rational design of advanced oxidation processes using ... · (1st order reaction) 1. plug flow...
TRANSCRIPT
Daisuke Minakata1,
Ke Li2,
John C. Crittenden1,
1 Brook Byers Institute for Sustainable Systems and School of Civil and Environmental EngineeringGeorgia Institute of Technology2 Faculty of Engineering, University of Georgia
GCOE Symposium, July 21, 2011 Kyoto Univ. Kyoto, Japan.
Rational Design of Advanced Oxidation Processes using Computational Chemistry
1
Emerging contaminants: detection of trace organic contaminants and uncertain human health and ecological effects
Concern about emerging contaminants necessitate a task to assess their removal efficiency during water treatment
Advanced Oxidation Processes (AOPs) are attractive and promising water treatment technologies because of the capability of mineralization of organic compounds.
AOPs may be used to control the emerging contaminants.
Non-selectivity of HO• and radical chain reactions make AOPs complex processes as well as diversity and complexity of structure of a large number of emerging contaminants.
Introduction to Advanced Oxidation Processes
2
Modeling AOPs
1. Simplified Pseudo-Steady-State Model
2. AdoxTM – AOP simulation software
3. Conventional Approach for Kinetic Model
4. Computer-based Kinetic Model
3
3. H2O2 is added to a water containing O3, i.e. [O3]0 is known.
4
2. H2O2 and O3 are added simultaneously
3 3O O
ss, 09 2 10 2 2 11 3 12 130 0 00 0
HOHO H O HCO R NOM
LK a P H
k k k k k
UV/H2O2 H2O2/O3 Simplified pseudo-steady-state model
R 12 ss, 0HOk k
H O2 2pH p
1 2 2 30 resss, 0
11 3 12 130 00
2 H O 10 OHO
HCO R NOM
Kk
k k k
2 2 2 2
-A
H O U-V H O
ss,0 -
10 2 2 0 11 3 0 12 0 13 0
2 P f (1- e )[HO ] =
k [H O ] + k [HCO ] + k [R] + k [NOM]
g
1. UV/H2O2
R R Rr k
5
Flow Models: Ideal flow at steady state (1st order reaction)
1. Plug Flow Reactor (PFR)
k
0C C e
2. Completely Mixed Flow Reactor (CMFR)
0CC
(1 k)
3. Tank-in-series (TIS)
0
n
CC
1 kn
1/ n
0C nQV 1
C k
0Q C CV
r
0CQV ln
k C
6
Flow Models: Non-ideal flow
4. Dispersed flow model closed system
A
2 2Ao
1 vL4a exp
2 EC
a vL a vLC (1 a exp (1 a exp) )2 E 2 E
a 1 4k (E / vL)
7
Group Contribution Method -Hypothesis-
Essence is same as the GCM for the gaseous phase Observed experimental reaction rate constant for a
given organic compound is the combined rate of all elementary reactions involving HO•, which can be estimated using Arrhenius kinetic expression.
The Ea consists of two parts: (1) Base part from main reaction mechanisms (i.e.,
H-atom abstraction, HO• addition to alkenes and aromatic compounds and HO• interaction with S, N, or P-atom-containing compounds).
(2) Functional group contribution partially from neighboring and/or next nearest neighboring functional group.
= aE
RTk Ae
Minakata, Li, Westerhoff, and Crittenden. ES&T 2009, 43, 6220-6227
8
1.0E+05
1.0E+06
1.0E+07
1.0E+08
1.0E+09
1.0E+10
1.0E+11
1.0E+05 1.0E+06 1.0E+07 1.0E+08 1.0E+09 1.0E+10 1.0E+11
kca
l (M
-1s-1
)
k exp (M-1s-1)
Ideal fit
kexp/kcal = 0.5 and 2.0
H-atom abstraction (Calibrated)
HO addition to alkene (Calibrated)
HO interaction with N,S,P (Calibrated)
HO addition to aromatic compounds
(Calibrated)
H-atom abstraction (Prediction)
HO interaction with N, S, P (Prediction)
HO addition to aromatic compounds
(Prediction)
• 257 rate constants (83% of 310 data) from calibration were within 0.5 ≤ kcalc/kexp ≤ 2.0.
