<>EPA
                           United States
                           Environmental Protection
                           Agency
EPA/540/SR-93/516
July 1993
                           SUPERFUND INNOVATIVE
                           TECHNOLOGY EVALUATION
                            Emerging Technology
                            Summary

                            Destruction of Organic
                            Contaminants in Air Using
                            Advanced  Ultraviolet Flashlamps
                             This summary describes a new pro-
                            cess for photo-oxidation of volatile or-
                            ganic compounds (VOCs) in air using
                            an  advanced  ultraviolet (UV) source,
                            and a  pulsed xenon flashlamp. The
                            flashlamps have greater output at 200
                            to 250 nm than medium-pressure mer-
                            cury lamps at the same power and,
                            therefore, cause much more rapid di-
                            rect photolysis of VOCs, including me-
                            thylene chloride (CH2CI2), chloroform
                            (CHCI3), carbon tetrachloride (CCg, 1,2-
                            dichloroethane  (1,2-DCA),  1,1,1-
                            trichloroethane (TCA), Freon 113, and
                            benzene. The observation of quantum
                            yields greater than unity indicate the
                            involvement of chain reactions for
                            trichloroethene (TCE), perchloroethene
                            (PCE), 1,1-dichloroethene (DCE), CHCI3,
                            and CH2CI2.
                             TCE was examined more closely be-
                            cause of its widespread occurrence and
                            very high destruction rate. Two full-
                            scale air emission control systems for
                            TCE were constructed at  Purus and
                            tested at a Lawrence Livermore National
                            Laboratory (LLNL) Superfund site. The
                            systems were operated at flash frequen-
                            cies of 1 to 30 Hz, temperatures be-
                            tween 33 and 60  °C, flows up to 300
standard cubic feet per minute (scfm),
and 100 scfm, at concentrations up to
260 part per million per volume (ppmv)
and 10,600 ppmv of TCE, respectively.
Residence times ranged from 5 to 75
sees. In all cases, except at the lowest
flash frequency, greater than 99% re-
moval of TCE  was observed.  Careful
attention  was paid to product forma-
tion and mass balances. The main ini-
tial photo-oxidation product of TCE was
dichloroacetyl chloride (DCAC), which
upon further photolysis was converted
in part to dichlorocarbonyl (phosgene
or DCC) and possibly formyl chloride,
and ultimately to HCI and CO2. Further
treatment of photo-oxidation products
is recommended  for full-scale opera-
tion.
  This Project Summary was developed
by EPA's Risk Reduction Engineering
Laboratory, Cincinnati, OH, to announce
key findings of the research project
that is fully documented in a separate
report of the same title (see Project
Report ordering information at back).

Introduction and Background
  Many environmental remediation sites
are polluted with  volatile organic  com-
                                                                       Printed on Recycled Paper

-------
pounds (VOCs). Some of these sites are
amenable to remediation by vacuum-in-
duced soil  venting and  groundwater  air
stripping methods. VOC air emission con-
trols for restoration activities, however, are
becoming required by regulatory agencies.
We report the application of a pulsed xe-
non lamp (flashlamp) as a UV light source
for the photo-oxidation of some VOCs in
air.
  Previously,  the  only  light source that
was routinely  used for UV  photolyses  on
a large scale was the mercury discharge
lamp  and  doped variations thereof.
Flashlamps discharge  electrical  energy
through  a fill gas  in short  micro second
(/js) pulses and remain  off for relatively
long periods of milliseconds  (ms). Because
flashlamps have   higher  temperatures
C>13,000 K) and pressures than continu-
ous lamps, the  emission  is  shifted  to
shorter wavelengths, Figure 1. The xenon
flashlamp has a maximum output at 230
nm  and a significant output at wavelengths
as low as 200 nm, whereas the mercury
lamp has most of its output at wavelengths
above 250 nm. A shift in peak output from
254 to 230 nm is  significant  because it
corresponds to a  1 to 2 order of magni-
tude increase in   absorptivity of  many
VOCs, thereby greatly enhancing the rates
of direct photolysis.
  Initially, laboratory experiments  were
performed on saturated and unsaturated
            chlorinated hydrocarbons in air to screen
            compounds for treatability. The kinetics of
            photo-oxidation were studied, and appar-
            ent quantum yields were determined for
            the disappearance. Efforts were made to
            characterize the photo-oxidation products
            of TCE.
              A full-scale photoreactor was  built for
            the photo-oxidation of TCE and was tested
            at LLNL Site 300  at Building Complex
            834. This summary contains information
            on the laboratory screening studies and
            performance  data  collected  at the LLNL
            site on the photochemical treatment pro-
            cess for TCE. The  TCE destruction effec-
            tiveness and  the yields of the main oxida-
            tion  products were characterized  under
            various operating  conditions, including
            flowrates  of  100 to 290  cfm and TCE
            concentrations of 30 to 10,000 ppmv.
              These results, combined with toxicologi-
            cal data, were used to estimate the oper-
            ating conditions suitable for  reducing the
            total toxicity  from  TCE  and  its  residual
            products by 99% with the use of UV pho-
            tolysis alone.

