Maximizing the magnetic anisotropy of Dy complexes by fine tuning organic ligands: A systematic multireference high-throughput exploration of over 30k molecules

This study employs automated multireference ab initio calculations to systematically screen over 30,000 dysprosium complexes, revealing that fine-tuning organic ligands in pentagonal bipyramidal structures can significantly enhance magnetic anisotropy by up to 100% compared to reference compounds, thereby demonstrating the power of high-throughput computational screening in overcoming chemically non-intuitive design challenges.

The Gist

Imagine you are trying to build the world's most stable, high-speed "magnetic memory stick" for a computer. To do this, you need a tiny molecule that acts like a magnet that never forgets its direction, even when the temperature rises or the environment gets noisy. Scientists call these Single-Molecule Magnets (SMMs).

The star player in this story is a specific type of atom called Dysprosium (Dy). Think of Dysprosium as a very sensitive, high-performance engine. But an engine doesn't run well on its own; it needs a perfectly tuned chassis and suspension to handle the speed. In chemistry, this "chassis" is the ligands (the organic molecules) that surround the Dysprosium atom.

Here is the simple breakdown of what this paper did, using some everyday analogies:

1. The Problem: Searching for a Needle in a Haystack

For decades, chemists have been trying to find the perfect arrangement of ligands to make Dysprosium magnets work better. They've built about 650 different versions and tested them.

• The Analogy: Imagine you are trying to find the perfect key to open a treasure chest. You've tried 650 keys you found in a drawer, but none of them fit perfectly. You know the right key exists, but the "drawer" (the world of known chemicals) is too small, and you've only looked at the top layer.
• The Reality: The paper points out that while there are billions of possible organic molecules in the world (like the ZINC database), chemists have only tested a tiny fraction of them for this specific job.

2. The Solution: The "Robot Chef"

Instead of waiting for humans to guess which new ligands to mix, the authors built a computational robot.

• Step A: The Archaeologist. First, the robot scanned every single crystal structure ever recorded in scientific databases (like a massive digital library of molecules). It cleaned them up, removed duplicates, and calculated how magnetic they would be.
  • Result: It found that while some existing molecules are good, none of them were the "record-breakers" the world was hoping for.
• Step B: The Inventor. Then, the robot decided to stop looking at old keys and start inventing new ones. It took a promising "template" molecule (a Dysprosium atom surrounded by five water molecules in a flat circle, with two ligands sticking up and down like a top) and swapped out the top and bottom ligands with 25,000 different organic molecules from a database of millions.
  • The Scale: This is like a robot chef trying 25,000 different spice combinations on a single dish in a few days, whereas a human chef might try 10 in a lifetime.

3. The Discovery: It's All About the "Second Sphere"

The most surprising finding wasn't just that they found better magnets, but why they worked better.

• The Old Rule: Scientists used to think you only needed to tune the atoms touching the metal directly (the "first sphere"). It's like thinking you only need to tune the engine bolts to make a car go fast.
• The New Rule: The study found that the second layer of molecules (the "second sphere") matters just as much.
  • The Analogy: Imagine the Dysprosium atom is a dancer. The ligands touching it are its hands. The researchers found that if the clothes the dancer is wearing (the outer ligands) have tiny bumps or shapes that gently push against the dancer's shoes, it forces the dancer to stand perfectly straight.
  • The Mechanism: They discovered that weak interactions between hydrogen atoms on the outer ligands and the water molecules near the metal act like "invisible glue." This glue locks the molecule into a perfect, symmetrical shape (called pentagonal bipyramidal). This perfect shape is what allows the magnet to hold its memory so strongly.

4. The Result: A Massive Leap Forward

By fine-tuning these outer "clothes" (the organic ligands), the team found molecules that were:

• 100% better than the previous best reference molecule.
• 30% better than any other known molecule with this specific shape.
• Close to the theoretical limit of what is physically possible for this type of magnet.

Why Does This Matter?

This paper is a game-changer because it proves that computers can do the heavy lifting in chemistry.