• The sample deviation was 0.92. Degree of Freedom was 164• 76 rate constants (62% of 124) from prediction were within 0.5 ≤
kcalc/kexp ≤ 2.0.
Minakata, Li, Westerhoff, and Crittenden. ES&T 2009, 43, 6220-6227
Modeling AOPs
1. Simplified Pseudo-Steady-State Model
2. AdoxTM – AOP simulation software
3. Conventional Approach for Kinetic Model
4. Computer-based Kinetic Model
9
• Reactor Options
– Completely mixed reactor (CMBR)
– Completely flow type reactor (CMFR) w/o TIS
– Plug flow reactor
– Non-ideal Mixing
• Main Process Operational Variables :
– Initial hydrogen peroxide concentration
– Incident UV-light intensity
• Water quality parameters:
– TOC concentration
– pH
– Concentrations of target compounds
– Byproducts can be included if they are known
Features of Current Version AdOxTM 1.0
10Crittenden et al., 1999. Wat. Res. 33, 2315-2328
Features of Current Version AdOxTM 1.0 for UV/H2O2
H2O2 + hν → 2HO•
H2O2 = quantum yield of H2O2 (=0.5) I0 = incident light intensity, einstein cm-2 sec-1 A=2.303b(εH2O2CH2O2+εRCR + εSCS + εHO2-CHO2-) b=pathlength, cm fH2O2= 2.303 b (εH2O2CH2O2 + εHO2-CHO2-)/A
H2O2/HO2- + HO H2O/OH- + HO2 2.7×107, 7.5×109
H2O2 + HO2/O2- HO + H2O/OH- + O2 3.0, 0.13
HO + HO H2O2 5.5×109
HO + HO2/O2- H2O/OH- + O2 6.6×109, 7.0×109
HO2 + HO2/O2- H2O2/HO2
- + O2 8.3×105, 9.7×107
R + hv Products
R + HO Products kMtBE=1.6×109, ktBA=6.0×108
HO + CO32-/HCO3
- CO3- + OH-/H2O 3.9×108, 8.5×106
HO + NOM Products 2.4×104 (mgC/L)-1s-1
NOM + hv Products 11
Concerns about UV/H2O2
• H2O2 has low UV light absorption
• Need to increase H2O2 concentration to lower EE/O
• High residual H2O2 will require treatment to reduce H2O2 such as chlorine
12
EE O
log i f
P t
V C C
Simulation Results (LPUV/H2O2) – NAIX + Dealkalization -
0.00
0.50
1.00
1.50
2.00
2.50
3.00
3.50
4.00
0.0
2.0
4.0
6.0
8.0
10.0
12.0
14.0
16.0
5 10 15 20
EE
O (k
Wh
/kg
al-
ord
er)
H2O
2re
sid
ua
l co
nc.
(m
g/L
)
H2O2 Dosage (mg/L)
Residual of H2O2 MtBE tBA
• 10 mg/L of H2O2 dosage would be the better choice although the optimum dosage of H2O2 is over 20 mg/L. This is because the cost required for over 20 mg/L of H2O2 dose overweigh the energy cost. • At 10 mg/L of H2O2 dosage, UV dose was calculated as 2,100 mJ/cm2. 13Li; Hokanson; Crittenden; Trussell; Minakata. Wat. Res. 2008, 43, 5045-5053
O3/H2O2 processfor MtBE removal and bromate mitigation
• Advanced oxidation process: O3/H2O2 process:
– HO2- + O3 → HO• + O2•- + O2 k = 2.2×106 M-1s-1
– H2O2 + O3 → O2 + H2O k = 0.0065 M-1s-1
– produce more HO• at higher pH due to pKa = 11.6 of H2O2
– H2O2 also scavenges HO•
• H2O2 + HO• → HO2• + H2O k = 2.7×107 M-1s
• optimum ratio: [H2O2]/[O3] = 0.5 mole/mole
• H2O2 addition keeps the O3 concentration low.