            Experimental Methods

            Laboratory Experiments: Pilot-
            Scale Photolyses
              Air mixtures were irradiated in  a 208-L
            steel, cylindrical reactor containing two
            small fans for mixing. A  high intensity, 6-
    15
    10
                               Xe Flashlamp
                               18.6% <300nm
                    Med  Pressure Hg Lamp
                       11.4% <300 nm
      180
                 200
220         240

     Wavelength (nm)
                                                    260
                                                                280
                                                                            300
Figure 1.  Emission spectrum fora mercury lamp versus a 6-in. Xenon flashlamp. (Both lamp
         outputs normalized to 3675 W input).
in. xenon  flashlamp was inserted in the
middle of the reactor through its side. All
photolyses were performed at atmospheric
pressure, and the gas temperature ranged
from 300 K to approximately 340 K.
  Known volumes of  reagents were in-
jected into the reactor by syringe, allowed
to mix, photolyzed, and analyzed by gas
chromatography with  photoionization or
electron capture detection. No reaction was
observed in laboratory  light or in the reac-
tor with the lamp off. CO2 measurements
were made in one run using a Horiba PIR-
2000 CO2 monitor.

Field Measurements:
Photoreactors
  In the field studies at LLNL, the process
stream was pumped from the extraction
wells, through a heat exchanger to cool,
and sent into two types of photoreactors.
The Air-2  reactor (not  shown) is  a large
steel cylinder, 4 ft in diameter by 8 ft in
length, with a  volume of 101  ft3. Four
xenon lamps are distributed about the cen-
ter of the cylinder and point radially in-
ward. The process stream flows from one
end  of the  cylinder to  the other. Air-3
(Figure 2) is a Purus-patented reactor con-
sisting  of four disc-shaped stainless steel
chambers. Each chamber is 42 in. in di-
ameter by 6.1 in. high with a volume of
4.1 ft3  exposed to the light source. The
lamps in Air-3 are positioned in the center
axis of each chamber. The  process flow
enters the bottom of each chamber, around
a deflection plate, in towards the lamp,
and then out the top.  The  chambers in
Air-3 are configured so  that the flow is
split through two sets of  chambers in se-
ries.  Some experiments  were conducted
with an enriched TCE  air stream by gas
injection from a pressurized liquid, TCE-
filled,  stainless-steel bubbler before the
reaction chambers.

Analyses
  Atypical sampling session involved set-
ting the process  flowrate, adjusting the
TCE concentration, and alternately taking
at least  three input  and  output TCE
samples while photolyzing. The impinger
samples were collected during the moni-
toring of the TCE concentration and were
connected to the  reactor port by 0.25-in.
i.d. Teflon tubing.
  TCE was analyzed by gas chromatog-
raphy with a 30-m J&W 624 capillary col-
umn  and  photoionization detection.
Samples  were drawn  by a  gas-tight sy-
ringe at septum-sealed sample ports where
the  process flow entered or exited the
reactor. TCE standards were prepared in
volume-calibrated, glass sampling  bulbs

