Keith T.
Butler
*a,
Yu
Kumagai
b,
Fumiyasu
Oba
bcd and
Aron
Walsh
ae
aCentre for Sustainable Chemical Technologies and Department of Chemistry, University of Bath, Claverton Down, Bath BA2 7AY, UK. E-mail: k.t.butler@bath.ac.uk
bMaterials Research Center for Element Strategy, Tokyo Institute of Technology, Yokohama 226-8503, Japan
cMaterials and Structures Laboratory, Tokyo Institute of Technology, Yokohama 226-8503, Japan
dCenter for Materials Research by Information Integration, National Institute for Materials Science, Tsukuba 305-0047, Japan
eGlobal E3 Institute and Department of Materials Science and Engineering, Yonsei University, Seoul 120-749, Korea
First published on 8th January 2016
The rapid progress in performance of solar cells based on hybrid halide perovskites means that devices based on these materials have reached a stage where research interest can now focus on development of robust technology. One of the key questions relating to these (and indeed any) devices is their lifetime and stability which in turn is often influenced by the quality of interfaces and junctions within the device. In this study we present a methodology which allows screening for mechanically stable, electronically suitable interface combinations – applying the technique to screen 175 common semiconductors for viability as electron and hole extracting contacts for CH3NH3PbI3. The screening method can be applied to any semiconductor junction problem and relies on easily obtained experimental or theoretical information – electron affinity, ionisation potential, lattice parameters and crystal structure. From the screening we rank the candidates according to a figure of merit, which accounts for band alignment and chemical/mechanical stability of the interface. Our screening predicts stable interfaces with commonly applied electron extraction layers such as TiO2 and ZnO as well giving insight into the optimal polymorphs, surfaces and morphologies for achieving good quality contacts. Finally we also predict potentially effective new hole and electron extraction layers, namely Cu2O, FeO, SiC, GaN, and ZnTe.
To date most of the research on hybrid perovskite solar cells has focused on the optimisation of the absorber layer. However, device optimisation also requires close attention be paid to contacting materials (for charge extraction) and the interfaces between materials. Indeed, single crystals have been found to have extraordinarily long carrier lifetime7 and it has been noted in several studies that the majority of efficiency loss in hybrid perovskite devices is related to electron–hole recombination in the interface region.8,9 In more mature technologies such as c-Si solar cells a lot of attention has been paid to the effects of orientation,10 doping11 and strain12 at interfaces. The challenge of designing new contact layers must address two crucial parameters: (i) good alignment of electronic energies at the interface,13 to allow for efficient charge extraction, while minimising energy loss; (ii) robust mechanical contacts with minimal interfacial defects, to ensure maximal durability and minimise interface carrier recombination sites.
Typical hybrid perovskite cells have been based on TiO2 electron extraction contacts and Spiro-oMeTAD hole extraction contacts.14,15 Empirically this architecture has been found to give the optimal efficiency although stability remains a concern.16 Nonetheless, based on the principles outlined above it is possible to imagine a host of alternative architectures with the potential to improve stability and device performance. Recently electron contacts based on ZnO17–20 and all solid-state inorganic cells with CuI hole extraction layers21 have been reported. In the case of the latter cells lifetimes and cyclability were found to be significantly improved by the inclusion of the CuI layer. There has also been active research for improved organic hole extraction materials, with modular analouges predicted to give improved band offsets with the absorber layer.22
The choice of contacting semiconductors is almost endless and the ability to efficiently screen potential candidates, based on easily obtained experimental or theoretical parameters would allow for a significant reduction in the trial-and-error studies required to establish new combinations. In this contribution we outline such a methodology and apply our scheme to a set of 175 semiconducting materials, identifying 17 candidate hole or electron contacts. Encouragingly our method predicts the solid-state materials most commonly applied in halide perovskite devices (organic materials are beyond the scope of the current methodology) as well as highlighting potential superior alternatives.
Interfacial phenomena can become extremely complicated and it is becoming clear that effects such as local ordering of dipoles23–25 and the role of chemical reactions26,27 and lattice distortions28 can be crucial for dictating the success or failure of a given system. The screening procedure outlined here cannot be considered to account for every possible effect of interface formation. However, the criteria we outline serve as necessary conditions for the formation of high quality interfaces, as such their application in the design of new architectures and materials combinations can lead to a focusing of the search on more relevant systems.
