Hollow platinum-gold and palladium-gold nanoparticles: synthesis and characterization of composition-structure relationship

Hollow palladium-gold (PdAu) and platinum-gold (PtAu) alloy nanoparticles (NPs) were synthesized through galvanic replacement reactions. PdAu NPs denoted PdAu-99.99 and PdAu-98 were produced using palladium precursors with different purity degree: Na2PdCl4 ≥ 99.99% and Na2PdCl4 98%, respectively. The effect of the addition time of the gold palladium precursor solution on the size of the generated NPs was evaluated. Two types of particles, with a rough and a smooth surface, were identified in the suspensions of PtAu and PdAu NPs by scanning electron microscopy (SEM), transmission electron microscopy (TEM), and scanning transmission electron microscopy (STEM). The atomic percentage of gold, platinum, palladium, and cobalt (atomic %) in the nanoparticles was determined by energy dispersive X-ray spectroscopy (EDX). PtAu NPs (26–42 nm) contain Pt (41 at%), Au (36 at%), and Co (23 at%). Two groups of hollow palladium gold NPs (30–50 nm) with a different residual cobalt content were produced. PdAu-99.99 NPs consisted of Pd (68 at%), Au (26 at%), and Co (6 at%), whereas PdAu-98 NPs were composed of Pd (70 at%), Au (22 at%), and Co (8 at%). The hollow structure of the NPs was confirmed by EDX line scanning. Selected area electron diffraction analysis (SAED) revealed the formation of PtAu and PdAu alloys and it was used in estimating the lattice parameters, too.


Introduction
Bimetallic nanoparticles have received great attention in the last years, as they exhibit outstanding properties in comparison with the corresponding monometallic forms. These unique properties include the tuned frequency shift of plasmon modes [1][2][3], the high catalytic activity towards hydrogen generation [4], the carbon dioxide reduction [5], and the reduction and oxidation of various chemical compounds such as nitrophenol [6], caffeine [7], benzyl alcohol [8], and formic acid [9]. Bimetallic nanoparticles with different shapes and structures have been reported such as nanoframes [9], nanoprisms [10], core-shell [1,11], nanoflowers [12], and hollow nanoparticles [13,14]. Within this group of nanoparticles with different morphologies, hollow nanoparticles offer additional advantages such as higher surface area, lower density, and higher catalytic activity due to the Abstract Hollow palladium-gold (PdAu) and platinum-gold (PtAu) alloy nanoparticles (NPs) were synthesized through galvanic replacement reactions. PdAu NPs denoted PdAu-99.99 and PdAu-98 were produced using palladium precursors with different purity degree: Na 2 PdCl 4 ≥ 99.99% and Na 2 PdCl 4 98%, respectively. The effect of the addition time of the gold palladium precursor solution on the size of the generated NPs was evaluated. Two types of particles, with a rough and a smooth surface, were identified in the suspensions of PtAu and PdAu NPs by scanning electron microscopy (SEM), transmission electron microscopy (TEM), and scanning transmission electron microscopy (STEM). The atomic percentage of gold, platinum, palladium, and cobalt (atomic %) in the nanoparticles was determined by energy dispersive X-ray spectroscopy (EDX). PtAu NPs (26-42 nm) contain Pt (41 at%), Au (36 at%), and Co (23 at%). Two groups of hollow palladium gold NPs (30-50 nm) with a different residual cobalt content were produced. PdAu-99.99 NPs consisted of Pd (68 at%), Au (26 at%), and Co (6 at%), whereas 1 3 Vol:. (1234567890) special arrangement of the atoms in their architecture [15][16][17][18][19].
A variety of hollow nanostructures has been synthesized through the deposition of two metals in templates which are later removed by thermal treatment [14], etching of bimetallic nanoparticles, or by galvanic replacement reactions. A starting core is oxidized by one or more noble precursors in a galvanic replacement reaction, due to the difference in Nernst potentials, and a shell of atoms of the noble precursor is formed [20]. Gold, palladium, and platinum have been used to generate hollow nanostructures, which show a higher catalytic activity than their solid counterparts [21][22][23].
Gold in bulk form does not react with molecular oxygen, and it has been regarded for a long time as a poor catalytic metal. However, in the last years the catalytic properties of bulk gold powder towards catalytic reactions of various chemical compounds such as carbene precursors, amines, benzyl alcohol, and isocyanides [24,25] have been reported. Furthermore, gold nanoparticles and their alloys such as AgAu, PtAu, and PdAu are able to promote several reactions such as the methanol oxidation [19], the reduction of 4-nitrophenol [26], the nitrite reduction [27], and the ethanol oxidation [28]. The performance of these alloys depends on the ratio between Au and the second component of the alloy and also on the arrangement of the atoms in the structure. This arrangement of the atoms is unique in the case of hollow nanoparticles. For that reason, a profound evaluation of structure and the composition of hollow PtAu and PdAu alloys are crucial.
In this study, hollow platinum-gold (PtAu) and palladium-gold (PdAu) alloy NPs were produced by galvanic replacement reactions using cobalt cores as templates. These NPs contain detectable levels of residual cobalt which have not been reported before. The PtAu and PdAu NPs were characterized by scanning electron microscopy (SEM), transmission electron microscopy (TEM), scanning transmission electron microscopy (STEM), energy dispersive X-ray spectroscopy (EDX), and selected area electron diffraction (SAED). PdAu NPs were prepared with two palladium precursors of different purity degrees. In this paper, we evaluate the influence of the purity degree of the palladium precursor on the final cobalt content of the PdAu NPs. Furthermore, the impact of the addition time of the reducing agent on the particle size was evaluated. In addition to this, the effect of the cobalt content on the crystal structure of the palladium gold and platinum gold NPs was evaluated. At the beginning of the study, PdAu NPs were synthesized with Na 2 PdCl 4 of high purity. Then, PdAu-98 NPs were produced with small changes in morphology and composition.

Synthesis of hollow PtAu and PdAu NPs
The PtAu and PdAu NPs were prepared by modification of the procedures described by Liu et al. [29] and Shang et al. [30]. For the synthesis of the hollow PtAu NPs, 15 mL of 5 mM CoCl 2 and 70 mL of water were mixed in a three-neck flask at room temperature. After that, 15 mL of 10 mM Na 3 C 6 H 5 O 7 solution were added to the flask under stirring, and the resultant solution was kept under nitrogen atmosphere and stirring for 1 h. Then, 20 mL of 0.03 M NaBH 4 solution were added turning it into a brown solution, while the nitrogen flux and the stirring were maintained. After 39 min, 4 mL of 2 mM K 2 PtCl 4 and 2.4 mL of 2 mM HAuCl 4 were dropped simultaneously in the solution. The mixture was kept under inert atmosphere and stirring for 1 h. After this time, the nitrogen flow was stopped, and the mixture was stirred for another 2 h. The resulting suspension was centrifuged at 8000 rpm and rinsed three times with water, and the NPs were resuspended in 10 mL of water.
For the synthesis of PdAu NPs, 15 mL of 6 mM CoCl 2 were added to a three-neck flask containing 70 mL of water, and the mixture was stirred. Afterwards, 15 mL of 10 mM Na 3 C 6 H 5 O 7 solution were added to the flask under stirring. The mixture was kept under inert atmosphere and stirred for 34 min. After this time, 40 mL of 0.03 M NaBH 4 solution were added to the flask, while the nitrogen flux and the stirring were maintained. Besides, 3.84 mL of 8 mM Na 2 PdCl 4 (≥ 99.99% or 98% purity degree) and 0.32 mL of 25 mM HAuCl 4 were mixed. Then, 4.16 mL of this mixture were dropped to the flask after different periods of time (179, 182, 191, and 196 min). The solution was stirred for 1 h under nitrogen atmosphere. Then, the nitrogen flux was stopped, and the mixture was stirred for 2 h. Finally, the NPs denoted that AuPd-99.99 and AuPd-98 were centrifuged, rinsed, and resuspended in 10 mL of water and dispersed by ultrasound treatment at 120 W by 20 min (Sonorex Digital 10P, DK 102 P, Bandelin).

