You are on page 1of 8

Article

pubs.acs.org/JPCC

Stabilities and Reconstructions of Clean PbS and PbSe Surfaces: DFT


Results and the Role of Dispersion Forces
Volker L. Deringer†,§ and Richard Dronskowski*,†,‡

Institute of Inorganic Chemistry, RWTH Aachen University, Landoltweg 1, 52056 Aachen, Germany

Jülich−Aachen Research Alliance (JARA-HPC), RWTH Aachen University, 52056 Aachen, Germany
*
S Supporting Information

ABSTRACT: Lead sulfide (PbS) and lead selenide (PbSe) are functional
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

materials with manifold applications. In particular, nanocrystals and other


nanostructures of PbS and PbSe are of mounting interest, which calls for a
Downloaded via SAINT MICHAEL'S COLG on September 17, 2023 at 21:23:13 (UTC).

thorough understanding of the underlying crystal surfaces. Here, we present a


comprehensive density-functional theory (DFT) survey of structures and
stabilities for the (001), (011), and (111) surfaces of rocksalt-type PbS and
PbSe. A representative set of possible reconstructions is explored for the polar
(111) surfaces, allowing us to suggest both 2 × 1 and “octopolar” 2 × 2 motifs
as favorable options; this agrees with previous experiments and could guide
future ones. With regard to methodology, we address the role of dispersion
interactions for studying PbS and PbSe surfaces: interestingly, the prediction of (001) surface energies depends crucially on the
use of dispersion corrections to DFT, and the latter directly influences the computed equilibrium shapes (Wulff constructions).
This study may serve as a starting point for future explorations of surfaces and nanocrystals of PbS, PbSe, and chemically related
chalcogenide materials.

■ INTRODUCTION
Lead chalcogenides have long been known as versatile
explored in 1999 already,19 and not much later, structures
and properties of clean lead chalcogenide (001) surfaces were
functional materials and today are of prime interest for studied by DFT.20,21 Further computations dealt with the
nanochemistry. New and ingenious synthetic approaches cast stability of PbSe nanocrystal facets,22 with surface oxidation as
both compounds into precisely controlled particles, nanowires, represented by cluster models,23 and with ligand binding to
and even nanocrystal superlattices with interesting and surfaces of both materials and how this influences nano-
sometimes unexpected properties.1−5 So enabled, applications morphology.24−29
of “miniaturized” PbS and PbSe encompass solar cells, sensors, Despite these advances, two theoretical questions remain
thermoelectrics, and diverse other fields.6−9 which both have important practical implications. First, to
The shape and properties of nanostructures are controlled by describe the morphology both of individual (say, epitaxially
the underlying crystal surfaces. For example, the particle grown) surfaces and of entire nanoparticles, one must consider
morphology of the homologous SnTe depends on the relative all relevant competing surfaces side-by-side, and include surface
stabilities of competing {001} and {111} facets, and tuning the reconstructions where appropriate. Second, the choice of DFT
chemical potential (or synthetic conditions) allows one to methods affects the prediction of absolute surface energies, with
modify the shape of SnTe nanostructures.10−13 Unfavorable the popular generalized gradient approximation (GGA)
surfaces change their atomic connectivity: they reconstruct, frequently underestimating them (i.e., erroneously making
which is not only an interesting phenomenon in itself but surfaces too stable).30−32 In the quest for more accurate DFT
directly affects the electronic nature of surface states.14 Going simulations, corrections for dispersion interactions33 have been
one or two length scales larger, the shape of PbSe nanocrystals recently established as useful and economic tools for inorganic
has recently been linked to charge-transport properties.15 All and materials chemistry, e.g., to achieve correct stability ranking
this is important for emerging electronic applications. of polymorphs,34−36 and were previously applied to surfaces of
The need is hence obvious to know as much as possible ternary Ge−Sb−Te phase-change materials.37
about the surfaces of lead chalcogenides. Besides a range of In this work, we report comprehensive DFT-based surface
experimental techniques, first-principles simulations based on studies for PbS and PbSe, expanding on our previous
density-functional theory (DFT) are today routinely employed investigations of the homologous PbTe38 and also on our
in the field of nanochemistry.16,17 We have very recently earlier work on germanium and tin telluride surfaces.10,39 We
reviewed experimental and theoretical work on IV−VI
chalcogenide surfaces,18 and here only reflect on the literature Received: March 1, 2016
for the two particular compounds PbS and PbSe. Adsorption of Revised: April 11, 2016
water molecules on PbS(001) has been computationally Published: April 12, 2016

© 2016 American Chemical Society 8813 DOI: 10.1021/acs.jpcc.6b02173


J. Phys. Chem. C 2016, 120, 8813−8820
The Journal of Physical Chemistry C Article

explore surface structures and stabilities, as well as the tendency Table 1. Lattice Parameters and Bulk Moduli of Rocksalt-
of PbS and PbSe (111) surfaces to reconstruct; ultimately, this Type PbS and PbSe
allows us to predict the equilibrium morphology of both
PbS PbSe
materials. In addition, the role of dispersion corrections for the
computation of PbS and PbSe surface energies is critically a (Å) B (GPa) a (Å) B (GPa)
addressed, for the first time to our knowledge. All this is experimentsa−d 5.9143(4)a 52.9b,c 6.1054(5)a 49.4c,d
expected to have important implications for realistic simulations 54.7c,d
of nanoscale lead chalcogenides. LDA (ref 52) 5.906 66.3 6.098 60.8

