This will not only affect the initial sample temperature,

but it will also decrease the cooling effect.

The other important aspect concerns the change in the relative positions of the energy

levels upon chlorination, as reported in the literature.22 In the case of PCDD and PCDFs, the

ionization potential rises while the energy level of the S1 state (which is resonantly excited by

35

the first photon absorbed) decreases. This may jeopardize the applicability of the simple (1+1)

single color REMPI scheme.

In Figures 11 and 12, REMPI spectra of 1,3,7,9-TCDD and 2,3,7,8-TCDD are shown.

These spectra are the first REMPI spectra of TCDD. Although they show sharp lines, the

two systems are very complicated and there is considerable overlap of the spectra for the two

molecules. We regard these spectra as preliminary and it is not yet clear whether we ionized

the molecules by excitation via the vibrationless S1 state or from the band origin, the S0

electronic groundstate. From Figures 11 and 12 it would appear possible that a simple single-

color REMPI scheme can be used for the detection of TCCD.

310 312 314 316 318 3200,0

0,1

0,2

0,3

0,4

0,5

0,6

Ion

Sig

nal [

a. u

.]

Wavelength [nm]

Figure 11: Wavelength dependent REMPI spectra for 1,3,7,9-TCDD. Mass 320 is monitored

36

309 310 311 312 313 314 315

0,0

0,2

0,4

0,6

Ion

Sig

nal [

a. u

.]

Wavelength [nm]

Figure 12: Wavelength dependent REMPI spectra for 2,3,7,8-TCDD. Mass 320 is monitored

DISCUSSION OF RESULTS

We suspect that the complexity of the TrCDD and TCDD wavelength spectra shown in

figures 9 through 12 is at least partly a result of insufficient cooling. The 0.5 mm diameter

gated inlet valve has been routinely heated to 100oC, and has appeared to produce adequate jet

cooling for a series of rigid organic molecules. However for these dibenzodioxin tests, to

avoid condensation of TrCDD and TCDD, we heated the sample lines, including the inlet

valve, to 225oC. This higher initial temperature results in the final temperature rising

approximately quadratically. Not only does the cooling process start from a higher initial

temperature, but the cooling effect itself is reduced because of the increased mean free path

resulting in fewer collisions in the expansion zone. Whereas a higher final temperature might

be acceptable for rigid molecules, in the case of dioxins, one has to assume that due to the

butterfly vibrations, the molecules may, to a significant extent, remain within these excited

states. Insufficient molecular cooling results in a finite population remaining within the exited

vibrational energy levels in the electronic ground state. This results in a crowded wavelength-

37

dependent REMPI spectrum and a reduction of sensitivity, because the excitation takes place

from a lot of different groundstate energy levels. The question of cooling and final

temperature will be checked in subsequent experiments by re-designing the inlet system.

These effects, however, do not necessarily influence whether TCDD can be ionized with a

single-color (1+1) REMPI. If the ground state S1 is less than half the ionization gap (as

discussed in /4/), one could still try to achieve ionization with a (1+1) single color system via a

vibrationally excited S1 state. This ionization would be unaffected by the sample temperature.

Since we were able to measure ion signals from TCDD with the present arrangement, one can

expect that sensitivity will only improve when the sample is more sufficiently cooled. We

conclude that monitoring of TCDD is possible with a (1+1) single-color REMPI.

A remaining question is how promising REMPI is as a CEM to measure PCDD and

PCDF congeners, their indicators, or their precursors in combustion flue gas. Sensitivity and

species selectivity are both important parts to this question. REMPI certainly has adequate

species selectivity, and data presented in this paper show that levels of chlorination up to

TCDD can be measured using the simple one-color (1+1) Jet-REMPI approach. Sensitivity is

affected by the conflicting requirements to heat the molecules in the sample line to avoid

condensation and cooling them in the expansion nozzle to facilitate spectral resolution. These

requirements begin to conflict at around TCDD and larger homologues. However, there are

several ways around this conflict.

First, the availability of real-time PCDD data may allow the establishment of correlations

between measured concentrations of lower-chlorinated dioxins and TEQ. This would allow

establishment of TEQ values using measurements of the more REMPI-accessible PCDD

homologues. Second, other molecules, such as PCDFs, when coupled with dioxin

measurements, may provide strengthened correlations with TEQ. PCDFs have a significantly

higher vapor pressure than dioxins (0.43 Pa for 2,3,7,8-TCDD at 150oC and about 6 Pa for

2,3,7,8-TCDF at the same temperature). Consequently, less heating of the sample lines is

required for PCDF. In addition, due to the relatively rigid C-C bond, PCDFs are less

susceptible to the effect of insufficient cooling. Moreover, it is known23 that PCDFs are

38

present in incinerator emissions generally more abundantly than PCDDs. This may lead to

situations when 50% of the total TEQ consists of a single congener, the 2,3,4,7,8-PeCDF.

