過渡金屬配合物展現多樣的結構和電子特性。這些特性可以通過調節配體施加的配
體場強度來改變過渡金屬配合物的性質。目前評估配體場強度主要依賴於特定對稱
過渡金屬配合物的實驗數據，這對其在各種配體上的廣泛應用存在限制。為了解決
這個問題，我們提出了一種利用從劍橋晶體結構數據庫獲取的過渡金屬配合物的晶
體結構來量化配體場強度的數據分析方法。通過應用此數據分析方法，我們利用一
種半監督深度生成模型框架，聯合樹變分自編碼器 (junction tree variational
autoencoder, JTVAE)，用於預測其配體場強度。我們的模型在訓練集上實現了平均
絕對誤差為0.047和均方根誤差為0.072。此外，該模型能夠生成具有所需配體場強
度值的新配體。此外，我們已將這種方法擴展到不僅生成配體，還能設計整個錯合
物，利用預先訓練的變分自編碼器 (variational autoencoder, VAE) 模型，我們實施了
一個人工神經網絡artificial neural network, ANN) 分類器，利用半監督變分自編碼器
(semi-supervised variational autoencoder, SSVAE) 的潛在分佈來預測錯合物的自旋狀
態，我們概述了兩種設計策略：第一種涉及單一突變，而第二種則使用種子生成。
這些策略可幫助我們在設計錯合物時，能朝特定自旋狀態進行生成，以及對特定錯
合物進行局部修改以產生類似自旋狀態且結構相似的錯合物。

Transition metal complexes display diverse structural and electronic characteristics.
These properties of a TM complex can be altered by modulating the ligand field strength
(LFS) inflicted by its ligands. Presently, assessing LFS relies heavily on experimental data
from a limited set of symmetric transition metal complexes, which poses constraints on its
broader applicability across various ligands. To address this, we propose a data-driven
approach using crystal structures of transition metal complexes sourced from the
Cambridge Structural Database (CSD) to quantify the ligand field strength. By employing
this data-driven metric, we demonstrate the effectiveness of a semi-supervised deep
generative model, specifically the junction tree variational autoencoder (JTVAE), in
predicting ligand field strength values. Our model achieves a mean absolute error (MAE)
of 0.047 and root mean squared error of 0.072 on the training set. Moreover, the model
enables the generation of new ligands with desired ligand field strength values.
Moreover, we've broadened this approach to not only generate ligands but also design
entire complexes. Leveraging a pretrained VAE model, we've implemented an artificial
neural network (ANN) classifier utilizing the latent distribution of the semi-supervised
variational autoencoder (SSVAE) to predict the spin state of a complex. With these models,
we're capable of generating new ligands possessing targeted properties and assembling
them into complexes exhibiting our desired spin states. We've outlined two design strategies:
the first involves single mutations, while the second employs seeded generation. These
strategies facilitate the deliberate evolution of parent complexes towards specific spin states
and localized modifications of seed complexes to produce similar spin states, respectively.

(1) Groom, C. R.; Bruno, I. J.; Lightfoot, M. P.; Ward, S. C. The Cambridge Structural Database. Acta Crystallogr. Sect. B Struct. Sci. Cryst. Eng. Mater. 2016, 72 (2), 171–179.
(2) CSD Statistics | CCDC. https://www.ccdc.cam.ac.uk/solutions/about-the-csd/csd-statistics/ (accessed 2023-12-13).
(3) Karimov, R. R.; Hartwig, J. F. Transition‐Metal‐Catalyzed Selective Functionalization of C(Sp 3 )−H Bonds in Natural Products. Angew. Chem. Int. Ed. 2018, 57 (16), 4234–4241.
(4) Darensbourg, D. J. Making Plastics from Carbon Dioxide:  Salen Metal Complexes as Catalysts for the Production of Polycarbonates from Epoxides and CO2. Chem. Rev. 2007, 107 (6), 2388–2410.
(5) Kalyanasundaram, K.; Grätzel, M. Applications of Functionalized Transition Metal Complexes in Photonic and Optoelectronic Devices. Coord. Chem. Rev. 1998, 177 (1), 347–414.
(6) England, J.; Britovsek, G. J. P.; Rabadia, N.; White, A. J. P. Ligand Topology Variations and the Importance of Ligand Field Strength in Non-Heme Iron Catalyzed Oxidations of Alkanes. Inorg. Chem. 2007, 46 (9), 3752–3767.
