Keywords

Introduction

Depending on the impact onto the binder metal and carbide, three main routes can be distinguished regarding the hard metal recycling techniques. Direct processes disintegrate the substrate so the carbide(s) and the binder are mixed in the reclaimed powdery product. The chemical composition of the treated feedstock remains basically unchanged, although minimal process-related contaminations occur. Additionally, the low consumption of energy and therefore the high cost efficiency are beneficial. [1, 2] Within the direct group the zinc process displays the main representative. For instance, in 2000 about 25% of the cemented carbide scrap passed through such facilities in the USA [3] and at least 10% of the raw material demand in Europe [4] was supplied by the same.

Indirect techniques decompose the scrap chemically in their elementary components where these are separated. The process scheme is similar to the primary route and accordingly extensive. For tungsten this implies dissolution, leach purification, crystallization, calcination, reduction and finally carburation in order to receive WC “tungsten carbide”. The process’s main advantage comprises that the reclaimed powders have the same quality as primary material. [2, 5] Two commercial practices exist inside this category, an oxidation process with chemical digestion or an alkaline salt fusion method. According to Shedd [3] a share of about 35% of the hard metal scrap in the USA is handled by these methods, with an even higher amount of 85% [5] in Germany. Hence, the last class, the semidirect methods, only plays a minor role in today’s recycling market. According to this recycling path, the present coatings as well as the binder metals are leached and leave behind a WC skeleton which can be pulverized more easily. Consequently, no extensive procedure should theoretically be necessary for obtaining a reusable pure WC powder. Also, common hydrometallurgical techniques retrieve the binder metal. [5,6,7]

However, several issues are reported in literature, such as partial oxidation of WC [6,7,8], slow dissolution of the binder [6, 8], non-removable films [9] and in addition references including powdered hard metal scraps [10] or at least an accompanying mechanical treatment [8, 9, 11]. Notably, the mechanical pulverization of cemented carbides causes troubles, since this category of materials serves for machining others. Thus, this report examines the leaching kinetics of a cutting insert in HCl (hydrochloric acid) and H2O2 (hydrogen peroxide) to provide the basis for overcoming the mentioned issues. In order to achieve this, a design of experiments was utilized to obtain statistically relevant data together with an adequate sample preparation. Altogether a simplification and elimination of side effects allowed a proper analysis of the influence of temperature and molar concentrations on the leaching behavior. Moreover, an empirical model equation of the reaction rate was developed to describe the lixiviation of cobalt out of the hard metal substrate.

Materials and Methods

The cutting inserts applied in these experiments were kindly supplied by CERATIZIT Austria GmbH and complied with the properties according to Table 1. As specified in ISO 1832, the mentioned geometry “SNUN120412” relies on a quadratic base with 12 mm comprising 1.2 mm rounded corners and exhibits a height of 4 mm.

Table 1 Attributes of the used cutting inserts
Full size table

A fundamental part of the experiments included the preparation of the test specimen since it eased the subsequent compilation of the model equation. In order to keep the reaction interface almost constant, the cutting insert was hot-embedded in the resin “Struers ClaroFast”. The pretreatment of the samples consisted of grinding and polishing, here the last step comprised a 1 μm diamond finish. A drill hole through the resin permitted the central placement in the reaction vessel by a glass bar. During the experiments the thermoplast remained stable in the applied corrosive media. No preferential leaching appeared at the juncture between the cutting insert and the mounting resin which could have affected the results. Corresponding to Fig. 1, the prepared sample only allowed the leaching in one direction.