■ COMPUTATIONAL METHODS
DFT computations were performed using the projector
PBE+SOC (ref 53)
HSE03+SOC (ref 53)
PBE (this work)
6.004
5.963
5.995
51.2
55.0
54.7
6.214
6.170
6.206
44.6
48.3
48.1
augmented-wave (PAW) method40 as implemented in the PBE-D3 (this work) 5.923 56.8 6.127 47.5
a
Vienna ab initio Simulation Package (VASP 5.3.5).41−44 The From ref 58 (X-ray diffraction; 120 K). bFrom ref 59 (at 295 K).
c
cutoff energy for the plane-wave expansion was 500 eV. Isentropic bulk moduli (BS) are quoted. dFrom ref 60 (at 300 K).
Electronic cycles were halted at ΔE < 10−6 eV cell−1, and
structures were optimized until forces on atoms fell below 10−2 the PBE+D3 agreement for a with previous low-temperature
eV Å−1. Exchange and correlation were treated in the GGA experiments 58 is excellent (to within 1.5% and 3.5%,
after Perdew, Burke, and Ernzerhof (PBE).45 Dispersion respectively; Table 1). Regarding bulk moduli, the “D3”
corrections were applied where noted, using the “D3” method correction leads to only minor changes which appear to
of Grimme and co-workers46 together with Becke−Johnson worsen the pure PBE results; the deviations are tiny, however,
damping;47,48 this approach is abbreviated as “PBE-D3” and likely within the error range of different experiments. The
throughout this paper. effect of dispersion corrections on the prediction of bulk
Surfaces were modeled using inversion-symmetric slabs moduli would be clearly more pronounced for other systems,
similar to our previous study on PbTe,38 with in-plane lattice e.g., for the TeO2 polymorphs which are strongly influenced by
dispersion forces.35


vectors locked to multiples of the respective optimized bulk
parameters (see below). The thickness of the slabs was >17 Å
(7 layers) for (001), > 16 Å (9 layers) for (011), and >23 Å (15 RESULTS AND DISCUSSION
layers) for (111) surfaces; vacuum regions of >11, >16, and >17 Pristine Surfaces. The initial step in theoretical surface
Å, respectively, were introduced to separate periodically science is to model the pristine, unreconstructed crystal faces;
repeated images. All atoms were fully relaxed in the slab the optimized structures are shown in Figure 1 for PbS. The
supercells. Reciprocal space was sampled on dense Γ-centered
Monkhorst−Pack meshes49 sized 16 × 16 for the (001), 16 ×
24 for the (011), and 9 × 9 for the (111) surface simulation
cells. To accommodate reconstructions, all (111) terminations
were modeled in 2 × 2 expansions of the conventional surface
unit cell, as done and discussed before.38,39 Structural models
were handled using wxDragon,50 and visualizations and Wulff
constructions were created with the help of VESTA.51
As a necessary reference, computations for bulk PbS and
PbSe were performed at the outset. Both bulk phases have been
widely studied by theory;52−56 exemplarily, we compare our
results to previous linearized augmented plane-wave (LAPW)
computations52 in the local density approximation (LDA), and
to a more recent work which explicitly includes spin−orbit
coupling (SOC).53 As discussed and justified earlier,38,57 SOC
is not taken into account in the present work: it has only small
effects on geometric structures but would increase computa- Figure 1. Low-index surfaces of PbS, structurally optimized at the
tional costs significantly. On the other hand, studies of PBE-D3 level of theory. Labels for interatomic distances d are given
electronic band structures, which have been done for bulk Pb that will be used in the following. The Pb- and S-terminated slabs
chalcogenides, e.g., in ref 53, would make the inclusion of SOC (panels c and d) are equally thick, but they have been shifted in the
mandatory and may in fact be a rewarding target for future drawings to visualize how both are derived from the same bulk
surface studies. structure. The respective, homologous PbSe surfaces are similar in
connectivity but naturally differ in bond lengths (cf. Figures 2 and 3).
Table 1 provides results for the bulk phases of the title
compounds and thereby compares the computational method-
ology we use both to experiment and to previous theoretical rocksalt-type (001) and (011) surfaces are “stoichiometric”,
studies. It is noteworthy that the simplistic LDA already that is, they bear equal amounts of cations and anions (“type I”
performs exceedingly well with respect to the a parameters of surfaces according to Tasker’s scheme; ref 61). By contrast, for
both compounds albeit it overshoots their bulk moduli (ref 52). the (111) surfaces two pristine terminations exist (Figure 1c−
The latter are better reproduced from gradient corrections d), and these are both polar unless reconstructed (Tasker “type
(PBE) and in particular from more accurate and costly hybrid- III”). Before looking at energies, we inspect the computed
functional computations (HSE03).53 The economic “D3” atomic structures of the pristine surfaces and refer to earlier
dispersion correction on top of PBE re-establishes LDA quality work where possible. The labeling convention we use is
for the lattice parameters together with reasonable bulk moduli; illustrated in Figure 1.
8814 DOI: 10.1021/acs.jpcc.6b02173
J. Phys. Chem. C 2016, 120, 8813−8820
The Journal of Physical Chemistry C Article

As there are previous reports on the structures of these surface rumpling are difficult to attain, and a final verdict may
particular (001) surfaces,20,21 we will here focus on a come from high-level computational techniques beyond
comparison between the pure PBE and the PBE-D3 results. DFT.31,62 In what follows, we will report structures at the
Nonetheless, we mention that at the PBE level, the computed dispersion-corrected PBE-D3 level and provide pure PBE data
changes in bond lengths are in line with earlier simulations: we in the Supporting Information as they do not lead to any
obtain an average shortening of d1 by ∼6%, and an expansion of changes in interpretation.
d2 by ∼4% for both compounds. The relaxations are slightly The pristine, dense (111) surfaces (Figure 1c−d) have been
stronger at PbS(001) than at PbSe(001). All this agrees less studied so far; recall that those are likely metastable with
reasonably with previous computations.20,21 regard to surface reconstructions (the latter will be discussed
Here, instead, we focus on a quantity characteristic for below). Figure 3 compiles computed, absolute distances at the
rocksalt (001) terminations: their rumpling, which measures the
shifts of cations and anions relative to each other at the relaxed
surface. We use the definition of the rumpling parameter Δzi
given by Satta and de Gironcoli:20
zianion − ziPb
Δzi =
d0
where zi denotes the respective position along [001] of an atom
located in the ith layer (the i = 1 atoms are directly at the
surface, i = 2 denotes the first subsurface layer, and so on). d0 is
the Pb−anion distance in the optimized bulk structure.
The results are provided in Figure 2. Interestingly, the
qualitative rumpling behavior is inverted between the two lead

Figure 3. Course of Pb−S and Pb−Se bond lengths near relaxed,


pristine lead chalcogenide (111) surfaces. Computational results at the
PBE-D3 level are shown; those based on PBE are qualitatively similar
(Supporting Information).

unreconstructed surfaces, which outline the formation of


characteristic bilayers (“short−long” alternation), similar to
what was found for α-GeTe(111) and PbTe(111).38,39 The
scattering in surface bond lengths spans a range of ∼0.4 Å both
for PbS and PbSe; the anion-terminated surfaces (squares in
Figure 3) show this effect more strongly than their Pb-
terminated counterparts (blue circles). While the present work
is concerned with slab modeling exclusively, we do mention in
passing that other studies have been reported on isolated, entire
particles: for example, PbSe nanocrystals were simulated by
Franceschetti.63 The distribution width of surface bond lengths
Figure 2. Surface rumpling at PbS(001) and PbSe(001) surfaces from in the latter work compares well to the present results from slab
DFT simulations at the PBE and PBE-D3 levels. The parameter Δzi is models.
described in the text and visualized in simplified cartoons. The more Let us now turn to surface energies, that is, stabilities which
Δzi deviates from zero, the more the atoms within a layer are shifted are the key determinants for nanocrystal morphology.17,18 The
relative to one another.20 surface energies γ are obtained using similar expressions as in
previous studies, e.g., on the homologous PbTe (see ref 38 and
chalcogenides:21 at PbS(001), the anions relax toward the references therein). We begin by discussing the neutral pristine
crystal interior, whereas they move to the outside at surfaces, (001) and (011). Here, there are a number of previous
PbSe(001); consequently, the Pb atoms relax outward (inward) PBE computations in the literature to which we compare our
for PbS (PbSe), respectively. This has previously been results in Table 2. The differences between previous studies and
rationalized at the hand of different ionic radii:20,21 it was ours are marginal; in fact, we have listed decimal places only to
argued that, compared to the small sulfide anion, its selenide make comparison feasible. At the PBE level, the (011) surfaces
counterpart is larger and more diffuse, and so it moves to the come out approximately twice as costly as (001), in line with
outside, away from the crystal lattice;20,21 similar behavior has previous work;22 this agrees well with the fact that no {011}
been seen and discussed for the heavier homologue PbTe.20,57 facets are observed in equilibrium.22 Using dispersion
Our simulations reproduce this contrast well; “D3” slightly but corrections (rightmost column in Table 2), the relative
consistently increases the predicted rumpling at PbS(001), and difference is smaller, but the surface energies still attain a
lowers it at PbSe(001), but using dispersion corrections does γ
ratio of γ011 ≈ 1.5; this is higher than what would be required
not change the qualitative conclusions. Unfortunately, precise 001
experimental benchmarks regarding the quantitative extent of for the {011} facets to appear in the equilibrium shape
8815 DOI: 10.1021/acs.jpcc.6b02173
J. Phys. Chem. C 2016, 120, 8813−8820
The Journal of Physical Chemistry C Article

Table 2. Computed Surface Energies of Pristine PbS and


PbSe (001) and (011) Surfaces
γ (meV Å−2)
a
Previous PBE PBE-D3
PbS (001) 13.7,b 11.2c 11.9 26.0
(011) 24.3 39.7
PbSe (001) 11.0,c 11.5d 11.6 25.5
(011) 19.8d 22.1 36.7
a
All previous studies employed the PBE level of DFT, and slightly
different but overall comparable computational parameters. bFrom ref
64. cFrom ref 65. dFrom ref 22.

γ
( γ011 = 2 ; see below). Accordingly, the (011) surfaces have
001
been neglected in the above discussion of pristine surfaces.
Looking at the computed absolute surface energies (Table 2),
there is a significant difference in the latter dependent on
whether dispersion corrections are added or not. The mere fact
that both results do differ is not surprising: for example, Ončaḱ
et al. computed γ = 56 meV Å−2 for the MgO(001) surface at
the PBE level, and γ = 80 meV Å−2 for the same surface using
PBE-D2.66 For comparison, an earlier quantum Monte Carlo
study of MgO(001) had arrived at γ = 74 meV Å−2 (ref 31),
much closer to the above dispersion-corrected result (−7%)
than to the uncorrected one (+32%). This observation is also Figure 4. Reconstructions of lead chalcogenide (111) surfaces. (a)
reminiscent of the fact that absolute surface energies depend on Schematic drawing (after refs 38 and 74) of candidate structural
the choice of exchange−correlation functional, as mentioned models. Smaller open and larger filled circles denote surface and
above.30−32 A more extreme influence of dispersion has subsurface atoms, respectively, and thus they can represent either
previously been encountered for “van der Waals” type (0001) cations or anions depending on the particular surface termination. (b)
surfaces of layered Ge−Sb−Te materials:37 there, switching on The Pb-terminated 2 × 1 reconstruction of PbS(111), relaxed at the
PBE-D3 level of theory and shown in top view. Two characteristic
dispersion corrections to GGA increases the computed γ’s for bond lengths di are labeled, and the boundary of the simulation box is
the stable “vdW” surfaces by an order of magnitude (or, to put indicated. (c) Same for the S-terminated octopolar reconstruction; the
it more plainly: the uncorrected GGA functional fails by a inset shows an isolated Pb4S4 motif at the surface. The complementary
factor of 10).37 In this regard, the lead chalcogenides seem to 2 × 1/S and oct/Pb reconstructions, as well as those at PbSe(111),
be intermediate between MgO (with which they share the show similar connectivities (with atomic species exchanged and
rocksalt structure) and the aforementioned layered chalcoge- structural parameters adjusted, of course).
nide materials.
(111) Surface Reconstructions. The (111) surfaces are
special in the rocksalt type, as they are densely covered by one scattering, 75 we do not know of atomically resolved
or the other atomic species exclusively. In ionic compounds observations (say, by scanning tunneling microscopy), and
such as MgO, this causes electrostatic instabilities which are again we resort to previous findings for the “textbook” surfaces
remedied by surface reconstructions67−69 or roughening.70 For of MgO. For example, the “octopolar” A4B4 moiety is a well-
the significantly less ionic GeTe39 and SnTe,10,14 this need not known way to alleviate surface polarity.67 Later, an additional
be the case throughout, and indeed an unreconstructed α- reconstruction pattern was proposed for the MgO(111)
GeTe(111)-(1 × 1) surface structure has recently been surface, which we label “MgO” here.68 The interested reader
observed in reflection high-energy electron diffraction is referred to ref 74 for an instructive survey of rocksalt-type
(RHEED) experiments.71 On the contrary, pure dense (111) reconstructions.
PbSe(111) surfaces did turn out to be unstable, both in Computed energies for all candidate reconstructions are
transmission electron microscopy (TEM) and in simula- summarized in Table 3. The “MgO” reconstruction does not
tions.22,72 Likewise, a recent study revealed that an as-prepared significantly stabilize the surfaces and in some cases has even
dense PbTe(111)-(1 × 1) surface does not prevail but slightly higher γ than the pristine alternatives (top row in Table
transforms into a stable (2 × 1) reconstruction upon heating.73 3); the “SV” motif does provide some stabilization but is not
Clarifying atomic mechanisms which might stabilize the PbS globally competitive. While the aforementioned two motifs are
and PbSe (111) surfaces now seems to be a natural task for polar (and thus their surface energies depend on the chemical
DFT. environment), the lower two, “2 × 1” and “oct” are not, and
To this end, we proceed to sample representative surface these two do significantly stabilize the surfaces. For PbSe, 2 ×
reconstructions (Figure 4a). The number of the latter is 1-reconstructed surfaces have been studied before at the PBE
necessarily finite, and limited by the size of simulation cells level;22 note that in the latter work, they have been dubbed
available; this makes the choice of models even more “(111)-Pb nonpolar” and “(111)-Se nonpolar”, respectively.
important, and the latter will ideally include patterns known There, and here, the Pb-terminated 2 × 1 reconstruction is
from earlier experiments. While evidence for PbSe(111) surface found visibly more stable than the anion-terminated alter-
reconstructions has indeed been found by Rutherford back- native;22 the same holds for PbS (Table 3) and PbTe (ref 38),
8816 DOI: 10.1021/acs.jpcc.6b02173
J. Phys. Chem. C 2016, 120, 8813−8820
The Journal of Physical Chemistry C Article

Table 3. Surface Energies of Pristine and Reconstructed For now, we focus on the two most favorable (111)
(111) Surfaces reconstructions so identified, which are practically degenerate
in their surface energies (Table 3). The optimized atomic
γ (meV Å−2), PBE-D3a
structures have been visualized in Figure 4b (“2 × 1”) and
PbS PbSe Figure 4c (“octopolar”). We collect characteristic bond lengths
term Pb poorb Pb richc Pb poord Pb richc for these in Table 4. For PbS in particular, the respective
pristine Pb 107 68 99 63
S/Se 62 102 52 88 Table 4. Structural Properties of Reconstructed Lead
SV Pb 63 44 60 42 Chalcogenide (111) Surfacesa
S/Se 52 71 47 65 d1 (Å) d2 (Å) d3 (Å)
MgO Pb 90 70 85 68
PbS (2 × 1) Pb-term 2.64 2.77
S/Se 83 103 74 91
S-term 2.63 2.78
2×1 Pb 38 38 36 36
oct. Pb-term 2.72 2.82 2.85
S/Se 44 44 43 43
S-term 2.73 2.78 2.95
oct Pb 39 39 38 38
PbSe (2 × 1) Pb-term 2.77 2.88
S/Se 39 39 38 38
a
Se-term 2.79 2.89
Results at the PBE level are given as Supporting Information (and are oct. Pb-term 2.83 2.95 2.95
also visualized in Figure 5). bDefined as μPb = E(bulk)
PbS − ES
(bulk) c
. Defined
Se-term 2.87 2.90 3.07
as μPb = EPb . Defined as μPb = EPbSe − ESe .
(bulk) d (bulk) (bulk)
a
The parameters d1, d2, and d3 are defined in Figure 4. All data given
and also when moving from the PBE level to the PBE-D3 level have been obtained at the PBE-D3 level of theory. The optimized
bond lengths in the bulk are 2.96 Å (PbS) and 3.06 Å (PbSe),
in Table 3.
respectively (Table 1).
It is again interesting to compare the results with and without
dispersion corrections, and the larger number of reconstruc-
tions allows us to probe this difference in a more systematic shortest bond (d1) is significantly compressed at the 2 × 1
way (Figure 5). All data refer to the “Pb-rich” case; that is, to reconstruction compared to the octopolar motif. Relative to the
bulk (Table 1), the computed d1 are shorter by ∼11% (2 × 1)
and ∼8% (octopolar) at PbS(111) surfaces. They are likewise
compressed at PbSe(111), by ∼10% at most.
Finally, the question that emerges is that of global stability:
can the reconstructed surfaces discussed here compete with
their important (001) counterparts?
Surface Phase Diagrams. The global stability of different
surfaces is addressed using surface phase diagrams.76 Results for
the lead chalcogenides are presented in Figure 6, side by side
such as to detect chemical trends. Regardless of computational
methodology, the pristine and polar (111) surfaces (dashed
lines) are higher in energy for PbS than for PbSe, due to the
higher electronegativity difference in the lighter sulfide
compound. By contrast, all charge-neutral surfaces (horizontal
lines) are practically indifferent to the anionic species. Both for
PbS(111) and PbSe(111), reconstructions are clearly favored
and significantly lower the surface energies; for clarity, we show
Figure 5. Scatterplot of PBE versus PBE-D3 computed surface only the most stable option in each of the surface phase
energies, combining data for pristine and reconstructed (111) surfaces diagrams. It should be stressed that the energy difference
of PbS (circles) and PbSe (squares). The dashed line is the result of a between the 2 × 1/Pb and both octopolar motifs is only 1−2
least-squares fit (y = 0.9759x + 17.33; R2 = 0.997). meV Å−2 both at the PBE and the PBE-D3 level of theory
(Table 3); the deviation is likely within the error range of the
computations. In short, both types of reconstructions appear to
μPb = E(bulk)
Pb , where neither anionic species enters in the be viable.
definition of γ; this makes side-by-side comparison straightfor- Wulff Constructions. The comprehensive set of surface
ward. stabilities thus obtained may be translated into equilibrium
Clearly, the outcomes of the two methods are closely shapes using the Wulff construction77,78 (Figure 7). One
correlated (dashed line in Figure 5). There is, however, not a observes mainly cube-like morphologies for both lead
scaling factor between the results, but an of fset: PBE-D3 chalcogenides, not unexpectedly, but the results are different
predictions are higher than those of PBE by ∼17 meV Å−2. This in detail depending on whether dispersion corrections are used
suggests that long-range dispersion interactions affect surfaces or not. The shapes predicted at the PBE level are perfect cubes,
of similar area in a roughly similar way, and are less sensitive to whereas those at the PBE-D3 level are truncated at the corners,
the particular local reconstruction. In turn, the dispersion slightly more strongly so for PbSe. This reflects an observation
correction affects the most stable surfaces (with lowest γ) most we have made above: the surface energies for the competing
strongly in relative terms. This will become important when {001} facets are notably higher when “D3” is switched on. It
performing Wulff constructions, because those precisely reflect would now seem interesting to tackle this question
the relative stability of competing {hkl} facets. experimentally, for example, by utilizing vapor−liquid−solid
8817 DOI: 10.1021/acs.jpcc.6b02173
J. Phys. Chem. C 2016, 120, 8813−8820
The Journal of Physical Chemistry C Article

Figure 7. Equilibrium shapes of PbS and PbSe nanoparticles, obtained


using the Wulff construction77 based on computed surface energies.
The {001} facets (gray) are predominant regardless of methodology.
Using PBE-D3 (lower panels), one obtains slightly truncated
polyhedra which expose {111} surfaces with a 2 × 1/Pb termination
(blue). No {011} facets are observed in any case.

nanoscale. Dispersion corrections to DFT do affect the


computed surface energies, in particular those of (001), and
thus they indirectly affect the computed equilibrium shapes.
Lead chalcogenide (111) surfaces are prone to reconstruct (as
observed in previous experimental and theoretical studies); our
surface phase diagrams clearly identify viable candidates,
namely, Pb-terminated (2 × 1) motifs as well as both cation-
and anion-terminated “octopolar” reconstructions. Structural
parameters and trends have been predicted and may be probed
by surface-sensitive experiments in the future. Finally, our
Figure 6. Computed surface phase diagrams for PbS (left) and PbSe
(right), excluding (a) and including (b) dispersion corrections to
findings for the clean surfaces may provide starting points for
DFT. Among the (111) reconstructions, only the respective most further endeavors: for simulations of few-layer nanosheets,
stable one is given. Note, however, that for both compounds and at reactivities of reconstructed (111) surfaces, their surface
both levels of theory, the respective pairs of “2 × 1” and “oct” motifs oxidation, and likely some more.
are degenerate within a few meV Å−2. For further candidate
reconstructions, as well as the definition of chemical potentials (“Pb-
poor”, “Pb-rich”), see Table 3.

*
ASSOCIATED CONTENT
S Supporting Information
The Supporting Information is available free of charge on the
(VLS) routes similar to those used previously for growing SnTe ACS Publications website at DOI: 10.1021/acs.jpcc.6b02173.
nanostructures.11−13,79 This would rule out effects of surface Computational results at the DFT-PBE level (as
ligands which have been described for PbSe nanocrystals,22 and supplements to Figure 3 and Tables 3 and 4) (PDF)


enable better comparison with the results of Figure 7. It should
also be noted that surface energies will additionally depend on AUTHOR INFORMATION
vibrational contributions at finite temperature, which have been
Corresponding Author
neglected here. These vibrational contributions have been
estimated to be on the order of ≤10 meV Å−2 in a seminal *E-mail: drons@HAL9000.ac.rwth-aachen.de. Telephone: +49
study on RuO2(110),76 but how they affect the particular 241 80 93642. Fax: +49 241 80 92642.
surfaces studied here remains to be explored before quantitative Present Address
§
comparison to experiment can be made. Finally, with wet- Engineering Laboratory, University of Cambridge, Trumping-
chemical syntheses in mind, the Wulff construction concept has ton Street, Cambridge CB2 1PZ, United Kingdom.
been nicely expanded to adsorbate-covered surfaces for the Notes
particular case of PbSe.27 It would now seem very interesting to The authors declare no competing financial interest.
explore the influence of dispersion corrections to DFT on the
outcome of such a model. ■ ACKNOWLEDGMENTS

■ CONCLUSIONS
We have studied crystal surfaces of PbS and PbSe by
We thank Philipp Konze for useful discussions and remarks on
this manuscript. This study received financial support from
Deutsche Forschungsgemeinschaft (DFG) through SFB 917
comprehensive first-principles simulations and have so obtained “Nanoswitches”. CPU time was allocated by the Jülich−Aachen
fundamental insight into the materials’ properties at the Research Alliance (Project jara0033).
8818 DOI: 10.1021/acs.jpcc.6b02173
J. Phys. Chem. C 2016, 120, 8813−8820
The Journal of Physical Chemistry C Article

■ REFERENCES
(1) Murray, C. B.; Sun, S. H.; Gaschler, W.; Doyle, H.; Betley, T. A.;
(21) Ma, J. X.; Jia, Y.; Song, Y. L.; Liang, E. J.; Wu, L. K.; Wang, F.;
Wang, X. C.; Hu, X. The geometric and electronic properties of the
PbS, PbSe and PbTe (001) surfaces. Surf. Sci. 2004, 551, 91−98.
Kagan, C. R. Colloidal synthesis of nanocrystals and nanocrystal (22) Fang, C.; van Huis, M. A.; Vanmaekelbergh, D.; Zandbergen, H.
superlattices. IBM J. Res. Dev. 2001, 45, 47−56. W. Energetics of Polar and Nonpolar Facets of PbSe Nanocrystals
(2) Talapin, D. V.; Lee, J.-S.; Kovalenko, M. V.; Shevchenko, E. V. from Theory and Experiment. ACS Nano 2010, 4, 211−218.
Prospects of Colloidal Nanocrystals for Electronic and Optoelectronic (23) Yashina, L. V.; Zyubin, A. S.; Püttner, R.; Zyubina, T. S.;
Applications. Chem. Rev. 2010, 110, 389−458. Neudachina, V. S.; Stojanov, P.; Riley, J.; Dedyulin, S. N.;
(3) Evers, W. H.; Goris, B.; Bals, S.; Casavola, M.; de Graaf, J.; van Brzhezinskaya, M. M.; Shtanov, V. I. The oxidation of the PbS(001)
Roij, R.; Dijkstra, M.; Vanmaekelbergh, D. Low-Dimensional Semi- surface with O2 and air studied with photoelectron spectroscopy and
conductor Superlattices Formed by Geometric Control over Nano- ab initio modeling. Surf. Sci. 2011, 605, 473−482.
crystal Attachment. Nano Lett. 2013, 13, 2317−2323. (24) Zhang, L.; Song, Q.; Zhang, S. B. Exceptionally Strong
(4) Boneschanscher, M. P.; Evers, W. H.; Geuchies, J. J.; Altantzis, T.; Hydrogen Bonds Affect the Surface Energy of Colloidal Nanocrystals:
Goris, B.; Rabouw, F. T.; van Rossum, S. A. P.; van der Zant, H. S. J.; Methylamine and Water Adsorption on PbS. Phys. Rev. Lett. 2010, 104,
Siebbeles, L. D. A.; Van Tendeloo, G.; et al. Long-range orientation 116101.
and atomic attachment of nanocrystals in 2D honeycomb superlattices. (25) Argeri, M.; Fraccarollo, A.; Grassi, F.; Marchese, L.; Cossi, M.
Science 2014, 344, 1377−1380. Density Functional Theory Modeling of PbSe Nanoclusters: Effect of
(5) Kovalenko, M. V.; Manna, L.; Cabot, A.; Hens, Z.; Talapin, D. V.; Surface Passivation on Shape and Composition. J. Phys. Chem. C 2011,
Kagan, C. R.; Klimov, V. I.; Rogach, A. L.; Reiss, P.; Milliron, D. J.; 115, 11382−11389.
et al. Prospects of Nanoscience with Nanocrystals. ACS Nano 2015, 9, (26) Kutana, A.; Erwin, S. C. PbSe nanocrystals remain intrinsic after
1012−1057. surface adsorption of hydrazine. Phys. Rev. B: Condens. Matter Mater.
(6) Talapin, D. V.; Murray, C. B. PbSe Nanocrystal Solids for n- and Phys. 2011, 83, 235419.
p-Channel Thin Film Field-Effect Transistors. Science 2005, 310, 86− (27) Bealing, C. R.; Baumgardner, W. J.; Choi, J. J.; Hanrath, T.;
89. Hennig, R. G. Predicting Nanocrystal Shape through Consideration of
(7) Nozik, A. J.; Beard, M. C.; Luther, J. M.; Law, M.; Ellingson, R. J.; Surface-Ligand Interactions. ACS Nano 2012, 6, 2118−2127.
Johnson, J. C. Semiconductor Quantum Dots and Quantum Dot (28) Zherebetskyy, D.; Scheele, M.; Zhang, Y.; Bronstein, N.;
Arrays and Applications of Multiple Exciton Generation to Third- Thompson, C.; Britt, D.; Salmeron, M.; Alivisatos, P.; Wang, L.-W.
Generation Photovoltaic Solar Cells. Chem. Rev. 2010, 110, 6873− Hydroxylation of the surface of PbS nanocrystals passivated with oleic
6890. acid. Science 2014, 344, 1380−1384.
(8) Konstantatos, G.; Sargent, E. H. Nanostructured materials for (29) Bielewicz, T.; Dogan, S.; Klinke, C. Tailoring the Height of
photon detection. Nat. Nanotechnol. 2010, 5, 391−400. Ultrathin PbS Nanosheets and Their Application as Field-Effect
(9) Androulakis, J.; Todorov, I.; He, J.; Chung, D.-Y.; Dravid, V.; Transistors. Small 2015, 11, 826−833.
Kanatzidis, M. Thermoelectrics from Abundant Chemical Elements: (30) Da Silva, J. L. F.; Stampfl, C.; Scheffler, M. Converged
