Property Prediction with Multiscale Simulations of Silicon Containing Polymer Composites

Transcription

1 Silicon-Containing Polymers and Composites ACS Division of Polymer Chemistry Omni Hotel, San Diego, CA Property Prediction with Multiscale Simulations of Silicon Containing Polymer Composites Dr. Andreas Bick and Dr. Lalitha Subramanian Dec 17, 2014

2 Overview Background Information Why Multiscale Simulation? MAPS Software Case Study: POSS in PI/PDMS Matrix Atomistic Simulation Mesoscale Simulation Conclusion

3 Background Information

4 Why Multiscale Simulations? Challenges with experimental development and characterization of composites High material and labor costs Difficult testing procedures under extreme conditions Lack of methods to fully probe molecular scale behavior Simulations can supplement and complement experimental information Atomistic scale can provide insight into chemistry that is inherent in the system Mesoscale can provide information on larger length and timescales without losing the chemistry information coming from atomistic level Allows screening of a number of different variables in the phase space used in materials design and optimization Focusses design of experiments towards the most successful path Provides key insights into the material not easily observable by experimental means

5 Materials and Process Simulations (MAPS) Build Simulate Analyze Nanoscale Mesoscale Monte Carlo Thermodynamic Analytics

6 About MAPS

7 MAPS Interface: Model of Epoxy Resin on Graphite

8 Multiscale Simulation of POSS in PI/PDMS Matrix

9 The Application Polyhedral Oligomeric Silsesquioxanes (POSS) are often used to generate polymer composites Mixtures with Polyimides (PI) are common for gas separation membranes Block copolymers of PI with Polydimethylsiloxane (PDMS) generate a more flexible material but with a certain amount of stiffness POSS generate a precise pore opening of 4.5 A 3 Increases selectivity of the membrane Flexible segments of PDMS improve contact with the filler PI used in this study is 6FDA-MDA

10 Multiscale Simulation Workflow Atomistic Build Simulate Analyze Mesoscale Build Simulate Analyze

11 Atomistic Level Simulation

12 Atomistic Simulation Workflow - Validate Sketch POSS Choose SciPCFF Optimize units Calculate Density, Solubility Parameter MD Simulation Build Xtl of POSS Compare with experimental data Validate model

13 Polyhedral Oligomeric Silsesquioxanes (POSS) Polyhedral Oligomeric Silsesquioxanes (POSS) can have various organic functionalizations We use Octamethyl silsesquioxane (OMS)

14 Polyhedral Oligomeric Silsesquioxanes (POSS) Simulated with SciPCFF Predicted density: 1.49 g/cm 3, experimental: 1.51 g cm 3 Predicted Hildebrand solubility parameter for PMS: 17.7 MPa 0.5 Unit cell parameters: a = b = A, c = A Force field capable of describing POSS crystal very well

15 Atomistic Simulation Workflow Predict for Pure Sketch 6FDA- MDA Build Amorphous PI MD Simulation MD Simulation Build POSS in PI Predict properties Predict Properties

16 Pure 6FDA-MDA Polyimid Simulated with SciPCFF Predicted density: 1.38 g/cm 3, experimental: 1.38 g/cm 3 Predicted Hildebrand solubility parameter for PI: MPa 0.5 Glass transition temperature(exp.) 570 K Youngs modulus predicted 2.5 GPa, experimental 2.9 GPa

17 Atomistic Simulation Workflow Predict for Composite Build POSS in PI MD Simulation Predict Properties Predict properties MD Simulation Build POSS in PI/PDMS Analyze all results

18 6FDA-MDA Matrix with 10 % POSS Simulated with SciPCFF Predicted density 1.41 g/cm 3 Youngs modulus predicted 3.1 GPa (experimental with PAHS 3.2 GPa) predicted by tensile strain up to 3 % (picture left)

19 6FDA-MDA Matrix with 10 % POSS (contd.) Glass transition temperature predicted 567 K (no measurable change as compared to pure polymer) Free Volume (Å 3 ) Tg Temperature (K)

20 6FDA-MDA Matrix with 10 % POSS (contd.) Free Volume 0.65 % as compared to 1.7 % for pure polymer Drastic reduction in Free volume Free Volume evenly distributed

21 6FDA-MDA-PDMS Block copolymer with 10 % POSS Repeating block 1 6FDA-MDA 15 PDMS Results in much more flexible matrix Should ensure even distribution of POSS Predicted density 1.19 g/cm 3 Free Volume goes down to 0.15 % Predicted Tg = 423 K

22 6FDA-MDA-PDMS Block copolymer with 10 % POSS (contd.) Youngs modulus predicted as 0.8 GPa Still high for a rubbery material

23 Mesoscale Level

24 Mesoscale Simulation Workflow Predict for Composite Determine Bead size Calculate Interaction Parameters Build mesoscale models of composite Analyze all results Predict Properties DPD Simulation

25 DPD Parameters from Atomistic Simulations From the cohesive energy and the volume provided by NPT LAMMPS simulations, the Hildebrand solubility parameter can be calculated: δ=( E i )1/ 2 V The factor converts (kcal/mol Å -3 ) 1/2 into (J/cm 3 ) 1/2 From the solubility parameter the Flory Huggins X- parameter can be deduced: χ AB = V b kt ( δ A δ B ) 2 Vb is the bead volume of the DPD simulation, kt with conversion Warren and Groot have observed a linear relationship between the Flory Huggins parameter and the DPD repulsive in a certain range of repulsion: χ a( for ρ=3) With the self interaction parameter a being defined as 25

26 6FDA-MDA-PDMS Block copolymer with 10 % POSS Filler - Mesoscale simulations Dissipative particle dynamics 340 chains with 10 blocks each, 85 filler particles 3 PDMS repeat units in 1 bead PI repeat unit represented by 2 beads POSS bead has 2 molecules System size 22 x 22 x 22 nm Repulsive parameters parameterized using solubility parameters from atomistic simulations Green = PDMS Red = PI Blue = POSS

27 6FDA-MDA-PDMS Block copolymer with 10 % POSS Filler - Mesoscale simulations Typical island-like structure of flexible/rigidblock copolymers observed PI domains in a matrix of PDMS Even distribution of POSS Distribution does not change when shear is applied (left)

28 6FDA-MDA-PDMS Block copolymer with 10 % POSS Filler - Mesoscale simulations POSS agglomerate as dimers and trimers mostly POSS located in PDMS rich areas Tan Delta is 0.8 Sinusoidal shear applied to the system Higher than typical particle filled rubber (Tan Delta < 0.4)

29 Summary Pure POSS, pure Polyimid, a mixed matrix composite PI/POSS and a mixed matrix composite PI-PDMS/POSS have been characterized by multiscale modeling (at the atomistic and at the mesoscale level) Glass transition temperature, Youngs modulus, Free Volume and POSS distribution have been simulated and compared with experimental values, where available Important properties of membranes can be predicted with atomistic and mesoscale modeling and can be used to design a gas separation membrane with tailor made properties

30 Concluding Remarks Systems and Processes modeled at Atomistic and Mesoscale Levels Annealing Stability Rheology Adsorption Solubility Reaction Miscibility Diffusion Absorption Corrosion Breakage Dispersion Separation

31 Thank you for your attention visit: contact: