Workshop on a “Drug Discovery” Approach to …web.mit.edu/dsadoway/www/InvitedTalks/HTE...
Transcript of Workshop on a “Drug Discovery” Approach to …web.mit.edu/dsadoway/www/InvitedTalks/HTE...
1
Workshop on a “Drug Discovery”Approach to Breakthroughs in Batteries
Introduction
Nick DelgassSchool of Chemical Engineering
Purdue University(Core expertise – heterogeneous catalysis)
2
The Problem with Discovery• NOT business as usual• Descriptor (search) space is huge – 106 – 1018
e.g. elements of the periodic table 4 at a time
• NOT only traditional scientific method• Breakthrough means NOT an extrapolation from where you are. Need new approach.
You are here
3
Navigating Descriptor Space
• Designed broad search – statistical analysisFast, helps identify descriptors
• Model-based search – incorporate knowledge, including fundamental
• “Only” need better, not best, performance• Intelligent search is an absolute necessity• Need a lot of data (information)
4
Navigating Descriptor Space• Experimental Data (parallel) - synthesis,
performance, and materials characterizationPrimary (high throughput, 1or 2 parameters)
Secondary(lower data rate, more parameters)Tertiary (lowest data rate, real conditions)
• Theory –Guide or drive models, calculate descriptors, cover space not accessible to experiments
• Informatics – Handling and using dataSearching, sharing, and archivingTools for knowledge extraction, model building
• TEAMS – Success requires all of the above
5
ExampleSearch for a new catalyst
W. Pryz, R. Vijay, J. Binz, JochenLauterbach, and D.J. Buttrey, “Characterization of K-Promoted RuCatalysts for Ammonia Decomposition Discovered Using High-Throughput Experimentation”, Topics in Catalysis, 2008, online
6
High-Throughput - Reactor
Reactant
Side view of Reactor16 sample capacity
RJ Hendershot, et al, Applied Catalysis A (2003)
Individual temperature monitoring
16-way Flow split
Effluent
16 stainless steel reactor tubes
Capillaries
7
Truly Parallel Screening
FTIRSpectrometer
MCTFPAMCTFPA
C.M. Snively, G. Oskarsdottir, and J. Lauterbach, Angewandte Chemie, 40(2001)Snively, C.M. and J. Lauterbach Applied Spectroscopy 59 (2005)
R. Hendershot, P. Fanson, C.M. Snively, and J. Lauterbach, Angewandte Chemie, 42(10); (2003).
128pixels
30 cm
Effluent In Effluent Out
4 0 0 0 3 5 0 0 3 0 0 0 2 5 0 0 2 0 0 0 1 5 0 00 .0
0 .2
0 .4
0 .61 .0
1 .2
Abs
orba
nce
W a ve n u m b e rs (c m -1)
8
The Next Step: Promotion of Ru/γ-Al2O3Catalysts
0 10 20 30 40 50 60 70 80 90 1000
10
20
30
40
50
60
70
80
90
100
Mod
el P
redi
cted
% N
H3 C
onve
rsio
n
Experimental % NH3 Conversion
■ Model Development Points
Term CoefficientsConstant 35.7
Ba -1.8K 22.3Cs -2.6
Ba*Ba -0.5K*K 5.6
Cs*Cs -3.0Ba*K -0.9Ba*Cs -1.0K*Cs -4.7
).....*()*(......)()()......()()( 212
22
1321 CsBaKBaKBaCsKBaCR λλββααα +++++++=
9
Promotion of Ru Catalyst
200 250 300 350 400 450 500
0
20
40
60
80
100%
NH
3 Con
vers
ion
T[°C]
4 Ru/12 K 4 Ru/12 Rb 4 Ru/12 Na 4 Ru/12 Cs 4 Ru/12 Ba 4 Ru
10
K-Promotion: Effect on Morphology
0
20
40
60
80
% N
H3 C
onve
rsio
n
4Ru 4Ru/12K
4Ru / γ-Al2O3
W. Pyrz, R. Vijay, J. Binz, J. Lauterbach, and D. Buttrey, Topics in Catalysis
T = 350oC
1 μm 1 μm
4Ru-12K / γ-Al2O3
KRu4O8
11
Our Research Philosophy - HT Catalytic Science
R. J. Hendershot, C.M. Snively, and J. Lauterbach,Chemistry –A European Journal,11; 806-814, 2005
Catalyst Characterization
20 25 30 35 40 45 50 55 60 65 70 75 80 85 90
♦
♦
♦•
♠
♦♦
♠♠
••
•
♠
♠♦ ♦
♦
♠
♠ ♠
•
•
♦
♦
Inte
nsity
[a.u
.]
2θ [°]
♦
•
♠
γ-Al2O3
5Co
1Pt/15Ba
5Co/15Ba
♠ − ΒaCO3
• − Co3O4
♦ − Al2O3
Design of Experiment
Modeling + UHV + Spectroscopy + Synthesis
Reactor
0
25
50
75 22002000
1800
1600
0.0
0.2
0.4
0.6
CO
N2O
NO
A
←
→
NO2
↑
t/sν/cm-1~
12
1- Hexene Polymerization by Titanium Catalysts
with Phenoxy based ligandsA large number of available substituted phenols allow tunability of steric and electronic variation of the catalyst.
