Atomically Resolved Scanning Tunneling Microscopy of Cleaved Chalcopyrite Surface

Abstract

Information on the microstructure of copper minerals is important for more detailed understanding and control of the Cu extraction processes such as flotation and direct leaching. For the first time, scanning tunneling microscopy (STM) observation with atomic resolution on CuFeS₂ surfaces has been achieved using cleaved natural chalcopyrite crystals at low temperatures. The surfaces with (011) and (012) orientations have been identified by STM observation, electron backscattering, and X-ray diffraction. (011) surfaces of two different types were observed. The one is a reconstructed surface, and the other surface has a structure that is close to that of a bulk terminated surface.

1. Introduction

Chalcopyrite (CuFeS₂) is the most important copper mineral in nature, counting for about 70% of copper reserves in the world.¹⁾ Currently, pure Cu metal is mostly produced from CuFeS₂-containing ores by an extraction process involving flotation, pyrometallurgy, and electrorefining.²⁾ As an alternative process, hydrometallurgy of CuFeS₂-containing ores have been studied intensively.^2–5) However, efficient copper leaching from CuFeS₂ is still challenge, and the hydrometallurgy for CuFeS₂-containing ores has not yet been widely put into practice.

Copper ores are declining in grade and becoming increasingly difficult to process. Thus, more detailed understanding and control of the Cu extraction process is required. Information on the microstructure of CuFeS₂ surface will help to understand the physicochemical phenomena in flotation (e.g., adsorption of collector on chalcopyrite surfaces)^6,7) and the mechanism of slow dissolution in hydrometallurgy (e.g., formation of reaction intermediates on chalcopyrite surfaces during dissolution).⁵⁾ However, there are relatively few studies on the nature of intrinsic surfaces of this material.

Previous studies show that the surfaces of this material are rather complex.^8,9) For example, Klauber has carried out X-ray photoemission spectroscopy (XPS) and measured the binding energy of photoelectrons using carefully prepared chalcopyrite surfaces and anticipated the formation of S₂ dimers on the surfaces without any leaching process.⁸⁾ He also proposed the possibility of the existence of pyrite (FeS₂) on chalcopyrite surface from the analysis of peak heights of photoelectrons. Regarding sulfur, Harmer et al., anticipated the formation of S polymers on fractured chalcopyrite surfaces.⁹⁾ It is obvious that the extensive diffusion and rebonding of surface atoms are required for the formation of pyrite or S polymers at the surfaces of chalcopyrite. Unfortunately, their XPS measurements provided no information on how pyrite and/or S polymers are distributed on the surface.

As expected, the formation of the above-mentioned complex structures has not been predicted in previously conducted theoretical simulations, which only showed some reconstructed structures and the formation of S dimers on some surfaces.^10–13) This may be because the capability of theoretical simulations to predict global minimum structures in an energy landscape remains limited owing to their high computational cost. Therefore, an experimental method to detect atomic arrangement on a surface is desirable to obtain certain knowledge about the surface.

Scanning probe microscopy (SPM) is another way to investigate the microstructures of surfaces. Although, SPM cannot be used to directly detect the elements of atoms, it can be used to observe the arrangement of atoms on a surface. By combining SPM results with theoretical calculations and the results of other experimental methods, SPM can be used to determine the structures of surfaces at the atomic level. Since SPM can observe local structure, it is also suitable for surfaces on which the local formation of other structures, such as S₂ dimer and/or pyrite, may occur.

So far, the investigation of surfaces of chalcopyrite by SPM has mainly been conducted with atomic force microscopy (AFM).^14–16) However, the spatial resolution in these trials was insufficient for investigating the atomic arrangement on CuFeS₂ surfaces. This is partially because in previous studies, the surfaces were prepared by mechanical polishing. In our experiments, we succeeded in preparing atomically flat surfaces by cleaving chalcopyrite-crystal at a low temperature. On such surfaces, extensive diffusion and rearrangement of atoms are expected to be suppressed. Therefore, the results obtained by STM can be compared with those of theoretical simulations and provide fundamental information for investigation of more complicated and large-scale atomic structures and chemical reactions on these surfaces.

2. Experimental Procedure

Chalcopyrite crystals of museum grade from Dashkesan mine, Azerbaijan were purchased and used as samples. We prepared a sample by cutting the crystal into small pieces of about 10 mm × 5 mm with a thickness of 1 mm. After introducing the sample into an ultrahigh vacuum (UHV) chamber, the sample was cooled by liquid nitrogen flow for 10–15 min. The nominal temperature during the cooling was 100 K. Then, the sample was cleaved using a push rod. After that, the sample was moved into the scanning tunneling microscopy (STM) unit that was kept at ∼77 K as quickly as possible. All STM observations were carried out under UHV and low-temperature conditions. We carried out ex-situ electron backscattering diffraction (EBSD) analyses (JEOL JXA-8530F) to determine the crystallographic orientation of the samples. For some samples, X-ray diffraction (XRD) analyses (Rigaku SmartLab, PANalytical X’Pert-ProMPD), were carried out to check the purity and crystallographic orientation after STM observations.

CuFeS₂ has a tetragonal crystal structure with the space group $\text{I}\bar{4}2\text{d}$ and lattice constants of a = 0.529 nm and c = 1.041 nm.¹⁷⁾ As shown in Fig. 1, its structure is very similar to the stacking of two zincblende structures in the c-direction, if one does not distinguish copper from iron. Moreover, the c/a ratio of this tetragonal structure is 1.968, which is very close to 2. In our EBSD analysis, we were unable to determine the crystallographic orientation of the tetragonal crystal; instead, we determined the orientation by assuming that CuFeS₂ has a zincblende structure. In this case, the obtained (h k l) surface in cubic notation corresponds to the (h k 2l), (l h 2k) or (k l 2h) surfaces in a tetragonal structure. In this paper, we describe such reduced orientation that is obtained by assuming the cubic structure as (h k l)_c by putting “c” after the closing parenthesis.

Fig. 1

Unit cell of CuFeS₂. Blue, brown, and yellow balls correspond to copper, iron, and sulfur atoms, respectively.

The OpenMX code was employed to simulate atomic arrangements and STM images of chalcopyrite surfaces using slab model.^18,19) We conducted spin-polarized calculation with generalized gradient approximation (GGA)-PBE functional. The pseudoatomic orbitals for sulfur, iron, and copper were S8.0-S3p3d1f1, Fe5.5H-s3p3d2, and Cu8.0H-s3p3d3, respectively. These are provided in the package of OpenMX. The cut-off energy and electronic temperature were set to 180 Ry and 500 K, respectively for good convergence of self-consistent field calculations. We relaxed the position of atoms until the maximum force acting on the atoms becomes less than 10⁻⁴ hartree/bohr. The energy dependences of the density of states and the local density of states were calculated using the final position of the relaxation with the Hubbard potential of 4.0 eV for iron and copper atoms.²⁰⁾

3. Results and Discussions

Figure 2 shows the typical result of powder XRD measurement of the purchased crystal after pulverization. The result is almost identical to the previous result and all observed peaks can be assigned to the peak of CuFeS₂.²¹⁾ The crystal is expected to not have secondary phases.

Fig. 2

XRD pattern of CuFeS₂ sample after pulverization. The reference data (red circle) are taken from PDF # 00-037-0471 for CuFeS₂.

Figure 3 is low-magnification SEM image of a fractured sample. Although river pattern can be seen, there is no large flat plane with specific orientation in this image. The overall slope of the surface gradually changes. However, as will be mentioned below, we found the surface of this material consists of small atomically flat surfaces.

Fig. 3

An example of SEM micrograph of cleaved CuFeS₂ crystal. The area of CuFeS₂ is indicated by white broken line.

In our STM observations, we found three different types of surface. In Figs. 4(a) and 4(b), we showed two of them. At first glance, these two are different; the surface in Fig. 4(a) consists of one-dimensional rows of bright spots whereas that in Fig. 4(b) appears flat with regular arrangements of atoms. The smallest spacing between the rows of bright spots in Fig. 4(a) is about 0.6 nm and a twofold spacing is most frequently observed. In some regions, a close-packed arrangement of these bright spots can be seen, as indicated in the inset of Fig. 4(a).

Fig. 4

Two types of surface observed on cleaved CuFeS₂. (a) The surface consists of rows of bright spots. Sample bias = −1.3 V. Tunneling current = 0.15 nA. Inset: Enlarged image showing close-packed arrangement of the bright spots. (b) Inside of the white oval, we can see an atomically flat and regular structure. Sample bias = −1.35 V. Tunneling current = 0.1 nA. (b) is a derivative image. Inset: Enlarged image of the structure shown inside the white oval. The black rectangles in both insets are the unit cells of the (011) surface.

The results of fast Fourier transform (FFT) of the images of these two types of surfaces are shown in Figs. 5(a) and 5(b), indicating that both types of surface have the same periodicity. Therefore, these are expected to be surfaces of the same orientation. The reciprocal lattice of these surfaces is defined with vectors of $|\overrightarrow{b_{1}}| = 2.149$ nm⁻¹, $|\overrightarrow{b_{2}}| = 1.67$ nm⁻¹, and θ₁₂ = 116.8° on average. We found that the observed periodicity corresponds to that of (011) within a 4% error. The (120) surface is another candidate for the surface, if we do not distinguish between copper and iron atoms. The accuracy of the length determined by STM is generally insufficient for distinguishing (011) from (120), because the c/a ratio in chalcopyrite is very close to 2. However, as will be demonstrated, we can expect that STM can distinguish the iron atoms from the copper atoms even if they are placed at subsurface. In this case, the period along ⟨001⟩_c ([001] for the (120) surface) becomes twice that of the observed one. As a result, the surface shown in Fig. 4 is concluded to be the (011) surface. From the structure of CuFeS₂, we can expect different types of (011) surface, namely, copper-terminated, iron-terminated and sulfur-terminated (011) surfaces if we assume an ideal bulk terminated surface (see Fig. 5(d)). It is natural to consider that the difference in the appearance between the images in Figs. 4(a) and 4(b) is due to the difference in termination, although the atomic structure of the observed surface is more complicated than ideal one especially for the case in Fig. 4(a).

Fig. 5

(a) FFT image of the surface that has the same structures as those in Fig. 4(a). (b) FFT image of the surface in Fig. 4(b). (c) Calculated lattice point in real space obtained from the average of several FFT images. The distances between lattice points are also indicated. The numbers in parentheses in (c) are the ideal value of distances on the (011) surface of CuFeS₂. (d) The red rectangle shows an example of the (011) surface. In this example, the topmost atoms are copper. We can make the surface in which the topmost atoms are iron or sulfur by cutting the crystal with the plane of the same orientation.

The last type of surface is shown in the STM image in Fig. 6(a). This surface shows an aggregation of very small flat terraces consisting of rows of atoms. The spacing between these rows is 0.54 nm, and the period along a row of atoms is 0.36 nm (Fig. 6(d)). The observed periodicity and symmetry of the surface correspond to those of the (110) or (012) surface (0.529 nm × 0.374 nm is the expected unit cell for (110), and 0.529 nm × 0.372 nm for (012)).

Fig. 6

Another type of surface obtained from our STM observations. (a) The surface consists of very small terraces. Each terrace consists of one-dimensional rows of atoms. The sample bias is +2.0 V. The tunneling current is 0.15 nA. We found that STM images of the surface show bias voltage dependence. (b) Typical STM image obtained under positive bias voltage. The bias voltage is +2.0 V (empty state image). (c) STM image obtained at the sample position as (b). The bias voltage is −2.0 V (filled state image). To compare the positions of dark lines, red arrows are put at the same positions of the surface in both (b) and (c). (d) Line profile along the white dotted line in (a). The vertical lines indicate the positions of surface atoms.

