molecular products design: challenges and...

1

Molecular Products Design:

Challenges and

Opportunities

Venkat Venkatasubramanian

Laboratory for Intelligent Process Systems

Purdue University

West Lafayette, IN, USA

LIPS Venkat Venkatasubramanian, Keynote Lecture,

European Congress in Chem. Engg, Copenhagen, Sept 2007 2

Outline Chemical Compounds Space

Engineer’s Perspective

Design of Molecular Products

Molecular Products: Industrial Case Studies

Fuel Additives Design (Lubrizol)

Rubber Products Design (Caterpillar)

Catalyst Design (ExxonMobil)

Seromycin Formulation for MDR-TB (Eli Lilly)

Focus on Practical Challenges

Data, Information and Knowledge Modeling Challenges

Summary

LIPS

Broad overview

Show some details but not all

Venkat Venkatasubramanian, Keynote Lecture,


Venkat Venkatasubramanian, Plenary Lecture,

Foundations of Computer-Aided Process Design, Princeton, July 2004.

Acknowledgements

Prof. J. F. Caruthers and Prof. W. N. Delgass Dr. K. Chan (Bear Stearns), Dr. A. Sundaram (ExxonMobil), Dr. P.

Ghosh (ExxonMobil), Dr. S. Katare (Dow), Dr. A. Sirkar (Intel), Dr. P. Patkar (ZS), Dr. S-H. Hsu (OSI Soft), S. Syal (Purdue), B. K. Balachandra (Amazon)

Funding Agencies NSF DOE Indiana 21st Century Research and Technology Fund GE, Bangalore, India Lubrizol Caterpillar ExxonMobil Equistar Cummins

Materials Design



Acknowledgements

Dr. C. Zhao (Bayer)

Dr. A. Jain (United)

Dr. S-H. Hsu (OSI Soft)

L. Hailemariam (Dow)

P. Akkisetty (Intel)

P. Sureshbabu (Dow)

I. Hamdan (Dow)

M. Chaffee (Purdue)

A. Shukla (Intel)

Funding Agencies:

NSF ERC

Indiana 21st Century R&T Fund

Eli Lilly

Pfizer

Abbott

Dr. Prabir Basu

Dr. Girish Joglekar

Prof. Ken Morris

Prof. G. V. Reklaitis

Information Technology@Purdue

Pharmaceutical Engineering

Venkat Venkatasubramanian FOCAPD 2004 Princeton

LUBRIZOL: Fuel Additive Design

EPA Performance Measure BMW Test for Intake Valve Deposit Stipulated to be less than 100 mg

over a 10,000 mile road test

Fuel-additives are added to gasoline to minimize IVD

Expensive testing Around $10K for a single datum

Molecular Product Design Problem: Design fuel-additives that meet desired IVD performance levels

Intake Valve and Manifold



About 1000

Rubber Parts in

Failure Critical

Functions

Tires, Treads,

Hoses, Shock

Absorbers, O-

rings, Gaskets,

Mounts …

Reliability and

Warranty

Problems

CATERPILLAR



Multi-Scale

Modeling

Quantum Mechanics

Kinetics

Heat Transfer FEA

Mechanics FEA

Part Design CAD

/CAM

Caterpillar’s Multi-Scale Modeling Challenge:

Design of Formulated Rubber Parts



ExxonMobil’s Grand Challenge:

Catalyst Design via Combinatorial Chemistry

Fundamental

Chemistry

Reaction

kinetics

Performance

measures

time

Rate

/Se

lectivity

C

A

B

F H

D

R

M

E G

N

k2

k7

k5

Reaction network

Electronic/Structural

descriptors

d band center

Electronegativity

. . .

Quantum calculations

Catalyst Structures Synthesis

Characterization

10/11/2005

10

Pharmaceutical Product

Development and Engineering

Preliminary product recipe

Drug is converted into Particles

Drug Synthesis

Raw Chemicals

Formulation

Process Development

Preliminary process recipe

Scale/up - Tech Transfer

Adjusted process recipe

Manufacturing

A Delayed Product



Molecular Products Design

Given a set of desired performance specifications, how does one rationally and efficiently identify the optimal product structures and formulations?

$200+ Billion/yr industry Areas of Application

Engineering Materials

Fuel and Oil Additives

Polymer Composites

Rubber Compounds

Catalysts

Solvents, Paints and Varnishes…..

Pharmaceuticals

Drug Design

Agricultural Chemicals

Pesticides

Insecticides

Zeolite!!



