Obtaining a hematite pigment by thermal transformation of the surface oxide of reinforcing steel bars

En esta investigación,se reporta la valorización de la cascarilla de óxido superficial de varillas de acero al carbono mediante su transformación térmica en un pigmento compuesto principalmente por hematita (ɑ-Fe2O3). Se utilizó la Fluorescencia de Rayos X (XRF) yla Difracción deRayos X (XRD) para determinar el contenido elemental del residuo procesado e identificar los óxidos de hierro involucrados en la calcinación, respectivamente. El residuo siderúrgico en polvo se compone mayoritariamente por Fe2O3(87.92 %), SiO2 (6.13%),CaO (1.88 %), Al2O3(1.30 %) y MnO (0.77 %). El contenido total de hierro tiene principalmente el siguiente contenido en óxidos de hierro: magnetita, maghemita, wustita, lepidocrocita, hematita y goetita. El tratamiento térmico del residuo a temperaturas de 750-850 °C y tiempos de sostenimiento de 0.5-1.50 h, evidenció una alta conversión de los óxidos de hierro precursores en hematita, con porcentajes de esta fase que oscilan entre 86.4 y 94.6 %. La mayor obtención de hematita se logró a una condiciónde 850 °C y 1.00 h.


Introduction
The great majority of iron-based alloys, despite of being widely used in engineering, have as their main disadvantage their low resistance to corrosion, so they must be protected by applying coatings [1]. Steel corrosion occurs mainly by two processes: i) high temperature corrosion when the metal reacts directly with the aggressive medium (O2, etc.) and ii) electrochemical corrosion when there is a contact between the metal and electrolytes (water, saline solutions, etc.) [2]. The corrosion of structural steels is a very worrying case since this gradually induces the formation of cracks in concrete, because of the presence of corrosion products (oxides, hydroxides and oxyhydroxides) is greatly favored by the long atmospheric exposure time of the reinforcing bar during the transport and storage before installation [3][4][5].
During hot rolling, the steel undergoes an oxidation process which produces an oxide scale consisted of a three-layered struture: thin outer hematite (ɑ-Fe2O3) layer, intermediate magnetite (Fe3O4) layer and thick inner layer of wustite (FeO) [6]. The composition of remnaining mill scale after steel descaling also maintains that structure [7].
In addition, the atmospheric corrosion of steel, which is a complicated electrochemical process influenced by variables such as relative humidity, temperature, chloride content, rain, etc., draws much attention due to the costs associated with the deterioration of various structures exposed to the open air [8]. The rust layer formed in carbon steels exposed to different atmospheric conditions, is mainly composed of lepidocrocite (γ-FeOOH) and goethite (ɑ-FeOOH); there is often also maghemite (γ-Fe2O3) and in marine atmospheres, it is common to find magnetite (Fe3O4) and akaganeite (β-FeOOH) [9].
Iron and steel manufacture generate wastes of various types and properties such as rolling mill scale, mainly composed of iron oxides [10]. This mill scale has been used to obtain hematite pigments through processes that involve a heat treatment step [11][12]. Mostly methods of preparing iron (III) oxides includes a thermal transformation final stage of the iron-containing material and in some cases, thermal processes can be used as direct routes of synthesis of iron oxide particles like hematite [13]. Other steelmaking wastes similar to this scale have been used as hematite precursors, which are also used by direct heat treatment under oxidizing conditions [14]. Balbuena et al. [15], test a residue from ship sandblasting operations to obtain a hematite photocatalytic material, used as an additive in concrete. Della et al. [16] calcined steel scrap powder to obtain hematite ceramic pigments.
The existing structural relationships between the main oxides and oxyhydroxides of iron generated as products of the carbon steel corrosion may be capable of producing hematite pigments by calcination, since this is one of the most thermodynamically stable phases and the final product of many oxidation transformation processes [17], as is summarized in Figure 1. Hematite, which can vary in color from light red to dark violet depending on its particle size, is widely used as a pigment in paints, varnishes, cosmetics , rubber, paper, glass, ceramics and tiles [18][19]. In this research, we sought to use a steelmaking residue from corrugated steel rods to obtain a hematite pigment by direct heat treatment.