(7) Pan, Y.; Lin, R.; Chen, Y.; Liu, S.; Zhu, W.; Cao, X.; Chen, W.; Wu, K.; Cheong, W.-C.; Wang, Y.; Zheng, L.; Luo, J.; Lin, Y.; Liu, Y.; Liu, C.; Li, J.; Lu, Q.; Chen, X.; Wang, D.; Peng, Q.; Chen, C.; Li, Y. Design of Single-Atom Co–N5 Catalytic Site: A Robust Electrocatalyst for CO2 Reduction with Nearly 100% CO Selectivity and Remarkable Stability. J. Am. Chem. Soc. 2018, 140 (12), 4218–4221.
(8) Chirik, P. J. Carbon–Carbon Bond Formation in a Weak Ligand Field: Leveraging Open-Shell First-Row Transition-Metal Catalysts. Angew. Chem. Int. Ed. 2017, 56 (19), 5170–5181.
(9) Varela-Álvarez, A.; Yang, T.; Jennings, H.; Kornecki, K. P.; Macmillan, S. N.; Lancaster, K. M.; Mack, J. B. C.; Du Bois, J.; Berry, J. F.; Musaev, D. G. Rh 2 (II,III) Catalysts with Chelating Carboxylate and Carboxamidate Supports: Electronic Structure and Nitrene Transfer Reactivity. J. Am. Chem. Soc. 2016, 138 (7), 2327–2341.
(10) Singh, S.; Brooker, S. Extension of Azine-Triazole Synthesis to Azole-Triazoles Reduces Ligand Field, Leading to Spin Crossover in Tris-L Fe(II). Inorg. Chem. 2020, 59 (2), 1265–1273.
(11) Dey, B.; Mondal, A.; Konar, S. Effect of Ligand Field Strength on the Spin Crossover Behaviour in 5-X-SalEen (X=Me, Br and OMe) Based Fe(III) Complexes. Chem. – Asian J. 2020, 15 (11), 1709–1721.
(12) Halcrow, M. The Effect of Ligand Design on Metal Ion Spin State—Lessons from Spin Crossover Complexes. Crystals 2016, 6 (5), 58.
(13) Sun, Q.; Mosquera-Vazquez, S.; Suffren, Y.; Hankache, J.; Amstutz, N.; Lawson Daku, L. M.; Vauthey, E.; Hauser, A. On the Role of Ligand-Field States for the Photophysical Properties of Ruthenium(II) Polypyridyl Complexes. Coord. Chem. Rev. 2015, 282–283, 87–99.
(14) Wagenknecht, P. S.; Ford, P. C. Metal Centered Ligand Field Excited States: Their Roles in the Design and Performance of Transition Metal Based Photochemical Molecular Devices. Coord. Chem. Rev. 2011, 255 (5–6), 591–616.
(15) Wacholtz, W. M.; Auerbach, R. A.; Schmehl, R. H.; Ollino, M.; Cherry, W. R. Correlation of Ligand Field Excited-State Energies with Ligand Field Strength in (Polypyridine)Ruthenium(II) Complexes. Inorg. Chem. 1985, 24 (12), 1758–1760.
(16) Buitrago Santanilla, A.; Regalado, E. L.; Pereira, T.; Shevlin, M.; Bateman, K.; Campeau, L.-C.; Schneeweis, J.; Berritt, S.; Shi, Z.-C.; Nantermet, P.; Liu, Y.; Helmy, R.; Welch, C. J.; Vachal, P.; Davies, I. W.; Cernak, T.; Dreher, S. D. Organic Chemistry. Nanomole-Scale High-Throughput Chemistry for the Synthesis of Complex Molecules. Science 2015, 347 (6217), 49–53.
(17) Tsuchida, R. Absorption Spectra of Co-Ordination Compounds. I. Bull. Chem. Soc. Jpn. 1938, 13 (5), 388–400.
(18) Jørgensen, Chr. K. The Interelectronic Repulsion and Partly Covalent Bonding in Transition-Group Complexes. Discuss Faraday Soc 1958, 26 (0), 110–115.
(19) Shimura, Y. A Quantitative Scale of the Spectrochemical Series for the Mixed Ligand Complexes of D6 Metals. Bull. Chem. Soc. Jpn. 1988, 61 (3), 693–698.
