Crystallographic texture can be rapidly determined by electrochemical surface analytics

Orientation affects application-de ﬁ ning properties of crystalline materials. Hence, information in this regard is highly-prized. We show that electrochemical jet processing (EJP), when coupled with accurate metrological appraisal, can characterise crystallographic texture. Implementation of this technique al- lows localised dissolution to be anisotropic and dependent on etch-rate selectivity, de ﬁ ned by the crystallography. EJP therefore, generates complex, but characteristic topographies. Through rapid surface processing and analysis, textural information can be elucidated. In this study, samples of polycrystalline Al and Ni have been subjected to EJP, and the resulting surfaces analysed to generate three-colour orientation contrast maps. Comparison of raw data acquired through our method with prior electron back-scatter diffraction data shows broad correlation and assignment (68% on a pixel-by-pixel basis), showcasing rapid large-area analysis at high ef ﬁ ciency.


Introduction
In polycrystalline materials, the distribution of grains and their absolute and relative crystallographic orientation has a profound impact on material performance, affecting factors such as fatigue [1], ultimate tensile strength [2] and creep [3]. Such information, regarding material texture and orientation is highly-prized as it serves to verify materials processing technologies. This understanding is especially important, particularly with the emergence of new manufacturing processes, seeking to further exploit orientation-specific material properties [4].
At present, the range of tools for determining the texture and orientation of material microstructure is limited. Orientation characterisation is commonly undertaken through diffractionbased techniques, such as electron back-scatter diffraction (EBSD) [5e7], transmission electron microscopy (TEM) techniques [8e10] and reflected electron techniques [11]. All of these rely upon electron diffraction and can characterise space group symmetries in an unambiguous manner, with lateral resolutions of <20 nm in some variants of EBSD [12], and smaller still with TEM techniques. However, their application is limited due the nature of analysis, and the use of electrons requiring high-vacuum equipment. In addition, sample preparation time and cost for EBSD and TEM specimens is often prohibitive and characterisation efficacy is dependent upon specimen quality. Large area analyses (mm scale) are impractical using TEM and encumbered by significant processing times in EBSD, despite increasing computational and processing efficiency [13]. Orientation characterisation of crystalline materials, under ambient conditions, is possible using X-rays to generate transmitted and back-reflected Laue patterns [14], although limited to single crystals. Novel methods for crystallographic characterisation, such as spatially resolved acoustic spectroscopy (SRAS), address these issues by eliminating vacuum equipment, reducing sample preparation time and are more suitable to large area appraisal than EBSD [15]. However, the resolution is presently limited by applied optical configurations and orientations for grains <50 mm have not been resolved, limiting the materials compatible with the technique.
Materials characterisation techniques involving optical measurements are more amenable to rapid manufacturing strategies than electron measurements [16]. This has led to notable contributions considering the adaption of optical measurement techniques to acquire valuable microstructural information such as grain boundary character [17] and partial orientation information [18]. Such techniques exploit the differential reflected intensity of light, acquired under different lighting conditions, or using a goniometer setup. Rapid orientation measurement has also been achieved for silicon-based materials through alkaline etching and optical imaging [19], where simulations were fit against micrographs to elucidate crystallographic information considering the highly selective etching characteristics of silicon in hydroxide solutions.
This study reports on the creation of a novel tool for determining partial grain orientation, when calibrated against a conventional approach. This method is based on electrochemical jet processing (EJP) [20] for the selective anisotropic electrochemical etching of differently oriented grains. When coupled with appropriate metrology and processing, the technique holds several advantages over conventional methods and other reported optical methods, which make application of the process attractive from an industrial perspective. The analysed surface is rapidly generated in the dissolution process, eliminating specimen preparation time. Modern 3D optical profilers are capable of characterising topographies from above in a single measurement, reducing analysis time and cost. No significant sample preparation is required before analysis and the process can be executed on rough and undulating materials. In addition, discrete areas can be selectively appraised, compared with chemical bath etching, allowing the method to be used to verify materials processing techniques. From a practical perspective, it is shown that orientation-dependent topographies can be generated using solutions of cheap non-toxic salts, without corrosive chemicals. Furthermore, the presented technique is entirely scalable should the practitioner seek to characterise larger objects; there is no fundamental limit to the topography generation process.