Characterization of the hematite pigment precursor residue
The surface oxide waste from reinforcing steel bars was put through a pretreatment described below: i) 200 mesh sieving (74 μm sieve opening), ii) drying in a convection oven at 60 °C for 24 hours and iii) pulverized in a bar mill with a content of 40 %, a speed of 32 rpm for a period of 7 h/kg [14][15][16]. Finally, the processed residue (PR) was characterized by using several analytical techniques. The chemical composition was obtained by X-Ray Fluorescence (XRF) with a Philips ® brand MagixPro PW-2440 brand Wavelength Dispersion spectrometer Fluorescence (WDXRF) [10]. The semi-quantification of the crystalline phases was determined by the softwareassisted analysis of the diffractogram obtained through X-ray Diffraction (XRD) with a Panalytical ® brand Empyrean 2012 model diffractometer operated with Co-Kα radiation [15,21].
In order to obtain micrographs and the elemental composition, Scanning Electron Microscopy/X-Ray Dispersive Energy Spectroscopy (SEM/EDS) was used by a Hitachi ® model SU3500 ® scanning electron microscope with an attached Oxford Instruments ® model detector X-MaxN [12,14]. The particle size distribution was determined by laser diffraction (LD) with a dry Cilas ® model 990 granulometer [15,22]. The CIELAB color coordinates of the RP dispersed in coconut oil between two glass lamellae were determined by colorimetry with a DataColor ® brand Check Plus model portable spectrophotometer, cofigured in a diffuse geometry with integrated/8° sphere (brightness excluded), angle from the 10° observer and light source D65 [16].

Obtaining of a pigment with high hematite content
As a preliminary heat treatment, a Rotable Central Composite Experimental Design (CCD) was used to calcinate 5 g of RP in a porcelain crucible using a Nabertherm brand LE 14/11 muffle model furnace. Two independent variables (factors) were considered in this study, the "Calcination temperature" (X1) and the "Holding time" (X2) with their corresponding levels summarized in Table 1, which affected the response variable, the "Hematite Percentage" (Y) [14][15][16][17].
The rotable CCD consists of: i) a 2 2 type factorial design, it is means 4 factorial runs; ii) 4 axial or star "runs" and iii) 2 runs in the center; also, each experimental combination between X1 (750-850 °C) and X2 (0.50-1.50 h) was replicated, for a total of 20 experimental runs [10][11][12]. The samples (S) of each pigment (P) obtained in the experimental design were characterized by XRD through the same analysis from a previous study [21] and by colorimetry in a similar way as was done with the PR. To produce the pigment (ABP325), 2.50 kg of the residue were put in three 25 by 40 cm AISI 304 stainless steel trays on an "Industrias Protón LTDA ® " brand HEMC5 model muffle furnace, under a calcination temperature of 850 °C and a holding time of 1.00 h, following the heat treatment scheme ( Figure 2). The pigment turned out to be modified by XRF, XRD, SEM/EDS, LD and colorimetry, as was done with PR. Furthermore, density and oil absorption were determined for the obtained pigment, taking into account the UNE-EN ISO 787-10: 1996 and ASTM D281-95 standards, respectively [23][24].
The PR particle size distribution obtained by laser diffraction was 0.3-75 µm, being the average particle size 10.10 µm. The SEM micrographs presented in Figure 4 confirmed the RP particle size and show an irregular morphology similar to that of the residue used by Sugrañez et al. [22]. Moreover, the results of the elemental analysis by EDS show that the percentages by weight of iron and oxygen were equivalent to 59.73 and 36.75 %, respectively. Likewise, the presence of silicon (2.47 %) and aluminum (1.04 %) were found. These results agree with the analyzes by chemical analysis by XRF and phase analysis by XRD.
The average color parameters values in the CIELAB coordinates for the RP dispersed in coconut oil were as follows: L*=15.00, a*=2.09 and b*=5. 18. These results can be compared with the CIELAB coordinates (L*= 46.02, a*=-0. 19 and b*=0.07) obtained for the steel scrap powder used by Della et al. [16]. The a* and b* coordinates of the PR are quite similar, but slightly higher, which may be mainly due to the contribution of the maghemite and lepidocrocyte presence. On the other hand, the PR is usually a bit darker than this steel scrap powder which has more of a metallic luster. It can be seen that the RP dispersed in coconut oil has a dark brown color ( Figure 5).   [17]. However, the values in the range of L* for these synthetic pigments are greater than the values obtained for all the samples of the experimental design, due to differences in the color measurement geometries used. The a* and b* values were affected mainly by the calcination temperature. For example, for a calcination temperature of 850 °C with a holding time of 1.00 h, a* and b* values are lower than those obtained at 800 ° C and 750 °C with the same holding time. For this reason, pigments obtained at higher temperatures are often more bluish (Table 2, Figure 5).