-------
Xe flashlamp
assembly
Cylindrical 1
photo-reactor J^
1 .V
Suprasil Y | f~~l
lamp hniJtintj T ^'r 1
Yonn/i 	 11M^ 1


1 8 1
1 — ' — 1 1 Valve 1 	 ' 1

iSo
A^\
ports
XXX Va/ve
Process flo w CX! ' 	
output
D_
-J 	 1

1 ' ' J
s
1 /6fc /
Va/ve J
A
TCE oufpuf
septum port
nput
m port
<
l?O Process flow -^ 	
/npuf
    Figure 2.  Schematic ofAir-3 Photochemical Reactor.
by injecting liquid TCE into the bulbs; the
detection limit was  approximately  0.01
ppmv.
  DCC was analyzed by EPA Method TO-
6: 3 L of gas were collected in a series of
impingers filled with 30 ml  of a 2% v/v
aniline/toluene solution, which reacts with
DCC  to give carbanilide. After solvent
evaporation  and take-up  in acetonitrile,
carbanilide was determined by high pres-
sure liquid chromatography  (HPLC) with
an octadecylsilyl column  and  UV detec-
tion at 254 nm. The  detection limit was
0.02 ppmv DCC, and the  average recov-
ery of carbanilide, based  on standard
samples run through the blowdown proce-
dure,  was 106 ± 19% (95% confidence
interval).
  DCAC was  determined  as methyl
dichloroacetate  after collection  and
derivatization in methanol impingers.
Analysis involved gas chromatography with
electrolytic conductivity detection. The de-
tection limit was approximately 0.05 ppmv
DCAC.
  Total HCI and hydrolyzable organic chlo-
rine were determined with the  use of wa-
ter impingers. The samples were analyzed
by using EPA Method 325.3, a titrimetric
method employing mercury  nitrate to de-
termine the chloride yield.
  In one set of experiments,  volatile or-
ganic compounds were determined by EPA
Method TO-14 with the use  of evacuated
stainless steel  (SUMMA) canisters and
were analyzed by Coast to Coast Analyti-
cal Services, San Luis Obispo,  CA. Agree-
ment was reasonable; the Purus analyses
agreed with the standards within 10% and
the Coast to Coast analyses within 25%.

Results and Discussion

Laboratory Experiments:
Photolysis Kinetics and
Quantum Yields
  Table 1  summarizes the results of the
laboratory experiments conducted in the
pilot-scale reactor. CCI4  was used as  an
actinometer assuming it has a disappear-
ance quantum yield of 1.0, based on the
literature data at 214 nm.  The apparent
quantum yields are  averaged over the
wavelength range of overlap of the com-
pound absorbances  and  the  emission
spectrum of the lamp. Benzene had a low
quantum yield, consistent with the ability
of aromatic compounds  to intersystem
cross, fluoresce,  and thermally  decay by
modes that do not result in bond cleav-
age. The first four halogenated compounds
in Table 1 have quantum yields near unity,
indicating that simple C-CI bond cleavage
is highly efficient, as expected by analogy
to CCI4. Nevertheless, these compounds
photolyzed relatively slowly because they
absorb light weakly. Even shorter wave-
lengths than those available from the cur-
 Table 1.  First Order Decay Coefficients and Wavelength-Averaged Disappearance Quantum
         Yields with a 2.756 kW Xenon Lamp
Compound
CCI4
CCIfCCIF2
Benzene
CH2CICH2CI
CFCI3
CCI3CH3
CH2CI2
CHCI3
1,1 -DCE
PCE
TCE
TCE + ethene
k(sec-1)"
0.00432
0.00093
0.0019
0.0024
0.0036
0.0041
0.0070'
0.0366'
1.24'
1.7'
5.5'
0.075
£ 'A,e\ 4
I \^voc
1.0
5.09
0.067
A/.D.#
1.18
0.79
4.60
1.79
0.0389
0.0134
0.0236
0.0236
kvoc
kcci4
1.0
0.22
0.44
N.D.
0.84
0.94
1.62
8.47
287
394
1300
17
Apparent
1.0
1.1
0.029
N.D.
0.99
0.74
7.5
15
11
5.3
31
0.4
 ' Initial rate constants are taken for non-log-linear curves.
 # N.D.= not determined.

-------
rent Purus flashlamps are  needed for a
commercially viable, direct, photolysis pro-
cess for these compounds.
  The rate constants highlighted with an
asterisk in Table 1 exhibited non-first or-
der behavior, i.e., after about an order of
magnitude loss, the rate constant started
to decline. This same set of compounds
photolyzed  more rapidly (especially  the
chloroolefins) and  had quantum yields
greater than one. These results all point
to the occurrence of a chlorine atom chain
reaction in these cases, and this conclu-
sion is  verified by literature studies using
chlorine gas to initiate chlorine atom reac-
tions.  Furthermore, when ethene  was
added  in  large excess to TCE, the TCE
loss became first-order and much slower
because ethene is an effective scavenger
of chlorine atoms.

Chain Photo-oxidation
Mechanism for TCE
  Below we show that DCAC is the main
initial photo-oxidation product of TCE in
air. The following chain mechanism is con-
sistent  with  all the  product and kinetic
information:
     HCIC=CCI2+ hv -» HCIC=CO + Cl»
       Cl« + HCIC=CCI2 -» HCI2C-CCI2'
     HCI2C-CCI2' + 02 -> HCI2C-CCI2OO'
   2HCLC-CCLOO -> 2HCLC-CCLO +  O.
    HCI2C-CCI20
HCI2C-CCIO
  (DCAC)
  Similar pathways can be written for the
other compounds that undergo chain de-
composition. A common  feature of the
mechanisms for chain reacting compounds
is that a Cl»  atom  reacts with  them to
generate a carbon-centered  radical that
has chlorine substitution,  which  can ulti-
mately cleave a Cl« atom and propagate
the chain. CCI4 and the freons photolyze
to the  same type of radical but cannot
form a chain because Cl» atoms do not
react with C-CI or C-F bonds.
  Experiments with  olefin mixtures dem-
onstrated that co-contaminants can cause
both sensitization  and inhibition of  pho-
tolysis. Thus,  TCE and PCE can be ex-
pected to sensitize the photo-oxidation of
the  DCE isomers  and vinyl chloride be-
cause the former are better light absorb-
ers and the latter react with Cl» atom more
readily. However, addition of chloroolefins
will not sensitize the photoreactions of the
chain  promoters CHCI3 and CH CI2 be-
cause the chloroolefins enhance the rate
of  Cl«  atom scavenging  as  well as Cl»
production.

Field  Experiments
  Figure 3 and Table 2 give TCE disap-
pearance data and product data  from ex-
periments in Air-3 at LLNL Site 300. A
larger data set over a broader range of
conditions in Air-2 gave similar results and
conclusions.
TCE Removal Efficiency
  Over the range  of experimental vari-
ables covered, TCE was photo-oxidized
to >99% conversion except at the lowest
flash frequencies and  number of lamps.
This conclusion held true even at TCE
concentrations up  to 10,000 ppmv and
flowrates up to 300 scfm (data not shown).
At the highest concentrations and optimal
conditions, conversions of 99.9996% were
achieved. Estimation of the TCE destruc-
tion  efficiency was often limited by the
0.01 ppmv detection limit of the gas chro-
matographic method.

TCE Oxidation  Products
  Figure 3 shows the  evolution  of prod-
ucts during the photolysis of TCE in Air-3
at the field site. The data are taken from
Table 2 and  converted to an exposure
time normalized to the standard condi-
tions of four  lamps operating at 30 Hz.
DCAC is the principal initial organic prod-
uct, formed in > 85% yield from the lost
TCE. With further exposure, DCAC was
consumed and  formed about 20%  DCC
(Figure 3) and  about  2% trichloroacetic
acid (TCAC; data not shown). The DCC
and TCAC yields did not equal the DCAC
loss,  presumably  because  these com-
pounds also photolyze  or because other
DCAC photolysis pathways  exist that  do
not form DCC or TCAC. The unidentified
                                                      1.5           2      4

                                                  Irradiation Time (seconds)
Figure 3.  Product yields from the photolysis of TCE in Air-3.