This investigation shows the surface topography exposed by EJP to be consistent with long-range crystallographic order, under appropriate conditions. For anodic dissolution techniques, such as EJP, material is removed atomistically [21], where the remnant material crystallography is not altered, due to the athermal nature of the dissolution process, making EJP an ideal candidate for characteristic topography generation. This contrasts with other etching methods, such as thermal etching, which may alter the grain size and stress condition of the workpiece, depending on the material. Removal anisotropy is related to passivation characteristics of the different planar families [22]. Orientation-dependent featuring has been observed in pitting corrosion studies, for both Al [23] and Ni [24], where pit formation and propagation is dependent on localised chemistry within pits, the potential [25,26] and the resulting dissolution flow [27,28].
In crystallographic pitting, information regarding the underlying microstructure is contained within the surface topography [29]. While material degradation mechanisms have been explored, the utilisation of this information for microstructural characterisation is yet to be undertaken. Pitting phenomena are strongly affected by the incident electrolyte, which has also been widely studied. Passivating electrolytes, such as NO 3 , most often used in EJP, are characterised by a tendency to generate oxide films during processing [30,31]. For micromachining, this allows enhanced dimensional accuracy [32], and brighter, pit-free surfaces [33], which is undesirable for the purpose of crystallography-driven topography generation. Conversely, aggressive electrolytes, such as Cl, lead to passivity breakdown, pitting, and reduced dimensional accuracy [34]. In this study, Cl-based electrolytes are shown in EJP to generate measurable orientation-specific topographies rapidly and across large areas in a controlled manner.

Specimen preparation
Commercially pure Al 1050 (Smith's Metals UK), as well as Ni 201 and 6082 Al alloy (Unicorn Metals UK) were used in this study (all !98% purity). Sample pieces were subsequently annealed (Ni 1050 C, Al 520 C) for 3 h prior to grinding and polishing (1 mm diamond). For samples appraised by EBSD, further polishing was undertaken using colloidal silica (0.3 and 0.06 mm sequentially), before washing successively in ethyl lactate, acetone, methanol and isopropanol.

Surface generation
Equimolar solutions of NaCl and NaNO 3 (2.3 M) were prepared in deionised water from salts (Sigma Aldrich UK, and Fisher UK, respectively), and used as the electrolytes in this study. A bespoke CNC 3-axis EJP platform was used in this study, outlined in previous work [20]. Pits were drilled into materials across a range of current densities (J noz ¼ 20e450 A/cm 2 ), where the total charge passed was fixed (0.56 C per pit), through adapting machine time for a given current density. In all experiments, inter-electrode gap was kept constant (500 ± 1 mm). Large-area analysis was undertaken through a planar machining approach at constant translation speed (500 mm/s) and J noz was maintained throughout machining (Al 6082, 50 A/cm 2 ; cp Ni 30 A/cm 2 ). Nozzle step-over was also maintained in each processing operation (Al 6082, 250 mm; cp Ni 500 mm). Samples were subsequently cleaned in deionised water in an ultrasonic bath, and characterised without further preparation.

Orientation mapping
A custom-built SRAS setup, was used to characterise the orientations of large areas of Al and Ni. In this case, a broadband Qswitched laser (1064 nm) and an optical mask, a grating pattern of known period, l g , were imaged onto each sample, matching the temporal frequency and spatial frequency of the laser and sample for narrow band excitation. The thermo-elastic effect caused by laser irradiance produces a packet of surface acoustic waves (SAWs) with a characteristic frequency, f s . Thus, the velocity of the SAW, which is a function of the mechanical properties of the sample is found by v s ¼ f s l g [35]. The surface perturbation due to the SAW packet was detected by a knife-edge detector, using a second continuous wave laser (532 nm). All scans in this work were conducted at 24 mm wavelength, l g , giving a patch size of z200 mm and a nominal resolution of z100 mm. EBSD (Oxford Instruments HKL Advanced EBSD system, with NordlysMax3) was used to characterise orientations of a large-area of Ni prior to EJP (field of view, FOV: 4.85 Â 3.63 mm, 4.5 mm/pixel).