Characterization of hollow PtAu and PdAu NPs
The size and the zeta potential of the NPs were determined by dynamic light scattering (DLS) and laser Doppler electrophoresis technique (LDET) using a Malvern Zetasizer Nano-ZS ZEN 3600 (MAL 500 677). The morphology and size of the NPs were evaluated by SEM in a focused ion beam system, FIB, model Auriga 60 Zeiss, and transmission electron microscopy (TEM) and STEM with a transmission electron microscope (TEM) type TECNAI 20S-Twin FEI. The samples for the TEM analysis were prepared by dropping a suspension of the NPs onto a carbon coated copper grid.
The elemental composition of the NPs was determined using EDX of the SEM and STEM. The crystallographic information of the NPs was obtained by SAED with the TECNAI microscope at 200 kV accelerating voltage.

Results and discussion
On the reaction mechanism The formation of the hollow PdAu and PtAu nanoparticles would involve three major steps. First, cobalt nanoparticles were synthesized through the reduction of cobalt ions by sodium borohydride under inert atmosphere, which prevent the oxidation of the newly formed nanoparticles [31]. These nanoparticles kept separated through electrostatic repulsion due to the negative charge of the citrate ions added previously. In the following step, cobalt nanoparticles were used as templates for the production of the hollow bimetallic nanoparticles. In this step, cobalt atoms on the surface of the nanoparticles were oxidized to cobalt ions (standard electrode potential, E°(Co 2+ /Co 0 ) = − 0.28 V) upon addition of AuCl 4 − , PdCl 4 2− , and PtCl 4 2− solutions which are solutions with high positive reduction potentials: AuCl 4 − (E°(Au 3+ /Au 0 ) = 1 V) and PdCl 4 2− (E° = 0.59 V) and PtCl 4 2− (E° = 0.76 V) [32]. Cobalt atoms were spontaneously replaced by gold and palladium or platinum atoms which nucleate and formed very small alloy particles (see below Eqs. 1 and 2). Due to the fact that the reduction potential of the gold precursor is more positive than the potentials of PdCl 4 2− and PtCl 4 2− , gold atoms would be reduced first [33]. This process led eventually to the formation of a thin shell around the cobalt templates [34], and according to Eqs. 1 and 2, the shell would be porous and incomplete because five atoms of cobalt (Van der Waals radius (rCo = 2 Å)) would be replaced by only two atoms of palladium (rPd = 2.1 Å) and two atoms of gold (rAu = 2.14 Å) or two atoms of gold and two atoms of platinum (rPt = 2.13 Å) [35]. It would allow at the early stage the diffusion of the gold, platinum, and palladium precursors and cobalt ions through the shell [30,36,37]. Finally, when the reaction mixture was exposed to air, the remaining cobalt cores were oxidized, and the nanoparticles became hollow structures.
Characterization of Hollow PtAu nanoparticles Figure 1 shows the intensity size distribution of the hollow PtAu particles obtained by DLS. The average size of the particles was 43 nm, and the zeta potential of the NPs determined by laser Doppler electrophoresis technique (LDET) was − 43 mV. The negative value of the zeta potential of PtAu NPs determined by DLS is associated with the negative charge of adsorbed citrate ions. Furthermore, the high absolute value of the zeta potential indicates that the suspension of the NPs was stable. According to SEM images, the bimetallic NPs have a spherical shape with diameters between 26 and 42 nm. Therefore, there is an apparent discrepancy between the size distribution of nanoparticles by intensity and the radius of particles obtained by SEM. However, the values reported in Fig. 1 correspond to the hydrodynamic radius of the nanoparticles which is larger than the radius of dried particles obtained by SEM. In addition, the presence of bigger nanoparticles and their agglomeration in the electrolyte contributes to stronger light scattering in the measurements with Zetasizer, thus shifting the distribution towards larger values. However, the radii obtained by SEM come from the sighting of individual dried particles.
EDX analysis revealed that PtAu NPs contain Pt, Au, and Co and an atomic Pt:Au ratio of 1:1 ( Table 1). The amount of residual cobalt (23 at%) found in the NPs is possibly due to the original composition of the core and could improve the catalytic effect of the particles towards different reactions such as the reduction of oxygen as has been demonstrated for PtCo NPs [38].
Two kinds of NPs were identified in the PtAu nanoparticle suspension by TEM. Some particles were spherical with a smooth surface, while others had a rough surface (Fig. 2). The sharp contrast between the central part and the outer surface of the NPs suggests the presence of inner cavities.