We found an interesting feature that helps determine the orientation of this surface, which are the dark lines on the terraces in Figs. 6(b) and 6(c). The spacing between these dark lines is about 1.49 nm. Such a long period in the ⟨110⟩_c direction cannot be realized in (110), but it is possible in the (012) surface. Interestingly, the width and position of these dark lines depend on the polarity of bias voltage; the width of a dark line is larger in the image obtained under a negative bias voltage than that in the image obtained under a positive bias voltage. The position of the center of the dark line is slightly shifted between them.

Thus far, a few theoretical investigations of the surfaces including (012) and (110) have been reported.^10,12) All theoretical simulations of the (012) or (110) surface predict only a relaxed structure; no reconstructed structure has been reported. This is also confirmed in our simulation shown in Fig. 7(a). The obtained structure is very similar to that obtained in previous studies.^10,12) A common feature of the chalcopyrite surface is that sulfur atoms tend to protrude to the vacuum.^10–12) The relaxed structure of the (012) surface is no exception; all the topmost atoms are sulfur atoms. Each sulfur atom has back bonding to two underneath metal atoms, namely iron or copper. We found that the height of sulfur atoms slightly differs depending on the species of metal atoms underneath (0.02 nm in maximum).

Fig. 7

(a) Side view of the slab model of (012) surface shows atom positions of relaxed and bulk terminated surfaces. The partial density of states at a constant height of ∼0.6 nm above the topmost sulfur atom are shown as a color map in (b) and (c) for the simulation of STM image. Interval of integration is from the Fermi level to +2.0 eV for (b) and from −2.0 eV to the Fermi level for (c). These correspond to the empty state (positive bias) and filled state (negative bias) STM images respectively.

The observed change in the contrast of the image depending on the polarity of the bias voltage indicates that we need to consider the energy dependence of the density of states in addition to the position of the atoms for the simulation of an STM image. Thus, we calculated the integration of the local density of states from the Fermi level to +2 eV (Fig. 7(b)) and from −2 eV to the Fermi level (Fig. 7(c)). Then, the STM image is roughly simulated by the integrated local density of states at the same height (∼0.6 nm above the highest sulfur atom). The simulated STM image shows similar patterns of dark (low-density site) and bright (high-density site) contrasts, to the experimental results; the changes in the width of the dark line and the position of the center of the dark line can be explained by the simulation. From this finding, we concluded that the surface shown in Fig. 6(a) is (012) rather than (110). Another point we can extract from this result is that we can distinguish copper from iron on this surface by STM. Actually, theoretical simulations have shown that copper atoms have a large density of states below the Fermi level and iron atoms have a large density of states above the Fermi level.^20,22,23)

Thinius et al., carried out a comprehensive study of surface energy and the atomic structure of chalcopyrite surfaces using density functional theory (DFT) calculation.¹²⁾ The relaxed surfaces presented in the order from the lowest to highest surface energy are (110), (012), (120), (122), (221), (121), (111), (011), (001), (112), and (010). They also estimated the energy of reconstructed surfaces and determined the following surfaces presented in the same order: (122), (120), [(110), (012)], (221), (121), (111), (011), (001), (112) and (010) (for (110) and (012), only the relaxed structure was found). Our experimentally obtained (012) surface reasonably corresponds to their results; the surface energy of (012) is higher than that of the most stable (122) by only ∼4% according to Ref. 12). On the other hand, the (011) surface did not show a low estimated energy. However, the (011) surface determined in our study has also been observed in a previous study as a distinct cleavage plane of this material.²⁴⁾ In that previous study, in addition to (011), (110)_c (including (012)) and (112) orientations were indicated as possible cleavage planes. Therefore, a search for an atomically flat (112) surface and observation by STM would be the target of future investigations. Regarding the (112) surface, Chen et al. predicted that the (112) and (-1-1-2) surface is unexpectedly stable in spite of the dipole moment perpendicular to the surface.¹⁰⁾

Figure 8 shows the scanning electron microscopy (SEM) micrograph and the result of EBSD of the surface shown in Figs. 4(a) and 6. For the surface in Fig. 6, we also determine the crystallographic orientation of the sample by single-crystal XRD analysis to check the relevance of our EBSD analysis. From the inverse pole figure obtained from the area inside the white rectangles in Figs. 8(a) and (c), we determined the crystallographic orientation of the sample’s normal in Fig. 4(a) as [0.29 0.24 0.93]_c and that in Fig. 6 as [0.57 0.26 0.77]_c by EBSD. Also, the orientation of sample’s normal in Fig. 6 is determined as [0.62 0.26 0.73]_c by single-crystal XRD. The difference between the orientation obtained by EBSD and that determined by single-crystal XRD is 3.9° for the surface in Fig. 6. The angle between the orientations of the surface assigned by STM observations and the sample normal is 16–18° for both surfaces.

Fig. 8

(a) SEM micrograph of the surface in Fig. 4(a). (b) Inverse pole figure of the surface in Fig. 4(a) obtained by EBSD analysis. (c) SEM micrograph of the surface in Fig. 6. (d) Inverse pole figure of the surface in Fig. 6. The orientation of surface normal obtained by single-crystal XRD is indicated by a star. The rectangles in (a) and (c) are the areas of EBSD analysis.

The SEM micrographs in Fig. 8 show a slightly wavy surface with river patterns. The mode of fracture is not a perfect cleavage, as pointed out by previous studies.²⁴⁾ Actually, we sometimes observed the surface with nanoscale corrugations on which no atomic resolution was obtained by STM. Therefore, the fracture of this material is expected to be a mixture of cleavage with an atomically flat surface and the fracture without defined planes.

Interestingly, even in the case of cleavage in ambient conditions, our observations showed that nearly atomically flat terraces of the same orientations (mainly (011)) appear on this material. The surface of the terraces has disordered arrangements of atoms with a thickness of 1–2 atomic layer (not shown here). These findings indicate that the surface obtained under conditions close to the actual process retain the structures similar to those reported in this article. Therefore, the results reported here will be valuable for not only the investigation of fundamental aspects of chalcopyrite surface but for the development of an effective method for the industrial mineral processing of CuFeS₂-contaning ores.

4. Conclusion

By fracturing chalcopyrite crystal at low temperatures, we found that atomically flat surfaces can appear on chalcopyrite samples. For the first time, atomic-resolution images of the surfaces of CuFeS₂ were obtained, and two orientations namely (011) and (012), were assigned. On (011), we found two different types of surface: one consists of rows of bright spots and the other appeared atomically flat. On the (012) surface, we explained the observed long period in the contrast of STM images as being due to the difference in the species of underlying atoms by comparing experimental results with the DFT calculation results.

Although the morphology of fractured surface of CuFeS₂ didn’t show large flat plane in the scale of low-magnification SEM observation, considerable fraction of the surface after fracture consisted of atomically flat surfaces. The findings in this study will be useful and basic information for more detailed understanding of flotation phenomena and leaching reactions of chalcopyrite and for investigating the origin of complex atomic re-arrangements on sulfide surfaces.

Acknowledgments

We appreciate kind support from Prof. T. Uda and Dr. Kazumi (EBSD and powder XRD, Kyoto Univ.), Dr. T. Yoshikawa and Mr. A. Nakagawa (Single crystal XRD, Shimadzu Techno-Research, Inc.).

This research was partially supported by JSPS KAKENHI Grant Number 20H02494.