(20) Cramer, S. P.; DeGroot, F. M. F.; Ma, Y.; Chen, C. T.; Sette, F.; Kipke, C. A.; Eichhorn, D. M.; Chan, M. K.; Armstrong, W. H.; Libby, E.; Christou, G.; Brooker, S.; McKee, V.; Mullins, O. C.; Fuggle, J. C. Ligand Field Strengths and Oxidation States from Manganese L-Edge Spectroscopy. J. Am. Chem. Soc. 1991, 113 (21), 7937–7940.
(21) Shimizu, K.; Maeshima, H.; Yoshida, H.; Satsuma, A.; Hattori, T. Ligand Field Effect on the Chemical Shift in XANES Spectra of Cu( II ) Compounds. Phys. Chem. Chem. Phys. 2001, 3 (5), 862–866.
(22) Reed, C. A.; Guiset, F. A “Magnetochemical” Series. Ligand Field Strengths of Weakly Binding Anions Deduced from S = 3/2, 5/2 Spin State Mixing in Iron(III) Porphyrins. J. Am. Chem. Soc. 1996, 118 (13), 3281–3282.
(23) Yarranton, J. T.; McCusker, J. K. Ligand-Field Spectroscopy of Co(III) Complexes and the Development of a Spectrochemical Series for Low-Spin D6 Charge-Transfer Chromophores. J. Am. Chem. Soc. 2022, 144 (27), 12488–12500.
(24) Syaima, H.; Rahardjo, S. B.; Suryanti, V. Facile and Rapid Synthesis of Tetrakis-2-Amino-5-Methylpyridinecopper(II) Chloride Pentahydrate. IOP Conf. Ser. Mater. Sci. Eng. 2020, 858 (1), 012019.
(25) Singh, S. K.; Eng, J.; Atanasov, M.; Neese, F. Covalency and Chemical Bonding in Transition Metal Complexes: An Ab Initio Based Ligand Field Perspective. Coord. Chem. Rev. 2017, 344, 2–25.
(26) F. Chilton, N.; Lei, H.; M. Bryan, A.; Grandjean, F.; J. Long, G.; P. Power, P. Ligand Field Influence on the Electronic and Magnetic Properties of Quasi-Linear Two-Coordinate Iron( Ii ) Complexes. Dalton Trans. 2015, 44 (24), 11202–11211.
(27) Moens, J.; Jaque, P.; De Proft, F.; Geerlings, P. A New View on the Spectrochemical and Nephelauxetic Series on the Basis of Spin-Polarized Conceptual DFT. ChemPhysChem 2009, 10 (5), 847–854.
(28) Ishii, T.; Tsuboi, S.; Sakane, G.; Yamashita, M.; Breedlove, B. K. Universal Spectrochemical Series of Six-Coordinate Octahedral Metal Complexes for Modifying the Ligand Field Splitting. Dalton Trans 2009, No. 4, 680–687.
(29) Brown, J. M.; Campbell, J. P.; Beers, A.; Chang, K.; Ostmo, S.; Chan, R. V. P.; Dy, J.; Erdogmus, D.; Ioannidis, S.; Kalpathy-Cramer, J.; Chiang, M. F.; for the Imaging and Informatics in Retinopathy of Prematurity (i-ROP) Research Consortium. Automated Diagnosis of Plus Disease in Retinopathy of Prematurity Using Deep Convolutional Neural Networks. JAMA Ophthalmol. 2018, 136 (7), 803.
(30) Gulshan, V.; Peng, L.; Coram, M.; Stumpe, M. C.; Wu, D.; Narayanaswamy, A.; Venugopalan, S.; Widner, K.; Madams, T.; Cuadros, J.; Kim, R.; Raman, R.; Nelson, P. C.; Mega, J. L.; Webster, D. R. Development and Validation of a Deep Learning Algorithm for Detection of Diabetic Retinopathy in Retinal Fundus Photographs. JAMA 2016, 316 (22), 2402.
(31) Rajpurkar, P.; Irvin, J.; Zhu, K.; Yang, B.; Mehta, H.; Duan, T.; Ding, D.; Bagul, A.; Langlotz, C.; Shpanskaya, K.; Lungren, M. P.; Ng, A. Y. CheXNet: Radiologist-Level Pneumonia Detection on Chest X-Rays with Deep Learning. 2017.