Fig. 1
figure 1

Specimen for the investigation of the reaction rate

Full size image

The experiments were executed in a 1000 ml double-walled borosilicate reaction vessel with a clamp fixed cap which enabled various access possibilities via four glass joints 29/32 and one 14/23 according to DIN 12242. This included a reflux condenser, a purging nitrogen supply, a specimen holder and a sampling port as well. Furthermore, a Pt100 sensor close to the specimen allowed an adequate temperature control. A magnetic stirring device at about 200 min−1 maintained a uniform intermixing of the liquid. The implanted design of experiments provided reliable results. Therefore, the software Modde 11 supported the creation of the CCF design as well as its final evaluation. Altogether the investigation compromised of 17 tests with the variable parameters in the same order as reported in Table 2. At the beginning 500 ml of leachant were heated to the desired temperature and subsequently the experiment started by the immersion of the specimen. The monitoring of the leaching process took place by extracting 2.5 ml samples after 5, 10, 20 and 30 min. An ICP-MS realized the analysis according to ÖNORM EN ISO 17294-2. A multi-element standard solution with a mass concentration of β = 10 mg/l for cobalt served for the calibration in the range of 0.50–100 µg/l. For the analysis a dilution of all samples occurred in following ratios: 1:100, 1:1000, 1:10,000.

Table 2 Results of the regression analysis for the k- and n-factor with corresponding coefficient of determination in relation to the variable parameters of the CCF design
Full size table

Additionally, a mathematic correction considered the binder metal removed by sampling before the linear regression analysis. For the assessment of the reaction process, a rate law is needed. Many diffusion controlled reactions obey the parabolic law. The general form complies with Eq. (1), based on Habashi. [12]

$$ k \left[ {\frac{\text{mg}}{{{\text{l}} \cdot { \hbox{min} }^{ 0.5} }}} \right]\cdot t^{0.5} \left[ {{ \hbox{min} }^{ 0.5} } \right] = y \, \left[ {\frac{\text{mg}}{\text{l}}} \right] $$
(1)

The factor k represents the reaction rate coefficient and t the elapsed time of the ongoing reaction, which together results in the concentration of the leached metal in the solution. As a result, some experiments disclosed a deviation from the parabolic law. Therefore, a modified empirical law according to Eq. (2) was implemented that applies the power factor (n) as a second parameter.

$$ k \left[ {\frac{\text{mg}}{{{\text{l}} \cdot { \hbox{min}^{\text{n}}}}}} \right] \cdot t^{n} \left[ {{ \hbox{min} }^{\text{n}} } \right] = y \, \left[ {\frac{\text{mg}}{\text{l}}} \right] $$
(2)

A simple logarithm permits a linear regression analysis which was executed by the function “LINEST” of Microsoft Excel 2010.

Results and Discussion

In order to compile the model equation, the determination of the reaction rate coefficient k and the power factor n in dependence of the variable parameters was conducted. The coefficient of determination indicates that most of the variance is accounted for in accordance with the underlying variables k and n as listed in Table 2.

The Fig. 2 reveals the relationship between the reaction rate coefficient and the power factor at different experimental conditions. The two not specified parameters in each diagram exhibit an intermediate adjustment in terms of the experimental boundaries. Surprisingly, the acid- and oxidant concentrations reduce the k-factor at higher amounts, while n is only slightly affected. In terms of temperature, n increases up to a peak at about 60 °C, whereas the reaction rate coefficient rises only insignificantly.

Fig. 2
figure 2

Prediction plot with 95% confidence interval estimation for the k- and n-coefficients in dependence of a variable parameter

Full size image

The contour diagram in Fig. 3 reports the attainable values at 2.5 mol/l H2O2 and variable temperatures as well as acid concentrations. Corresponding to the illustration, the power factor lies around 0.5 for a large area, but can deviate noticeably depending on the parameters.

Fig. 3
figure 3

Response contour plot for the k- and n-coefficients at 2.5 mol/l H2O2 in reference to temperature and acid concentration

Full size image

Conclusion

In agreement with the current investigation, the leaching of a cutting insert consisting of cemented carbide in an aqueous media of HCl and H2O2 follows the classical law of diffusion depending on acid, oxidant concentration and temperature adjustment. Nevertheless, a certain deviation of the power factor occurs, which may be caused by a complex leaching mechanism. Possible reactions include the dissolution of cobalt, complex formation and diffusion through the remaining WC skeleton. Additionally, the WC itself may loosen and therefore the diffusion zone does not grow according to the proceeding cobalt dissolution. A further side reaction includes the decomposition of the oxidant into oxygen and water. This report discloses the fundamental parameters which are able to positively affect the leaching characteristics of cemented carbides . The obtained results indicate a substantial relation within the investigated system.