• In the presence of Br-, H2O2/HO2- reacts with HOBr/OBr- to produce Br- resulting in BrO3- formation. In addition, there is significant contribution of the reaction of O3 with Br• in regard with BrO3- formation (von Gunten and Oliveras, 1998).
14
Simulation – O3 with H2O2 process–
Simulation parameters
• MtBE = 300 ug/L
• pH = 7.6
• NOM = 1.4 mg/L
• Alkalinity = 318 mg/L as CaCO3
• initial O3 = 3 mg/L
• molar ratio: [H2O2]/[O3] = 0.5 (optimum), 0.75, 1.0
• H2O2 = 1.06, 1.59, 2.12 mg/L
• Br- = 300 ug/L
• Completely Mixed Batch Reactor (CMBR)
15
Simulation results – O3 with H2O2 process–
0.0
0.5
1.0
1.5
2.0
2.5
3.0
[O3]
(mg
/L)
[H2O2]/[O3]=0.5
[H2O2]/[O3]=0.75
[H2O2]/[O3]=1.0
0.0
0.5
1.0
1.5
2.0
0 20 40 60 80 100 120
[H2O
2]
(mg
/L)
Time (sec)
0
5
10
15
20
25
30
0
50
100
150
200
250
300
[tB
A]
(ug/
L)
[MtB
E]
(ug
/L)
MtBE tBA
0
1
2
3
4
5
6
7
0 20 40 60 80 100 120
[BrO
3-]
(u
g/L
)
Time (sec)16
17
Comparison of Adox (UV/H2O2) to Simplified-pseudo steady state model
The impact of H2O2 dosage on EE/O for trichloroethylene (TCE) destruction using H2O2/UV process (operating conditions: [TCE]0 = 100 g/L, Alkalinity = 100 mg/L CaCO3, [NOM] = 0 mg/L, UV Light Intensity = 1.04x10-6 einstein L-1 s-1 at 254 nm, reactor size = 70 L with 15.8 cm of the effective path length and the total lamp power is 160 W (assuming 20 % efficiency)
Modeling AOPs
1. Simplified Pseudo-Steady-State Model
2. AdoxTM – AOP simulation software
3. Conventional Approach for Kinetic Model
4. Computer-based Kinetic Model
18
Conventional Approach for Kinetic Model in AOP
1.One “hot” parent organic chemical contaminant
2.Conduct batch experiments to measure byproducts using GC/MS etc.
3.Obtain rate constant/Measure rate constant (e.g., pulse radiolysis)
4.Build kinetic model based on both experimental findings and literature-reported rate constants. Unknown rate constants are usually obtained by fitting with experimental data.
19
• UV irradiation ranging from 200 nm to 300 nm inan interval of 1 nm
• Molar extinction coefficients of TCE, H2O2 and 4types of main byproducts are measured andsimulated in an interval of 1 nm from 200 to 300nm
• pH change of the system
• Dissociation of 8 organic/inorganic acids
• 22 literature reported rate constants are used in the model
The model is a kinetic model and considers:
TCE UV/H2O2 Modeling
20
Li, Stefan, and Crittenden, 2004, ES&T 38, 6685-6693Li, Stefan, and Crittenden, 2007, ES&T 41, 1696-1703
0.E+00
2.E-03
4.E-03
6.E-03
8.E-03
1.E-02
1.E-02
0 5 10 15 20 25 30 35
Time (min)
Co
nc
en
tra
tio
n (
mo
l/L
)
exp
mod
0.E+00
2.E-04
4.E-04
6.E-04
8.E-04
1.E-03
1.E-03
0 10 20 30 40
Time (min)
Co
nc
en
tra
tio
n (
mo
l/L
)
mod
exp
0.E+00
1.E-06
2.E-06
3.E-06
4.E-06
5.E-06
6.E-06
7.E-06
8.E-06
9.E-06
0 10 20 30 40
Time (min)
Co
nc
en
tra
tio
n (
mo
l/L
)
exp
mod
0.E+00
2.E-06
4.E-06
6.E-06
8.E-06
1.E-05
1.E-05
0 10 20 30 40
Time (min)
Co
nc
en
tra
tio
n (
mo
l/L
)
mod
exp
Comparison of Modeled and Experimental Concentration Profiles 1/2
TCE