Characterization of synthetized hematite pigments in the experimental design
The behavior described above coincides with the obtained hematite pigments trend from steel scrap powder calcination proved by Della et al. [16]. The pigments are also more bluish as the holding time increases (0.50, 1.00 and 1.50 h) for the same calcination temperature (800 °C). Furthermore, this trend of obtaining hematite pigments with a distinctive blue hue is the product of the influence of a relatively thick particle size, which is in accord with Schaufler et al. [25].
From the semi-quantitative analysis of the diffractogram obtained by XRD for ABP325, a composition of 91.4% hematite, 3.1% maghemite and 5.5% quartz were determined, which demonstrate a high conversion of the precursor phases that constituted around 88% of the PR (Figure 3). On the other hand, the pigment has a morphology characterized by an irregular shape and size particle ( Figure 6). This is complemented by the particle size distribution results obtained through LD, with values mainly between 0.3 and 45 μm (average particle size of 14.57 μm), which is in accordance with the specifications of ASTM D185-07 (2019) to be used as a pigment [26].
Elemental analysis by EDS exhibits iron and oxygen concentrations equivalent to 55.69 % and 40.71 %, respectively. The presence of silicon (2.70%) and aluminum (0.91%) is also observed. These results confirms a high amount of iron and oxygen estimated at 96.40 % of the total sample analyzed, being the remaining 3.60 % attributed to silicon and aluminum impurities. These results are quite similar to the elemental chemical analysis by EDS of the PR, since there was only a transformation of the different crystalline phases of precursor iron oxides into hematite and maghemite.
The density values for two replicas of the obtained pigment PAB325 (3.84 and 3.85 g/ml) are close to the lower value of the specific gravity range (4.0-5.0), reported for synthetic red iron oxide pigments with a purity of ~85 %Fe2O3 [27] . The density of the pigment obtained was influenced by irregular shape, relatively high particle size and low expected surface area, as has been indicated by Oyarzún [28]. The oil absorption values for two replicas of obtained pigment (ABP325) were 15.36 and 16.34 g/100 g of pigment, respectively. These values are consistent with the oil absorption range for hematite pigments (15)(16)(17)(18)(19)(20)(21)(22)(23)(24)(25)(26)(27)(28)(29)(30) and are highly dependent on the large particle size (14.57 μm) of ABP325 and even on the nature of the obtained product [29][30]. Oil absorption values generally depend on the pigment production method [11], since they get bigger with increasing density [31] or decreasing particle size [32]. The high purity pigment (95.3 %Fe2O3) synthesized from mill scale by Quddus et al. [12], had an oil absorption value of 29.99 g/100 g of pigment compared to a value of 28 g/100 g of pigment for a recognized commercial pigment (Bayferrox® 110) and 19.77 g/100 g of pigment for another cheaper commercial pigment.
The CIELAB coordinates of ABP325 dispersed in coconut oil ( Figure 5) are: L*=20.3; a*=15.7 and b*=10.8. These results differ with the obtained results for the samples of the experimental design synthesized to the same calcination condition (850 ° C and 1.00 h), because of the large-scale change did not achieve the same chemical purity and phase composition [32]. The bluishred color pigment ABP325 coincides with a* (13.3 and 21.1) and b* (5.0 and 8.0) values of some high purity hematite pigments sythetized through magnetite calcination by Schaufler et al. [25]. Red hematite can be converted to purple hematite by modifying its particle size sufficiently, as is explained by Cornell and Schwertmann [17], due to there is evidence that heating hematite to temperatures above 800 °C causes a color change towards purple, due to the transformation of associations of small hematite crystals into oriented aggregates.
Synox ® commercial synthetic hematite pigments dispersed in coconut oil can be used as a reference for the pigment obtained ( Figure 5), which has the following color coordinates: Red 130B (L*=23.83, a*=29.09 and b*=24.66) and Red 130S (L*=23.30, a*=28. 94 and b*=23.73). These commercial reference pigments has much higher values of the a* and b* coordinate which indicates a yellowish-red color very different from the bluish tone of ABP325. This apparent discrepancy is mainly due to the much smaller particle size of these commercial pigments [25][26].