Different Products, But Common Challenges

Lots of Data Uncertain/Noisy Data Complex chemistry, lots of species Nonlinear systems and processes Incomplete, Uncertain, Mechanisms Combinatorially large search spaces Need Multi-scale Models Hundreds of Differential and Algebraic Equations to

formulate and solve Laborious and time consuming model development

process Limited human expertise



Traditional Approach

Revise Design

No

Designer

Hypothetical Molecule

Synthesize

Evaluate

Design Objectives

Yes

Solution

Meets Design Objectives ?

Drawback: Protracted and Expensive Design Cycle

Need a Rational, Automated Approach: Discovery Informatics

EUREKA!!!



Need a New Paradigm Cyberinfrastructure Methods and

Tools

Ontology-based Discovery

Informatics

LIPS Venkat Venkatasubramanian, Plenary Lecture, Pharmaceutical

Sciences World Congress, Amsterdam, April 2007 15

What is an Ontology?

Originated in Philosophy

Study of Existence

Computer Science and Artificial Intelligence

A formal, explicit specification of a shared conceptualization

Knowledge representation in AI

Logic, Semantic Networks, Frames, Objects, Ontology Web-driven development

Semantic Richness: Description Logic (DL) provides

foundation for formal reasoning

Easily create, share and reuse knowledge

Semantic Search

16

Ontology: A Simple Example

Excipient Properties

Material

API

Is a Is a

has

Concept Instance Property

Roles Composition

Range

has

has

Flow aid

Filler

Lubricant

Minimum

Maximum

Average

Is an instance of has

Contact angle

Moisture content

Is a

Material Composition

has

Web Ontology Language: OWL



Catalyst Design Problem – The Grand Challenge

Fundamental

Chemistry

Reaction

kinetics

Performance

measures

time

Rate

/Se

lectivity

C

A

B

F H

D

R

M

E G

N

k2

k7

k5

Reaction network

Electronic/Structural

descriptors

d band center

Electronegativity

. . .

Quantum calculations

Catalyst Structures Synthesis

Characterization



1 A° 100 A° 10 m 1 cm

10-15 s

10-12 s

10-6 s

1 s

102 s

Quantum

Mechanics

DFT

Atomistic

Simulation

Molec. Dyn.

Statistical

Mechanics

TST

Continuum

Models

Transport Models

Elect.

Struct. Thermo.

Props.

Molec.

Struct.

Rate

Consts.

Overall

Rxn Rates

Conc.

Profiles

Microscopic Mesoscopic Macroscopic

Modeling at Different Length and Time Scales



Product Formulation and Design

Forward Problem: Prediction

– Estimate Product Performance from Formulation

Inverse Problem: Design

– Determine a set of products that satisfy desired performance criteria

Forward Problem

Prediction

Design

P1, P2, P3, ...Pn

Inverse Problem

Product Formulation Product Performance



Paraffin Aromatization

Identify a catalyst formulation for light paraffin aromatization that is superior to Ga/H-ZSM-5 in terms of:

Higher Benzene, Toluene, Xylene (B/T/X) selectivity

Higher Hydrogen selectivity

Microkinetic model development for the kinetic description of the system

Venkat Venkatasubramanian, Plenary Lecture, Foundations of Computer-Aided Process Design, Princeton, July 2004.

Reaction Network – Propane Aromatization on HZSM-5

• 31 gas phase species + 29 surface species + 271 reaction steps

• Model with 31 ODEs, 29 algebraic equations

• 13 parameters with up to 10 orders of magnitude bounds on each

Hydride transfer

Cx-y= -scission/Oligomerization

Cx+ Cy

+

H2 Cx-y

Cx= Cy

=

H+

CxH+

Cx

Paraffin

Dehydro-

cyclization

Aromatics



Ontological Informatics: Modeling Super Highway

• DAE System: 30 to 1000 species and 10 to 50 parameters

• Parameter search space is highly nonlinear with numerous local minima

GA based pseudo global parameter estimator

Fast and efficient coverage of the parameter space

• Optimization: Traditional least-squares and features-based objectives

Chemistry

Compiler

Reaction Description

Language Plus (RDL+)

Equation

Generator

Automatic generation of

differential algebraic

equations (DAEs)

Parameter

Optimizer

Solution of DAEs

Least squares

Features

Advanced

Statistical

Analyzer

Sensitivity Analysis

Uncertainty Analysis

Error Propagation

Reactions

Grouping

Chemistry

Rules

time

Rate

/Se

lectivity

Performance

Curves

Data



Modeling Super Highway: Reaction Modeling Suite (RMS)

Reaction Network

Reaction Description Language Plus English Language Rules

C

A

B

F H

D

R

M

E G

N

k2

k7

k5

Mathematical Equations

1

/

/

541

1

CBA

AB

AA

BkDkCkdtdC

CkdtdC

100’s of DAE’s

. . .