REFERENCES

• 1)   S.  Wang: JOM 57 (2005) 48–51. doi:10.1007/s11837-005-0252-5
• 2)  M.E. Schlesinger, M.J. King, K.C. Sole and W.G. Davenport: Extractive Metallurgy of Copper, fifth edition, (Elsevier, 2011).
• 3)   E.  Córdoba,  J.  Muñoz,  M.  Blázquez,  F.  González and  A.  Ballester: Hydrometallurgy 93 (2008) 81–87. doi:10.1016/j.hydromet.2008.04.015
• 4)   E.  Córdoba,  J.  Muñoz,  M.  Blázquez,  F.  González and  A.  Ballester: Hydrometallurgy 93 (2008) 88–96. doi:10.1016/j.hydromet.2008.04.016
• 5)   Y.  Li,  N.  Kawashima,  J.  Li,  A.  Chandra and  A.  Gerson: Adv. Coll. Int. Sci. 197–198 (2013) 1–32. doi:10.1016/j.cis.2013.03.004
• 6)   Y.  Jia,  K.  Huang,  S.  Wang,  Z.  Cao and  H.  Zhong: Miner. Eng. 137 (2019) 187–199. doi:10.1016/j.mineng.2019.04.015
• 7)   A.  Fredriksson,  A.  Holmgren and  W.  Forsling: Miner. Eng. 19 (2006) 784–789. doi:10.1016/j.mineng.2005.09.037
• 8)   C.  Klauber: Surf. Interface Anal. 35 (2003) 415–428. doi:10.1002/sia.1539
• 9)   S.  Harmer,  A.  Pratt,  W.  Nesbitt and  M.  Fleet: Am. Mineral. 89 (2004) 1026–1032. doi:10.2138/am-2004-0713
• 10)   V.  Chen,  G.  Mallia,  R.  Martínez-Casado and  N.  Harrison: Phys. Rev. B 92 (2015) 155426. doi:10.1103/PhysRevB.92.155426
• 11)   C.  Oliveira,  G.  Lima,  H.  Abreu and  H.  Duarte: J. Phys. Chem. C 116 (2012) 6357–6366. doi:10.1021/jp300713z
• 12)   S.  Thinius,  M.  Islam and  T.  Bredow: Surf. Sci. 669 (2018) 1–9. doi:10.1016/j.susc.2017.11.003
• 13)   H.  Zhao,  Y.  Zhang,  M.  Sun,  P.  Ou,  Y.  Zhang,  R.  Liao and  G.  Qiu: Hydrometallurgy 187 (2019) 18–29. doi:10.1016/j.hydromet.2019.05.002
• 14)   G.  Qian,  C.  Gibson,  S.  Harmer-Bassell and  A.  Pring: Minerals 10 (2020) 485. doi:10.3390/min10060485
• 15)   H.  Petrus,  T.  Hirajima,  K.  Sasaki and  H.  Okamoto: Colloids Surf. A 389 (2011) 266–273. doi:10.1016/j.colsurfa.2011.08.015
• 16)   M.  Farquhar,  P.  Wincott,  R.  Wogelius and  D.  Vaughan: Appl. Surf. Sci. 218 (2003) 34–43. doi:10.1016/S0169-4332(03)00538-5
• 17)   R.  Adams,  R.  Beaulieu,  M.  Vassiliadis and  A.  Wold: Mater. Res. Bull. 7 (1972) 87–91. doi:10.1016/0025-5408(72)90264-4
• 18)   T.  Ozaki and  H.  Kino: Phys. Rev. B 72 (2005) 045121. doi:10.1103/PhysRevB.72.045121
• 19)   K.  Lejaeghere et al.: Science 351 (2016) aad3000. doi:10.1126/science.aad3000
• 20)   R.  Khaledialidusti,  A.  Mishra and  A.  Barnoush: AIP Adv. 9 (2019) 065021. doi:10.1063/1.5084308
• 21)   S.  Conejeros,  P.  Alemany,  M.  Llunell,  I.  Moreira,  V.  Sańchez and  J.  Llanos: Inorg. Chem. 54 (2015) 4840–4849. doi:10.1021/acs.inorgchem.5b00399
• 22)   M.  Zhou,  X.  Gao,  Y.  Cheng,  X.  Chen and  L.  Cai: Appl. Phys. A 118 (2015) 1145–1152. doi:10.1007/s00339-014-8930-1
• 23)   R.  Martínez-Casado,  V.  Chen,  G.  Mallia and  N.  Harrison: J. Chem. Phys. 144 (2016) 184702. doi:10.1063/1.4947080
• 24)   G.  Wolff and  J.  Broder: Acta Crystallogr. 12 (1959) 313–323. doi:10.1107/S0365110X59000949