Releasing chemical energy in spatially programmed ferroelectrics

Chemical energy ferroelectrics are generally solid macromolecules showing spontaneous polarization and chemical bonding energy. These materials still suffer drawbacks, including the limited control of energy release rate, and thermal decomposition energy well below total chemical energy. To overcome these drawbacks, we report the integrated molecular ferroelectric and energetic material from machine learning-directed additive manufacturing coupled with the ice-templating assembly. The resultant aligned porous architecture shows a low density of 0.35 g cm−3, polarization-controlled energy release, and an anisotropic thermal conductivity ratio of 15. Thermal analysis suggests that the chlorine radicals react with macromolecules enabling a large exothermic enthalpy of reaction (6180 kJ kg−1). In addition, the estimated detonation velocity of molecular ferroelectrics can be tuned from 6.69 ± 0.21 to 7.79 ± 0.25 km s−1 by switching the polarization state. These results provide a pathway toward spatially programmed energetic ferroelectrics for controlled energy release rates.

The parameterization of the data was done following the IsoMap approach for non-linear dimensionality reduction, or as used in this case for parameterization of complex data. This approach generates a graph connecting data points on a high dimensional space to their nearest neighbors, mapped out in the high dimensional space, and then fit a low dimensional manifold [2][3][4][5] .The Isomap algorithm maps the distribution of elements in the high dimensional space, represented by the set of data points {xi} ϵ R n , onto a convex nonlinear manifold M d of lower dimension d < n and through dimensionality reduction, obtain a two or three dimensional embedding of the elements into a weighted graph. The mapping is carried out such that the geodesic distances between the elements in the higher dimensional manifold are preserved when it is mapped onto the lower-dimensional graph.
In order to construct the initial graph in R n , we used K nearest-neighbors (KNN), which graphs each point connected by an edge to its '' k nearest neighbors. The choice of k was optimized by statistically determining the smallest value that could minimize the residual variance | dM -dG |, while providing the maximum number of alternative paths. This ensures that the resulting graph is neither over-connected, leading to loss of pairwise geodesic distances, nor are critical neighbors disconnected.

Development of Robust High-Throughput Model:
The IsoMap analysis provides a new parameterization, which serves as a more efficient representation of the data, while reducing dimensionality and the risk of over-fitting, as well as reducing noise and data sensitivity in the analysis. The regression approach employed was partial least squares (PLS) [2][3][4][5][6][7] . In PLS the training data is converted to a data matrix with orthogonalized axes, which are based on capturing the maximum amount of information in fewer dimensions. The relationships discovered in the training data can be applied to a test dataset based on a projection of the data onto a highdimensional hyperplane within the orthogonalized axis-system. With PLS, the properties of the composites can be modeled as a function of the chemical and additive descriptors independent of each other. Typical linear regression models do not properly account for the co-linearity between the descriptors, and as a result, the isolated impact of each descriptor on the property cannot be accurately known. However, by projecting the data onto a high-dimensional space defined by axes which are comprised of a linear combination of the composite descriptors and also orthogonalized, the impact of the descriptor on the property can be identified independent of all other descriptors. This analysis, therefore, provides a unique approach that incorporates multiple steps, including the development of a feature set providing a multi-scale of information, the parameterization of the data through a non-linear manifold learning approach, and the development of a high-throughput regression linking the parameters and the property space. This approach is applied here to a specific problem, but it is anticipated to be general for a variety of relevant design objectives.  Fig. 13. P-E and I-E loops of EIP at room temperature (100 Hz). The maximized current at the coercive field is known as direct evidence of ferroelectricity as the current peak is generated by the dipole reversal rather than electrical conductivity. The experimentally reported structure of [Hdabco]ClO4 crystal at 295K 8 is an orthorhombic unit cell which is used as our initial structure. The unit cell remains orthorhombic after geometry optimization. After an initial optimization, two phonon modes were found to have imaginary phonon frequencies of -100 cm -1 , which indicated the corresponding configuration is a saddle point of the energy surface. The eigenvectors of those phonon modes at the Γ point correspond to the motion of C and N atoms which result in the twist of the dabcoH + cation. By incrementally moving the atoms in the directions proportional to the directions in the eigenvectors, a double-well can be found. We then performed another geometry optimization using the configuration at the minimum of the double-well as the new starting point, and the final optimized structure was found to be monoclinic. The phonon dispersion of the monoclinic crystal only has negligible negative eigenvalues at the Γ point. It should also be noted that there is a discontinuity along A-Γ-B, which is also observed in other literature 9,10 and is likely due to the LO-TO splitting caused by the nonanalytical term correction 11 . The long-range nature of Coulomb interaction, together with the high polarizability of The ice-templating method (right side) results in an aligned architecture with a porous structures.

HP-DSC
Results for high-pressure DSC of [Hdabco]ClO 4 under N 2 show that enthalpy of decomposition increases with pressure but reach a maximum at approximately 200 psi ( Supplementary Fig. 23a). The data clearly demonstrate that suppression of dabco volatilization by application of pressure results in a higher enthalpy of decomposition, presumably due to oxidation of dabco by O 2 from ClO 4decomposition. complete by 350 °C only dabco and its pyrazine pyrolysis products are observed at and below that temperature. Little evidence of dabco oxidation by the perchlorate is observed. Given that the oxygen balance for dabco is -185%, this is not very surprising. However, the lack of oxidation products also suggests that when in an unconfined condition under ambient pressure, the fuel component of the has a higher ratio of intact dabco relative to pyrazine decomposition products suggesting that cellulose somehow influences the distribution of dabco-related products. Assuming that dabco decomposition proceeds through a radical mechanism, it is likely that dabco fragment radicals react with the cellulose matrix, thereby contributing to the black residual material observed at the conclusion of the analysis (Supplementary Fig. 28).
GC/MS result are also telling in that they show very little with respect to chlorine-containing volatiles. For any perchlorate, one would expect to see significant levels of HCl or Cl 2 among the products. While a relatively small amount of HCl is observed when a SIC for m/z = 36 is extracted from the 400 °C pyrolysis results for the 3D printed [Hdabco]ClO 4, the level is far below that expected. It is proposed that Cl radicals formed on decomposition of ClO 4reacted with available fuel (i.e., the cellulose matrix and any dabco that did not evaporate) and contributed to the observed black residue.