The average size of the nanoparticles with smooth and rough surfaces is very close because the particle size depends on the size of the template. Due to the fact that the newly formed atoms are deposited on the surface of the template, both the rough and smooth nanoparticles would have a slightly larger size than the original template [39,40].
Cobalt cores were oxidized to cobalt ions when PtCl 4 2− and AuCl 4 − were added. As a final effect, pores were formed on the surface where the nucleation of PtAu NPs starts (see Eq. 1). The final amount of cobalt will depend on the porosity of the initial PtAu core shell [36]. Smooth surfaces identified in TEM images of PtAu NPs would possibly be obtained by the replacement of cobalt atoms on cobalt cores placed away from the ion concentration center, as it has been suggested by Schwartzberg et al. [41] for hollow gold NPs. Rough surfaces would be the result of particle overgrowth on a longer time scale. The PtAu NPs formed at the site of the addition of the metal precursors would act as seeds for the attachment of gold and platinum atoms after the reduction of PtCl 4 2− and AuCl 4 − by citrate ions [41,42]. The formation of particles with rough surfaces could also be explained by the large lattice mismatch between gold and platinum (3.9%) which favors the island growth mode [43]. The rough surface of the NPs would be appropriate for catalytic purposes because of their larger surface area and higher number of defects.
The overgrowth of the nanoparticles that lead to rough surfaces would be limited by the availability of gold and palladium precursors which depends on the stirring which was relatively high. Therefore, the particle size would not increase considerably.
According to Fig. 3a, the NPs with a smooth surface contain both Pt and Au. The characteristic line  of copper was observed due to the carbon-coated copper grid. According to the line profile, the amount of gold and platinum was higher at the starting position of the nanoparticle and at the opposite side of it in comparison with the central area of the nanoparticle (resolution for the probe beam of 1 nm). This profile is similar to that of the second particle, although, now it contains a lower amount of Au and Pt. The profiles of the NPs resembled the characteristic profile of hollow particles as reported by Rades et al. [44] for hollow silica particles with a higher amount of platinum and gold at the starting position and the opposite side of the NPs. Figure 3b shows the EDX line scan through a PtAu NPs, which exhibit a rough surface. The EDX profile showed the characteristic line of gold and platinum, even though, it suggested that the second type of particles has a higher content of gold.
The distribution of atoms in the shell was analyzed by SAED. TEM images and their corresponding SAED patterns are shown in Fig. 4. The pattern with concentric rings and spots confirmed that both groups of NPs shown in Fig. 4a and c were polycrystalline. The rings were assigned to the (100), (200), (220), (311), and (420) planes of a fcc structure. The interplanar distance (d) for each plane family was calculated from the rings of both groups of NPs. The d-values are given in Table 2, together with the reference values for the planes in pure gold and platinum obtained from ICDD PDFs 00-004-0784 and 00-004-0802, respectively. According to the SAED analysis of PtAu NPs, the interplanar spacing obtained for the planes in the first group of PtAu nanostructures ( Fig. 4a and  b) were intermediate values between the distances in elemental gold and platinum ( Table 2). The lattice parameter for PtAu was calculated (3.982 Å) from the interplanar spacing between (200) planes. The lattice parameter of pure Au and pure Pt (ICDD PDFs 00-004-0784 and 00-004-0802) were 4.079 Å and 3.924 Å, respectively, and the lattice parameter calculated for PtAu lies between these values. Both features, intermediate values of interplanar spacings and of the lattice constant, confirmed that the first group of PtAu NPs (particles with smooth and rough surfaces) was PtAu alloys.
For the second group of NPs, two d-spacing values (220) and (420) are intermediate values to those reported in pure gold and platinum. Three d-spacing values smaller than those from a fcc platinum structure and lower lattice parameter (3.894 Å, value calculated from the SAED data) than those corresponding to pure gold and platinum could imply a contraction of the lattice due to the presence of cobalt atoms, which have a smaller atomic radius than Au and Pt. Likewise, it can be stated that the second group of NPs are alloys of Pt and Co [45,46], due to  the large mismatch between gold and platinum, which was mentioned previously. Furthermore, Su et al. [12] have shown by density functional theory (DFT) calculations that Pt is more prone to disperse on cobalt surfaces than to form Pt on a cobalt core because the adsorption energy of Pt dispersed on cobalt is more negative, and therefore, the formation of more stable PtCo alloys is facilitated than the formation of platinum packed structures over cobalt. In addition, the alloy would have a compressive effect of 1.02% which could contribute to make these particles more active for catalytical purposes.
