Bulk materials with Intrinsic Intermediate Band: data study and applicability to
solar energy conversion
Intermediate band (IB) materials are of great interest when considered the possible applications of its
unique properties. The existence of a narrow band of energy in the surroundings of the Fermi level is intriguing
and still presents multiple questions to answer, mainly for the less common compounds that present
this property naturally in the bulk form. Solar cell materials are being studied for decades as great candidates
for clean energy production and have been subject to constant development. At least since the 1990
decade, the concept of IB materials applied for solar cells has gained widespread interest. The possibility of
searching for materials with intrinsic IB to be applied in solar cells was propelled by recent advancements
in computational calculations and the creation of materials repositories. Using DFT (Density Functional
Theory) based simulations and computational screening, we could select metallic IB compounds that satisfied
the conditions for efficient application in solar energy conversion, presenting 3 previously synthesized
candidates (Bi2Rh2O7, Ca5FeN6, and OsTb6I10). Through machine learning and data mining, we could expand
the number of metallic IBs, identifying prototypical structures able to provide minimal features to build
new metallic IB materials, reporting 68 novel stable results and relating ligand field splitting to its origins.
The existence of this intermediate band for both metals and semiconductors was associated with variables
that could lead to more confined states as bigger radii of separators in ternaries, lower dimensionality and
packing, besides the presence of highly electronegative elements. The presence of d states and the number
of electrons were identified as important in order to separate IB materials in terms of their occupation level.
Between the semiconductor IBs, we found compounds with the expected characteristics for thermoelectric
applications. The information collected was enough to use classification techniques that could provide combinations
of elements and formulas leading to the prediction of materials with higher chances of having IBs,
reaching rates above 80%.