High-Performance Nanostructured PbSe−PbS. J. Am. Chem. Soc. 2011, properties of clean metal surfaces by all-electron first-principles
133, 10920−10927. calculations. Surf. Sci. 2006, 600, 703−715.
(10) Deringer, V. L.; Dronskowski, R. Stability of Pristine and (31) Alfè, D.; Gillan, M. J. The energetics of oxide surfaces by
Defective SnTe Surfaces from First Principles. ChemPhysChem 2013, quantum Monte Carlo. J. Phys.: Condens. Matter 2006, 18, L435−
14, 3108−3111. L440.
(11) Li, Z.; Shao, S.; Li, N.; McCall, K.; Wang, J.; Zhang, S. X. Single (32) Finocchi, F.; Goniakowski, J. The effects of exchange and
Crystalline Nanostructures of Topological Crystalline Insulator SnTe correlation on the computed equilibrium shapes of wet MgO
with Distinct Facets and Morphologies. Nano Lett. 2013, 13, 5443− crystallites. Surf. Sci. 2007, 601, 4144−4148.
5448. (33) Grimme, S. Density functional theory with London dispersion
(12) Safdar, M.; Wang, Q.; Mirza, M.; Wang, Z.; He, J. Crystal Shape corrections. Wiley Interdisciplinary Reviews: Computational Molecular
Engineering of Topological Crystalline Insulator SnTe Microcrystals Science 2011, 1, 211−228.
and Nanowires with Huge Thermal Activation Energy Gap. Cryst. (34) Guńka, P. A.; Dranka, M.; Piechota, J.; Ż ukowska, G. Z.;
Growth Des. 2014, 14, 2502−2509. Zalewska, A.; Zachara, J. As2O3 Polymorphs: Theoretical Insight into
(13) Shen, J.; Jung, Y.; Disa, A. S.; Walker, F. J.; Ahn, C. H.; Cha, J. J. Their Stability and Ammonia Templated Claudetite II Crystallization.
Synthesis of SnTe Nanoplates with {100} and {111} Surfaces. Nano Cryst. Growth Des. 2012, 12, 5663−5670.
Lett. 2014, 14, 4183−4188. (35) Deringer, V. L.; Stoffel, R. P.; Dronskowski, R. Thermochemical
(14) Wang, J.; Liu, J.; Xu, Y.; Wu, J.; Gu, B.-L.; Duan, W. Structural Ranking and Dynamic Stability of TeO2 Polymorphs from Ab Initio
stability and topological surface states of the SnTe (111) surface. Phys. Theory. Cryst. Growth Des. 2014, 14, 871−878.
Rev. B: Condens. Matter Mater. Phys. 2014, 89, 125308. (36) Bachhuber, F.; von Appen, J.; Dronskowski, R.; Schmidt, P.;
(15) Kaushik, A. P.; Lukose, B.; Clancy, P. The Role of Shape on Nilges, T.; Pfitzner, A.; Weihrich, R. The Extended Stability Range of
Electronic Structure and Charge Transport in Faceted PbSe Phosphorus Allotropes. Angew. Chem., Int. Ed. 2014, 53, 11629−
Nanocrystals. ACS Nano 2014, 8, 2302−2317. 11633.
(16) Groß, A. Theoretical Surface Science: A Microscopic Perspective; (37) Deringer, V. L.; Dronskowski, R. DFT Studies of Pristine
Springer: Berlin, 2009. Hexagonal Ge1Sb2Te4(0001), Ge2Sb2Te5(0001), and
(17) Barnard, A. S. Modelling of nanoparticles: approaches to Ge1Sb4Te7(0001) Surfaces. J. Phys. Chem. C 2013, 117, 15075−15089.
morphology and evolution. Rep. Prog. Phys. 2010, 73, 086502. (38) Deringer, V. L.; Dronskowski, R. Stabilities and Reconstructions
(18) Deringer, V. L.; Dronskowski, R. From Atomistic Surface of PbTe Crystal Surfaces from Density-Functional Theory. J. Phys.
Chemistry to Nanocrystals of Functional Chalcogenides. Angew. Chem. C 2013, 117, 24455−24461.
Chem., Int. Ed. 2015, 54, 15334−15340. (39) Deringer, V. L.; Lumeij, M.; Dronskowski, R. Ab Initio
(19) Wright, K.; Hillier, I. H.; Vincent, M. A.; Kresse, G. Dissociation Modeling of α-GeTe(111) Surfaces. J. Phys. Chem. C 2012, 116,
of water on the surface of galena (PbS): A comparison of periodic and 15801−15811.
cluster models. J. Chem. Phys. 1999, 111, 6942−6946. (40) Blöchl, P. E. Projector augmented-wave method. Phys. Rev. B:
(20) Satta, A.; de Gironcoli, S. Surface structure and core-level shifts Condens. Matter Mater. Phys. 1994, 50, 17953−17979.
in lead chalcogenide (001) surfaces. Phys. Rev. B: Condens. Matter (41) Kresse, G.; Hafner, J. Ab initio molecular dynamics for liquid
Mater. Phys. 2000, 63, 033302. metals. Phys. Rev. B: Condens. Matter Mater. Phys. 1993, 47, 558−561.