Characterization of AuPd-99.99 and AuPd-98 NPs The sizes of hollow PdAu NPs prepared at different addition times with the less pure palladium precursor were determined by DLS (Table 3) at least 3 days after the preparation day.
The DLS measurements of PdAu-98 shown that in order to obtain smaller NPs, the optimal time for the addition of the gold-palladium salts solution was 191 min (Table 3). Smaller NPs should exhibit better catalytic properties because they contain a higher number of atoms on their surface as has been proved for the oxidation of methanol and ethanol by PtAu/ TiO 2 NPs and for the ethylene oxidation by Pt NPs [22,47]. In addition, when the NPs are smaller properties such as the relative position of the center of the d-band (Fermi level shift), the d-band vacancy and the density of states are modified. So, Zhang et al. [48] reported that when the diameter of gold NPs decreases from 10.3 to 1.9 nm, the energy of the d-band center shifts towards higher values, and the d-band vacancy increases. Furthermore, Bai et al. [49] have shown that when the diameter of Pt NPs decreases from 2.4 to 0.7 nm, the occupation of the antibonding states is lower. These changes in the aforementioned properties are important because they lead to repulsive or attractive interactions between the NPs and potential analytes.
Shorter waiting times for the addition of the gold and palladium precursor solutions probably did not allow that all the borohydride ions added in the previous step produce the maximal number of cobalt cores. Otherwise, longer times could allow the agglomeration of the cobalt cores forming, bigger gold-palladium particles. In addition, the peak with average size of 4429 nm shown in the size distribution of PdAu-98 NPs (Fig. 5b) could be associated with some NPs agglomerated observed in the SEM micrographs.
The size distribution of one sample of PdAu-98 for 191 min (smaller NPs) is shown in Fig. 5b. 97% of PdAu NPs had an average size of 52 nm and the other 3%, 4429 nm. In addition, the zeta potential of these NPs was measured, and the average value obtained was − 44 mV.
PdAu NPs were also synthesized with Na 2 PdCl 4 ≥ 99.99% with the same addition time for comparing average sizes (Fig. 5a). The average size of these NPs was now 48 nm, and their zeta potential is − 43 mV.
The rather negative values of the zeta potential of the PdAu-98 and PdAu-99.99 indicated that these suspensions of NPs should be stable in media of moderate ionic strength.
SEM image of PdAu-99.99 NPs (Fig. 6) revealed that they have a spherical shape with a diameter range between 30 and 50 nm. EDX spectrum confirmed that the particles consist of palladium and gold, in the atomic ratio of 2.7:1 ( Table 4), and that they contain a residual amount of cobalt from the cobalt cores used as precursors.
The morphology of the PdAu-98 is similar to the particles described previously. The suspension contains spherical particles with diameters between 30 and 50 nm. The EDX spectrum confirmed that the particles are made up of palladium and gold (Table 5); however, the ratio between Pd and Au is slightly higher (3:1) than PdAu-99.99. The particles also contain a small amount of residual cobalt but in a higher percentage, 2% higher than in the first group of NPs.
TEM images showed that PdAu-99.99 (Fig. 7) and PdAu-98 (Fig. 8) NPs have two types of characteristic morphology: thin shells with a void inner space and particles with a rough surface. EDX line scans were acquired through some PdAu-99.99 NPs (Fig. 9) and PdAu-98 NPs (Fig. 10) samples, in order to check if the NPs have a hollow structure.
PdAu-99.99 NPs with a rough surface were analyzed by EDX line scan along the white arrows in Fig. 9a and b. EDX analysis confirmed that NPs are composed of gold, palladium and cobalt. It was observed that the NPs have a higher amount of gold and palladium at the extremes of the line in comparison with the central position. The profiles also show the presence of a very low amount of cobalt and also of carbon, the later coming from the TEM grid.