• Before: Chemists would mix chemicals, wait for them to crystallize, and hope for the best. It was slow, expensive, and relied on "gut feeling."
• Now: We can simulate thousands of possibilities, find the ones that should work based on physics, and then tell the chemists: "Go build these specific ones."

The Big Picture

Think of this as moving from hand-crafting a single perfect watch to 3D printing millions of watch designs to find the one that keeps perfect time. The authors showed that by using high-speed computer simulations to explore the "chemical universe," we can discover materials that human intuition would never have guessed. They didn't just find a better magnet; they found a new way to design the future of magnetic technology.

Technical Summary

1. Problem Statement

The design of Single-Molecule Magnets (SMMs), particularly those based on Dysprosium (Dy(III)) ions, relies heavily on achieving large magnetic anisotropy and slow magnetic relaxation. These properties are governed by the crystal field splitting (CFS) of the $J=15/2$ ground state multiplet.

• Current Limitations: While synthetic strategies (e.g., using strong axial donors and weak equatorial ligands) have been established, the exploration of the chemical space for mononuclear Dy complexes has been extremely limited. Existing databases (like SIMDAVIS) contain only ~650 entries, representing a tiny fraction of the available chemical space (compared to millions of organic molecules in the CCSD).
• Knowledge Gap: There is a lack of systematic understanding regarding how the second coordination sphere (the organic ligands surrounding the first coordination shell) influences magnetic properties. Current design rules focus almost exclusively on the immediate binding atoms, potentially missing crucial structural perturbations that could maximize anisotropy.
• Challenge: Traditional experimental screening is slow, and computational methods for magnetic properties (requiring multireference methods) are computationally expensive and difficult to automate for high-throughput screening.

2. Methodology

The authors developed a fully automated, high-throughput computational framework to screen existing crystallographic data and generate novel molecular prototypes.

A. Screening of Existing Crystallographic Databases

• Data Collection: Extracted all mononuclear Dy complexes from the Crystallographic Open Database (COD), Cambridge Structural Database (CCSD), and SIMDAVIS.
• Preprocessing:
  • Removed disordered atoms and polynuclear complexes.
  • Used Periodic DFT (pDFT) to determine the correct charge and spin multiplicity of the unit cell, ensuring the isolated molecule's electronic state was physically consistent.
• Electronic Structure Calculation:
  • Employed CASSCF (Complete Active Space Self-Consistent Field) calculations to compute magnetic properties.
  • Convergence Strategy: Developed a novel workflow to ensure stable convergence of CASSCF for Dy complexes. This involved using DFT natural orbitals as an initial guess, automatically rotating orbitals to ensure a dominant $4f$ character (at least 30%), and constructing active spaces from ligand framework orbitals combined with isolated Dy(III) orbitals if necessary.
  • Calculated the energy gaps between Kramers Doublets (KDs), specifically $\Delta E_{01}$ (ground to 1st excited) and $\Delta E_{07}$ (ground to 7th excited), and effective $g$-tensors.

B. High-Throughput Generation of Novel Molecules

• Template Selection: Selected a stable, pentagonal bipyramidal (PBP) reference complex: $[(tBuPO(NHiPr)_2)_2Dy(H_2O)_5]^{3+}$. This geometry is known to be promising for high anisotropy.
• Ligand Library: Extracted ~200,000 organic ligands (anions and neutrals) from the QM9star database.
• Sampling: Used Principal Component Analysis (PCA) on atomic fingerprints (bispectrum components) of the ligand binding atoms to select a diverse, representative subset of ~34,000 ligands.
• Assembly & Optimization:
  • Automated assembly of $[Dy(H_2O)_5L_2]^{n-}$ complexes.
  • Two-step DFT Optimization: First, constrained to maintain PBP symmetry; second, unconstrained to find the global minimum geometry.
  • Replaced the Dy core with Yttrium (Y) during DFT geometry optimization to avoid convergence issues with $f$-electrons, then swapped back to Dy for CASSCF calculations.
• Final Dataset: Generated and simulated 25,024 new Dy complexes.

