High Accuracy Barrier Heights, Enthalpies, and Rate Coefficients for Chemical Reactions

This Zenodo repository contains the data presented in Spiekermann, K. A.; Pattanaik, L.; Green, W. H.* High Accuracy Barrier Heights, Enthalpies, and Rate Coefficients for Chemical Reactions, Sci. Data 9, 417, (2022). We recommend people refer to this dataset as RDB7 i.e. a diverse reaction database whose transition states contain up to 7 heavy atoms.

Atom-mapped SMILES, barrier heights, reaction enthalpies, and Reaction Mechanism Generator (RMG) reaction family for each reaction are listed in the comma-separated values files b97d3.csv, wb97xd3.csv, ccsdtf12_dz.csv, and ccsdtf12_tz.csv. ccsdtf12_dz_individual_heats_of_formation.csv containing the individual heats of formation for each stable species (i.e., reactant and product). The values in all of these files are in kcal/mol. Q-Chem output files from the reoptimized products are provided for 16,302 reactions at B97-D3/def2-mSVP and for 11,926 reactions at ωB97X-D3/def2-TZVP level of theory. For convenience, these also include the original log files for the reactant, transition state, and non-reoptimized products from Grambow et al. (10.5281/zenodo.3715478) since they were used to calculate barrier heights, enthalpies, and rate constants in this work. The numbering of reaction indices matches that from the originally published dataset to facilitate easy comparison. MOLPRO output files from the single point calculations are provided for 11,926 reactions at the CCSD(T)-F12/cc-pVDZ-F12 level of theory as well as for the 15 validation reactions run at CCSD(T)-F12/cc-pVTZ-F12. The raw log files for all calculations are stored in b97d3.tar.gz, wb97xd3.tar.gz, ccsdtf12_dz.tar.gz, and ccsdtf12_tz.tar.gz. Each archive contains a separate folder for each reaction, which contains log files for the reactant, transition state, and product/s. The Q-Chem log files contain the output from a geometry optimization and harmonic vibrational analysis while the MOLPRO log files contain output from an energy calculation. Transition state theory rate constants, fitted Arrhenius parameters, and average percentage error between the calculated and fitted rate constants can be found for the rigid reactions in ccsdtf12_dz_rigid.csv. The list of 50 temperatures (K) used during Arrhenius fitting is provided in arkane_temperatures.csv, and the raw Arkane outputs are provided in ccsdtf12_dz_rigid.tar.gz.

The improvement from fitting bond additivity corrections at B97-D3/def2-mSVP, ωB97X-D3/def2-TZVP, CCSD(T)-F12/cc-pVDZ-F12//ωB97X-D3/def2-TZVP, and CCSD(T)-F12/cc-pVTZ-F12//ωB97X-D3/def2-TZVP is shown in b97d3_def2msvp_BAC.csv, wb97xd3_def2tzvp_BAC.csv, ccsdtf12_ccpvdzf12__wb97xd3_def2tzvp_BAC.csv, and ccsdtf12_ccpvtzf12__wb97xd3_def2tzvp_BAC.csv respectively. The files contain the experimental and calculated enthalpies for the reference species from the RMG-database used for fitting. The correction values are publicly stored on the RMG-database GitHub on the AEC_BAC branch, though they are also provided in fitted_corrections.pkl for convenience. Further validation of the BACs at the double zeta level was done by comparing to experimental values from the Pedley set since over half of these molecules were not in the RMG-database training set used for fitting. The comparison is shown in ccsdtf12_dz_vs_Pedley_experimental.csv.

本Zenodo仓库收录了Spiekermann, K. A.; Pattanaik, L.; Green, W. H.* 发表于《Scientific Data》9卷417期（2022年）的论文《High Accuracy Barrier Heights, Enthalpies, and Rate Coefficients for Chemical Reactions》所配套的数据集。我们建议将本数据集命名为RDB7，即一类过渡态包含最多7个重原子的多样化反应数据库。

各反应的原子映射SMILES、能垒高度、反应焓，以及反应机理生成器（Reaction Mechanism Generator, RMG）反应族信息，均列于逗号分隔值（Comma-Separated Values, CSV）文件b97d3.csv、wb97xd3.csv、ccsdtf12_dz.csv与ccsdtf12_tz.csv中。另有ccsdtf12_dz_individual_heats_of_formation.csv文件，包含各稳定物种（即反应物与产物）的生成焓。所有文件中的数值单位均为千卡每摩尔（kcal/mol）。

针对16302个反应，我们提供了B97-D3/def2-mSVP理论级别下重新优化产物的Q-Chem输出文件；针对11926个反应，则提供了ωB97X-D3/def2-TZVP理论级别下的同类文件。为便于使用，这些文件还包含了Grambow等人（10.5281/zenodo.3715478）工作中反应物、过渡态与未重新优化产物的原始日志文件——本研究的能垒高度、焓与速率常数正是基于这些文件计算得到。反应索引的编号与原始发表数据集保持一致，便于进行对照比较。

另有11926个反应在CCSD(T)-F12/cc-pVDZ-F12理论级别下的单点能计算MOLPRO输出文件，以及15个在CCSD(T)-F12/cc-pVTZ-F12理论级别下运行的验证反应的同类文件。所有计算的原始日志文件分别存储于b97d3.tar.gz、wb97xd3.tar.gz、ccsdtf12_dz.tar.gz与ccsdtf12_tz.tar.gz压缩包中。每个压缩包均包含对应每个反应的独立文件夹，内含反应物、过渡态与产物的日志文件。其中Q-Chem日志文件包含几何优化与简谐振动分析的结果，MOLPRO日志文件则包含能量计算结果。

刚性反应的过渡态理论速率常数、拟合得到的阿伦尼乌斯参数，以及计算值与拟合速率常数间的平均百分比误差，均可在ccsdtf12_dz_rigid.csv文件中查询。阿伦尼乌斯拟合过程中使用的50个温度（单位：开尔文）列表存储于arkane_temperatures.csv文件中，原始Arkane输出文件则存放于ccsdtf12_dz_rigid.tar.gz压缩包内。

在B97-D3/def2-mSVP、ωB97X-D3/def2-TZVP、CCSD(T)-F12/cc-pVDZ-F12//ωB97X-D3/def2-TZVP以及CCSD(T)-F12/cc-pVTZ-F12//ωB97X-D3/def2-TZVP理论级别下拟合键加和校正（bond additivity corrections, BAC）所带来的性能提升，分别展示于b97d3_def2msvp_BAC.csv、wb97xd3_def2tzvp_BAC.csv、ccsdtf12_ccpvdzf12__wb97xd3_def2tzvp_BAC.csv与ccsdtf12_ccpvtzf12__wb97xd3_def2tzvp_BAC.csv文件中。上述文件包含了用于拟合的RMG数据库参考物种的实验焓与计算焓。校正值已公开存储于RMG数据库GitHub仓库的AEC_BAC分支，为方便使用，本仓库同时提供了fitted_corrections.pkl文件。我们还通过与Pedley数据集的实验值对比，对双zeta级别下的BAC进行了进一步验证——其中超过一半的分子并未纳入用于拟合的RMG数据库训练集，相关对比结果展示于ccsdtf12_dz_vs_Pedley_experimental.csv文件中。