In the year 2043, Nostradamus predicts that the planet Hercolubus will arrive on Earth (fake news, but funky way to start a story). QuantumRun predicts that teleportation of the first piece of physical mail will be achieved. I predict that the first of January 2043 will be a sunday – it is correct, if you are wondering. Now, what about quantum computing? Will we have quantum computers in our phones? Let’s not get there. Let’s talk instead about one of my favorite topics — computational chemistry.
While I was writing a short historical overview of how we went from solving the Schrödinger equation by hand to running the first quantum chemical calculations on a computer, I stumbled upon a pretty cool recent article by Stefan Grimme and Peter Schreiner — Computational Chemistry: The Fate of Current Methods and Future Challenges.
A lot has been written about the challenges of theoretical methods applied to chemistry, but in their article, Stefan and Peter take an intriguing perspective as they travel 25 years in the future – to 2043.
They have both been actively working in the field – in fact, I’ve studied several articles and applied methods developed by Stefan Grimme. Trust me, we can trust them: They’re well-equipped to read through the computational crystal ball and make prophecies about what lies in our far tomorrows.
But before we dive in, let’s clear something up real quick.
Quantum Chemistry vs. Computational Chemistry – A Flash Detour
Long story short: quantum chemistry focuses on the application of quantum mechanics to the study of chemical systems and describes the behavior of electrons in atoms and molecules using mathematical models based on quantum mechanics, such as the Schrödinger equation.
Computational chemistry is an umbrella term that covers a wide range of computational methods from quantum mechanics to classical physics-based approaches, such as molecular mechanics and molecular dynamics.
And now, without any further ado…
Computational Chemistry in 2043
– We Will Go “Full” Quantum
Nowadays, when it comes to studying massive systems like DNA/RNA chains, (membrane) proteins, polymers or even plastics, we’ve gotta switch up our methods, from quantum to classical.
Why, you ask?
Well, these are hundred of thousands of atoms at once and as the system gets larger, the less quantum our descriptions can be.
So, how do we tackle these gargantuan systems?
We turn to classical newtonian physics. Instead of getting all up in the atoms and electrons and solving Schrodinger-like equations, we treat those beasts as it they were made of point particles, like little balls attached by springs. We call this ensemble of parameters “force fields,” and they’re our go-to for studying large molecules.
In the year 2043, we’ll be going full quantum. Quick and dirty quantum methods will replace these classical approaches. Force fields will still be used, but for microscale processes, like cellular phenomena or entire battery cells.
In a nutshell: by the year 2043, the classical in computational chemistry will be increasingly replaced by the quantum.
– Wiggling Atoms everywhere
Atoms can move in so many complex ways – like wiggling or jiggling – and this can have a huge impact on how a reaction proceeds, a protein behaves, or a material absorbs light and conducts electricity. By understanding these dynamics, we can predict what will happen, control it, or design better processes.
Pretty neat, right?
Well, here is the catch: to simplify the electronic problem and make computations more efficient (or in some cases, even possible), we, quantum chemists, make use of the Born-Oppenheimer approximation, which assumes that the nuclei are fixed while the electrons move. Think of the nuclei as the poles, and the electrons, well, as the sexy performers who twirl around them.
Quantum chemistry focuses on the description of electrons, while nuclear dynamics is treated with statistical or with force-field-based methods, the balls and springs we talked about before.
We can do that, certainly, but that’s not entirely accurate. Key aspects like quantum tunneling and precise spectroscopy (needed, for instance, in astrochemistry) require us to consider the quantum nature of nuclei.
Now, here where the fun starts: by the year 2043, nuclear dynamics will be included accurately and routinely in our quantum chemical simulations.
How cool is that?
Daunomycin-DNA intercalation
This classical force-field based simulation (molecular dynamics) is a test calculation that my co-workers and I performed to support our investigation on Copper−1,10-Phenanthroline Complexes Binding to DNA. It contains 20.000 atoms (including water and ions). Environmental molecules are not shown for clarity.
– Quantum Nanoreactors
Again I ask you to open your (non-existing?) third eye and see the future. A chemist is sitting at their computer, dreaming of discovering new chemical reactions without even setting foot in a lab. Sounds like science fiction, right? Well, it is to some extent, but it’s becoming more and more of a reality every day.
Thanks to the ab initio nanoreactor approach, chemists can already use computers to simulate chemical reactions before ever carrying them out in their beakers. This has led to some exciting discoveries, like the creation of glycine from basic materials.
So, what does the future hold?
By 2043, we will regularly use quantum chemistry not just to interpret experimental results, but as a tool for discovery.
– Intelligent Computational Chemistry
Picture this: You open your ChemGPT webapp (oh, that’s a good idea). You give it a 3D structure as an input and say: I want to estimate the electrical conductivity or want to get an accurate IR spectra, or know which reactions would happen at a certain temperature. Now, your ChemGPT will evaluate the structure (is it a molecule? a surface? a protein?), identify the challenges to calculate the desired property, choose the correct quantum/classical method to employ for the calculation, run it, and give you a simple answer.
Maybe we won’t ever have such a smart chemGPT, but it may be smart enough to adapt to the situation and use different methods depending on what it needs to calculate. Basically, our chemGPT could combine a super detailed expert system with a multi-level method, so that it can decide to use a quantum accurate approach for the layer of the molecular system that needs it, but switch to something simpler, like a quick and dirty quantum level, for layers that are farther away from the chemical center of interest.
By 2043, chemoinformatics, machine learning, and other intelligent methods will become an integral component of computational chemistry.
A Final Personal Touch
It’s 2043. I wake up in the morning, my back aches while I brew myself a refreshing cup of coffee. I step out onto the porch of my beach house. The sun is shining, and my electrical motorbike is parked in the shade of a palm tree. I grab my coffee and join my colleagues for a virtual discussion on the latest features of our simulation platform. We are running our calculations on cloud-based hybrid quantum-classical computers to improve the performance of a carbon capture system.
Welcome to the future of computational chemistry, where even systems of thousands, milions, or, who knows, billions, can be fully and accurately modeled in their dynamics, thermodynamics, and kinetics. Where automated quantum simulations will predict the outcome of reactions and material processes to address real-world scientific challenges.
Let’s be clear: there are mountains to climb ahead. Peaks of immense heights. But, oh boy, how beautiful is the landscape on the other side.