8819 DOI: 10.1021/acs.jpcc.6b02173


J. Phys. Chem. C 2016, 120, 8813−8820
The Journal of Physical Chemistry C Article

(42) Kresse, G.; Furthmüller, J. Efficient Iterative Schemes for Ab (66) Ončaḱ , M.; Włodarczyk, R.; Sauer, J. Water on the MgO(001)
Initio Total-Energy Calculations Using a Plane-Wave Basis Set. Phys. Surface: Surface Reconstruction and Ion Solvation. J. Phys. Chem. Lett.
Rev. B: Condens. Matter Mater. Phys. 1996, 54, 11169−11186. 2015, 6, 2310−2314.
(43) Kresse, G.; Furthmüller, J. Efficiency of ab-initio total energy (67) Wolf, D. Reconstruction of NaCl Surfaces from a Dipolar
calculations for metals and semiconductors using a plane-wave basis Solution to the Madelung Problem. Phys. Rev. Lett. 1992, 68, 3315−
set. Comput. Mater. Sci. 1996, 6, 15−50. 3318.
(44) Kresse, G.; Joubert, D. From ultrasoft pseudopotentials to the (68) Finocchi, F.; Barbier, A.; Jupille, J.; Noguera, C. Stability of
projector augmented-wave method. Phys. Rev. B: Condens. Matter Rocksalt (111) Polar Surfaces: Beyond the Octopole. Phys. Rev. Lett.
Mater. Phys. 1999, 59, 1758−1775. 2004, 92, 136101.
(45) Perdew, J. P.; Burke, K.; Ernzerhof, M. Generalized Gradient (69) Zhang, W.-B.; Tang, B.-Y. Stability of MgO(111) Polar Surface:
Approximation Made Simple. Phys. Rev. Lett. 1996, 77, 3865−3868. Effect of the Environment. J. Phys. Chem. C 2008, 112, 3327−3333.
(46) Grimme, S.; Antony, J.; Ehrlich, S.; Krieg, H. A consistent and (70) Madey, T. E.; Chen, W.; Wang, H.; Kaghazchi, P.; Jacob, T.
accurate ab initio parametrization of density functional dispersion Nanoscale surface chemistry over faceted substrates: structure,
correction (DFT-D) for the 94 elements H-Pu. J. Chem. Phys. 2010, reactivity and nanotemplates. Chem. Soc. Rev. 2008, 37, 2310−2327.
132, 154104. (71) Wang, R.; Boschker, J. E.; Bruyer, E.; Di Sante, D.; Picozzi, S.;
(47) Johnson, E. R.; Becke, A. D. A post-Hartree-Fock model of Perumal, K.; Giussani, A.; Riechert, H.; Calarco, R. Toward Truly
intermolecular interactions: Inclusion of higher-order corrections. J. Single Crystalline GeTe Films: The Relevance of the Substrate
Chem. Phys. 2006, 124, 174104. Surface. J. Phys. Chem. C 2014, 118, 29724−29730.
(48) Grimme, S.; Ehrlich, S.; Goerigk, L. Effect of the damping (72) Schapotschnikow, P.; van Huis, M. A.; Zandbergen, H. W.;
function in dispersion corrected density functional theory. J. Comput. Vanmaekelbergh, D.; Vlugt, T. J. H. Morphological Transformations
Chem. 2011, 32, 1456−1465. and Fusion of PbSe Nanocrystals Studied Using Atomistic
(49) Monkhorst, H. J.; Pack, J. D. Special points for Brillouin-zone Simulations. Nano Lett. 2010, 10, 3966−3971.
integrations. Phys. Rev. B 1976, 13, 5188−5192. (73) Wu, H.; Si, J.; Yan, Y.; Liao, Q.; Lu, Y. Reconstructions and
(50) Eck, B. wxDragon version 2.0.4, Aachen, 1994−2015. stabilities of PbTe(111) crystal surface from experiments and density-
(51) Momma, K.; Izumi, F. VESTA 3 for three-dimensional functional theory. Appl. Surf. Sci. 2015, 356, 742−746.
visualization of crystal, volumetric and morphology data. J. Appl. (74) Franchini, C.; Bayer, V.; Podloucky, R.; Parteder, G.; Surnev, S.;
Crystallogr. 2011, 44, 1272−1276. Netzer, F. P. Density functional study of the polar MnO(111) surface.
(52) Wei, S.-H.; Zunger, A. Electronic and structural anomalies in Phys. Rev. B: Condens. Matter Mater. Phys. 2006, 73, 155402.
lead chalcogenides. Phys. Rev. B: Condens. Matter Mater. Phys. 1997, 55, (75) Kimura, K.; Nakajima, K.; Fujii, Y.; Mannami, M. Observation of
13605−13610. the PbSe(111) surface using high-resolution Rutherford backscattering
(53) Hummer, K.; Grüneis, A.; Kresse, G. Structural and electronic spectroscopy. Surf. Sci. 1994, 318, 363−367.
properties of lead chalcogenides from first principles. Phys. Rev. B: (76) Reuter, K.; Scheffler, M. Composition, structure, and stability of
Condens. Matter Mater. Phys. 2007, 75, 195211. RuO2(110) as a function of oxygen pressure. Phys. Rev. B: Condens.
(54) Zhang, Y.; Ke, X.; Chen, C.; Yang, J.; Kent, P. R. C. Matter Mater. Phys. 2001, 65, 035406.
Thermodynamic properties of PbTe, PbSe, and PbS: First-principles (77) Wulff, G. Zur Frage der Geschwindigkeit des Wachstums und
study. Phys. Rev. B: Condens. Matter Mater. Phys. 2009, 80, 024304. der Auflösung von Krystallflächen. Z. Kristallogr. - Cryst. Mater. 1901,
(55) Skelton, J. M.; Parker, S. C.; Togo, A.; Tanaka, I.; Walsh, A. 34, 449−530.
Thermal physics of the lead chalcogenides PbS, PbSe, PbTe from first (78) Moll, N.; Kley, A.; Pehlke, E.; Scheffler, M. GaAs equilibrium
principles. Phys. Rev. B: Condens. Matter Mater. Phys. 2014, 89, 205203. crystal shape from first principles. Phys. Rev. B: Condens. Matter Mater.
(56) Li, W.-F.; Fang, C.-M.; Dijkstra, M.; van Huis, M. A. The role of Phys. 1996, 54, 8844−8855.
point defects in PbS, PbSe and PbTe: a f irst principles study. J. Phys.: (79) Shen, J.; Cha, J. J. Topological crystalline insulator
Condens. Matter 2015, 27, 355801. nanostructures. Nanoscale 2014, 6, 14133−14140.
(57) Hoang, K.; Mahanti, S. D.; Jena, P. Theoretical study of deep-
defect states in bulk PbTe and in thin films. Phys. Rev. B: Condens.
Matter Mater. Phys. 2007, 76, 115432.
(58) Noda, Y.; Masumoto, K.; Ohba, S.; Saito, Y.; Toriumi, K.; Iwata,
Y.; Shibuya, I. Temperature Dependence of Atomic Thermal
Parameters of Lead Chalcogenides, PbS, PbSe and PbTe. Acta
Crystallogr., Sect. C: Cryst. Struct. Commun. 1987, 43, 1443−1445.
(59) Peresada, G. I.; Ponyatovskii, E. G.; Sokolovskaya, Z. D.
Pressure Dependence of the Elastic Constants of PbS. phys. stat. sol.
(a) 1976, 35, K177−K180.
(60) Medzhidov, R. A. Thermodynamic and elastic properties of lead
chalcogenides. High Temp. High Press. 1989, 21, 643−646.
(61) Tasker, P. W. The stability of ionic crystal surfaces. J. Phys. C:
Solid State Phys. 1979, 12, 4977−4984.
(62) Schimka, L.; Harl, J.; Stroppa, A.; Grüneis, A.; Marsman, M.;
Mittendorfer, F.; Kresse, G. Accurate surface and adsorption energies
from many-body perturbation theory. Nat. Mater. 2010, 9, 741−744.
(63) Franceschetti, A. Structural and electronic properties of PbSe
nanocrystals from first principles. Phys. Rev. B: Condens. Matter Mater.
Phys. 2008, 78, 075418.
(64) Choi, H.; Ko, J.-H.; Kim, Y.-H.; Jeong, S. Steric-Hindrance-
Driven Shape Transition in PbS Quantum Dots: Understanding Size-
Dependent Stability. J. Am. Chem. Soc. 2013, 135, 5278−5281.
(65) Kim, C.-E.; Tak, Y.-J.; Butler, K. T.; Walsh, A.; Soon, A. Lattice-
mismatched heteroepitaxy of IV-VI thin films on PbTe(001): An ab
initio study. Phys. Rev. B: Condens. Matter Mater. Phys. 2015, 91,
085307.

8820 DOI: 10.1021/acs.jpcc.6b02173


J. Phys. Chem. C 2016, 120, 8813−8820

You might also like