The main disadvantage of PCDFs seems to be that polychlorinated congeners require a

two color (1+1) REMPI /4/, making a CEM somewhat more complicated. However, we will

continue to investigate the use of PCDFs as a surrogate for TEQ-related CEM measurements.

39

CHAPTER 8. TECHNICAL ISSUES

PARTICLE MEASUREMENTS

During Phase I we invited a CEM stack-interface specialist (Mr. Herb DeFrieze) to discuss

ideas on how to introduce particle-bound dioxins in the Jet-REMPI analyzer. These ideas

focused on ways to vaporize dioxin from the surface of small particles and separate the

particles from the vapor prior to analysis. Two important requirements are simplicity (to

accomplish a reliable, workable design), and dioxin chemistry (not altering the subject during

the measurement). We also want to preserve as much of the real-time measurement capability

as possible. Our best concept is to thermally desorb dioxins from the particles and physically

separate the vapor from the particles using a non-contact filter, then concentrate and analyze

the vapor.

SENSITIVITY

Although sensitivity requirements vary from 500ng/dscm to sub-ng/dscm, depending on

the application for the CEM, it appears that a Jet-REMPI CEM will be more useful with at

least the capability for detection below the ng/dscm level. We recommend a sample

concentrator such as a cold finger, to concentrate the vapor prior to injection into the Jet-

REMPI analyzer.

SPECIES - WHAT TO DETECT?

Studies of full scale-operational plants have shown correlations between PCDD/F

emissions and concentrations of precursor substances such as chlorophenols and

chlorobenzenes.24,25 Unfortunately, prediction of actual TEQ values from measured precursor

concentrations has so far turned out to be not very accurate. An explanation is that the

process of heterogeneous PCDD and PCDF formation on fly ash clearly depends not only on

precursor concentrations but also on the number of active sites and the presence of a catalyst.

Consequently, a correlation between TEQ and precursor concentrations may strictly hold only

for otherwise unchanged conditions.

40

In order to predict TEQ values more directly we could attempt to use a particular

PCDD and/or PCDF congener as a surrogate. In addition, it may be appropriate to discern

between PCDDs and PCDFs because ample evidence may be found in the literature that these

groups of compounds are formed through different mechanisms. The same probably holds

for their destruction; e.g., reflected by the different PCDD/PCDF ratios that are found when

sampling at various positions in the exhaust gas train of an incinerator. Similarly, this ratio

changed when, in a laboratory experiment, different inhibitors were added to the fly ash or

when in an experiment on heterogeneous combustion of acetylene simply the temperature was

changed.

This study focussed on detection of PCDD. For the choice of a suitable PCDD

surrogate (or surrogates), several criteria have to be met. The untreated flue gas distribution of

the homologues is generally in favor of the more highly chlorinated dioxins.26 To use

octachlorodibenzodioxin (OCDD) as an indicator is, however, unfeasible due (1) to its

extremely low vapor pressure which would lead to severe condensation problems in any

sampling line and (2) to the REMPI sensitivity which in this case is probably extremely low for

spectroscopic reasons.27,28 This suggests use of a lower-chlorinated PCDD as a surrogate.

These congeners are normally not measured because they do not, by definition, contribute to

the TEQ.

Future research on the formation mechanisms of PCDD and PCDF is needed to

provide information on the relationship between observed low-chlorinated homologue yields

and the tetra- through octa-CDD and tetra- through octa-CDF compounds that comprise

TEQ. These tests should be conducted in a well-controlled, alterable parameter, test facility to

enable a wide range of conditions to be simulated with certainty. Establishment of these

correlations will indicate how well specific, lower-chlorinated dibenzodioxins and

dibenzofurans serve as surrogates for TEQ and will provide a potential opportunity for use of

Jet-REMPI techniques in on-line monitoring. A technically easier application for Jet-REMPI

monitoring is as a research tool for studying fundamental questions relating to PCDD and

41

PCDF formation. By studying correlations between low-chlorinated compounds and TEQ we

hope to find surrogates that might be more easily monitored by Jet-REMPI.