We also develop a figure of merit to rank the materials considered in our screening procedure for their application as photovoltaic contacts. This figure of merit ranks the interface matching out of 100, where a score of 100 represents an ideal interface of the material with itself, having a zero band offset and perfect epitaxial matching.
The electronic alignment is included in the figure of merit using an exponential function of the offset. This is physically motivated for both positive and negative offsets at the interface. For barriers to charge transport it is well established that contact resistance depends exponentially on the barrier height. It has also been shown by Niemegeers and co-workers30 that a negative offset has the same exponential effect on the interface thermionic current. At a barrier of 0.5 eV cell performances are sharply decreased. The offset contribution is subtracted from the overall figure of merit so we set the condition that this contribution equals 100 at an offset of 0.5 eV, meaning a figure of merit of zero above this.
For the purposes of our figure of merit we assume that dislocations act as recombination centres. The number of dislocations increases linearly with the interface strain, up to a critical value, after which the interface becomes fully incoherent and impractical for device application. Materials above 5% mismatch will form incoherent interfaces and therefore are not considered in our study, up to 5% the number of defects formed will be proportional to the strain and we introduce this into the denominator of the figure of merit.
Considering these contributions we have a final ELS figure of merit for the interface of
(1) |
A selection of IP, EA, and Eg values from experiment and high-level theory, as well as some estimated by electronegativity arguments,33–76 for 175 binary and ternary and quaternary semiconductors was assembled for the purposes of this screening. The full list of screened values is available in the ESI.†
The nature of the contact will also depend on the conduction mechanism of the semiconductor (p- or n-type), which is determined by the doping limits77 and growth conditions. This level of prediction is beyond the scope of the screening procedure carried out in our study and the dopability of candidate materials may be assessed in further studies or by reference to the literature after our full screening is applied.
This lattice matching is characterised by the precision of the match between cells and the area of the cells required to obtain that precision. For example one side of the interface (A) may have a lattice constant of 5.129 Å and the other a lattice (B) has a constant of 3.840 Å we then find that an expansion of 3 × A and 4 × B gives a lattice mismatch of around 0.2%. One could imagine that by applying such supercell expansions matching of almost any lattice pair is possible and therefore we set an upper boundary on the size to be considered, in our case we allow the lattice to expand up to five times in either surface vector.
The precision of the match gives a lower bound to the amount of strain required to form chemical bonds across the interface. For example if a precision of 1% is found for a pair of iso-structural materials, this means that the minimal distance that atoms on either side of the interface should shift to achieve bonding is ±0.5%. In general the strain induced by this shifting is alleviated by the formation of dislocations. Above a certain threshold of matching the strain can no longer be accommodated and incoherent interfaces, with extremely high defect concentrations, occur. As we wish to avoid such a situation, in our study we set a precision threshold of 5%.
The first problem when assessing the fit of two surfaces cleaved from a crystal is how to define the surface lattice. For example two similar lattices may be a rotation of one another. The lattices can be defined by their primitive vectors, however the choice of primitive vectors is not unique. Zur and McGill presented a reduction scheme which selects a set of primitive vectors for a surface in an almost unique way, selecting a set of primitive vectors independent of any rotations or reflections of the lattice.
The lattice matching technique applied here has previously been used to search for epitaxial heterointerfaces. It has been shown to robustly predict experimentally verified interface orientations of CdTe on GaAs,78 CdTe on sapphire,78 transition metal silicides on Si,79 PtNi3 on PtCo380 and Si on GaP.80
We begin by cleaving all of the low index surfaces of the materials identified in the electronic screening step, that is (h, k, l) h, k, l = 1 or 0, h + k + l > 0. We then apply the reduction scheme of Zur and McGill to obtain the primitive vectors (Fig. 2). Next we compare these primitive vectors with those of (110) and (100) surface of the pseudo-cubic CH3NH3PbI3 structure, using the supercell expansion and precision criteria outlined above. We note that at 300 K MAPI can exhibit a tetragonal distortion with c/a ≃ 1.01;81 however, the pseudo-cubic phase has also been reported for thin films at this temperature.82
Fig. 2 Methodology for screening for structural epitaxy. (a) Surface cells are defined based on all low index terminations. (b) The algorithm of Zur and McGill78 is applied to ensure that all surface cells are reduced to a reduced set of primitive translations. (c) Screening of supercell expansions of each material (up to a certain threshold) for lattice matching within a certain mismatch percentage. |
To further assess the chemical viability of an interface we search for interfaces with the maximal overlap of atomic positions. We begin our procedure by defining the atom positions in the surface plane of two materials (see Fig. 3). In many cases for complex materials there is more than one surface termination possible for a given plane. We enumerate and consider all surfaces which have non-polar terminations (type-I or type-II in Tasker classification83); this enumeration is performed using the METADISE84 code.