3. Key Contributions

1. First Large-Scale Multireference Screening: Successfully performed CASSCF calculations on over 30,000 Dy complexes (631 from databases + 25,024 generated), a feat previously unattainable due to computational cost.
2. Automated Workflow: Created a robust pipeline for extracting magnetic properties from crystallographic data and generating new molecules, handling charge/multiplicity determination and orbital convergence automatically.
3. Discovery of Second-Sphere Effects: Demonstrated that fine-tuning the second coordination sphere (organic ligand backbone) is as critical as the first coordination sphere in maximizing magnetic anisotropy.
4. Identification of Stabilization Mechanisms: Identified specific supramolecular interactions (H···H contacts) that lock the local geometry into an ideal $D_{5h}$ symmetry, which is essential for record-breaking anisotropy.

4. Key Results

A. Analysis of Existing Databases

• Performance: The best molecule in existing databases was the dysprosocenium $[Dy(Cpttt)_2]^+$, with $\Delta E_{01} \approx 570 \text{ cm}^{-1}$ and $\Delta E_{07} \approx 1703 \text{ cm}^{-1}$.
• Symmetry Trends: Compounds with $D_{5h}$ symmetry showed the highest $\Delta E_{01}$ values. However, even among $D_{5h}$ candidates, large variations existed, suggesting that symmetry alone is insufficient without specific structural details.
• Limitation: No new record-breaking molecules were found in the existing crystallographic databases; the "best" known molecules were already identified.

B. Results from Generated Molecules

• Record Breaking: The high-throughput generation identified molecules with $\Delta E_{07} > 1600 \text{ cm}^{-1}$.
  • This represents a ~100% increase in magnetic anisotropy compared to the reference compound.
  • A ~30% increase over any known pentagonal bipyramidal Dy complex.
  • Values approaching those of pseudo bi-coordinated Dy ions (the current theoretical limit).
• Ligand Influence:
  • Axial Donors: Oxygen-based axial donors were found to be superior to Nitrogen-based ones for maximizing splitting.
  • Structural Parameters: The study revealed that the angle between axial ligands ($\Theta$) and the planarity of the equatorial water molecules (MPP score) are critical.
• Stabilization Mechanisms: Three distinct classes of ligand interactions were identified that stabilize the ideal $D_{5h}$ geometry:
  1. (O-type): Interaction between water protons and axial oxygen atoms.
  2. (MC-type): Interaction between water protons and multiple axial carbon atoms.
  3. (SC-type): Interaction between water protons and hydrogens from a single carbon atom (often in a $CH_2-CH_2$ group).
  • The (SC-type) interactions yielded the highest energy gaps ($\Delta E_{01} > 630 \text{ cm}^{-1}$).

C. Magneto-Structural Correlations

• High $\Delta E_{01}$ correlates strongly with near-ideal $g_z$ values (close to the theoretical limit of 20 for the ground state) and minimal $g_x/g_y$ mixing, indicating pure $M_J$ states.
• The study confirms that while $D_{5h}$ symmetry is necessary, it is not sufficient; the specific supramolecular interactions (H-bonding networks) within the second coordination sphere are the "missing key" to achieving ideal magnetic anisotropy.

5. Significance

• Paradigm Shift in SMM Design: The work challenges the conventional wisdom that only the first coordination shell matters. It establishes that supramolecular interactions in the second sphere are a powerful, untapped degree of freedom for tuning magnetic properties.
• Accelerated Discovery: The study proves that automated, high-throughput multireference simulations can rapidly explore chemical spaces that are inaccessible to traditional experimental trial-and-error or intuition-based design.
• General Applicability: The computational framework is not limited to Dy or SMMs; it can be extended to other lanthanides, actinides, and magnetic properties (e.g., excited states), bridging the gap between quantum chemistry and materials discovery.
• Future Outlook: The authors suggest that future work should integrate crystal environment modeling and synthetic feasibility scoring (e.g., via machine learning) to move from "in silico" design to practical synthesis.

In summary, this paper provides a landmark demonstration of how computational high-throughput screening, combined with advanced multireference quantum chemistry, can uncover non-intuitive design principles (specifically the role of second-sphere supramolecular interactions) to push the boundaries of magnetic anisotropy in molecular magnets.