The EDX scan lines of PdAu-98 NPs ( Fig. 10a and b) disclosed that they contain more palladium than gold and a low amount of cobalt. Although the profile of both types of nanoparticles correspond to a typical hollow nanoparticles profile, they are slightly different in comparison to the profile obtained for PdAu-99.99 NPs. Figure 9 shows that the content of gold and palladium in the center of the particles is significantly different in comparison to the extremes of the lines; however, for PdAu-98 NPs, the content of both metals in    the center of the nanoparticles and at the extremes is only slightly different which would indicate that their cavity is smaller and their shell thickness is higher at the central position.
Gold and palladium have a face-centered cubic (fcc) crystal structure. The lattice constants of Au and Pd are 4.079 Å and 3.890 Å, respectively, and they are miscible in the whole range of    [50]. SAED diffraction patterns were obtained to confirm, that, as expected, PdAu NPs produced in this work are alloys and not segregated phases. Figure 11a and c show the STEM image of the PdAu NPs, and Fig. 11b and c show the corresponding SAED patterns. Two groups of PdAu particles with different surface roughness were analyzed. NPs in Fig. 11c have predominantly rough surfaces, whereas Fig. 11 a shows similar amounts of NPs with rough and smooth surfaces. The concentric rings from inside to outside were assigned in the SAED pattern of both groups of NPs to (111), (200), (220), (311), and (420) planes, which are related to a fcc structure.
The lattice parameters of both groups of NPs (3.954 Å and 3.946 Å) were calculated from the (200) d-spacings ( Table 6).
Four of the d-spacing values obtained from Fig. 11a and b (Table 6) are intermediate values between the d-values of the pure metals; however, for the planes (111), the interplanar distance is smaller than the reference value for palladium.
The calculated d-spacing values of the second group of NPs (Fig. 11c) are mostly intermediate values between the ones of gold and palladium. Only the d-spacing value of the (220) plane is equal to the d-value of palladium.
Comparing to the reference values of Pd and Au (Table 6), the lattice parameters obtained for the two groups of PdAu-99.99 NPs lie among the values of gold and palladium. These changes in the lattice parameter indicate that when a second metal is added to the palladium structure, its crystal structure is modified, and therefore, also its catalytic properties change as well as its stability, as it has been proved for the oxidation of methanol with PdCo, PdNi, and PdCu supported on reduced graphene oxide [51]. However, the smaller value of the interplanar distance between (111) planes of the first group of PdAu NPs could be the result of the incorporation of cobalt atoms (smaller atomic radio) in the (111) planes preferentially.
All the d-spacing values of the NPs of Fig. 12a are intermediate values between the d-values of the pure metals ( Table 7). The lattice parameter, calculated from the (200) d-spacing, was 3.964 Å. The second group of NPs (Fig. 12c)  For PdAu-98 NPs, it can be concluded that both groups of PdAu NPs are alloys because almost all of the d-spacings and the lattice parameters are intermediate values between the corresponding values of the pure metals.

Conclusions
Hollow alloys of platinum-gold and palladium-gold were synthesized by galvanic replacement reactions. The optimal addition time of Na 2 PdCl 4 for the production of the gold-palladium NPs was 191 min. PtAu and PdAu NPs with smooth and rough surfaces were produced. NPs with a rough surface would be the result of particle overgrowth on a longer time scale. The percentage of cobalt is higher in PtAu NPs. PdAu-98 NPs had more residual cobalt than PdAu-99.99. EDX line scan confirmed the hollow structure of platinum gold and palladium gold NPs and shows that PtAu NPs with a rough surface have a relatively higher content of gold. SAED analysis suggested the formation of some alloys of platinum and cobalt due to the relatively high content of cobalt of PtAu NPs. PdAu-99.99 and PdAu-98 NPs exhibit quite similar characteristics; although SAED pattern confirmed that PdAu-98 NPs are alloys, but it could not be verified for all the PdAu-99.99 NPs.
Funding Open Access funding enabled and organized by Projekt DEAL. This study was funded by the Schlumberger Foundation Faculty for the Future program in the form of a PhD fellowship.

Data availability
The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.

Conflict of interest
The authors declare no competing interests.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated 1 3 Vol.: (0123456789) otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.