Conclusiones
Crystalline phases characterization by XRD of the processed residue "PR" confirmed the presence of all the corrosion products reported on reinforcing steel bars by Zitrou et al. [3]. Furthermore, mainly iron oxides coincide with the atmospheric corrosion products identified in the oxide scale of a carbon steel structure characterized in a very similar way by Velázquez and Jaramillo [33].
The nature of surface oxide waste from reinforcing carbon steel bars allowed its use as hematite precursor through direct thermal treatment achieved with simplicity after the correct choice of the experimental conditions involved in the calcination [13]. The obtained hematite showed properties that qualify it as a pigment [27], therefore it can be implemented in various applications [19], such as in the field of paints [29].

Suggestions
It is considered interesting to investigate other aspects related to the heat treatment and management of the recovered iron and steel waste, as well as the implementation of the obtained hematite pigments in various fields. Likewise, the grinding time of similar steelmaking residues will be increased to produce a hematite pigment precursor, with a more micronized particle size, since this will has greatly influence in the final properties of target products, it is means the pigment will have values of density, surface area, oil absorption, colorimetric coordinates, etc., closer to the other commercial hematite pigments.

Acknowledgements
The autors thank to the Universidad del Atlántico, for providing the necessary resources that allowed this experimentation and especially to the Bioprocesos group for financing this project in the "Primera Convocatoria Interna para apoyo al desarrollo de trabajos de grado en investigación formativa -nivel pregrado y posgrado 2018" offered by the institution. Moreover, we thank to Dr. Vidal Barrón head of laboratorio de Edafología at the Universidad de Córdona (Spain) and to Dr. Luciano Cuesta for objectively advising us on the technical aspects for the correct development of the pigment obtained, especially in heat treatment.
In addtition, we thank to ULTRACEM company for offering the provision of services to analyze the granulometry of the processed waste and pigments obtained, especially to Eng. David Daza for allowing access to this advice and to Eng. Wilmar Betancourt for his collaboration with the samples analysis. Special thanks to the AQUATERRA S.A.S company for offering its facilities, raw materials, equipment and personnel to contribute significantly to the completion of this project, especially to Eng. Gustavo Correa for making the company's resources available to us and engineers Juan Arias, Liliana Higuera and Stiven Becerra for their collaboration and their theoretical-practical teachings in the colorimetric analysis of the processed residue and obtained pigments.