Image analysis data for the study of the reactivity of the phases in Nd-Fe-B magnets etched with HCl-saturated Cyphos IL 101

Three phases can be distinguished in Nd‒Fe‒B permanent magnets: a Nd2Fe14B matrix grain phase, a Nd-rich grain boundary phase and a Nd-oxide phases. Common reaction models for leaching, such as the shrinking-particle model, cannot simply be applied to composite Nd‒Fe‒B permanent magnets because of the different chemical reactivities of the crystalline phases mentioned above. Etching the surface of a Nd‒Fe‒B magnet to expose its microstructure to electron microscopy is a necessary practice to correlate the microstructure itself to the specific properties of the magnets. Aqueous solutions of mineral acids are often used for etching purposes. However, these solutions are too low viscous to easily control the etching front and they show little selectivity in the etching process. In our work, the ionic liquid Cyphos IL 101 was used to etch bulk magnets instead of aqueous HCl solutions. The bulk Nd‒Fe‒B magnets were first polished, then exposed to a solution of 3 M HCl in Cyphos IL 101 for different times and at different temperatures. Afterwards, the etched Nd‒Fe‒B magnets were washed with ethanol and acetone. The results were examined via scanning-electron microscopy and image analysis. A commercial software, ImageJ®, was employed for image analysis. The latter technique was used to correlate the etched area (%area) or the grain and oxide size to the etching temperature or the etching time. The grain or the oxide size were calculated as Feret diameter. Image analysis revealed to be a necessary tool to support and correct the findings first suggested by the simple scanning-electron microscopy. The data presented in this article might be reused to corroborate a new reactivity order of the three Nd‒Fe‒B phases, different from that traditionally reported in literature, which is – from the most to the least reactive – grain boundary > oxides > the Nd2Fe14B grain phase.


a b s t r a c t
Three phases can be distinguished in Nd-Fe-B permanent magnets: a Nd 2 Fe 14 B matrix grain phase, a Nd-rich grain boundary phase and a Nd-oxide phases. Common reaction models for leaching, such as the shrinking-particle model, cannot simply be applied to composite Nd-Fe-B permanent magnets because of the different chemical reactivities of the crystalline phases mentioned above. Etching the surface of a Nd-Fe-B magnet to expose its microstructure to electron microscopy is a necessary practice to correlate the microstructure itself to the specific properties of the magnets. Aqueous solutions of mineral acids are often used for etching purposes. However, these solutions are too low viscous to easily control the etching front and they show little selectivity in the etching process. In our work, the ionic liquid Cyphos IL 101 was used to etch bulk magnets instead of aqueous HCl solutions. The bulk Nd-Fe-B magnets were first polished, then exposed to a solution of 3 M HCl in Cyphos IL 101 for different times and at different temperatures. Afterwards, the etched Nd-Fe-B magnets were washed with ethanol and acetone. The results were examined via scanning-electron microscopy and image analysis. A commercial software, Im-ageJ®, was employed for image analysis. The latter technique was used to correlate the etched area (%area) or the grain and oxide size to the etching temperature or the etching time. The grain or the oxide size were calculated as Feret diameter. Image analysis revealed to be a necessary tool to support and correct the findings first suggested by the simple scanning-electron microscopy. The data presented in this article might be reused to corroborate a new reactivity order of the three Nd-Fe-B phases, different from that traditionally reported in literature, which is -from the most to the least reactive -grain boundary > oxides > the Nd 2

Value of the Data
These data give a new insight in the reactivity of the three phases of Nd-Fe-B magnets, the stoichiometric grain phase, the oxide phase and the grain boundary phase. Materials scientists and engineers involved in the production of Nd-Fe-B magnets can benefit of these data for the design of new magnets and/or for the study of Nd-Fe-B magnets corrosion. These data can be integrated with similar ones from the corrosion of Nd-Fe-B magnets in other media, such as nitric acid, aqueous saline solutions, to give a general scheme for the reactivity of the three phases of Nd-Fe-B magnets. A technique cheaper than those commonly used to reveal the Nd-Fe-B magnet surface has been proposed in this paper.

Data Description
A simple scheme of the microstructure of Nd-Fe-B magnets is shown in Fig. 1: the grain boundary η (red), the Nd-rich oxides n (white) and the stoichiometric Nd 2 Fe 14 B phase ϕ (grey).
The elemental composition of the treated Nd-Fe-B magnets has been previously obtained [1] . Table 1 reports the composition per major elements of the phases ϕ (Fe, Nd and O -the latter being a contamination) and n (Nd, Dy and O), obtained via EDS; Fig. 2 is a back-scattered electron image of an unetched Nd-Fe-B magnet.