Orientation-topographic analysis
Secondary electron and back-scatter electron microscopy (both Philips XL-30 SEM) were used to appraise surface textures after EJP. Focus-variation microscopy (Alicona G5 Infinitefocus, 50Â and 100Â objectives, FOV 324 Â 324 mm and 160 Â 160 mm, respectively) was used to qualitatively appraise topography including form and to evaluate root-mean-square (rms) surface roughness, Sq, per current density level. Sq was appraised after form removal (2nd order polynomial) and subjected to high-pass filter (at cut-off spatial wavelength of 80 mm). A coherence scanning interferometry (CSI) instrument (Zygo NewView™ 8300, 50Â objective, FOV 167 Â 167 mm) was used to obtain accurate topographical information, from which surface gradient information was calculated. The characteristics of the processed surface determined by the orientations of crystalline grains were analysed based on the magnitude and direction of the gradient of surface height. The CSI system has been carefully calibrated, of which the surface topography repeatability is of the order of 0.1 nm and the step height accuracy is below 0.3% of the measured height [36e38]. Given accurately measured topography, the gradient vector of the surface can be easily obtained from the directional derivatives of the height map as where h is the height function of a surface, g x and g y are the partial derivatives of the height function in x and y directions, respectively. The direction of the gradient, at a given point on the surface, calculated as is the orientation in which the gradient has the largest value (gradient magnitude) calculated as In this work, the calculation of the gradient is done by numerically convolving the Sobel kernel with the height map of the surface [39]. The gradient direction, q, ranges from À180 to 180 measured anticlockwise from the positive x-axis. Then the 2D map of the surface slope angle is calculated from the magnitude of the gradient as tan À1 ðjgradðhÞjÞ.

EJP generates large-area orientation characteristic topographies
To characterise EJP-generated orientation-dependent surfaces, samples of polycrystalline Al and Ni were appraised using SRAS, to output crystallographic orientation maps [35], shown in Fig. 1b and c. From these datasets, discrete grains were identified, such that a pit was selectively machined into a grain vicinal to each of the three principal cubic orientations relative to the normal direction (ND) (Fig. 1a), using Cl electrolyte at a constant current density (20 A/ cm 2 ). Location coordinates were imported into computer aided manufacturing software to generate a machine toolpath, with which the samples were subsequently processed. Secondary electron micrographs show textural variation associated with different orientations. Where grain orientation tends towards <001>, smooth surfaces are generated in Al in the central pit region (Fig. 1d) and a small number of cubic crystallographic pits are observed. Large gradients, approaching 90 , relative to the transverse direction (TD), (Fig. 1a), are seen in this orientation (Supplementary Fig. 1), resulting from retardation of dissolution in <001> directions, correlating with previous studies [40,41]. Here, resulting topography is defined by inhibited-etching directions. In Ni, surfaces are generated from repeating inverted squarepyramidal pits, when grains vicinal to [001] are machined ( Fig. 1 e and f). The pyramidal walls generally feature gradients of between 30 and 40 , relative to the ND, indicating that for Ni, preferential removal is likely to be arrested by <111> directions (theoretically, tan À1 1 ffiffi ffi 2 p , from ND, Supplementary Fig. 2).
For grains oriented towards <101> relative to the ND, repeating stepped features are generated in both Al and Ni ( Fig. 1g and h), however the topographies are characteristic for each material. In Al, the inter-step angle is approximately 90 in this orientation, again indicating material removal retardation in the <001> directions. Conversely for Ni, grains tending towards [101] feature relatively large plateaus, which are indicative of a tendency towards an equilibrium removal condition. These areas are truncated by stepped features (Fig. 1h), the angle between the two surfaces being >90 , generally between 125 and 140 ( Supplementary Fig. 3). Blunted diamond-like pits shown for Ni z[101] surfaces (Fig. 1i), previously reported in Ni [42], appearing to result from inhibited dissolution along <111> directions, theoretically tan À1 1 ffiffiffiffiffi ffi 0:5 p , from ND ( Supplementary Fig. 3). This indicates that dominant material removal follows <001>, and dissolution is inhibited along the <111>, and possibly <101> directions. Lower selectivity between <111> and <101> directions complicates surface topographies of Ni compared with Al, although the surfaces remain characteristic.