L2
L1
M
P
CH2 n-Bu
C.I.+
C.I. = [MeB(C6F5)3]-
[B(C6F5)4]-
M = Ti, Zr, Hf
L1 =-
CH3
CH3
CH3CH3
CH3
-{ }-
O
Ph
O
O
O
O
IO
BrO
Cl
O
F
O
F
FF
O
O
O
Br
O
O
Br
O
O
Br
O
O
Br
O
O
O
PhPh
O
PhPh
Br
O
PhPh
Ph Ph
O
PhPh
Br
Ph Ph
Steric
L2 ligand
O
PhPh
Ph Ph
O
MeMe
Elec
tron
ic
13
Structure-Activity Correlation in Single-Site, Aryl-Oxide, Olefin Polymerization Catalysts
FamilyFamily--dependent correlation of dependent correlation of propagation rate versus Epropagation rate versus EIPSIPS
Universal Universal correlationcorrelation
DFTDFT--Based Structure ModelBased Structure Model: The solid: The solid--angle available for angle available for monomer approach (monomer approach (44πγπγ) ) to Tito Ti--center is related to the OArcenter is related to the OAr-- and and CpCp’’--ligand exclusion coneligand exclusion cone--angles (angles (θθ) and a steric factor (f) ) and a steric factor (f) related to the counterrelated to the counter--anion exclusion.anion exclusion.
θθCpCp’’
θθOArOAr
““ff””
The propagation rate constant correlates with ionThe propagation rate constant correlates with ion--pair pair separation energy (Eseparation energy (EIPSIPS ) in a univeral manner:) in a univeral manner:
T.A. Manz, K. Phomphrai, G. Medvedev, B.B. Krishnamurthy, S. ShaT.A. Manz, K. Phomphrai, G. Medvedev, B.B. Krishnamurthy, S. Sharma, J. Haq, K.A. rma, J. Haq, K.A. Novstrup, K.T. Thomson, W.N. Delgass, J.M. Caruthers, M.M AbuNovstrup, K.T. Thomson, W.N. Delgass, J.M. Caruthers, M.M Abu--Omar, Omar, J. Am. Chem. Soc. J. Am. Chem. Soc. 2007, 129, 3776.2007, 129, 3776.
14
Computer-aided Product Formulation/Design
(Discovery Informatics)
• Forward Problem: Prediction– Estimate Product Performance from Descriptors
– Quantitative forward model• Inverse Problem: Design
– Determine a set of products that satisfy desired performance criteria
– Guided stochastic search (e.g. genetic algorithm)
Forward ProblemPrediction
Design
P1, P2, P3, ...Pn
Inverse Problem
Product Descriptors Product Performance
15
Component “length” directlyindicative of stability of additive
Breakage of this bond removes “dirt” carrying capacity totally
Chemical nature of this component (polar/non-polar) controls “dirt” removingcapacity
Breaking of these bonds control “length”
The first-principle model tracks the structural distribution of fuel-additive with time due to reactive degradation
Search for New Surfactant MoleculesForward model: First-Principles Model of Additive
Degradation + neural net model of engine performance
1619thNAM May 26, 2005
Run Rank/Identifier FitnessPredicted IVD
(PLS-NN Model)Structural Description
1, I-1 0.997 11.4 mg Novel Structure. Infrequently used linker.
2, I-2 0.996 11.5 mgNovel Structure. Same tails as beststructure, different heads and linkersI
6, I-6 0.993 12.0 mgVariant of structure found in the BMWdatabase. Same head and linkers, differenttails
1, II-1 0.999 10.1 mgNovel Structure. Different from I-1 .Infrequently used linker component.
2, II-2 0.989 12.6 mgSlight variant of additive structure foundin BMW and HONDA databases.Different tails but same head and linker
II
4, II-4 0.983 13.2 mgMinor variation of structure II-2 above.Slight modification of the head
1, III-1 1.00 8.9 mg Novel Structure. Different from I-1 and II-1 . Commonly used components
2, III-2 0.994 11.9 mgVariant of III-1 . One linker and tailmodified.III
3, III-3 0.993 12.1 mgVariant of structure II-2 above. Slightmodification of head. A linker and tailinserted.
Results of GA Inverse Search
Objective:Determine a structure with IVD < 10 mg
Population Size: 25; Generations: 25
17
E. Christoffersen, P. Liu, A. Ruban, H.L. Skiver, and J. K. Norskov, “Amode Materials for Low-Temperature Fuel Cells: A Density Functional Theory Study”. J. Catal., 199, 123-131 (2001).
Norskov d-band Center Descriptor (DFT)
18
E. Christoffersen, P. Liu, A. Ruban, H.L. Skiver, and J. K. Norskov, “Amode Materials for Low-Temperature Fuel Cells: A Density Functional Theory Study”. J. Catal., 199, 123-131 (2001).
Correlation of ΔHads with d-Band Center
19
Drug Discovery Approach to Breakthroughs in Batteries
Informatics in Drug Discovery Ernst R. Dow, Eli Lilly
Electrochemical energy storage and extended-range electric vehicles Mark Verbrugge, GM
Data handling and informatics tools for model-based discovery James Caruthers, Purdue
New High Energy/Power Devices Ralph J. Brodd, Broddarp
High-throughput ab-initio computing and data mining methods for the prediction of crystal structure Gerbrand Ceder, MIT
Materials Informatics: An “omics” approach to materials based design for battery technology Krishna Rajan, Iowa State
20
Discussion at dinner tonightBreakout Groups (Tuesday)
• Two discussion sessions• One writing session
Objective• Report:
Define and justify opportunities for significant advancement in battery technology.
Suggest guidelines for program structure• Website: Maintain as a resource for the community. Start by uploading slides of talks
21
E. Christoffersen, P. Liu, A. Ruban, H.L. Skiver, and J. K. Norskov, “Amode Materials for Low-Temperature Fuel Cells: A Density Functional Theory Study”. J. Catal., 199, 123-131 (2001).
Computed surface segregation energies