MCADCA
H2O2
TCE UV/H2O2 Modeling
21
Comparison of Modeled and Experimental Concentration Profiles 2/2
0.E+00
1.E-05
2.E-05
3.E-05
4.E-05
5.E-05
6.E-05
7.E-05
8.E-05
9.E-05
1.E-04
0 5 10 15 20 25 30 35
Time (min)
Co
nc
en
tra
tio
n (
mo
l/L
)
exp
mod
Oxalic acid
0.E+00
5.E-05
1.E-04
2.E-04
2.E-04
3.E-04
3.E-04
4.E-04
4.E-04
5.E-04
0 10 20 30
Time (min)
Co
nc
en
tra
tio
n (
mo
l/L
)
exp
mod
Formic acid
0
1
2
3
4
5
6
7
0 10 20 30 40
Time (min)
pH
exp
mod
pH
TCE UV/H2O2 Modeling
22
23
Parent compoundMajor byproducts
(yield 10-30 mole%)
Minor byproducts
(yield 2-5 mole%)
Acetone
*Stefan and Bolton, 1999
acetic, pyruvic, and oxalic acids
pyruvaldehyde
formic and glyoxylic acids
hydroxyacetone, formaldehyde
MtBE
* Stefan et al., 2000
acetone, acetic acid, formaldehyde,
tert-butyl formate (TBF), pyruvic acid,
tert-butyl alcohol (TBA), 2-methoxy-2-
methyl propionaldehyde (MMP), formic,
methyl acetate
hydroxy-iso-butylaldehyde,
hydroxyacetone,
pyruvaldehyde and hydroxy-
iso-butyric, oxalic acid
Dioxane
*Stefan and Bolton, 1998
1,2-ethanediol diformate, formic acid,
oxalic acid, glycolic acid, formaldehyde,
1,2-ethanediol monoformate
methoxyacetic acid glyoxal
TCE
* Stefan and Bolton, 2002formic acid, oxalic acid
dichloroacetic acid,
mono-chloroacetic acid
A few studies examined entire reaction mechanisms
24
Parent compound
H-atom abstraction by HO•or HO• addition
O2 addition
Uni/Bi molecular decay
HydrolysisHO• reactions
Β-scission, 1,2-H shift
Peroxyl radical mechanisms
Carbon centered radical
PeroxylRadical
Oxy radical
Intermediates (aldehydes, alcohols etc.)
Carbon centered radical
Carboxyl acid
CO2/Minerals
*Bolton, J.R.; Carter, S.R. (1994) in Aquatic and Surface Photochemistry, Lewis Publishers, Boca Raton, USA., 467-490.
General Reaction Mechanisms in AOP
The most significant observed by-products are the carboxylic acids, due to
the fact that the second order rate constants for these compounds are
much lower than those for most organics. However, if adequate reaction
time is provided, all by-products (>99% as measured by a TOC mass
balance) are destroyed.
Modeling AOPs
1. Simplified Pseudo-Steady-State Model
2. AdoxTM – AOP simulation software
3. Conventional Approach for Kinetic Model
4. Computer-based Kinetic Model
25
Computer based kinetic model Establish a computer-based first-principle kinetic model
of reactions that are initiated by HO• in aqueous AOPs.
Reaction Pathway Generator (Graph theory)
Reaction Rate Constant Predictor(QSARs, Quantum mechanical calculations)
Ordinary Differential Equations (ODEs) Solver
26
Reaction Pathway Generator
Reaction Rate Constant Predictor
ODE Generator and Solver
ToxicityProfiles
Toxicity Estimator
ConcentrationProfiles
A Computer-Based Kinetic Model in AOPs
Why so important?
We now know the general reaction mechanisms that HO• initiates in the aqueous phase AOPs based on a number of experimental studies.
We have a large number of datasets for reaction rate constants in the aqueous phase AOPs.
