Materials scientists' bread and butter has always been deciphering the structural features of molecules found in nature or manufactured in the lab. However, as science and technology improve, the task of discovering novel materials with highly desirable qualities has grown even more ambitious. Materials scientists use advanced simulation tools that involve quantum mechanics, the same rules that govern molecules themselves, to perform such a feat systematically.
The
simulation-based technique has shown to be extremely effective, to the point
where an entire branch of study called materials informatics has been created
to investigate it. However, there have been some instances of failure.
Disiloxane, a silicon (Si)-containing molecule having a Si-O-Si bridge with
three hydrogen atoms at each end, is a prominent example. The construction is
simple enough, but estimating how much energy is required to bend the Si-O-Si
bridge is notoriously difficult. Due to the sensitivity of the calculated
properties to parameter selections and level of theory, experimental results
have been inconsistent, and theoretical calculations have returned wildly
disparate values.
Fortunately,
this difficulty has recently been solved by an international research team lead
by Dr. Kenta Hongo, Associate Professor at Japan Advanced Institute of Science
and Technology. The team accomplished this feat in their study, which was
published in Physical Chemistry Chemical Physics, by employing a cutting-edge
simulation technique known as the "first-principles quantum Monte Carlo
method," which finally overcame the challenges that other standard
techniques had failed to overcome.
Is it all
just a matter of better simulations? That's not the case. "It's not
surprising to get a result that differs from the experimentally determined
value. With more thorough, and often more expensive, simulations, the agreement
can be improved. With disiloxane, however, more precise simulations actually
make the agreement worse "Dr. Hongo argues. "Rather, what our method
has achieved is good results with little reliance on the adjustment parameters,
so we don't have to worry about whether the adjusted values are adequate."
The
researchers compared the first-principles quantum Monte Carlo method to other
conventional techniques like DFT computations and the "coupled cluster
method with single and double substitutions and noniterative triples"
(CCSD(T)), as well as actual observations from prior investigations. The key
difference between the three approaches was their sensitivity to the "completeness"
of basis sets (a set of functions used to define the quantum wavefunctions).
The choice
of basis set affects the amplitude as well as the places of zero amplitude for
the wavefunctions in DFT and CCSD(T), but only the zero amplitude positions in
quantum Monte Carlo. This allowed the amplitude to be tweaked until the
wavefunction shape resembled that of an exact solution. "In computing the
bending energy barrier, this self-healing property of the amplitude works well
to reduce basis-set reliance and lower the bias coming from an incomplete basis
set," Dr. Hongo explains.
While this is a noteworthy development in and of itself, Prof. Hongo emphasises the larger picture. "New medications and catalysts are frequently designed using molecular simulations. Eliminating the fundamental obstacles in employing them makes a significant contribution to the design of such materials. The way we utilised in our work could become a typical strategy for tackling such challenges, thanks to our powerful supercomputers "he declares.
References:
Adie Tri Hanindriyo et al, Diffusion Monte Carlo evaluation of disiloxane linearisation barrier, Physical Chemistry Chemical Physics (2022). DOI: 10.1039/D1CP01471D
.jpg)
0 Comments