(32) Gómez-Bombarelli, R.; Wei, J. N.; Duvenaud, D.; Hernández-Lobato, J. M.; Sánchez-Lengeling, B.; Sheberla, D.; Aguilera-Iparraguirre, J.; Hirzel, T. D.; Adams, R. P.; Aspuru-Guzik, A. Automatic Chemical Design Using a Data-Driven Continuous Representation of Molecules. ACS Cent. Sci. 2018, 4 (2), 268–276.
(33) Moor, J. The Dartmouth College Artificial Intelligence Conference: The Next Fifty Years. AI Mag. 2006, 27 (4), 87–87.
(34) Choi, R. Y.; Coyner, A. S.; Kalpathy-Cramer, J.; Chiang, M. F.; Campbell, J. P. Introduction to Machine Learning, Neural Networks, and Deep Learning. Transl. Vis. Sci. Technol. 9 (2), 14.
(35) Jordan, M. I.; Mitchell, T. M. Machine Learning: Trends, Perspectives, and Prospects. Science 2015, 349 (6245), 255–260.
(36) 128279 128279
(37) Jin, W.; Barzilay, R.; Jaakkola, T. Junction Tree Variational Autoencoder for Molecular Graph Generation. 2018. https://doi.org/10.48550/ARXIV.1802.04364.
(38) Taylor, M. G.; Yang, T.; Lin, S.; Nandy, A.; Janet, J. P.; Duan, C.; Kulik, H. J. Seeing Is Believing: Experimental Spin States from Machine Learning Model Structure Predictions. J. Phys. Chem. A 2020, 124 (16), 3286–3299.
(39) Coley, C. W.; Rogers, L.; Green, W. H.; Jensen, K. F. SCScore: Synthetic Complexity Learned from a Reaction Corpus. J. Chem. Inf. Model. 2018, 58 (2), 252–261.
(40) Paulsen, H.; Duelund, L.; Winkler, H.; Toftlund, H.; Trautwein, A. X. Free Energy of Spin-Crossover Complexes Calculated with Density Functional Methods. Inorg. Chem. 2001, 40 (9), 2201–2203.
(41) Cirera, J.; Via-Nadal, M.; Ruiz, E. Benchmarking Density Functional Methods for Calculation of State Energies of First Row Spin-Crossover Molecules. Inorg. Chem. 2018, 57 (22), 14097–14105.
(42) Radoń, M. Benchmarking Quantum Chemistry Methods for Spin-State Energetics of Iron Complexes against Quantitative Experimental Data. Phys. Chem. Chem. Phys. 2019, 21 (9), 4854–4870.
(43) Vidal, D.; Cirera, J.; Ribas-Arino, J. Accurate Calculation of Spin-State Energy Gaps in Fe(III) Spin-Crossover Systems Using Density Functional Methods. Dalton Trans. 2021, 50 (47), 17635–17642.
(44) Drosou, M.; Mitsopoulou, C. A.; Pantazis, D. A. Spin-State Energetics of Manganese Spin Crossover Complexes: Comparison of Single-Reference and Multi-Reference Ab Initio Approaches. Polyhedron 2021, 208, 115399.
(45) Leto, D. F.; Ingram, R.; Day, V. W.; Jackson, T. A. Spectroscopic Properties and Reactivity of a Mononuclear Oxomanganese(IV) Complex. Chem. Commun. 2013, 49 (47), 5378–5380.
(46) Börzel, H.; Comba, P.; Hagen, K. S.; Lampeka, Y. D.; Lienke, A.; Linti, G.; Merz, M.; Pritzkow, H.; Tsymbal, L. V. Iron Coordination Chemistry with Tetra-, Penta- and Hexadentate Bispidine-Type Ligands. Inorganica Chim. Acta 2002, 337, 407–419.
(47) O’Boyle, N. M.; Banck, M.; James, C. A.; Morley, C.; Vandermeersch, T.; Hutchison, G. R. Open Babel: An Open Chemical Toolbox. J. Cheminformatics 2011, 3 (1), 33.
(48) Pracht, P.; Bohle, F.; Grimme, S. Automated Exploration of the Low-Energy Chemical Space with Fast Quantum Chemical Methods. Phys. Chem. Chem. Phys. 2020, 22 (14), 7169–7192.
(49) Grimme, S. Exploration of Chemical Compound, Conformer, and Reaction Space with Meta-Dynamics Simulations Based on Tight-Binding Quantum Chemical Calculations. J. Chem. Theory Comput. 2019, 15 (5), 2847–2862.