Concern about emerging contaminants
Greater than 50 million chemical compounds registered in CAS and more than 40 million chemicals are available
Is it feasible?
We now have robust tool (e.g., computational chemistry) and computational resources
We can make it feasible
27
Reaction Pathway Generator – HO• with CH3CH2OH
Pathway
Running...
CCO
----->
C0(O0(H0)C0(H0H0H0)H0H0)
9 done
initial size is : 2
Generation started, be patient pls.
below is the species list
C(O(H)C(HHH)HH)
|
O*(H)
|
C*(O(H)C(HHH)H)
|
O(HH)
|
O*(C(C(HHH)HH))
|
C*(C(O(H)HH)HH)
|
O*(O(C(O(H)C(HHH)H)))
|
the following is the reaction list
C(O(H)C(HHH)HH) + O*(H) = C*(O(H)C(HHH)H) + O(HH) [HA]
|
C(O(H)C(HHH)HH) + O*(H) = O*(C(C(HHH)HH)) + O(HH) [HA]
|
C(O(H)C(HHH)HH) + O*(H) = C*(C(O(H)HH)HH) + O(HH) [HA]
|
C*(O(H)C(HHH)H) + O(//O) = O*(O(C(O(H)C(HHH)H))) [OA]
|
O*(C(C(HHH)HH)) = C*(O(H)C(HHH)H) [OT]
|
O*(C(C(HHH)HH)) = C(//OHH) + C*(HHH) [BS]
|
C*(C(O(H)HH)HH) + O(//O) = O*(O(C(C(O(H)HH)HH))) [OA]
|
O*(O(C(O(H)C(HHH)H))) = C(//OHC(HHH))+ *OOH [PH]
|
C(//OHH) + H2O = C(O(H)O(H)HH) [HC]
|
C*(HHH) + O(//O) = O*(O(C(HHH))) [OA]
|
2*O*(O(C(C(O(H)HH)HH))) = O*(C(C(O(H)HH)HH)) *2 + O2 [PB3]
|
2*O*(O(C(C(O(H)HH)HH))) = O(C(C(O(H)HH)HH)O(C(C(O(H)HH)HH))) + O2 [PB4]
|
2*O*(O(C(C(O(H)HH)HH))) = C(C(//OH)O(H)HH) *2 + H2O2 [PB2]
|
2*O*(O(C(C(O(H)HH)HH))) = C(C(//OH)O(H)HH) + C(O(H)HHC(O(H)HH)) + O2 [PB1]
|
C(//OHC(HHH)) + H2O = C(O(H)O(H)C(HHH)H) [HC]
Total number of reactions is: 2921
Total number of species is: 243
Species lists
Reaction lists
28
Li and Crittenden, 2009 ES&T 43, 2931-2837
29
6
7
8
9
10
11
6 7 8 9 10 11
log
kH
O•in
aq
eou
s p
hase
(M
-1s-1
)
log k HO• in gaseous phase (M-1 s-1)
alkane
alcohol
ether
ester
aldehyde
Ketone
carboxylic
Nitrile
Alkylhalide
y=x
y = 0.57x + 4.39R² = 0.97
y = 0.61x + 3.54R² = 0.58
y = 0.55x + 4.1R² = 0.89
y = 0.97x + 0.026R² = 0.82
y = 4.56-36.7R² = 0.83
y = 0.98x + 0.077R² = 0.77
y = 1.88x - 8.25R² = 0.66
y = 0.65x + 2.69R² = 0.99
y = 0.96x - 0.53R² = 0.70
Rate Constant Prediction for the Aqueous Phase
Significant mechanistic differences between the gaseous and aqueous phases (e.g., effect of hydrogen bond to the oxygenated compounds, cage effect, solvation, etc.)
GCM is limited because:1. Many datasets are required to calibrate parameters2. Only deal with molecules for which all required group rate
constants and group contribution factors have been calibrated before.
