Atomistic Solution to the Doping Bottleneck in Compound Semiconductors

Activation and tuning of the useful properties of semiconductors intended for photovoltaic, electronic, electrochemical and other applications hinges on their successful release of charge carriers, a process typically achieved in appreciable quantity via substitutional incorporation of foreign ions into the host crystal structure. As it stands, the rate-limiting step in designing efficient semiconductors is the identification of the most viable dopant ions to achieve n- and/or p-type conductivity – a process severely clouded by our incomplete understanding of chemical bonding behavior in solids. Today, doping studies rely exclusively on resource-intensive first-principles calculations, with little chemical guidance available to computationalists on where to focus their resources. An atomistic understanding of the factors underlying ion substitutions in solids would provide such guidance, and would further enable rapid calculations to be made for host/dopant combinations prior to their synthesis in lieu of expanding costly resources in low-return compositional spaces.<br/><br/>Three major factors underlie doping difficulties in semiconductors: [1] dopant insolubility, [2] dopant un-ionizability, and [3] dopant compensation. Factor [3] is terminal whereby formation of compensated defects is caused by the charge carrier itself, and not the identity of the dopant. However, difficulties associated with factors [1] and [2] may be circumvented via appropriate dopant selection, whose ease of substitution in the host crystal structure dictates dopant level depth in the band structure. Factors known to influence dopant level depth include similarities in atomic orbital energies, size of the substituting ions, and local bonding geometry – all of which amenable to guidance via an adequate, systematized understanding of crystal-chemical behavior.<br/><br/>By and large, the primary reason underlying our inability to model ion substitution in solids (and therefore semiconductor doping) has been our collective failure to systematically describe, rationalize and quantify pairwise chemical bonding behavior across the periodic table of elements. We recently remediated to this problem via wide-scale analysis of chemical bonding in solids in the form of bond-length dispersion analyses. Results were reported for cations bonded to O^2- and N^3-,^[1] covering 177,446 reliable bond lengths hand-picked from 9210 crystal-structure refinements for oxides, and 6,770 bond lengths from 720 crystal-structure refinements for nitrides; equivalent analysis for cations bonded to S^2-, Se^2- and Te^2- is underway. These data were subsequently used to identify and quantify structural, electronic and/or bond-topological mechanisms underlying anomalous bonding behavior across the periodic table.<br/><br/>Having quantified the most elusive of crystal-chemical variables for the very first time, i.e. chemical bonding behavior, we successfully developed an atomistic model of ion substitution which instantaneously and quantitatively predicts solubility limits between ion configurations at ambient conditions. Our model quantitatively reproduces doping profiles for all popular compound semiconductors (e.g. III-V, II-VI semiconductors), and can further make instantaneous predictions for the doping of ternary, quaternary and higher-dimensional systems. Thus, our model resolves the doping bottleneck in compound semiconductors while enabling the pursuit of complex, bespoke semiconductor design with bulk properties adapted to end-user needs.<br/><br/><u>References</u>:<br/>[1] Gagné, O. C. & Hawthorne, F. C. (2016). <i>Acta Cryst</i>. B72, 602-625; Gagné, O. C. (2018). <i>Acta Cryst</i>. B74, 49-62; Gagné, O. C. & Hawthorne, F. C. (2018a). <i>Acta Cryst</i>. B74, 63-78; Gagné, O. C. & Hawthorne, F. C. (2018b). <i>Acta Cryst</i>. B74, 79-96; Gagné, O. C. & Hawthorne, F. C. (2020). <i>IUCrJ</i> 7, 581-629; Gagné, O. C. (2021). <i>Chem. Sci</i>. 12, 4599-4622.