SYSTEM COST/COMPLEXITY

The laboratory prototoype Jet-REMPI system used in these experiments cost

approximately $350K for components. This system has considerable flexibiliy necessary for

laboratory and field experiments, e.g. laser tunability over a broad wavelength region. This

capability would not be duplicated in production prototypes, which would be expected to cost

less than $200K for components, and would become cheaper still if only a few specific

molecules were to be measured.

Although the Jet-REMPI system is sophisticated, we believe the complexity can be

significantly reduced as performance requirements are simplified. Our goal is to eventually

build a “black box” that can be operated without any technical background in spectroscopy.

42

CHAPTER 9. WHAT NEXT?

A minimum detection limit of 30 ng/m3 is many orders of magnitude more sensitive than

other CEM techniques, but still four to seven orders of magnitude less sensitive than what is

ideally needed to directly measure the current proposed emissions standard of 0.2 ng/dscm

Toxic Equivalency (TEQ). If a CEM were to directly measure TEQ, we would ideally need

detection limits between 0.001 ng/m3 (for the most toxic congener (2,3,7,8 TCDD)) to 1.0

ng/m3 (for the least toxic octo- DD), depending on the Toxic Equivalency Factor (TEF) of

each of the 17 TEQ congeners.

No doubt, our ability to develop a dioxin CEM thus rests on our ability to improve

sensitivity. From this test, we identified several possible improvements, such as insuring

sample cooling in the expansion nozzle. The current nozzle may not have cooled adequately

because of the increased heating in the preceding sample line necessary to avoid condensation

of the heavier chlorinated compounds. The problem can be avoided by redesigning the inlet

valve assembly. Other hardware and software modifications can probably lead to a total of

two orders of magnitude improvement in Jet-REMPI detection limits.

The next step would be to design a sample pre-concentrator and a particle desorber /

separator. A sample preconcentrator can concentrate the sample as much as is needed to

make the measurement. A particle desorber, placed upstream of the sample concentrator, will

desorb dioxins from particles. The separator will separate out the particles, sending only the

vapor-phase dioxins to the analyzer.

With these enhancements, it may be possible to measure dioxins from a combustion-type

environment, in near real time, at sub-ng/m3 concentrations. To achieve this, we believe,

would require a several million dollar development program. Indeed, simply developing the

REMPI capability for all 17 congeners in the TEQ (necessary for an “ideal” direct TEQ

monitor) would take thousands of hours in a research laboratory. Eventually, this can and

probably will be achived, but in the meantime, there are more practical applications for a

dioxin CEM that can be accomplished in the near-tern for far less money.

43

Even with little or no sample pre-concentration, the next-generation REMPI system

capable of measuring vapor-phase 2,3,7,8 TCDD in low ng/m3 levels would be a remarkable

achievement. Such an instrument would be very useful in research laboratories studying dioxin

formation and control. These laboratory experiments for studying dioxin formation and

control usually operate at total dioxin concentrations around 100 to 1000 ng/m3 total.

Depending on how many of the 210 total congeners are present, detection limits around 5 to

10 ng/m3 for several congener of interest would be more than sufficient.

By combining data from carefully controlled experiments designed to isolate dioxin

formation mechanisms, we can begin to identify surrogates that indicate the presence or

imminent formation of dioxins. Hopefully these surrogates will be much easier to measure

than the full suite of 17 congeners in the TEQ, thereby allowing a much more usable

compliance instrument. In addition, this mechanistic information will lead to control

strategies to prevent dioxin formation in the first place.

CHAPTER 10. CONCLUSIONS

This test program has for the first time provided a measured detection limit for a DCDD

species and obtained the first known spectra for TrCDD and TCDD species. The detection

limit of DCDD (30 ng/dscm) is too high to make use of Jet-REMPI as a compliance CEM

but we reasonably anticipate two orders of magnitude improvement in sensitivity. With less

stringent sensitivity limits, Jet-REMPI will be applicable to combustion process control and

research studies on PCDD and PCDF formation. As research efforts improve our

understanding of PCDD and PCDF formation, it is likely that correlations between indicators

or precursors, including the lower chlorinated species measured in this work, will enable us to

predict TEQ values from these more REMPI-measurable species. In this manner, a

PCDD/PCDF compliance CEM can be developed that will derive continuous measurements

from correlative, rather than direct, measurements.