For grains tending towards <111> relative to the ND, repeating pyramidal-type textures are observed in Al (Fig. 1j), which is expected considering slowest etching directions aligning with <001>. For a periodic pyramidal surface, we expect three dominant slopes, with 120 inter-separation. For Ni, surfaces are composed of plateaued triangular features, where the angle of each triangular plate z60 ( Supplementary Fig. 4). This corresponds to removal arrested by <111> faces, which are equilateral when sectioned through this orientation. Hexagonal pit morphologies, reflective of the packing of this orientation, are observed ( Fig. 1k and l).
The surface textures generated in both materials here are intuitive, when considering their origins in the respective geometries of the principal orientations, and are therefore characteristic for a given parent-grain orientation. In Al, implicit in the exposure of cubic and pyramidal features is arrested material removal along <001> crystallographic directions [43]. For Ni, characteristic topographies can be explained considering removal arrested mainly by <111> directions. These relationships can be quantified in a repeatable manner and underlying orientation information can be output, relating the apparent topography to the principal cubic orientations.

Manipulating topography
Grain-dependent topography succinctly demonstrates anisotropic material removal and is dependent on the material, applied energy, and electrolyte used. For Al and Ni, it is shown that the result of using aggressive Cl electrolytes in EJP, is grain-dependent texturing at low applied current density (J noz <100 A/cm 2 ), where J noz is a tophat approximation of the current density distribution Fig. 2a). Grain-dependent texturing here results from crystallographic corrosion phenomena across the machined area. Textural relationship with J noz is demonstrated by differential surface texturing (roughness, Sq, over multiple grains), which was measured for both materials in aggressive Cl and passivating NO 3 electrolytes (Fig. 2b and c). For Al and Ni processed with Cl, a power-type regression between surface amplitude and current density is followed, across J noz ¼ 20e450 A/cm 2 . Amplitude depends on diffusion layer thickness in anodic dissolution, whereby electropolishing is achieved at high current densities, and asperities act as field-focussing elements, increasing localised current to enhance their dissolution rate [45,46]. At low current densities (<100 A/cm 2 ), where the surface resulting from EJP is dominated by microstructure effects, relatively large amplitudes are generated using Cl, indicative of the increasing scale of microstructuredefined topography. The scale of grain-dependent texturing is inversely proportional to the applied current density in EJP, between J noz ¼ 20e200 A/cm 2 , beyond which the amplitude does not decrease significantly. For Al and Ni, we can enhance lattice-defined texturing at current densities <100 A/cm 2 , hence operation within this domain is preferred. We did not observe equivalent amplitude relationships for these materials processed with NO 3 , indicative of reduced dependence on the specimen microstructure. Topographic reconstructions, acquired after we machined across current densities: J noz ¼ 20, 50, 100, 300 A/cm 2 , using Cl (Fig. 2der) correlate with the amplitude analysis, resulting in a gradual transition from microstructure-defined removal to electropolishing. Using NO 3 electrolyte under equivalent processing, we observed no lattice-defined topography in Al, and a limited extent in Ni, although less than with Cl ( Supplementary Fig. 5), likely to be a result of the more positive oxidation potential: Ni < Al (vs. standard hydrogen electrode) [47]. In Al, anodic oxide-generating NO 3 delocalises material removal from the microstructure, hence etching is isotropic. Thus aggressive electrolytes, stimulating anisotropic removal in EJP, are preferable when characterising these materials.
The SEM micrographs (Fig. 2ees) acquired at identical current densities (20, 50, 100, 300 A/cm 2 ), show single grain topographies. In this study, lattice-defined removal is demonstrated, resulting in cubic geometric structuring at current densities <100 A/cm 2 (J noz ), for both materials, the scale of which decreases as J is increased. For both materials studied here, processing at high current density (J noz !300 A/cm 2 ) results in negligible periodic topography at this length scale. Transpassive dissolution of materials has previously been shown to have limited dependence on microstructural factors, such as orientation [48]. It is assistive to the technique that latticedefined topography is largely insensitive to applied current density over a broad range (20e100 A/cm 2 ), even where transpassive dissolution modes are predicted, (Ni, J 20 A/cm 2 ) [49].
When machining, the dissolution profile is dictated by the current density distribution. Classically, this appears Gaussian for a cylindrical nozzle, (Fig. 2a). Localised variation in microscale topography is observed as a result of the feature form, such as for Al vicinal to [001], where inhibited-etching directions, defined by the material-process response, follow <001>. A pit section is shown in Fig. 3a, where the central region appears reflective, indicative of lower surface amplitude, compared with the walls (Fig. 3b). At low Fig. 2. The operational domain is defined by current density. Current density, J, exhibits a Gaussian-type distribution, where J noz is a tophat approximation considering current, I, and nozzle area, A (a). RMS roughness, Sq, shown as approximate power regression with J noz , for Al machined with Cl, but not upon machining with NO 3 (b), also observed for Ni (c).