Next we expand the lattices by the ratios determined in the previous screening step for epitaxy. In our case we are examining systems with different crystal structures, so we will not necessarily obtain complete overlap of lattice sites. Instead we define a cutoff radius (5% of the CH3NH3PbI3(110) u vector, or 0.3 Å), if we obtain overlap within that radius we count the sites as coincident (filled burgundy circles in Fig. 4(b)). We apply translations of the contacting surface with respect to the CH3NH3PbI3 surface, selecting the configuration with maximum overlap to calculate our atomic site overlap (ASO) ratio. The ASO ratio is defined as
(2) |
Fig. 4 Conditions for the identification of atomic site overlap (ASO) of two surface cells (a). The atomic positions of the two cells are overlayed. (b) If the centres are within a threshold separation they are counted as coinciding (filled green circles). (c) The lattices are translated with respect to one another in order to maximise the ASO factor (eqn (2)). |
We note that we have taken no account of the nature of the atoms occupying the lattice sites, simply treating all points as equal. For a more rigorous assessment of the suitability of a combination of materials surfaces one could take into account the formal charges of the species at each site, such that positive and negative ions would compensate for one another. Furthermore, quantities such as ionic radii and chemical hardness/softness of the species occupying a site can give further insight into the propensity for two surfaces to form mechanically stable interfaces. In the context of this methodology, which is primarily designed to provide a swift and simple screening procedure for the identification of possible promising interface matches, however, we neglect the difference between sites. To test the validity of our approximations we have screened possible interfaces of GaN/ZnO, the procedure is outlined in detail in the supporting information, the combination of lattice and site screening correctly identifies the experimentally observed interfaces, filtering out less favourable combinations that have not been experimentally observed.
Material | Surface | ΔVBM (eV) | ΔCBM (eV) | Multiplicity A | Multiplicity B | Strain (%) | ASO | ELS | |||
---|---|---|---|---|---|---|---|---|---|---|---|
u | v | u | v | u | v | ||||||
MAPI | 110 | ||||||||||
Ce2O3 | 110 | 0 | 1 | 4 | 1 | 3 | 1.8 | 2.4 | 0.3 | 10 | |
Ce2S3 | 101 | 0.0 | 3 | 1 | 2 | 2 | 2.4 | 1.9 | 0.3 | 10 | |
CoO | 110 | −0.4 | 3 | 3 | 2 | 2 | 1.5 | 1.8 | 0.3 | 8 | |
Cu2O | 110 | 0.0 | 3 | 3 | 2 | 2 | 1.6 | 1.9 | 0.3 | 10 | |
CuI | 001 | 0.1 | 1 | 3 | 1 | 2 | 3.8 | 2.3 | 0.3 | 8 | |
CuI | 110 | 0.1 | 1 | 1 | 1 | 1 | 3.8 | 3.5 | 0.6 | 12 | |
FeO | 100 | −0.3 | 3 | 2 | 2 | 1 | 2.6 | 0.3 | 0.3 | 11 | |
FeO | 110 | −0.3 | 3 | 3 | 2 | 2 | 2.6 | 2.8 | 0.3 | 7 | |
GaN | 100 | −0.1 | 2 | 5 | 1 | 3 | 1.3 | 2.7 | 0.6 | 18 | |
GaN | 110 | −0.1 | 5 | 5 | 1 | 3 | 3.0 | 3.6 | 0.4 | 8 | |
In2O3 | 110 | 0.3 | 3 | 3 | 5 | 5 | 3.6 | 3.4 | 0.1 | 2 | |
LiNbO3 | 100 | 0.2 | 5 | 2 | 4 | 3 | 2.8 | 4.0 | 0.3 | 7 | |
MnTiO3 | 100 | 0.0 | 5 | 3 | 4 | 5 | 1.9 | 3.6 | 0.3 | 8 | |
NaNBO3 | 110 | 0.2 | 5 | 5 | 3 | 3 | 3.3 | 3.6 | 0.2 | 4 | |
NiO | 110 | −0.4 | 3 | 3 | 2 | 2 | 0.4 | 0.2 | 0.3 | 18 | |
Pr2S3 | 101 | 0.1 | 3 | 1 | 2 | 2 | 3.5 | 2.6 | 0.3 | 7 | |
SiC | 110 | 0 | 5 | 5 | 3 | 3 | 0.2 | 0.0 | 0.5 | 45 | |
Sm2S3 | 110 | 0.3 | 3 | 3 | 4 | 5 | 0.2 | 3.7 | 0.4 | 12 | |
SnS2 | 100 | −0.2 | 5 | 3 | 3 | 2 | 3.8 | 0.7 | 0.3 | 9 | |
SrTiO3 | 110 | −0.1 | 5 | 5 | 3 | 3 | 3.3 | 3.5 | 0.4 | 10 | |
Tb2S3 | 110 | 0.3 | 3 | 4 | 5 | 5 | 0.0 | 2.3 | 0.3 | 11 | |
TiO2 r | 011 | −0.2 | 4 | 5 | 3 | 3 | 2.6 | 2.5 | 0.3 | 8 | |
ZnO | 100 | −0.2 | 2 | 5 | 1 | 3 | 3.2 | 2.2 | 0.6 | 14 | |
ZnTe | 100 | −0.1 | 1 | 3 | 1 | 2 | 3.1 | 3.0 | 0.3 | 8 | |
ZnTe | 110 | −0.1 | 1 | 1 | 1 | 1 | 3.1 | 2.9 | 0.4 | 11 | |
ZnTiO3 c | 011 | −0.3 | 5 | 5 | 4 | 3 | 0.1 | 3.0 | 0.3 | 9 | |
MAPI | 100 | ||||||||||
Ce2S3 | 100 | 0.0 | 3 | 2 | 2 | 5 | 2.1 | 0.0 | 0.2 | 8 | |
CoO | 100 | −0.4 | 3 | 3 | 2 | 2 | 1.9 | 1.5 | 0.3 | 8 | |
Cu2O | 100 | 0.0 | 3 | 3 | 2 | 2 | 2.0 | 1.6 | 0.3 | 10 | |
CuI | 100 | 0.1 | 1 | 1 | 1 | 1 | 3.4 | 3.8 | 0.8 | 17 | |
CuI | 110 | 0.1 | 1 | 3 | 1 | 4 | 3.4 | 2.1 | 0.4 | 11 | |
FeO | 100 | −0.3 | 3 | 3 | 2 | 2 | 2.9 | 2.6 | 0.3 | 7 | |
FeO | 110 | −0.3 | 3 | 1 | 2 | 1 | 2.9 | 3.3 | 0.7 | 13 | |
GaN | 100 | −0.1 | 2 | 5 | 1 | 4 | 1.7 | 3.0 | 0.6 | 17 | |
In2O3 | 100 | −0.3 | 3 | 3 | 5 | 5 | 3.3 | 3.7 | 0.1 | 2 | |
In2S3 | 100 | 0.0 | 4 | 1 | 5 | 5 | 2.9 | 2.7 | 0.1 | 4 | |
NaNbO3 | 100 | 0.2 | 5 | 5 | 3 | 3 | 3.7 | 3.3 | 0.4 | 8 | |
Nd2S3 | 110 | 0.3 | 3 | 2 | 4 | 5 | 1.1 | 1.4 | 0.2 | 7 | |
NiO | 100 | −0.4 | 3 | 3 | 2 | 2 | 0.1 | 0.4 | 0.3 | 19 | |
Pr2S3 | 100 | 0.1 | 3 | 2 | 2 | 5 | 3.1 | 0.8 | 0.2 | 6 | |
SiC | 100 | 0.0 | 2 | 5 | 1 | 4 | 1.8 | 0.3 | 0.6 | 27 | |
Sm2S3 | 110 | 0.3 | 3 | 2 | 4 | 5 | 0.1 | 2.4 | 0.2 | 7 | |
SnO2 | 001 | 0.0 | 4 | 4 | 3 | 3 | 0.7 | 0.3 | 0.4 | 25 | |
SrTiO3 | 100 | −0.1 | 5 | 5 | 3 | 3 | 3.6 | 3.3 | 0.4 | 10 | |
Tb2S3 | 010 | 0.3 | 3 | 3 | 5 | 5 | 0.3 | 1.7 | 0.2 | 9 | |
TiO2 a | 001 | −0.2 | 5 | 5 | 3 | 3 | 0.5 | 0.2 | 0.4 | 27 | |
TiO2 a | 100 | −0.2 | 5 | 2 | 3 | 3 | 0.5 | 0.7 | 0.4 | 23 | |
TiO2 a | 101 | −0.2 | 5 | 3 | 3 | 5 | 0.5 | 2.4 | 0.2 | 7 | |
TiO2 r | 001 | −0.2 | 4 | 4 | 3 | 3 | 2.4 | 2.8 | 0.4 | 10 | |
ZnO | 100 | −0.2 | 2 | 5 | 1 | 4 | 3.6 | 3.4 | 0.6 | 12 | |
ZnTe | 100 | −0.1 | 1 | 1 | 1 | 1 | 2.7 | 3.1 | 0.5 | 13 | |
ZnTe | 110 | −0.1 | 1 | 3 | 1 | 4 | 2.8 | 2.8 | 0.3 | 8 |
ZnO has also been reported as an electron extraction contact,18,19 allowing for lower temperature synthesis of working devices. Indeed the performance of devices containing TiO2, ZnO and CdS electron extraction layers has been reported.19 The authors report that TiO2 and ZnO cells show comparable performance, but the CdS based cells show much diminished efficiencies. Based on the alignment of energy levels there is little difference between the three materials for efficient electron extraction, indeed all three remain after the first step in our screening. However, based on interface epitaxy CdS was ruled out in our screening as it could not meet the criteria for mechanical strain. The strain at the CdS/absorber interface could lead to deformation of the band energies or the formation of defects at the interface, both of which would contribute to the diminished performance reported. The mismatch factor for ZnO, however is not as favourable as the best TiO2 interfaces, this, coupled with the chemistry of the interface, as recently suggested by De Angelis and co workers23 may also have a bearing on the stability.