Model Generator

Beta Scission Label-site c1+ (find positive carbon)

Label-site c2 (find neutral-carbon attached-to c1+)

Label-site c3 (find neutral-carbon attached-to c2)

Forbid (primary c3)

Forbid (less-than (size-of reactant) 9)

Disconnect c2 c3)

Increase-order-of (find bond connecting c1+ c2)

Add-charge c3

Subtract-charge c1+

Beta Scission Label-site c1+ (find positive carbon)

Require (c1+ primary and product)

set-k k1

Label-site c2+ (find positive carbon)

Require (c2+ secondary and product)

set-k 20*k1

Label-site c3+ (find positive carbon)

Require (c2+ tertiary and product)

set-k 60*k1

Chemistry

8. Beta Scission

transforms a carbenium ion into a

smaller carbenium ion and an olefin

Grouping

8. a. Formation of a secondary carbenium ion

is 20 times faster than a primary carbenium ion

b. Formation of a tertiary carbenium ion

is 60 times faster than a primary carbenium ion

. . .

. . .

. . .

. . .



Ontological Informatics for Catalyst Design

High

Throughput

Experiments

Chemistry

Rules

Human

Expert

High

Throughput

Experiments

Chemistry

Rules

Human

Expert

Feature

Extraction

Reaction

Modeling

Suite

Reaction

Modeling

Suite

time R

ate

/Se

lectivity

Performance

Curves

time R

ate

/Se

lectivity

Performance

Curves

time

Formulation of Experiments Formulation of Experiments

Model Refinement Model Refinement



Results

Over prediction of C2s and under prediction of aromatics Space time x 104 [hr]

Wt

%

C2H6 C3H6

C3H8 Aromatics

Data*

* Lukyanov et. al., Ind. Eng. Chem. Res. , 34, 516-523, 1995

Model



Results – Model Refinement

Refined model adds alkylation step to convert lighter alkanes to higher ones Space time x 104 (hr)

Wt

%

C2H6 C3H6

C3H8 Aromatics

Data Original Model

Refined Model



Ontological Informatics for Catalyst Design: Summary

Chemistry Rules

High Throughput Experimental

Data

Reaction Knowledge

Base

Kinetic Modeling Toolkit

(Reaction Modeling Suites)

Chemistry Rules

High Throughput Experimental

Data

Reaction Knowledge

Base

Kinetic Modeling Toolkit

(Reaction Modeling Suites)

Represents reaction chemistries explicitly

Customizable for different catalyst chemistries

Results

Real-time evaluation of 100s of reaction pathways and 1000s of DAEs

Saved months of model development effort

Improved mechanistic understanding and first principles models:Insight

Application


LUBRIZOL: Fuel Additive Design

EPA Performance Measure BMW Test for Intake Valve Deposit Stipulated to be less than 100 mg

over a 10,000 mile road test

Fuel-additives are added to gasoline to minimize IVD

Expensive testing Around $10K for a single datum

Molecular Product Design Problem: Design fuel-additives that meet desired IVD performance levels

Intake Valve and Manifold



Modeling Philosophy

All models are wrong…. … but some are useful!

-- George Box (U. Wisconsin)

Hybrid Models Combine First-principles and

Data-driven models


Hybrid Models in Discovery Informatics

Ab Initio

Methods

Group

Contribution

Topological

Techniques

Molecular

Mechanics

Molecular

Structure

Primary

Model

Descriptors

Properties

Neural Networks

Secondary

Model

Regression

GFA


Component “length” directly

indicative of stability of additive

Breakage of this bond

removes “dirt” carrying

capacity totally

Chemical nature of this component

(polar/non-polar) controls “dirt” removing

capacity

Breaking of these bonds control “length”

The first-principle model tracks the structural distribution of fuel-additive with time

due to reactive degradation

First-Principles Model for Additive Degradation


Population Balance Equations


Inverse Problem


Forward and Inverse Problem

Forward Problem: Prediction

– Estimate Product Performance from Formulation

Inverse Problem: Design

– Determine a set of products that satisfy desired performance criteria

Forward Problem

Inverse Problem

Prediction

Design

Product Formulation Product Performance

P1, P2, P3, ...Pn


Fitness Calculn, Parent Selectn

Fit

nes

s

Genetic Algorithms (GA) GAs are stochastic evolutionary search procedures based on

the Darwinian model of natural selection

Operators New

Population

“ Survival of the fittest ”

Evolution

Initial Population (random)


Computer-Aided Design of Fuel-Additives

Hybrid Model

Additive

Performance

Intake Valve

Deposit

Additive Structure

Fuel Property

Engine Conditions

Genetic Algorithms

.