Reconstructions of EJP processed areas acquired using focus-variation (FV) microscope (50Â objective), and subjected to 2nd order polynomial form filter [50]. SE micrographs of single-grain areas of the surface, machined with Cl. At low current density (J noz ¼ 20 A/cm 2 ), grain-dependent topography is generated in Al (d), and repeating microscale structuring is observed (e). For Ni, the same effect is shown (f), and repeating topographies are evident (g). Upon increasing current density (J noz ¼ 50 A/cm 2 ), grain-dependent topographic contrast is still evident in Al, although finer in scale (h), with ordered surface structuring (i). The same effect is observed for Ni (j), although the scale of the periodic structuring is reduced (k). At higher current density (J noz ¼ 100 A/cm 2 ) topography amplitude is reduced, but still apparent in both materials, such as for Al (l), and finer-scale repeatable surfaces are shown in SEM, although appearing to delocalise from high-level ordering associated with lower current densities (m), this effect is observed in Ni (n), and (o). At higher current densities (300 A/cm 2 ), no grain-dependent topography is observed in Al (p), and smooth surface textures are observed (q), the same is shown for Ni, although intergranular attack is evident (r), and some ordering is apparent (s). Electropolishing mechanisms dominate !100 A/cm 2 . Error bars are the standard deviation of surfaces appraised, from three discrete pits for each data point. Black scale bars are 50 mm, white scale bars are 10 mm.
J, smooth surfaces are not generally expected in EJP [46]. Despite differential amplitudes, similarities in the textural morphologies at the pit base and the side wall are observed Fig. 3b c and d).
An interpretation of the surface-generation mechanism for Al is shown in the idealised schematic in Fig. 3e. As dissolution achieves steady-state and a side wall of the pit becomes exposed, additional <001> inhibited-etch directions are exposed. Therefore, the topography generated is inconsistent with that subject to attack from the normal direction. While the topography created throughout the pit is not uniform, this does not present significant challenge to the succeeding analysis method, which accommodates this variation. This is evidenced by the complementary gradient magnitude plots in Fig. 3f and g, wherein a large lowangle peak <10 is observed regardless of the feature form, correlating with the flat exposed <001> facets. Dominant slope angles are similar at the pit base and the side walls, as these are defined by equivalent inhibited-etch directions, however surface dispersion of cubic pits is lower at the base. Material removal not occurring in the normal direction in this case, leads to periodic surface stepping, scattering more light, therefore appearing dull to the observer. Since characteristic slope angles are the same at the base and side of the pit, they remain representative of the underlying grain orientation. The dependency of the microtopography on the feature form is therefore reduced, under anisotropic etching conditions. When creating striations through jet translation, the same effect is observed, shown for Ni, where the orientation was evaluated by SRAS prior to EJP (see Fig. 4a). Translation EJP reveals grain-dependent topography, shown in the optical micrograph in Fig. 4b, where surfaces of grains vicinal to [001] appear dull to the observer, due to the theoretical inverted pyramidal microscale topography (see Fig. 1e), and surfaces of grains vicinal to [111] appear bright, due to the expected plateaued areas. Periodic form over the surface correlates with nozzle diameter and is controlled by the Gaussian energy distribution (see Fig. 4c). Topographies from a single-grain area, corresponding to an  Fig. 4def). Pitting frequency is inversely related to the applied current density, however it should be noted that the succeeding analysis routine does not consider discrete pits, rather the larger area microscale surface structuring. Similar topographies are observed despite the total surface amplitude being greater in the low J region. Consistency of micro-topography generated, largely independent of form, allows significant areas to be rapidly processed and characterised (with no fundamental limit).