(50) Grimme, S.; Bannwarth, C.; Shushkov, P. A Robust and Accurate Tight-Binding Quantum Chemical Method for Structures, Vibrational Frequencies, and Noncovalent Interactions of Large Molecular Systems Parametrized for All Spd-Block Elements (Z = 1–86). J. Chem. Theory Comput. 2017, 13 (5), 1989–2009.
(51) Bannwarth, C.; Ehlert, S.; Grimme, S. GFN2-xTB—An Accurate and Broadly Parametrized Self-Consistent Tight-Binding Quantum Chemical Method with Multipole Electrostatics and Density-Dependent Dispersion Contributions. J. Chem. Theory Comput. 2019, 15 (3), 1652–1671.
(52) Ioannidis, E. I.; Gani, T. Z. H.; Kulik, H. J. molSimplify: A Toolkit for Automating Discovery in Inorganic Chemistry. J. Comput. Chem. 2016, 37 (22), 2106–2117.
(53) Bannwarth, C.; Caldeweyher, E.; Ehlert, S.; Hansen, A.; Pracht, P.; Seibert, J.; Spicher, S.; Grimme, S. Extended Tight-Binding Quantum Chemistry Methods. WIREs Comput. Mol. Sci. 2021, 11 (2), e1493.
(54) Neese, F. Software Update: The ORCA Program System—Version 5.0. WIREs Comput. Mol. Sci. 2022, 12 (5), e1606.
(55) Caldeweyher, E.; Bannwarth, C.; Grimme, S. Extension of the D3 Dispersion Coefficient Model. J. Chem. Phys. 2017, 147 (3), 034112.
(56) Hehre, W. J.; Ditchfield, R.; Pople, J. A. Self—Consistent Molecular Orbital Methods. XII. Further Extensions of Gaussian—Type Basis Sets for Use in Molecular Orbital Studies of Organic Molecules. J. Chem. Phys. 2003, 56 (5), 2257–2261.
(57) Hay, P. J.; Wadt, W. R. Ab Initio Effective Core Potentials for Molecular Calculations. Potentials for K to Au Including the Outermost Core Orbitals. J. Chem. Phys. 1985, 82 (1), 299–310.
(58) HANDY, N. C.; COHEN, A. J. Left-Right Correlation Energy. Mol. Phys. 2001, 99 (5), 403–412.
(59) Perdew, J. P.; Burke, K.; Ernzerhof, M. Generalized Gradient Approximation Made Simple. Phys. Rev. Lett. 1996, 77 (18), 3865–3868.
(60) Tao, J.; Perdew, J. P.; Staroverov, V. N.; Scuseria, G. E. Climbing the Density Functional Ladder: Nonempirical Meta–Generalized Gradient Approximation Designed for Molecules and Solids. Phys. Rev. Lett. 2003, 91 (14), 146401.
(61) Staroverov, V. N.; Scuseria, G. E.; Tao, J.; Perdew, J. P. Comparative Assessment of a New Nonempirical Density Functional: Molecules and Hydrogen-Bonded Complexes. J. Chem. Phys. 2003, 119 (23), 12129–12137.
(62) Becke, A. D. Density-Functional Thermochemistry. V. Systematic Optimization of Exchange-Correlation Functionals. J. Chem. Phys. 1997, 107 (20), 8554–8560.
(63) Reiher, M.; Salomon, O.; Artur Hess, B. Reparameterization of Hybrid Functionals Based on Energy Differences of States of Different Multiplicity. Theor. Chem. Acc. 2001, 107 (1), 48–55.
(64) Yanai, T.; Tew, D. P.; Handy, N. C. A New Hybrid Exchange–Correlation Functional Using the Coulomb-Attenuating Method (CAM-B3LYP). Chem. Phys. Lett. 2004, 393 (1), 51–57.
(65) Chai, J.-D.; Head-Gordon, M. Systematic Optimization of Long-Range Corrected Hybrid Density Functionals. J. Chem. Phys. 2008, 128 (8), 084106.
(66) Schäfer, A.; Huber, C.; Ahlrichs, R. Fully Optimized Contracted Gaussian Basis Sets of Triple Zeta Valence Quality for Atoms Li to Kr. J. Chem. Phys. 1994, 100 (8), 5829–5835.