• Chemicals
The ionic liquid trihexyl(tetradecyl)phosphonium chloride (Cyphos® IL 101, [C101][Cl]) was purchased from Cytec (Brussels, Belgium). Concentrated HCl (37 wt.%) was purchased from VWR (Haasrode, Belgium), while D 2 O (99.99%) from Sigma-Aldrich (Diegem, Belgium). Technical grade ethanol, propanol and acetone were purchased from Merck (Darmstadt, Germany). Nd-Fe-B magnets were kindly provided by Magneti Ljubljana. They were never magnetized because re-   [2] , with the exception that isostatic pressing is not applied and that the sintering is carried out together with a heating treatment.  • Etching tests Cyphos IL 101 saturated with HCl served as the etching agent. The solution was prepared by contacting 50 mL of HCl (37%) with 5 g of Cyphos IL 101. This mixture was biphasic and part of the HCl was extracted to the ionic liquid (IL) with a final concentration of 3.0 M acid in the IL phase. This phase was separated and used as etching agent. The effect of the acid concentration was studied by also testing Cyphos IL 101 containing 1.2 M HCl. In the latter case, the etching agent was prepared by volumetrically diluting 10 times 37 wt. % HCl in the IL. In this way, a homogeneous mixture is obtained, which is applied as etching agent. Besides the effects of the concentration of the acid, the effect of the time and of the temperature on the etching mechanism were also investigated.
Before etching, the magnets were polished using standard metallographic methods to expose the microstructure to the etching agent and to the microscope. The surface was first scratched with a Struers Rotopol-15 machine using Struers waterproof SiC polishing papers with grit (in order of use): P#220, P#600, P#1000, P#2400 and P#4000. A step of finer polishing was then performed, using a Struers Labopol-5 and a diamond DP-paste P of grain size (in order of use) 3, 1 and ¼ μm. The conditions applied with both devices were: water-off polishing, ω = 150 rpm and propanol as lubricant. Whole magnets were polished without any further treatment, since they are big enough to be easily handled. To minimize the oxidation of the surface by air in between the different steps, the samples were kept under vacuum.
For the etching tests, the polished surface of a magnet was put into contact with a droplet of the etching agent, Cyphos IL 101 saturated with HCl, in a Petri dish for a fixed time (from 1.5 min to 4 h, at room temperature) and then washed with acetone. For the temperature tests, the Petri dish was heated to 40 or 60 °C for 3 min, controlled by a thermometer. The samples were analyzed via electron microscopy and image analysis. Scanning electron microscope (SEM) pictures and energy-dispersed spectra (EDS) were collected with a JEOL JSM 5800 microscope, operating at 20 kV. The polished samples were made conductive by spraying a carbon layer on them using a Balzer SCD 050 sputter coater. The EDS analysis were collected as average on at least 3 points per each SEM picture.
• Image analysis Image analysis supported the mere observation of SEM pictures and the EDS analysis to understand the reactivity of the three phases η, n and ϕ of the Nd-Fe-B magnets. Interestingly, the literature never considered image analysis to validate or discuss the data about corrosion of Nd-Fe-B magnets obtained from electron microscopy. A commercial software, ImageJ®, was used for the image analysis [3] . Two data were analysed: the Feret diameter and the percent- age of etched area, %area. The Feret diameter is defined as the distance between the two parallel planes that delimit the object and are perpendicular to a specific direction, with respect to which the measure is given. It is determined using the projections of a three-dimensional object on a two-dimensional plane. A graphical explanation of the Feret diameter is reported in Fig. 10 . The SEM images were converted to 8-bit grayscale, from 0 to 255 number of grey ranges. Simple linear scaling was applied. A threshold was set to define the area to analyse: when the Feret diameter was measured, the threshold referred to the particles, whereas when the %area was measured the threshold referred to the void spaces. In the latter case, dimples that might have been originated by the polishing rather than by the etching were unavoidably considered as well. The error was accounted to be mostly ≤5%. However, for measurements uniformity, the dimples area was excluded using the "Set Measurements" options. Three SEM pictures for each set of temperature or time conditions were examined to minimize errors due to the quality of the pictures or to the sensitivity of the operator. Moreover, two operators carried out the image analysis and their results were averaged.
The scale set for the measurement was based on the scale of the SEM analysis, paying attention on the fact that the images were in 1:1 aspect ratio. Area fraction images in threshold mode is the fraction, in percentage, of pixels highlighted by the threshold command. The fitting of the curves Feret diameter or %area vs time or temperature was done using the Origin® software. The curves were split in two fragments, based also on the SEM images, to better fit them with a super-linear, power law or a linear law.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships which have, or could be perceived to have, influenced the work reported in this article.