Surface analytics
The topography of jet processed surfaces varies within singlegrains, due to the lattice orientation relative to the dissolution front. However, surface characteristics unique to given orientations in both Al and Ni, are revealed by the gradients of the resulting topography. Using CSI, the gradient vector of the surface can be obtained from the directional derivatives of the topographic dataset (see methodology), showing a statistical distribution of the characteristic slopes for a given surface. Subsequently, the twodimensional map of the slope angle is calculated from the magnitude of the resulting gradient vector, and the gradient direction at   Fig. 1b) and for z[111] oriented grains, the second and third peaks are clearly observed, still separated by 120 (h). each point on the surface that is pointing in the direction of the steepest slope at that point. Here, the Sobel kernel [39] is used to conduct the numerical calculation, and the gradient direction ranges from À180 to 180 , where the angle measured anticlockwise from the positive x-axis (defined as 0 ) is positive and the angle measured clockwise is negative.
From this distribution, the gradient directions can be quantified, yielding a distribution of peaks unique for an orientation in a given material. For Al grains oriented vicinal [001], micro-topographies comprised of cubic recesses are generated under anisotropic electrochemical etching conditions, (repeating flat areas). This means small slope angles (<10 ) are prominent (Fig. 5c). In this case, the gradient direction varies and trend observation is difficult. For orientations vicinal to [101] and [111], further analysis is necessary for differentiation, through analysis of slope direction, as the histograms for both of these surfaces show similar prominent slope angles which are >45 . For a stepped topography representative of that observed for etched grains orientated vicinal [101], the theoretical difference of peaks is 180 (Fig. 5e). When grains vicinal [111] are subjected to EJP, periodic pyramidal topographies are observed. The theoretical angle in this case will be 120 (Fig. 1j). Where the peak separation is >160 -<190 , the sample surface assigned to [101], and where the peak separation is >100 -<130 , the sample surface is assigned to [111]. For surfaces where neither of these criteria are satisfied, such as a strongly hybridised texture, or an area of unprocessed surface, the sampling area is set as a nonmeasured point (see Fig. 6).
For Ni, EJP-generated surfaces are more complex, resulting from lower anisotropic etching, hence contrasting is more challenging. The inverted square-pyramidal pitting seen resulting from EJP of orientations vicinal [001], means that the dominant slope angle will be 45 . In addition, the slope directions of the pyramidal features are separated by 90 . For grains oriented vicinal to [101], the histogram shows a dominant peak between 10 and 20 , along with a smaller peak between 30 and 40 . In addition, the separation of the slope directions is >120 . For grains oriented vicinal to the [111], the histogram shows a dominant slope angle <10 , corresponding to plateaued areas. Characterisation of the slope directions in this case shows three directions with a separation of 120 . Both [101] and [111] surfaces show six major slope directions for high-slope surfaces (60e90 ).
Through applying this method, a grain of Al or Ni can be readily assigned to the closest principal orientation. Automating data acquisition and analysing topographical information allows planar response profiles to be plotted to generate three-colour orientation texture maps. Equivalent orientations generate identical characteristic slopes from identical EJP treatment and the aforementioned topographical analysis. However, since this characterisation intervention is far less than conventional methods, the technique is deployable upon high-value assets beyond the laboratory. Nonetheless, it is intuitive that reciprocal orientations lead to machined surfaces with characteristic alignment relative to the normal and transverse directions. Analysis of absolute gradient directions may be applied to increase the measurement fidelity.

Crystallographic texture mapping, validation and error
Grain orientation texture maps were generated for Al and Ni specimens and are shown in Fig. 7. Comparing with prior SRAS orientation data (Supplementary Fig. 6a) in Al, surface processing reveals grain-dependent topography evidenced by the optical micrograph in Fig. 7a. This surface was measured using CSI (<20 s per 167 Â 167 mm acquisition) and the aforementioned characterisation routine, to output an orientation contrast map (Fig. 7b). To generate this map, each acquisition area was split into 16 discrete analysis regions, corresponding to a pixel size of 42 mm. SEM analysis of grain topographies show textures expected from the respective assignments in Al (Fig. 7cef), demonstrating large-area method repeatability. In addition, clear boundaries are observed, where intergranular corrosion, where topography is not representative of the underlying orientation, is <2 mm wide ( Fig. 7g and h), less than the sampling distance of this technique. For Ni, prior EBSD data (Fig. 7i), was down sampled to three-colours (Euler angles for each interrogated grain are shown in Supplementary Table 1), against which to benchmark the novel approach (Fig. 7j). The resulting output orientation map using this method is shown in Fig. 7k (42 mm per pixel). Validation was undertaken by down sampling EBSD data to the same pixel size as the output map. The two datasets were subsequently overlaid in Matlab, wherein they were subtracted; any value that did not correspond to zero was allocated mis-assigned. Conversely, zero values relate to direct correlation. The map is in broad agreement with prior EBSD data on a pixel-by-pixel basis (68% pixels correlate to EBSD) ( Supplementary Fig. 7c) and crystallographic contrast is observed in the output map. Correlated pixel assignment was greatest for [001] textured grains (75%), followed by [101] (65%) and [111] (62%).