SnO2 does not, in general, meet the criteria for good band alignment with CH3NH3PbI3; however, in this material the effect of surface termination on the EA and IP can be very pronounced.87 In this case the initial screening suggests that SnO2 does not form a good electronic contact. For our analysis we have included the (001) orientation, which has an IP calculated to be 7.47 eV and an EA of 4.04 eV, this is almost perfectly aligned to CH3NH3PbI3 for electron extraction without energy loss. In addition the (001) orientation shows minimal strain with CH3NH3PbI3 as well as good atomic site overlap. These findings suggest that growth of CH3NH3PbI3 on (001) oriented SnO2 could lead to excellent quality devices with clean interfaces, however, we would expect architectures based on SnO2 to be more sensitive to the growth conditions than those based on TiO2.
CuI has recently been reported as a hole extraction material to replace the usual organic material,21 with excellent photocurrent stability and high carrier mobility. The devices did display higher interface recombination than Spiro-OMeTAD based devices, which can be related to the barrier in band energy between the perovskite and CuI I. However, the lattice matching and ASO both suggest mechanically stable interface formation; experimentally the cells were found to be stable without encapsulation for 54 days.21 The screening methodology outlined here can be applied in the search for hole extraction materials with better band offsets and equivalent mechanical stability.
The new material that displays the most obvious potential for contacting CH3NH3PbI3 on the basis of the ELS figure of merit is SiC. SiC can act as a p- or n-type semiconductor depending on doping, it crystalises as many different polytypes. Due to this polytypism the ELS figure of merit is sensitive to growth conditions. The common 2H hexagonal polymorph shows excellent interfacial matching with pseudo-cubic CH3NH3PbI3. Details of the IP and EA would also be expected to be sensitive to the crystal structure88 and this system merits further investigation for potential application as a contacting layer in halide perovskite cells. The SiC band gap of 2.36 eV means that it is not totally transparent, therefore in terms of device design a back layer of SiC is preferable to a front contact.
Transition metal oxides (TMOs) are commonly applied in organic photovoltaics (OPV) as hole transport layers, so a direct application in the same context for CH3NH3PbI3 may seem reasonable. Of the TMOs used in OPV devices NiO is by far the most promising from our analysis. It has a reasonable match of IP with CH3NH3PbI3 and the matching of the (110) and (100) surfaces of the two materials is extremely good, in both cases the matching with NiO is as good as the matching with TiO2 and NiO has a high ELS value, we note that there have been initial reports of the successful application of NiO in halide perovskite solar cells.89,90 FeO also has promising ELS values for certain orientations, the (100) surface matches to CH3NH3PbI3(110) and vice versa. CoO has a promising band alignment with CH3NH3PbI3, but poor site matching at the interface means that the ELS value is relatively low. Other TMOs such as MoO3,WO3 and V2O5, which are commonly applied in organic photovoltaics, have too large values of IP to align well with CH3NH3PbI3.