.

Selection Recombination

Physical Model

Statistical/Neural-Net

Correlation



Rank/Identifie

r

Fitness

Predicted IVD

Structural Description

III-1

1.000

8.9 mg

Novel Structure

New Chemistry

I-2

0.996

11.5 mg

Novel Structure

New Chemistry

I-6

0.993

12.0 mg

Found by the

traditional approach

also Found new candidates in minutes

Opened up new possibilities: New Insights

Think “Out of the Box”

Objective: Determine a structure with IVD < 10 mg


Tires, Treads,

Hoses, Shock

Absorbers, O-

rings, Gaskets,

Mounts ……..

About 1000 Rubber

Parts in Failure

Critical Functions

Reliability and

Warranty Problems

CATERPILLAR



Multi-Scale

Model

Quantum Mechanics

Kinetics

Heat Transfer FEA

Mechanics FEA

Part Design CAD

/CAM

Caterpillar’s Grand Challenge: Design of

Formulated Rubber Parts


Vu

Time

Accelerator

Sulfur

Retarder

Other

Curatives

Inputs Kinetic Model

Cure time

Curing temperature

Natural Rubber

Other rubber systems

like SBR, Chlorobutyl rubber,

etc.

Other Accelerator

and Retarder systems

Curing Curve

Carbon

Black

Occluded Volume,

Mixing Effects

Rubber Compounds Formulation: Forward Model

Error Analysis


Hyperelasticity

Homogeneous deformation

modes

Non-homogeneous

deformations

Finite Element solution

Of complex systems

Stress-strain Predictions

Curing curve from

Kinetic Model

Vu

Time

G

Forward Model (Contd..)

Error

Analysis


Rubber, RH

BtS

S8 S8

S7 RS-Sy+1

RS-Sy+8

S8

S8

S7 Bt Sz+1

S8

S8 Bt Sz+8

+ Bt-SH

+ Bt-Sz

Crosslink Precursor

RSx –SBt, RSx-Zn-SBt

Crosslinks

RSSy -R

Rubber, RH

Sulfurating Species

BtS-Sx-SBt, BtS-Zn-Sx-SBt

Persulfenyl Radical

RS-Sy

Rubber, RH Bt Sz+2

Rubber, RH

RS-Sy

Cyclic Sulfide Main- Chain Modification,

Conjugated Dienes/Trienes,

Inactive thiols

RSSy-1R + BtSZnSSBt

(Desulfuration)

(Degradation)

ZnO BtSZnSBt

BtS-ZnSx-SBt

(Scorch Delay)

CDB + Pthalimide

(Retarder Action)

S8 … S7 S8 BtS-SBt

BtSNR2

BtSSSBt BtSSxSBt BtSS8SBt

BtS-SBt BtSSxSBt

+

BtSSySBt

BtS-SBt BtSSxSBt +

BtSSySBt

+ CTP -R2NH

Lumped S8 Pickup

Sequential S8 Pickup

Both

Mechanisms

BtSH

820 Reactions, 107 Species

Complex Reaction Network Synthesis

Cipac Presentation, Nov 29 2000

Population Balance Equations: 101 Coupled Nonlinear ODEs

EBxBxAxkkkkf

Axkf

SAkf

VuBxkkf

BxBxAxkkkf

BxBxAxkkkf

Axkkkkf

E

MBT

S

Vu

Bx

Bx

Ax

xx

,,,,,,,

,

,,

.

.

.

,,,

.

.

.

,,,,,

.

.

.

,,,,,

.

.

.

,,,,

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

.