In order to realise the origins of uncorrelated pixels in the validation study, the Ni surface was inspected with SE microscopy (see Fig. 7leo). Different causes have been identified leading to correlation discrepancy: 1) Material removal, where the revealed microstructure is not representative of the near-surface microstructure, measured by EBSD ( 50 mm of Ni was removed, Supplementary Fig. 8b), leading to shifting boundaries during etching, also the complete corrosion of grains. 2) Hybrid topographies of unpolarised grains not vicinal to any principal orientation. Bisecting grains are assigned as non-measured points in the output maps, although in some instances surfaces appear representative of the more dominant removal directions (such as Ni <001>). This could explain the bias of mis-assigned pixels from [101] grains to [001], ahead of slow-dissolving [111] (Supplementary Fig. 7d). 3) Grain-boundary effects, such as where stepped edges are generated, exhibiting [101]-characteristic topographies.
The difficulty in validating the novel method with conventional materials characterisation techniques such as EBSD is further evidenced in Fig. 8 (z10.5 mm/pixel). The prior EBSD data was down sampled to the same pixel size accordingly (see Fig. 8a). Upon EJP, the shifting of the grain boundary zone and the removal of a grain is clearly shown in the subtracted overlay (see Fig. 8c). This contributes to the uncertainty, in correlation with prior EBSD data. In addition, it should be noted that a small proportion of the subtractive mis-assignment arises from non-measured points in the prior EBSD data (in the sample area, approximately 10% of the pixels in the prior EBSD data have returned as non-measured).

Conclusions
In this study, a scalable procedure has been defined allowing rapid assignment of crystalline texture under ambient conditions with minimal intervention, making the technique particularly amenable to large-area industrial application. Automation of data acquisition has been demonstrated, resulting in the ability to differentiate grains according to the three cubic principle crystallographic directions.
In addition, a route through which orientation-dependent topographies can be rapidly and selectively generated using nontoxic solutions is presented, without altering the crystallography of the material. In this regard, it is suggested that the EJP surface preparation technique could be assistive to other developing optical characterisation methods such as directional reflectance   zones between grain 1 and 3 (g), and grain 3 and 4 (h), in both cases, the intergranular crevice is <2 mm wide. For Ni, appraised by prior EBSD orientation measurement (RGB dataset), CSI acquisition area black inset (50x objective), with labelled grains (i). EBSD data downsampled to three-colours, through bin assignment to the closest principal orientation for comparison, grains labelled as per inverse pole figure (j). Output planar texture map (42 mm/pixel, colour convention as Al), showing orientation contrast and correlation (68%) with prior EBSD data on a pixel-by-pixel basis (k). Three causes of non-correlated pixels are identified here: surface destruction, resulting in apparent grain boundary shifting (l) and complete corrosion of prior measured grains e the annealing twin within grain 8 is removed, and not assigned in output map (m); bisection of parentgrain orientation -texture shown likely to be grain 11 (after complete corrosion of grain 12), which has an orientation bisecting [001] and [101]. The method has assigned to [001], largely as a consequence of the inverted pyramidal topography (n). Mis-assignment caused by boundary effects: similar z[111] grain boundaries mis-assigned to [101]-type texturing, due to the large intergranular step (o). (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.) microscopy [18]. Although a large degree of uncertainty is acknowledged and remains the subject of ongoing research within the group, it is hoped that the materials community will assist with the development of this technique into a more comprehensive method.
Enhancements in output data quality and correlation with EBSD can be achieved using simple filtering operations, although care was taken to present only raw datasets in this study. Further enhancement of the polar resolution and crystal class identification is possible by considering absolute gradient directions, as well as hybridisation characteristics of unpolarised (to the principal cubic directions) grain topographies.