GaN shows potential as an electron extraction layer suggested from our screening. It has excellent band alignment with CH3NH3PbI3 and the (100) surface matches well with the (110) surface of CH3NH3PbI3. It has ELS values similar to CuI and superior to ZnO. It also has the obvious advantage that it is well known as an n-type semiconductor91 and can be controllably doped with Si or O.
Cu2O has excellent band alignment with CH3NH3PbI3, suggesting potentially barrier-less contact. The lattice mismatch is reasonable and the ELS of 10 means that it is potentially useful as a contact material. Cu2O is also well known to have p-type conductivity controllable by the synthesis conditions92 and is earth-abundant making it highly suited in the context of photovoltaic energy generation.
ZnTe, although less earth-abundant than Cu2O, also shows promising interface mismatch and band alignment to CH3NH3PbI3. It is also a p-type semiconductor and is commonly applied in a similar role in the context of CdTe solar cells, where it is used as a hole extraction layer.
Ce2O3,Ce2S3 and Sm2S3 were all identified as having ELS values of 10 or greater; however, for reasons of resource scarcity or supply risk they are not considered any further. Ce is exceedingly rare with a very low crustal abundance, whilst Pr and Sm are both considered critical materials and have a high supply risk. The flexibility of our current screening procedure means that concerns of supply and abundance could easily be facilitated in the figure of merit, in a similar way to a recent study which screened earth abundant thermoelectrics.93
Oxide perovskites have been considered as electron extracting layers (in the case of SrTiO394). Sharing the same crystal structure as CH3NH3PbI3 means that they can have potentially superior matching of atomic sites at the interface. The perovskites studied here with cubic structures (SrTiO3 and NaNbO3) have a very good match in terms of EA for electron extraction from CH3NH3PbI3. In terms of interface mismatch they are a reasonable match for CH3NH3PbI3, with as good matching as, for example, commonly applied ZnO. Also, given the wide range of chemical formulas which conform to the Goldschmidt tolerance factor for cubic perovskites, it is possible to imagine designing of better matched alternatives. ZnTiO3 and MnTiO3 both crystallise in the rhombohedrally distorted perovskite structure; the R space group (number 148), nonetheless they have interface matching (with the (110) CH3NH3PbI3 surface) to the cubic species mentioned. In addition they have promising band alignment, with the caveat that these values are not from experiment or high level theory, but were estimated on the basis of Mulliken electronegativity of the compounds.
We applied this methodology to screen 175 common semiconductor materials for application as contacts (both hole and electron extracting) in hybrid perovskite solar cells. From our initial set 16 materials were identified as good candidates for contacting – including all of the materials in our test set which have been successfully applied in the hybrid perovskite devices previously. In addition we suggest possible new architectures, which could lead to improved performance or enhanced device stability.Those identified materials which were not ruled out due to supply concerns along with their ELS values are presented in an energy-band-alignment diagram in Fig. 5.
Fig. 5 Materials identified as having a favourable ELS value. The valence and conduction bands of each material are shaded according to the ELS value (colourbar to the right). Electron extraction layers are placed to the left of CH3NH3PbI3 (MAPI) and hole extraction layers are placed to the right. All electron energies are in (eV). Note several materials with ELS ≥10 are not included due to scarcity of constituent species. All values are presented in Table 1. |
It is important to emphasise that the ELS method presented identifies candidate contacting materials that meet a number of necessary conditions, it does not guarantee that a material will form a good contact. Other factors should also be assessed in order to discriminate amongst the promising materials. Importantly conductivity should be considered, this is a combination of carrier mobility and concentration. Both of these values are potentially accessible from first principles calculations and could be appended as a further computational screening step. Processability of a material is also an important practical concern, however this is difficult to predict from calculations and is best determined by the expert practitioner. Nonetheless, the application of our method significantly focuses the space within which these more detailed assessments must be performed.
The study presented was limited to a set of well known and studied semiconductor materials. However, with the advent of computational high-throughput screening of materials such a study can easily be extended to the search for new materials or novel combinations of old materials. The simplicity of the concepts and ease of computation mean that this methodology can be easily applied by researchers in the field of semiconductors with access to a reasonable desktop computer and a basic knowledge of modern computer scripting languages. As such we hope that this tool may be of use to experimental and modelling groups alike in the continuing search for functional materials combinations.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5tc04091d |
This journal is © The Royal Society of Chemistry 2016 |