THE JOURNAL OF CHEMICAL PHYSICS 141, 054709 (2014)

Thermodynamic stability and structures of iron chloride surfaces:A first-principles investigation

Sherin A. Saraireh1 and Mohammednoor Altarawneh2,a)

1Physics Department, Faculty of Sciences, Al-Hussein Bin Talal University, Ma’an, Jordan2School of Engineering and Information Technology, Murdoch University, Perth, Australia

(Received 20 March 2014; accepted 17 July 2014; published online 5 August 2014)

In this study, we report a comprehensive density functional theory investigation of the structureand thermodynamic stability of FeCl2 and FeCl3 surfaces. Calculated lattice constants and heats offormation for bulk FeCl2 and FeCl3 were found to be in relatively good agreement with experimentalmeasurements. We provide structural parameters for 15 distinct FeCl2 and FeCl3 surfaces alongthe three low-index orientations. The optimized geometries for all surfaces are compared withanalogous bulk values. Ab initio atomistic thermodynamic calculations have been carried out toassess the relative thermodynamic stability of FeCl2 and FeCl3 surfaces under practical operatingconditions of temperatures and pressures. The FeCl2 (100-Cl) surface is found to afford the moststable configuration at all experimentally accessible gas phase conditions. © 2014 AIP PublishingLLC. [http://dx.doi.org/10.1063/1.4891577]

I. INTRODUCTION

The chemisorption of halogenated species on iron-basedsurfaces is of importance for numerous technological appli-cations, such as magnetic sensors and recording media, cat-alyst, and metallurgical process.1–4 Moreover, iron (ferric)chlorides are commonly used in environmental applications:for instance, ferric(III) chloride is commonly used in watertreatment as a coagulant to remove organic content.5 Ferricchlorides are also deployed as a potential leaching agent inchemical beneficiations of many metals.6 The functionality offerric chlorides stems from their reported high reactivity andstability at elevated temperatures. However, ferric chloridesalso act as aggressive corrosive materials, which in turn re-duces the lifetime of mechanical equipment and pipelines.7 Inaddition, ferric chlorides are found as prominent heavy metalpollutants resulting from the incineration of municipal solidwaste.8, 9

Despite the importance of ferric chlorides in industrialand environmental applications, an atomic-based descriptionof ferric chlorides surfaces is still lacking in the literature.Previously, experimental10–15 and theoretical studies16 havefocused on the interaction of chlorine molecules with differ-ent clean iron surfaces. In a recent theoretical study,17 we per-formed detailed density functional theory (DFT) periodic-slabinvestigation of the interaction between atomic chlorine andthe clean Fe (100) surface. Our results indicated that the for-mation of iron chlorides via substitutional adsorption is ther-modynamically unfavorable. Our results were in accord withearlier experimental measurements with regard to the absenceof chlorine diffusion into bulk Fe.

Here we investigate the structural, electronic, and ther-modynamic properties of a range of iron chloride surfaces

a)Author to whom correspondence should be addressed. Electronic mail:M.Altarawneh@Murdoch.edu.au. Telephone: (+61) 8 9360-7507. Onleave from Chemical Engineering Department, Faculty of Engineering, Al-Hussein Bin Talal University, Ma’an, Jordan.

using density functional theory calculations. We use the ro-bust technique of ab initio atomistic thermodynamics18, 19 toaddress the relative thermodynamic stability of ferric chloridesurfaces: the essence of this technique is to deploy energiesfrom first principles calculations on microscale systems to ob-tain thermodynamic values such as the Gibbs free energy ofadsorption. As Fe(I)Cl is not stable in its solid state and onlyexists in the vapor phase,14 we limit our analysis to FeCl2and FeCl3 surfaces. Our aim in this study is twofold: first, weaim to provide optimized geometries for all plausible perfectFeCl2 and FeCl3 surfaces. Second, we aim to determine themost stable configurations under realistic operational condi-tions of varying temperatures and pressures.

II. CALCULATION METHODOLOGY

A. Density functional theory calculationsand convergence

The spin-polarized generalized gradient approximationbased on the Perdew-Wang functional (PAW-GGA) was de-ployed in all structural and energetic calculations as imple-mented in the VASP code.20–22 FeCl2 and FeCl3 surfaceswere modelled using super-cells containing symmetric slabs(with inversion symmetry). FeCl2 surfaces generally con-tained slabs with 6–18 atomic layers (ranging from 16 to34 atoms) and a 20–28 Å vacuum region between adjacentvertical slabs. In calculations of FeCl3 surfaces, 10–24 atomiclayers (containing 32–46 atoms) with a 20–30 Å vacuum re-gion were used. All surfaces were fully relaxed while keepingthe deepest (2–8) layers fixed in their corresponding bulk po-sitions. Dipole corrections were applied along the z-direction.

Calculations were performed using an energy cut-offof 320 eV. The Brillouin zone (BZ) integration was per-formed using Monkhorst-Pack (MP) grids.23 This schemehas resulted in 4 and 5 k-points in the irreducible part ofthe BZ for the (001-Fe)/(100-Cl)/(100-Fe) and the (001-Cl)/(101)/(110)/(111) directions, respectively, for the FeCl2

0021-9606/2014/141(5)/054709/11/$30.00 © 2014 AIP Publishing LLC141, 054709-1

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

134.115.2.117 On: Tue, 26 Aug 2014 06:11:14

054709-2 S. A. Saraireh and M. Altarawneh J. Chem. Phys. 141, 054709 (2014)

surfaces. The irreducible part of the BZ grid for the FeCl3surfaces consisted of 5 k-points in all directions. Tests on(100-Fe) for the FeCl2 system using an energy cutoff 400 eVand a 13 irreducible k-points changed its total energy by only1.6 MeV, indicating adequate convergence with respect tothese parameters. A similar test on the FeCl3 (101) surfacechanged its total energy by only 1.2 meV. The total energyconverged to an accuracy of 10−4 eV, and the forces on eachion to an accuracy of 0.02 eV Å−1.

B. Ab initio atomistic thermodynamics

DFT total-energy calculations described in Sec. II Aare used as input for the ab initio atomistic thermodynam-ics calculations.18, 19 This approach has been comprehensivelydemonstrated in numerous studies.24–28 Here, we only presenta brief overview. In these calculations, the most stable sur-face structure or configuration at a given temperature (T) andpressure (P) is the one that minimizes the surface free energy,γ (T, P):

γ (T , P ) = 1

A

[GSurf (0 K, 1 atm) − NFeg

BulkFeClx(0 K, 1 atm)

−(NCl − NFe)μCl(T , P )].

In this equation, GSurf(0 K, 1 atm), A, NCl, NFe signify the totalenergy for FeCl2 or FeCl3 structures obtained by default at 0K and 1 atm, surface area of the structure, and number of chlo-rine and iron atoms in FeCl2 or FeCl3 structures, in that order.The quantity gBulk

FeClx refers to energies of bulk FeCl2 or FeCl3per unit formula. The chlorine chemical potential, μCl(T, P)is calculated as

μCl(T , p)= 1

2

[Etotal

Cl2+μ̃Cl2

(T , 1 atm) + kBT ln

(P

1 atm

)],

where EtotalCl2

, kB, and μ̃Cl2(T , 1atm) denote energy of an

isolated chlorine molecule, Boltzman constant, and theT-dependent chlorine chemical potential at the ambient atmo-spheric pressure. The latter is obtained from standard thermo-dynamic tables.

Finally, the change in chlorine chemical potential(�μCl(T, p)) is expressed as

�μCl(T , p) = μCl(T , p) − 1/2ECl2.

Considering that positive values of �μCl(T, p) imposeunrealistic conditions with regard to the tendency of Cl2 to de-compose into its constituent atomic chlorine, practical condi-tions necessitate deploying negative values, entailing that Cl2reforms from atomic chlorine.

III. RESULTS

A. Bulk structure and unit cells

Iron chloride exists in two forms: iron(II) chloride (fer-rous chloride, FeCl2), and iron(III) chloride (ferric chloride,FeCl3). Under normal conditions, bulk FeCl2 occurs in a mon-oclinic structure with Cl octahedral around Fe, while bulkFeCl3 exhibits a hexagonal unit cell. These unit cells are de-picted in Fig. 1. Calculations of the lattice constant of the unit

FIG. 1. Unit cells of (a) FeCl2 and (b) FeCl3. Fe and Cl atoms are orangeand green colored, respectively. Direction of lattice vectors is shown for bothunit cells.

cells and the heat of formation were carried out using an en-ergy cutoff 600 eV and automatic generation of k-points sam-pling of (12 × 12 × 12) centered at the point. The heat offormation Ef of the FeCl2 and FeCl3 was calculated accord-ing to

Ef = EBulkFeCl

n− EBulk

Fe − n

2ECl2(g).

In which EBulkFeCl

n

and EBulkFe refer to the total energies of

bulk iron chloride and bulk iron per unit formula, correspond-ingly.

Optimized lattice constants and calculated values for theheats of formation are presented in Table I along with thecorresponding experimental data.29, 30 The energy of forma-tion (Ef) of FeCl2 and FeCl3 was calculated to be −2.31and −5.67 eV, respectively. These values are seen to slightlydeviate from analogous experimental values of −3.55 and−4.15 eV, respectively. The lattice constants a and b of FeCl2are in a reasonable agreement with corresponding experimen-tal measurements at 3.585 Å. However, our predicted c lat-tice constant significantly overshoots the experimental value.This deviation could be attributed to the well-documentedshortcoming of the standard DFT methods in describing long-range interactions. The a and c lattice constants of FeCl3 over-shoot the experimental values by 0.22 Å and 0.55 Å. Near-est atomic distances in bulk FeCl2 and FeCl3 are given inTable II. Bearing in mind the paucity of experimental mea-surements for lattice constants and Ef for FeCl2 and FeCl3,our calculated values in Table I serve as initial predictionstoward more accurate values. The noticeable difference be-tween calculated and experimental c lattice constant may con-tribute to the deviation of calculated Ef from their analogousexperimental values.

TABLE I. Calculated and experimental values from Refs. 29 and 30 for thethree lattice constants (a, b, c) and enthalpies of formation (Ef) for FeCl2 andFeCl3.

Ef (eV) a (Å) b (Å) c (Å)

FeCl2 (this work) −2.31 3.512 3.512 7.118Experimental −3.55 3.585 3.585 5.733FeCl3 (this work) −5.67 6.961 6.575 7.242Experimental −4.15 6.690 6.690 6.690

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

134.115.2.117 On: Tue, 26 Aug 2014 06:11:14

054709-3 S. A. Saraireh and M. Altarawneh J. Chem. Phys. 141, 054709 (2014)

TABLE II. Nearest atomic distances (in Å) for optimized bulk of FeCl2 andFeCl3.

Cl–Cl Cl–Fe Fe–Fe

FeCl2 3.54, 4.18 2.47, 5.59 2.54, 3.55FeCl3 3.20, 3.63 2.26, 4.13 6.55

B. Geometries of low-index of FeCl2and FeCl3 surfaces

We consider various surface terminations for FeCl2 andFeCl3. FeCl2 exhibits five inequivalent low-index orienta-tions, (001), (100), (101), (110), and (111). Each of the (001)and (100) surfaces exhibits two distinct terminations depend-ing on whether they terminate with Fe or Cl atoms in theiroutermost layer. Fig. 2 presents side and top views for theoptimized structures of FeCl2 (001-Cl) and (001-Fe), and theprimary structural parameters for these two surfaces are givenin Table III. Similarly, Fig. 3 shows side and top views of theoptimized structures of FeCl2 (100-Cl) and FeCl2 (100-Fe)and with structural parameters shown in Table IV. The opti-mized geometries for the remaining surfaces of FeCl2 that weconsidered, (101), (110), and (111), are shown in Fig. 4 andTable V.

Surface relaxations and reconstructions were found tobe minimal for (001-Cl), (001-Fe), and (100-Cl). Optimizedstructures of the other four FeCl2 configurations indicate thatFe-terminated surfaces are not stable. For instance, the FeCl2(100-Fe) surface initially contains only Fe atoms at its top-most layer, whereas the optimized minimum energy structuredisplays that the surface becomes terminated with Cl atoms.The initial structure of the FeCl2 (110) surface contains Feand Cl atoms in its outermost layer, however, the outermostlayer of the optimized surface constitutes of Cl atoms only.

TABLE III. Summary of optimized geometries for 001-Cl and 001-Fe sur-faces of FeCl2. All distances are in Å and all bond angles are in degree (◦).Labels of atoms are based on Fig. 2.

Bond length 001-Cl 001-Fe Bond angle 001-Cl 001-Cl

Cl1–Cl2 3.51 3.51 Fe1–Cl1–Fe2 54.94 90.86Cl3–Cl4 3.51 3.51 Fe2–Cl2–Fe3 54.94 90.86Cl1–Cl3 3.46 4.89 Fe1–Cl3–Fe2 90.80 . . .Cl2–Cl5 3.51 3.51 Fe2–Cl4–Fe3 90.80 . . .Fe1–Cl1 4.29 2.31 Cl2–Fe3–Fe4 44.57 89.99Fe2–Cl1 3.46 2.31 Fe3–Cl2–Cl5 135.43 90.00Fe3–Cl2 2.46 2.31 Fe3–Cl2–Fe4 90.86 56.45Fe1–Fe2 3.51 3.51 Fe3–Cl5–Fe4 26.39 56.45Fe3–Fe4 3.51 3.51 Fe1–Cl1–Cl3 35.08 149.43. . . . . . . . . Cl3–Cl2–Cl4 30.0 42.04

The downward displacement of Fe atoms was also observedfor FeCl2 (101) and (111) surfaces.

In the case of FeCl3 we investigated eight surfaces fromseven distinct orientations: (001), (010), (011-Cl), (011-Fe)(100), (101), (110), and (111). The (001) and (100) orienta-tions (i.e., parallel to the xy- and yz-planes, respectively) of-fer equivalent structural compositions. The optimized FeCl3geometries are shown in Figs. 5–7: (001) and (010) surfacesare shown in Fig. 5; (011-Cl), (011-Fe), and (100) surfacesare shown in Fig. 6; and (101), (110), and (111) are shownin Fig. 7. Tables VI–VIII list the primary structural featuresfor all FeCl3 surfaces based on atomic symbols given inFigures 5–7.

Seven of the eight FeCl3 surfaces investigated exhibitminimal reconstructions with respect to their correspondingbulk positions. The exception was the (110-Fe) surface, whichwas initially terminated with only Fe atoms and converged toa metastate in which the outermost layer consisted of only Clatoms.

FIG. 2. (001-Cl) and (001-Fe) surfaces of FeCl2 · Fe and Cl atoms are black and green colored, respectively. Sizes of atoms in “top view” vary to reflect theirpositions along the z-direction where atoms in the topmost layer are the biggest.

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

134.115.2.117 On: Tue, 26 Aug 2014 06:11:14

054709-4 S. A. Saraireh and M. Altarawneh J. Chem. Phys. 141, 054709 (2014)

FIG. 3. (100-Cl) and (100-Fe) surfaces of FeCl2 · Fe and Cl atoms are black and green colored, respectively. Sizes of atoms in “top view” vary to reflect theirpositions along the z-direction where atoms in the topmost layer are the biggest.

We are now in a position to make some general remarksregarding the geometrical characteristics of optimized FeCl2and FeCl3 surfaces:

� The FeCl2 (001-Fe) surface is the only structure that istruncated with only Fe atoms in its outermost layer.The presence of a second chlorine atomic layer be-neath the first iron layer acts to inhibit Fe atoms fromthis particular surface. All of the other structures thatinitially contained Fe only atoms in their outer atomiclayer underwent a displacement of Fe atoms toward thebulk and converged to structures in which their outer-most layers consist of Cl atoms only.

� The co-existence of Fe and Cl atoms in the outermostlayers occurs only on the FeCl3 (101) surface.

TABLE IV. Summary of optimized geometries for 100-Cl and 100-Fe sur-faces of FeCl2. All distances are in Å and all bond angles are in degree (◦).Labels of atoms are based on Fig. 3. Numbers in parentheses refer to valuespertinent to the (100-Fe) surface.

Bond length 100-Cl 100-Fe Bond angle 100-Cl 100-Fe

Cl1–Cl2 4.64 4.65 Fe1–Cl1–Cl2 107.48 130.79Cl1–Cl3 3.57 3.63 Fe1–Cl1–Cl3 50.72 105.40Cl2–Cl4 3.51 3.96 Cl2–Fe2–Cl4 92.36 87.79Cl1–Fe1 2.31 1.39 Fe2–Cl4–Fe4 100.06 52.27Cl3–Fe1 2.76 4.22 Cl2–Cl5–Fe2 32.18 23.46Cl2–Fe2 2.21 2.88 Cl2–Cl5–Fe5(7) 90.00 . . .

Cl4–Fe2 2.63 2.82 Fe2–Fe5(7)–Cl2 90.00 127.53

Fe2–Fe5(7) 3.51 . . . Fe2–Fe5(7)–Cl5 . . . 127.5

Cl5(7)–Fe2 4.15 5.74 Fe3–Cl3–Fe5 . . . 104.15

Cl2–Fe5(7) 4.15 . . . Fe4–Cl4–Fe6 . . . 82.58

Cl5–Fe5(7) 2.21 2.88 . . . . . . . . .

Cl4–Fe6 . . . 3.18 . . . . . . . . .

� A comparison of the geometries of bulk FeCl2(Table II) with our optimized FeCl2 surfaces revealsthat all FeCl2 surfaces retain the qualitative character-istics of bulk FeCl2. For example, the shortest Fe–Fedistances in FeCl2 surfaces are found to change byonly 2.5% in reference to bulk FeCl2 (2.54 Å). Theshortest Fe–Cl distances in FeCl2 surfaces are within−10.5% to +16.3% with respect to the shortest Fe–Cl distances in bulk FeCl2 (i.e., 2.47 Å). Likewise, theshortest Fe–Cl spacings in FeCl3 deviates marginally(−4.4% to +2.6%) in reference to the analogous valuein bulk FeCl3 (i.e., 2.26 Å). Generally, all FeCl3 Cl–Cl distances are within ±15.0% of their correspondingbulk values.

� Certain FeCl2 and FeCl3 surfaces share similar geo-metrical features. For example, both FeCl2 (100-Cl)and FeCl3 (001-Cl) configurations display adjacent,separated sheets of iron and chlorine atoms.

C. Thermodynamic phase diagrams

In order to address the stability of FeCl2 and FeCl3surfaces we applied the technique of ab initio atomistic ther-modynamics. The fundamental assumption of this approachis that FeCl2 and FeCl3 surfaces are formed because of theexistence of an Fe bulk in a chlorine gas phase reservoir im-posed by a given Cl chemical potential. We estimated surfacefree energies as a function of chlorine chemical potential, asdemonstrated in Sec. II B. In practice, the chlorine chemicalpotential varies only within certain limit: a lower limit (termedthe Cl-poor limit) signifies the decomposition of the ironchloride into its constituent elements, iron metal, and gaseouschlorine. It follows that reasonable Cl-poor limit values forFeCl2 and FeCl3 can be set at their Ef values, −2.31 eVand −5.67 eV, respectively. Correspondingly, the upper

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

134.115.2.117 On: Tue, 26 Aug 2014 06:11:14

054709-5 S. A. Saraireh and M. Altarawneh J. Chem. Phys. 141, 054709 (2014)

FIG. 4. (101), (110), and (111) surfaces of FeCl2 · Fe and Cl atoms are black and green colored, respectively. Sizes of atoms in “top view” vary to reflect theirpositions along the z-direction where atoms in the topmost layer are the biggest.

limit (Cl-rich) values denote a very rich gaseous chlorineenvironment, i.e., zero chlorine chemical potential. Table IXpresents calculated values of γ (T, P) at the two definedboundaries alongside with ratios of chlorine to iron atoms(RCl/Fe) for all FeCl2 and FeCl3 surfaces. A thermodynamicequilibrium is established between these two boundary limits.Thus, it is instrumental to explore trends in surface energieswhen as chlorine chemical potential gradually increases fromCl-poor limit to Cl-rich limit. Fig. 8 plots the dependenceof surface energies on values of �μCl(T, p) for FeCl2 and

FeCl3 surfaces. Selected T-P bar lines are inserted in Fig. 8to demonstrate the variation of γ (T, P) values with thechange of these two operational parameters. Most of thestable FeCl2/FeCl3 constructions are highlighted in Fig. 8.

It is evident from the energy phase diagram in Fig. 8(a)that the thermodynamic stability of the FeCl2 system is dom-inated by two terminations, namely, the 001-Fe and 100-Clsurfaces. As the most chlorine-deficient termination (RCl/Fe= 1.5), the 001-Fe surface is the only configuration thatis associated with a positive slope. As a result, the 001-Fe

TABLE V. Optimized geometries for 101, 110, and 111 surfaces of FeCl2. All distances are in Å and all bondangles are in degree (◦). Labels of atoms are based on Fig. 4. Numbers in parentheses refer to values pertinent tothe (111) surface.

Bond length 101 110 111 Bond length 101 110 111

Cl1–Cl2 3.51 5.97 5.96 Cl1–Fe1 2.48 6.25 2.25Cl1–Cl3 6.16 4.73 6.08 Cl2–Fe2 2.48 . . . 2.25Cl3–Cl4 3.51 4.39 4.80 Cl1–Cl5 3.71 5.29 3.40

Cl2–Cl4 6.16 4.71 7.94 Bond angle 101 110 111

Cl1–Fe1 2.48 6.25 2.25 Cl1–Cl2–Cl4 106.48 36.81 . . .Cl5–Cl6 3.51 . . . 6.08 Cl1–Cl3–Fe1 20.41 121.38 . . .Fe1(3)–Cl5 2.44 4.45 2.76 Cl3–Fe1(2)–Cl5(2) 30.03 . . . 1.03.8

Fe2–Cl6(2) 2.44 . . . 2.25 Cl2–Cl4–Fe2 20.41 . . . . . .

Fe1–Fe2 3.51 2.89 6.08 Cl4(5)–Fe2–Cl6(2) 38.03 . . . 127.7

Fe2–Cl3 7.02 2.41 2.25 Cl1–Fe1(3)–Cl2 . . . 72.12 101.41

Fe3(4)–Cl4(3) . . . 5.89 2.41 Fe2–Cl3–Cl4 89.87 163.46 105.80

Fe2–Fe4 . . . . . . 2.41 Cl1–Fe1–Cl4 . . . 34.35 . . .Cl5–Fe2 . . . 2.46 2.23 Cl4–Cl2(3)–Fe3(4) . . . 106.42 129.57

Cl5–Fe4 . . . . . . 3.98 Fe1–Fe3–Cl1 . . . . . . 55.56Cl2(1)–Fe3 . . . 2.45 2.41 Fe1–Cl1–Cl2 134.82 21.75 110.37

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

134.115.2.117 On: Tue, 26 Aug 2014 06:11:14

054709-6 S. A. Saraireh and M. Altarawneh J. Chem. Phys. 141, 054709 (2014)

FIG. 5. (001) and (010) surfaces of FeCl3 · Fe and Cl atoms are black and green colored, respectively. Sizes of atoms in “top view” vary to reflect their positionsalong the z-direction where atoms in the topmost layer are the biggest.

surface displays a very high stability in regions below the Cl-poor limit (i.e., <−3.95 eV). Between the Cl-poor and Cl-rich limits the prevailing stability of the chlorine-rich 100-Clsurface is easily recognized. All of the other terminations in-cur very high surface energies. The noticeable stability of the

100-Cl surface sources from its high chlorine-rich termina-tion (RCl/Fe = 3. 0) which serves to favor its formation underCl-rich surrounding gas phase. The facts that geometries ofthe 100-Cl surface largely exhibit corresponding geometriesof bulk FeCl2 also contribute to its high stability ordering.

FIG. 6. (011-Cl) (011-Fe) and (100) surfaces of FeCl3 · Fe and Cl atoms are black and green colored, respectively. Sizes of atoms in “top view” vary to reflecttheir positions along the z-direction where atoms in the topmost layer are the biggest.

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

134.115.2.117 On: Tue, 26 Aug 2014 06:11:14

054709-7 S. A. Saraireh and M. Altarawneh J. Chem. Phys. 141, 054709 (2014)

FIG. 7. (101), (110), and (111) surfaces of FeCl3 · Fe and Cl atoms are black and green colored, respectively. Sizes of atoms in “top view” vary to reflect theirpositions along the z-direction where atoms in the topmost layer are the biggest.

The phase diagram for FeCl3 is depicted in Fig. 8(b)where all γ (T, P) values attain negative slopes. This figureshows that for values of the chlorine chemical potential less

TABLE VI. Summary of optimized geometries for (001) and (101) surfacesof FeCl3. All distances are in Å and all bond angles are in degree (◦). Labelsof atoms are based on Fig. 5. Numbers in parentheses refer to values pertinentto the (010) surface.

Bond length 001 010 Bond length 001 010

Cl1–Cl2 3.81 2.23 Fe1–Fe5(4) 1.17 2.36

Cl1–Cl3 4.80 6.87 Cl1–Cl10 1.08 . . .Cl4–Cl5 3.70 4.25 Cl5–Cl11 1.02 . . .Cl6–Cl7 3.63 2.39 Fe1–Cl6 4.27 2.42Cl8–Cl9 3.56 . . . Fe2–Cl7 . . . 2.22Cl1–Cl4 3.50 4.09 Fe5(4)–Cl5 . . . 5.33

Cl4–Cl6(5) 3.79 4.25 Bond angle 001-Cl 001-Cl

Cl6(3)–Cl8 3.90 6.95 Fe1-Cl1–Cl2 142.37 44.48

Cl2–Cl5 3.27 . . . Fe2–Cl4–Cl5 162.24 . . .Cl5–Cl7(6) 3.56 4.17 Cl1–Cl2–Cl3 154.11 158.62

Cl7(6)–Cl9 3.89 2.46 Fe3–Cl6–Cl7 120.39 165.14

Cl1–Fe1 2.32 2.83 Fe3–Cl8–Cl9(3) 112.69 42.91

Cl4(2)–Fe2 2.67 . . . Cl5(3)–Cl2(8)–Cl11(7) 14.35 . . .

Cl6(11)–Fe3 4.13 2.24 Fe1–Cl1–Cl10 133.51 . . .

Cl8–Fe4(3) 2.67 2.24 Fe1–Fe2–Cl2 . . . 44.48

Fe1–Fe2 2.55 2.23 Fe1–Cl2–Fe2 30.28 9.87Fe2–Fe3 2.49 3.20 Fe3–Cl8–Fe2(3) 48.92 56.21

Fe3–Fe4 2.34 . . . Cl6–Fe1–Cl7 . . . 32.34

than −4.0 eV, the (101) surface is thermodynamically themost stable construction. As the Cl chemical potential in-creases over a short range of −4.0 to −3.0 eV, the (001)surface becomes the most favorable configuration in thisrange. As the chlorine chemical potential become greater than−3.0 eV (toward the Cl-rich limit), the surface favors the for-mation of the (011-Cl) structure.

Combining the two-phase diagrams in Fig. 8 reproducesthe stability ordering of FeCl2 in Fig. 8(a). While FeCl3 con-figurations are more “chlorine-rich” than FeCl2 surfaces (i.e.,higher RCl/Fe ratios), the FeCl2 (100-Cl) termination has agreater thermodynamic stability than all FeCl3 surfaces overthe entire range of chlorine chemical potentials that were con-sidered. This finding is in agreement with the experimentalfindings of Hino and Lambert,14 who found that the low-pressure interaction of chlorine results in the formation ofFeCl2 bulk. Their thermochemical analysis indicated a prefer-ential desorption of FeCl2, rather than FeCl or FeCl3. Puttingthis into the context presented in the Introduction: while ap-preciable loads of FeCl2 have often been detected in prod-ucts of municipal waste incinerations,31 there are no definitemeasurements pertinent to FeCl3. We note, however, that theoverall distributions of FeCl2 and FeCl3 do not only dependon their thermodynamic stability, but kinetic factors can alsobe of quite significant importance.

To the best of our knowledge, there are no previousreports investigating the thermodynamic stability of othermetallic chlorides that we may compare our data with.

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

134.115.2.117 On: Tue, 26 Aug 2014 06:11:14

054709-8 S. A. Saraireh and M. Altarawneh J. Chem. Phys. 141, 054709 (2014)

TABLE VII. Summary of optimized geometries for 011-Cl, 011-Fe, and 100 surfaces of FeCl3. All distancesare in Å and all bond angles are in degree (◦). Labels of atoms are based on Fig. 6. Numbers in parentheses referto values pertinent to the (100) surface.

Bond length 011-Cl 011-Fe 100 Bond length 011-Cl 100-Fe 100

Cl1–Cl2 3.81 3.79 5.07 Cl11–Fe4 . . . . . . 2.26Cl2–Cl3 3.06 3.11 . . . Cl8–Fe2 . . . . . . 6.81Cl3–Cl4 2.85 3.62 5.56 Cl3–Fe3(4) . . . 3.73 2.24

Cl4–Cl5 2.07 3.72 3.35 Cl4–Fe3(2) . . . 2.21 4.47

Cl5–Cl6 3.09 3.24 5.84 Cl5–Fe3(2) . . . 2.23 . . .

Cl7–Cl8 3.31 3.31 . . . Bond angle 100–Cl 100-Fe 100

Cl1–Fe1 5.85 5.83 2.16 Cl1–Cl2–Fe1 154.93 154.34 20.07Cl2–Fe1 2.18 2.18 6.04 Cl1–Fe1–Cl2 16.01 16.33 53.77Cl3–Fe1 2.18 2.21 . . . Fe1–Cl3–Cl4 100.61 101.97 . . .Cl5(4)–Fe2 . . . 2.34 2.47 Cl5–Fe2(1)–Cl6 91.69 88.21 130.30

Cl6–Fe2 2.16 2.31 . . . Cl11–Fe2(4)–Cl12 81.91 83.47 120.52

Cl8–Fe1(2) 5.41 5.41 6.81 Cl8–Cl9–Fe1 115.78 115.78 . . .

Cl9–Fe1(2) 2.27 2.27 2.46 Cll0–Fe4–Cl11 . . . . . . 120.52

Cl10–Fe4 . . . . . . 4.32 . . . . . . . . . . . .

However, in an analogy to the profound stability of thechlorine-covered FeCl2 (100-Cl) termination, ab initio atom-istic studies on Fe2O3,32 PdO,24 and CuO27, 28 have consis-tently concluded that oxygen-covered terminations afford themost stable surfaces. The phase diagrams in Fig. 8 indicatethat facets of iron chloride crystals are more likely to be dom-inated by the FeCl2 (100-Cl) at all experimentally accessiblegas phase conditions. In our recent study of the interaction ofchlorine with the clean Fe (100) surface,17 we have pointedout that substituted adsorption of chlorine atoms with a cov-erage of 2.0 produces the FeCl2 (001-Cl) surface. However,that configuration was found to be among the lowest stablestructures in the system of Cl and Fe(100). In an agreementwith our previous conclusion,17 the FeCl2 (001-Cl) is signif-icantly higher in energy than the most stable termination of

FeCl2 (100-Cl). It is insightful to assess the effect of defects orFeCl clustering on the structure and the stability of the FeCl2(100-Cl) surface.

D. Density of state

Having discussed the energetic and geometric features ofa broad range of low-index iron chloride surfaces, we nowconsider the electronic structures via calculating projecteddensity of states (PDOS) for selected surface reconstructions.Fig. 9 shows the PDOS for the bulk FeCl2 and the FeCl2 (001-Cl) surface. Our calculation for bulk FeCl2 [Fig. 9(a)] indi-cates that the surface state band resides just above the Fermilevel (at zero eV) where one region is highlighted. This regionextends from 0.5 to 1.0 eV. Two distinct bands in the valence

TABLE VIII. Summary of optimized geometries for 101, 110, and 111 surfaces of FeCl3. All distances are inÅ and all bond angles are in degree (◦). Atoms labels are based on Fig. 7. Numbers in parentheses refer to valuespertinent to the (110) surface.

Bond length 101 110 111 Bond angle 101 110 111

Cl2–Cl3 2.24 3.82 6.41 Cl2–Fe2–Cl3 24.54 . . . . . .Cl3–Cl4 3.27 4.55 8.43 Cl7–Fe2–Cl8 35.02 . . . . . .Cl5–Cl6 2.24 4.28 4.47 Cl9–Fe5–Cl10 . . . 95.40 . . .Cl7–Cl8 3.70 4.55 4.49 Cl8–Cl3–Cl9 24.03 . . . . . .Cl8–Cl9 8.33 3.74 . . . Cl7(1)–Cl8(2)–Fe4(1) . . . 25.33 . . .

Cl8–Cl2 5.73 . . . . . . Cl5(7)–Fe4(3)–Cl6 24.54 39.20 . . .

Cl8–Cl3 3.39 . . . . . . Cl6(5)–Cl5(6)–Fe3 45.82 25.33 . . .

Cl8–Cl4 2.99 . . . . . . Cl4(3)–Cl3(4)–Fe2 43.67 24.54 . . .

Fe1–Cl2 2.28 5.29 2.17 Cl3–Fe2–Fe1 43.32 . . . . . .Fe2–Cl3 2.29 2.26 . . . Cl5–Fe2–Cl6 . . . . . . 150.48Fe3–Cl4(6) 5.85 5.30 . . . Cl7–Fe3–Cl8 . . . . . . 173.80

Fe3–Cl5(7) 2.28 2.27 2.24 Cl4–Cl5–Fe2 . . . . . . 47.30

Fe5–Cl7(9) 2.27 2.32 . . . Cl1–Cl2–Fe1 . . . . . . 164.46

Fe5–Cl10 . . . 2.27 . . . Cl1–Cl2–Cl3 . . . . . . 29.54Fe1–Cl3 . . . . . . 2.17 Cl1–Fe1–Cl2 157.47 52.12 . . .Fe3–Cl8 . . . . . . 2.25 . . . . . . . . . . . .

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

134.115.2.117 On: Tue, 26 Aug 2014 06:11:14

054709-9 S. A. Saraireh and M. Altarawneh J. Chem. Phys. 141, 054709 (2014)

TABLE IX. Values of γ (T, P) at Cl-poor and rich limits in eV/Å2 at corre-sponding ratios of chlorine to iron (RCl/Fe). Stoichiometric surfaces of FeCl2and FeCl3 will have values of RCl/Fe at 2.0 and 3.0, respectively.

γ (T, P) γ (T, P)

F.eCl2 RCl/Fe Poor Rich FeCl3 RCl/Fe Poor Rich

001-Cl 2.0 0.43 0.00 001 2.67 0.88 0.09001-Fe 1.5 0.67 0.45 100 3.6 1.98 − 0.04100-Cl 3.0 0.22 − 0.15 010 3.0 1.21 0.01100-Fe 2.0 0.34 0.06 011-Cl 4.5 1.06 − 0.13101 1.84 0.51 0.16 011-Fe 3.6 1.06 − 0.05110 2.0 0.31 0.05 101 2.0 0.82 0.29111 1.67 0.38 0.19 110 3.2 1.45 − 0.02. . . . . . . . . . . . 111 3.2 1.21 − 0.03

region can be observed: the first one is between −0.5 and−2 eV, while the other distinct band stretches from −2.7 to−4 eV. These bands predominantly belong to both Fe (3d) andCl (3p). Further investigation of the PDOS of the FeCl2 (001-Cl) structure shows two main regions in the valence band be-low the Fermi level. The first band occupies the region of−3.2 to −4 eV, while the wide band is extending from −5 to−9 eV, these two bands can be attributed to the hybridized Fe3d–Cl 3p. A wide band gab of 1.0 eV is also observed. ThePDOS of the FeCl2 (001-Cl) shows generic features similarto that of bulk FeCl2. An additional small unoccupied Fe (3d)band can be seen at +2.5 eV, reflecting an incomplete fillingof the 3d shell of iron. This surface does not exhibit metalliccharacter.

FIG. 8. Surface free energies for the inequivalent low index (a) FeCl2 and (b) FeCl3 surfaces as a function of the chlorine chemical potential.

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

134.115.2.117 On: Tue, 26 Aug 2014 06:11:14

054709-10 S. A. Saraireh and M. Altarawneh J. Chem. Phys. 141, 054709 (2014)

FIG. 9. Projected density of state (a) bulk FeCl2 and (b) FeCl2 (001-Cl) surface, (c) bulk FeCl3, and (d) FeCl3 (100) surface (Fermi level at 0 eV).

Considering the PDOS of the bulk chloride FeCl3[Fig. 9(c)] we can see two distinct bands in the unoccupiedstate in the ranges of 3–4 eV, and 1–2 eV. These two bandsshow predominantly both Fe (3d) and Cl (3p) characters. Thehybridized Fe 3d-Cl 3p band is near the Fermi level and ex-tends to −4 eV. In this case, the distinct band at the Fermilevel is evidence of metallic character for the bulk material.Comparing the PDOS of the FeCl3 (100) structure with thePDOS of the bulk FeCl3 [Fig. 9(d)] we see a shift and a broad-ening of the bands. Most notably, the hybridized Fe 3d–Cl 3pband extends to a lower energy (away from the Fermi Level)and becomes wider at 5.5 eV. The disappearing of the distinctpeak at the Fermi level indicates that unlike the bulk, the (100)surface does not exhibit metallic character.

IV. CONCLUSIONS

We performed detailed DFT calculations to investigatethe atomic structure, energies, and thermodynamics stabili-ties of various perfect FeCl2 and FeCl3 surfaces. We havecalculated the surface free energies by incorporating the ef-fect of temperature and pressure into the total DFT energy for

all plausible low-index surfaces with various terminations andpresented these energies as a function of the chloride chemicalpotential. The FeCl2 (100-Cl) is thermodynamically the moststable configuration under all practical conditions of varyingtemperatures and pressures. Electronic structures in terms ofPDOS were addressed for selected structures. The PDOS ofthe FeCl2 (001-Cl) shows a similar generic feature to that ofthe bulk FeCl2, where the latter is found to exhibit metalliccharacter. The PDOS of the FeCl3(100) surface shows a shiftand a broadening of the hybridized Fe 3d-Cl 3p band to alower energy away from the Fermi Level.

ACKNOWLEDGMENTS

This study has been supported by a grant of computingtime from the National Computational Infrastructure (NCI),Australia, and the iVEC supercomputing facilities, as wellas funds from the Australian Research Council (ARC). Wegratefully acknowledge S. R. Schofield for help with thepreparation of the manuscript.

1M. Kiguchi, T. Yokoyama, S. Terada, M. Sakano, Y. Okamoto, T. Ohta, Y.Kitajima, and H. Kuroda, Phys. Rev. B. 56, 1561 (1997).

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

134.115.2.117 On: Tue, 26 Aug 2014 06:11:14

054709-11 S. A. Saraireh and M. Altarawneh J. Chem. Phys. 141, 054709 (2014)

2Y. Nakakura, G. Zheng, and E. I. Altman, Surf. Sci. 401, 173 (1998).3D. V. Potapenko, S. E. Sysoev, A. V. Ermakov, B. J. Hinch, D. R. Strongin,A. P. Wright, and C. Kuivila, Phys. Rev. B. 68, 075408 (2003).

4M. R. Vogt, F. A. Moller, C. M. Schilz, O. M. Magnussen, and R. J. Behm,Surf. Sci. 367, L33 (1996).

5P. McMahon, D. L. Cheung, and A. Troisi, J. Phys. Chem. Lett. 2(21), 2737(2011).

6G. El Samrani, B. S. Lartiges, E. Montargès-Pelletier, V. Kazpard, O.Barrès, and J. Ghanbaja, Water Res. 38(3), 756 (2004).

7Z. Wang, J. Che, and C. Ye, Hydrometallurgy 105(1–2), 69 (2010).8T. Kawai, H. Nishihara, and K. Aramaki, Corros. Sci. 38(2), 225 (1996).9M. J. Quina, J. C. Bordado, and R. M. Quinta Ferreria, Waste Manage. 28(11), 2097 (2008).

10R. C. Burshtein and N. A. Shurmovskaya, Surf. Sci. 2, 210 (1964).11D. Gay, M. Textor, R. Mason, and Y. Iwasawa, Proc. R. Soc. London, Ser

A 356, 25 (1977).12P. A. Dowben and R. G. Jones, Surf. Sci. 84, 449 (1979).13P. A. Dowben and R. G. Jones, Surf. Sci. 88, 348 (1979).14S. Hino and R. M. Lambert, Langmuir 2, 147 (1986).15R. G. Jones, Surf. Sci. 88, 367 (1979).16Š. Pick, Surf. Sci. 602(24), 3733 (2008).17M. Altarawneh and S. Saraireh, Phys. Chem. Chem. Phys. 16, 8575

(2014).

18M. Scheffler and J. Dabrowski, Philos. Mag. A 58, 107 (1988).19K. Reuter and M. Scheffler, Phys. Rev. Lett. 90, 046103 (2003).20J. P. Perdew, J. A. Chevary, S. H. Vosko, K. A. Jackson, M. R. Pederson,

D. J. Singh, and C. Fiolhais, Phys. Rev. B 46, 6671 (1992).21P. E. Blöchl, Phys. Rev. B 50, 17953 (1994).22G. Kresse and D. Joubert, Phys. Rev. B 59, 1758 (1999).23J. Monkhorst and J. D. Pack, Phys. Rev. B 13, 5188 (1976).24J. Rogal, K. Reuter, and M. Scheffler, Phys. Rev. B 69, 075421 (2004).25W-X. Li, C. Stampfl, and M. Scheffler, Phys. Rev. B 65, 075407 (2002).26A. Soon, M. Todorova, B. Delley, and C. Stampfl, Phys. Rev. B 73, 165424

(2006).27A. Soon, M. Todorova, B. Delley, and C. Stampfl, Phys. Rev. B 75, 125420

(2007).28A. Soon, M. Todorova, B. Delley, and C. Stampfl, Surf. Sci. 601, 5809

(2007).29R. D. Bach, D. S. Shobe, H. B. Schlegel, and C. J. Nagel, J. Phys. Chem.

100(21), 8770 (1996).30C. Vettier and W. B. Yelon, J. Phys. Chem. Solid 36, 401 (1975).31A. J. Chandler, T. T. Eighmy, O. Hjelmar, D. S. Kosson, S. E. Sawell, J.

Vehlow, H. A. van der Sloot, and J. Hartlén, Municipal Solid Waste Incin-erator Residues,1st ed. (Elsevier, 1997).

32W. Bergermayer, H. Schweiger, and E. Wimmer, Phys. Rev. B 69(19),195409 (2004).

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

134.115.2.117 On: Tue, 26 Aug 2014 06:11:14