-------
Table 2.  Summary of Field Results with the Air-3 Photoreactor '
Freq.
(Hz)
30
30
30
30
15
15
5
5
1
1
No. of
3.7-kW
Lamps
4
4
4
2
4
2
4
2
4
2
• Res.
Time
(sec)
9.6
10.1
10.4
4.6
10.1
4.8
10.4
4.8
9.3
4.8
TCE
Destruction
(%)
>99.99
>99.99
>99.99
99.92
>99.99
>99.99
>99.99
>99.99
99.16
86.57
Mole%
DCC*
N.D.s
19.6
26.0
17.3
21.3
12.4
8.3
9.3
12.3
6.9
Mole %
DCAC*
25.8
24.4
34.6
53.7
N.D.
64.5
72.2
75.2
81.8
86.2
Mole %
Cl-
61.6
89.9
91.4
55.3
68.2
43.2
41.9
38.6
35.8
35.8
Chlorine
Balance
(Mole%)
78.8
106.2
114.5
91.1
N.D.
86.2
90.0
88.8
90.3
93.3
* Flowrate = 100 dm, initial [TCE] = 100 ppmv.
t Dichlorocarbonyl (phosgene).
* Dichloroacetyl chloride.
* Not detectable.
carbon compounds must contain either no
chlorine or only hydrolyzable forms of chlo-
rine, because a good  mass balance on
chlorine is obtained (93 ±23 %). The total
chlorine recovery was defined as:
     Fraction  chlorine recovered =
     (moles chloride  + 2 x moles
     DCAC) / (3 x moles TCE lost)
because the measured chloride was the
sum of gaseous HCI  and  hydrolyzable or-
ganic chlorine, such as DCC, formyl chlo-
ride, and  the carbonyl chloride of  DCAC.
Thus formyl  chloride is a likely unidenti-
fied form of both chlorine and carbon. TO-
14 whole air sampling  tests verified that
the concentrations  of nonhydrolyzable
chlorine compounds  are  low: the  chloro-
form yield was 0.65%  of the TCE input;
the carbon tetrachloride yield, 0.15%; and
the methylene chloride yield,  0.05%.  A
mass balance for carbon cannot be deter-
mined  based on the measurements per-
formed in this study; however, a prelimi-
nary CO2 measurement  indicated that most
of the  carbon  is  converted to CO2 with
enough light exposure.

Estimation of Parameters to
Achieve Recommended
Treatment Levels at LLNL Site
300
  Because of the formation of toxic prod-
ucts, the  efficacy of treatment at LLNL
Site 300  cannot be  measured simply  in
terms of TCE removal. The  major product
DCAC has approximately  40 times greater
long-term toxicity than TCE,  and DCC
(phosgene)  exhibits acute toxicity. Thus,
these products would need to be removed
by further photolysis or other post-treat-
ment before emission to the atmosphere.
A risk assessment indicated that the most
one can emit is 0.025% DCAC or 0.45%
DCC of the initial TCE concentration  to
reach the goal of 99% reduction  in initial
toxicity. Our data indicate that a  flowrate
of 13 to 20 cfm in Air-2 or Air-3 with four
3.7-kW lamps can achieve the  desired
DCAC reduction. It is uncertain if the DCC
concentration would be low enough when
this DCAC treatment goal is reached. DCC
could  easily  be removed with   a water
scrubber, where it would would rapidly
hydrolyze  to CO2 and HCI. Traces  of
DCAC, however, would also hydrolyze to
dichloroacetic acid (DCAA) and HCI, and
with reasonably low  water flows, the re-
sidual DCAA would  still  be at  least an
order  of magnitude above the proposed
drinking water limit of 0.2 ppb. Use  of a
relatively  dry scrubber, such as slaked
lime is a possibility because it would trap
both DCC and DCAC. However, DCAA is
likely  to  leach out when  the  lime  is
landfilled. Promising  approaches include
using very small flows of water and treat-
ing them by incineration or other thermal
processes.