5251427

16

80015

634

4323

4212

4102011

8

d

dt

Initial Conditions:

Ax(>0)(0) = Bx(0) = B*x(0) = Vux(0) = MBT(0) = E(0) = 0

A0(0) = [MBS]0 S8(0) = [Sulfur]0

Rate-Constants ordered in accordance with the polysulfidic bond-strengths and relative reactivities

Most simple model that explains all the essential features of sulfur vulcanization


Common Accelerators

Need for Quantum Chemistry Models

Large number of accelerators

Different Zn-ligand complexes

Different polysulfidic lengths

Mixed accelerator systems

Different elastomers

100s to 1000s of species

100s of rate constants


Quantum Chemical Calculations

Ab initio quantum chemistry calculations

Determine bond strengths - free radical disassociation

Rate constant ratios from bond disassociation energy

Ax

Polysulfidic Chain Zinc

Reactants

Products

En

erg

y

H Ea

Rxn coordinate

Decreases the load on experimentation

as well as on parametric optimization


Estimation of Kinetic Parameters

Example: thiuram & benzothiazole accelerators

Me 2 N

S

S x

S

NMe 2 MBS

N

C

S

S

N

C

S

S S x TMTD

Ax: Variation of BDE with bond position

(Accelerator=MBS)

0

20

40

60

80

100

120

1 2 3 4

Bond relative to the benzothiazole group

BD

E (

kc

al/

mo

l) A0

A1

A2

A3

A4

DFT results (Gaussian 98)

S x S C

S

N

cL


Heat-Transfer Effects in the Mold

Part

Curing a Part in the Mold Thermal Profile in Part Temp, F

Heat transfer

FEA code

Undercured Regions

85-87

89-91

91-93

93-95

95-97

> 99

97-99

87-89

Overcured Regions Temperature Profile

+

Kinetic Model

Vulcanization Profile in Part


Rubber parts

Product Design of Cab Mount

FEA via ABAQUS

Cross-sectional view of the assembly showing

contact between rubber and metal

What is the optimum

part shape?

material composition?

Requirements:

mechanical stiffness

damping

long-term performance


Part Design: Chemistry Material Behavior

Manufacturing Performance

Data

Prediction

Predictions are in reasonable

agreement with the data!

Simulation of the Cab Mount under

an axial deformation history

Material parameters predicted using

kinetic model for the rubber formulation

used in the mount!

Axial Stiffness of the Cab Mount


Quantum

Chemistry

Need Multi-Scale Modeling Environment

Chemical

Process

Chemical

Kinetics

Solid

Mechanics

CAD/CAM

Business

Model

Fluid

Mechanics

FEA

Heat

Transfer

FEA

Molecules

Materials

Market

+

Hierarchical Models


[1.75 1.25 0.05 0 315]

[0.50 4.00 0.10 0 310]

[1.75 1.50 0.05 0 310]

[0.75 2.75 0.10 0 315]

[3.00 0.75 0.10 0 315]

[2.75 1.00 0.25 0 330]

0.9999

0.9999

0.9996

0.9994

0.9991

0.9990

15

16

17

13

17.5

10.5

69.92

71.57

75.75

67.88

65.34

70.46

0.91

0.92

0.97

0.88

0.85

0.93

1.63

1.65

1.75

1.58

1.52

1.67

Formulation Fitness Tmax Vumax 100 200

Time, min

Co

nce

ntr

atio

n o

f C

ross

lin

ks,

mo

l/m

3

New Formulations

Inverse Problem Solution

Better Formulations

Designed in hours instead

of about 2 weeks taken by

human experts

Used daily at Caterpillar

Hybrid models at work!


Interactive Software Used At Caterpillar on a Daily Basis

10/11/2005

53

Pharmaceutical Product

Development and Engineering

Preliminary product recipe

Drug is converted into Particles

Drug Synthesis

Raw Chemicals

Formulation

Process Development

Preliminary process recipe

Scale/up - Tech Transfer

Adjusted process recipe

Manufacturing

A Delayed Product

LIPS

Drug Recalls: Systemic Failures in

Manufacturing Quality Control

•Johnson & Johnson Recalls More Products (Jan 14, 2011)

•Can Johnson & Johnson Get Its Act Together? (Jan 15, 2011)

–McNeil Consumer Healthcare unit of J.& J. (2010)

–Recalled ~288 million drug products

–136 million bottles of liquid Tylenol, Motrin, Zyrtec and Benadryl for infants and children

55

Purdue Ontology for

Pharmaceutical Engineering (POPE)

Information

Equation

Editor

MathML

Model

Ontology

Engine

Operation

Ontology

Interface

Solver

Experiment/

Existing

Database

ModLAB

OntoMODEL



Summary

Reviewed Modeling and Informatics Challenges in Molecular Products Design

Fuel Additives, Rubber Compounds, Catalysts, Drug Products

Need for Cyberinfrastructure Concepts, Methods, and Tools

Ontologies

Variety of Optimization Methods

Hybrid Models

This is only a beginning….Miles to go before we sleep….

57

Thank You for Your

Attention!

Any Questions?

molecular products design: challenges and...

Documents