3. Limited number of literature-reported experimental rate constants for other reaction mechanisms than HO•
4. Assumed that a functional group has approximately the same interaction properties under a given molecule, so that the GCM disregards the changes of the functional properties that can arise from the intramolecular environment by electronic push-pull effects, hydrogen bond formation, or by steric effects.
Application of computational chemistry (quantum mechanical calculations) for predicting reaction rate constants in aqueous phase
Looking at reaction energies (transition states) 30
Challenges still remains
31
rxn,aq aq reactants,aqG G G
G≠aq is a quasithermodynamic quantity, kcal/mol, that indicates the free
energy of the transition state Greactants,aq is the molar free energy of reactants, kcal/mol
rxn,aq rxn,gas rxn,solvationG G G
,0 0
rxn,solvation solvation solvation reactants,solvation reactants,solvationG G G G G
where
rxn,aqln k a G b
Correlation between rate constant and free energy of activation
From transition state theory (TST)
H-atom abstraction from a C-H bond of Aliphatic Saturated Compound
32
based on calculated ∆G≠based on experimental ∆G≠
Minakata and Crittenden, ES&T 2011, 45, 3479-3486
Minakata, Song, and Crittenden, ES&T, 2011, 45, 6057-6065
8
9
10
11
12
15 14
1
6
4
y = -0.418x + 21.271r² = 0.681
0
5
10
15
20
25
-6 -4 -2 0 2 4 6 8 10 12
ln k
H
∆ Gactaq,calc, kcal/mol
Neutral (G3 + COSMO-RS)-ref.16
Ionized (G4 + SMD) - This study
y = -1.386x + 27.723R² = 0.849
0
5
10
15
20
25
0.0 2.0 4.0 6.0 8.0 10.0
ln k
∆GactI , kcal/mol
Neutral
Ionized
Calibration (N=26)13 out of 14 from prediction was within difference of factor of 5
33
Oxygen addition to carbon centered radical
exp Dk k
04 /1000D lk D r N From Smoluchowski’sequation
Ф (=0.25): modification parameter
5
1.14 0.589
13.26 10l
b
DV
*CH3*CH2CH3 *CH2OH
CH3C*HOH*CH2CH2OH
*CCl3
*CH(OH)2
*CH2COCH3
CH3COOC*H2
*CH2COOCH30.0E+00
2.0E+09
4.0E+09
6.0E+09
8.0E+09
1.0E+10
1.2E+10
1.4E+10
1.6E+10
1.8E+10
2.0E+10
0.0E+00 1.0E+10 2.0E+10
kca
lc, M
-1s-1
kexp, M-1s-1
perfect fit*CH3
*CH2CH3*CH2OH
CH3C*HOH*CH2CH2OH*CCl3
Radius of molecule and molar volume are quantum mechanically calculated at B3LYP/6-311++G(3df,3pd)
34
HO• addition to C=C bond of Alkenes Peroxyl Radical Bi-decay
1-Hydroxyethylperoxyl
2-Hydroxyethylperoxyl
Hydroxymethylp
eroxyl
1-Hydroxy-1-methylethylperoxyl
Carboxymethylpe
roxyl, anion
2-
Oxopropylperoxyl
Acetoxymethylpe
roxyl
y = -0.389x + 21.492R² = 0.806
0
5
10
15
20
25
0 2 4 6 8 10 12 14ln
kDelta Gact
aq
Summary • Four different approaches have been shown to model aqueous phase advanced oxidation processes.
• A simplified pseudo-steady-state model is precise enough to examine the feasibility of the process.
• AdoxTM (UV/H2O2) simulation software can be used to design AOPs not assuming constant pH, steady-state.
• Conventional approach to establish kinetic model in AOP is time consuming and does not consider all possible by-products.
• Our under-developing computer-based kinetic model is robust for rational design of AOPs but there are still some challenges remains.
35
Acknowledgement
• National Science Foundation (NSF): 0607332 and 0854416
• Georgia Tech College of Computing, Office of Information Technology and CEE IT Services for high computational resources.
• High Tower Chair and Georgia Research Alliance at Georgia Tech
• Brook Byers Institute for Sustainable Systems
• University of Notre Dame Radiation Center and Department of Energy
36