Conclusions

Kinetics of VOC Photo-
oxidations
  The  low-wavelength emission of the
pulsed xenon lamps allows direct  photoly-
sis of many VOCs, particularly chlorinated
compounds and freons, that is  not  pos-
sible with commercial mercury lamps. Nev-
ertheless, light absorption by some VOCs
is still weak enough at 230 nm that either
photosensitization or an even lower-wave-
length source is needed  for the  photoly-
ses to be rapid enough for commercializa-
tion at present. On the other hand,  very
rapid and efficient destruction is observed
for compounds that  undergo chain reac-
tions initiated by light, notably TCE, PCE,
and DCE, in order.

TCE Photo-oxidation Products
  The main product (> 85%)  from  the
chain photo-oxidation of TCE is  DCAC.
Further oxidation of DCAC is approximately
100 times slower than  the photolysis of
TCE and forms DCC in about 20% yield,
TCAC U 2%),  CHCI3 (-0.65%), CCI4
(-0.15%), CH2CI2 (-0.05%), and possibly
formyl chloride and  DCAA. Evidence was
found that the carbon-containing products
are eventually  converted to  CO,,  with
enough exposure.

Estimation Of Process
Parameters for Remediation
  Although  both  full-scale reactors dem-
onstrated very efficient  removal of TCE,
the formation of undesirable intermediates
required that their toxicity be taken  into
consideration. A  reduction in toxicity for
TCE  of 99% requires  that the residual
DCAC concentration be 0.025%  of  the
TCE  input  concentration,  and  the DCC
concentration  must be 0.45% of the TCE
input concentration. The maximum flowrate
that meets  the DCAC  reduction goal at
LLNL using four  3.7-kW lamps  was esti-
mated to be between 13 and 20 cfm. At
this level of treatment, the DCC concen-
tration may still be  excessive and addi-
tional treatment may be needed.  Scrub-
bing  with water  under  these conditions
would rapidly  hydrolyze the DCC  to CO
and HCI and the DCAC to DCAA and HCI.
The accumulation of even a trace of DCAA
may,  however, result in a disposal prob-
lem for the water because the  expected
EPA drinking water  limit for DCAA is so
low (-0.2 ppb).

Recommendations
  Further studies on the use of low-wave-
length lamps for the destruction of VOCs
should be directed at (1) verifying  the ef-
fectiveness  of  dry  or  wet scrubbers to
remove  acidic photo-oxidation  products,
(2) developing thermal or  other methods
for post-treatment of products  such  as
DCAA present in the water after scrub-
bing, and (3) examining the use of shorter-
wavelength  UV lamps or catalysts for pho-
tolysis of a broader range of VOCs. Purus
will examine some  of  these issues to-
gether with  Argonne National Laboratory
in  continued demonstrations  at the  De-
partment of Energy Savannah River site.
  The full report was submitted in fulfill-
ment of Cooperative Agreement No. CR-
818209-01-0 by  Purus, Inc., under  the
sponsorship of the U.S. Environmental Pro-
tection Agency.
                                                                     •&U.S. GOVERNMENT PRINTING OFFICE: 1993 - 750-071/80034

-------

-------

-------
Mark D. Johnson, Werner Haag, and Paul G. BIystone are with Purus, Inc., San
  Hose, CA 95134. Paul F. Daley is with Lawrence Livermore National
  Laboratory, Livermore, CA 94550.
Norma Lewis is the EPA Project Officer (see  below).
The complete report, entitled "Destruction of Organic Contaminants in Air Using
    Advanced Ultraviolet Flashlamps," (Order No. PB93-205383; Cost: $17.50,
    subject to change)  will be available only from:
      National Technical Information Service
      5285 Port Royal Road
      Springfield, VA 22161
      Telephone: 703-487-4650
The EPA Project Officer can be contacted at:
      Risk Reduction Engineering Laboratory
      U. S. Environmental Protection Agency
      Cincinnati, OH 45268
  United States
  Environmental Protection Agency
  Center for Environmental Research Information
  Cincinnati, OH 45268

  Official Business
  Penalty for Private Use
  $300
     BULK RATE
POSTAGES FEES PAID
         EPA
   PERMIT No. G-35
  EPA/540/SR-93/516

-------