44

REFERENCES

1. Ballschmiter, Bacher, Dioxine, VCH, Weinheim, Germany, 1996.

2. Fred Chanania, EPA Office of Solid Waste, personal communication, 2/97.

3. H. Oser, R. Thanner, H.H. Grotheer, B. Gullett, N. Bergan French, D. Natschke, “DLRJet-REMPI as a Continuous Emissions Monitor: Measurements of ChlorinatedDibenzodioxins”, 1997 Int. Conf. On Incineration and Thermal Treatment Technologies.May, 1997. To be published.

4. U. Boesl, R. Zimmermann, C. Weickhardt, D. Lenoir, K.-W- Schramm, A. Kettrup, E.W.Schlag, Chemosphere 29, 1429 (1994)

5. E. A. Rohlfing, 22nd Symp. (Int.) on Combustion, The Combustion Institute, 1843 (1988)

6. B.A. Williams, T.N. Tanada, and T.A. Cool, 24th Symp. (Int.) on Combust., TheCombustion Institute, 1587 (1992)

7. Grotheer, Oser, Thanner, Patent No: DE-4441972

8. H. Oser, R. Thanner, and H.H. Grotheer, 1996 Int. Conf. on Incineration and ThermalTreatment Technologies. Proceedings p. 387-392

9. H. Oser, R. Thanner, and H.H. Grotheer, 8th Int. Symp. on Transport Phenomena inCombustion. San Francisco, July 1995, Proceedings, Vol. 2, pp.1646-56.

10. H. Oser, R. Thanner, and H.H. Grotheer, European Symposium on Optics forEnvironmental and Public Safety, Munich, June 1995. SPIE 2504,15 (1995)

11. H. Oser, R. Thanner, and H.H. Grotheer, Fourth International Congress on ToxicCombustion Byproducts, June 1995. Comb. Sci. and Tech. 116, 567 (1996)

12. R. Zimmermann, U. Boesl, D. Lenoir, A. Kettrup, Th.L. Grebner, H.J. Neusser, Int. J.Mass Spectr. and Ion Phys. 145, 97 (1995)

13. R. Zimmermann, D. Lenoir, A. Kettrup, H. Nagel, U. Boesl, 26th Symp. (Int.) onCombustion, The Combustion Institute (1996), in press

14. A. E. O'Keeffe, G. C. Ortman, Anal. Chem. 38, 760 (1966)

15. A. P. Altshuller, I.R. Cohen, Anal. Chem. 32, 802 (1960)

45

16. H. Oser, R. Thanner, and H.H. Grotheer, 1996 Int. Conf. on Incineration and ThermalTreatment Technologies. Proceedings p. 387-392.

17. R. Zimmermann, D. Lenoir, A. Kettrup, H. Nagel, and U. Boesl, 26th Symp. (Int.) onCombustion, Naples, July 1996. In Press

18. R. Zimmermann, U. Boesl, D. Lenoir, A. Kettrup, Th.L.Grebner, and H.J. Neusser, Int. J.Mass Spectr. and Ion Phys. 145, 97 (1995)

19. C. Weickhardt, R. Zimmermann, U. Boesl, and E. W. Schlag, Rapid Comm. Mass. Spectr.7, 183 (1993)

20. C. Weickhardt, R. Zimmermann, K.-W. Schramm, U. Boesl, and E.W. Schlag, RapidComm. Mass. Spectr. 8, 381 (1994)

21. B.F. Rordorf, Thermochemica Acta 112, 117 (1987)

22. R. Zimmermann, U. Boesl, D. Lenoir, A. Kettrup, Th.L. Grebner, H.J. Neusser, Int. J.Mass Spectr. and Ion Phys. 145, 97 (1995)

23. R. Addink, K. Olie, Env. Sci. and Tech. 29, 1425 (1995)

24. C. Weickhardt, R. Zimmermann, U. Boesl, and E. W. Schlag, Rapid Comm. Mass.Spectr.7, 183 (1993)

26. C. Weickhardt, R. Zimmermann, K.-W. Schramm, U. Boesl, E.W. Schlag, Rapid Comm.Mass. Spectr. 8, 381 (1994)

27. ARGUS - Arbeitsgruppe Umweltstatistik: Koordination, Erfassung und Auswertung vonDioxinmessungen an Abfallverbrennungsanlagen. TU Berlin, April 1991

28. H. Oser, R. Thanner, and H.H. Grotheer, 1996 Int. Conf. on Incineration and ThermalTreatment Technologies. Proceedings p. 387-392.

29. R. Zimmermann, U. Boesl, D. Lenoir, A. Kettrup, Th.L.Grebner, and H.J. Neusser, Int.J. Mass Spectr. and Ion Phys. 145, 97 (1995)