A robotic platform for high-throughput electrochemical analysis of chalcopyrite leaching †

Cu extraction from chalcopyrite ores is typically a slow process that involves aggressive chemical reagents with signi ﬁ cant environmental impact. Ionic liquids (IL) have been proposed as a potentially more benign solution, but the sheer number of IL variants complicates the search for the most e ﬃ cient solvent systems. Here, we present an automated electrochemical platform that allows for screening of 180 and more leaching samples in parallel with minimal solvent consumption. In a proof-of-concept study, we screen 25 samples with di ﬀ erent IL and water contents, and ﬁ nd two orders of magnitude di ﬀ erence in leaching performance within this array. The best performing system is then applied in a tank leaching con ﬁ guration, with real-time electrochemical monitoring of Cu evolution in solution. All electrochemical data is found to be in excellent agreement with o ﬀ -line ICP-AES data.

The industry is also increasingly challenged with processing lower-grade chalcopyrite ores, 6 driving the development of new solution-based processes, applicable to both (bio-)heap and tank leaching. 7Conventional acid-oxidant chemical leaching systems, including the commonly employed Fe 2 (SO 4 ) 3 -H 2 SO 4(aq) system, are established and cheap (see recent reviews 6,8 ).However, the overall leaching process is slow at moderate temperatures, whilst also ultimately limited by surface passivation 9 or other kinetic effects. 10,11Hence, there is still significant potential towards improving the leaching performance. 6ccordingly, improved Cu extraction from CuFeS 2(s) has been achieved, e.g. by using alternative acid-oxidant combinations, microorganisms, ultra-fine grinding, elevated temperatures and pressures. 12However, the benefit of such methods is offset by increased cost and energy consumption, incompatibility with existing workflows or more generally higher environmental impact.
In the search for alternative lixiviant systems, ionic liquids (ILs) have emerged as an interesting alternative.ILs have welldocumented benefits as solvents in synthetic chemistry, 13 and it is conceivable that their chemical structure could be tailored in such a way to avoid the formation of kinetic barriers and maintain a high Cu extraction efficiency over time.To this end, notable dissolution enhancement, compared to ferric H 2 SO 4 -based lixiviants, has been reported for first generation alkylimidazolium hydrogen sulfate ionic liquids towards CuFeS 2(s) .Maximum Cu recovery of 86.8% from a Cu concentrate (∼70% CuFeS 2(s) ) by 100% [C 4 C 1 im][HSO 4 ] (bmim•HSO 4 ; pH ∼ −1 [ESI †]) presented a ∼60% enhancement compared to 1 M H 2 SO 4(aq) benchmark solutions ( pH −0.3), both containing excess Fe 2 (SO 4 ) 3 oxidant. 14However, as noted above, these cited enhancements have not been rigorously normalised for medium pH, such that there may be no IL-related Cu extraction enhancement.A subsequent kinetic study of [C 4 C 1 im]-[HSO 4 ] (aq) leaching applied to chalcopyrite ore (∼20% Cu), focused on the effect of temperature and agitation, measuring an Arrhenius-type activation energy from chalcopyrite dissolution of 69.4 kJ mol −1 . 15hile such studies are a promising first step, further improvement is highly likely, given the structural diversity of ILs.However, the complex nature of the dissolution process renders 'ab initio' rational design of optimal lixiviant systems out of reach.On the other hand, screening a large number of ILs or lixiviant compositions simultaneously and at a small scale appears to be a more realistic option.Such a combinatorial approach would furthermore allow for the optimisation of the leaching performance over several generations of leaches, where the best performing lixiviants could then be upscaled for more in-depth studies.Similar strategies have yielded excellent results in protein design and other areas. 16ere, we demonstrate that such a combinatorial methodology is indeed a powerful tool for screening and improving the Cu extraction performance of IL-based lixiviants, Fig. 1.Similar robotic electrochemical workstations have been applied in other contexts before, namely for automated combinatorial electrochemistry in large sample arrays for microelectrode studies 17,18 and in endothelial cell NO x excretion screening. 19ur platform allows for the screening of up to 180 ‡ samples at the same time, with electrochemical, in situ monitoring of Cu extraction from CuFeS 2(s) .We found that the detection capabilities of cupric ion-selective electrodes (ISE) were insufficient to detect Cu 2+ in solution in the presence of ILs at the initial stages of the leach ([Cu] < ∼1 mMsee ESI Fig. S4 †), but that Cu electrodeposition/anodic stripping improved the detection limit by approximately one order of magnitude, based on the conditions used here, see below.
Comparison with ICP-AES as a standard ex situ method, yielded a 1 : 1 correspondence with the electrodeposition/ anodic stripping results.After calibration and testing with model samples, we have applied the robotic screening platform to an array of 25 leaching samples in a proof-of-concept study.We find a 100-fold variation of the leaching performance between the best and worst performing lixiviants within the array.Subsequently, we examined the best-performing sample via electrochemical monitoring in a 120 mL scale tank leaching reactor over approximately 6 days.

Results and discussion
Atomic emission spectroscopy (ICP-AES) is a standard quantification method for metal ions in solution.However, ex situ solution sampling can be time-consuming and disruptive to leaching processes.Additionally, careful calibration for matrix effects may be required in complex solution environments, which extends to IL variations. 20,21On the other hand, our remit demands a technique allowing for high throughput automated study, alongside real-time monitoring of [Cu 2+ ] in diverse solution environments at all stages of the leaching process.In this study, ICP-AES was thus used only as an independent benchmark for electrochemically derived [Cu] measurements.
Since Cu ion-selective electrodes (ISE) appeared to be a facile and straightforward real-time detection approach for Cu 2+ in solution, we first tested Cu ISE suitability for the task at hand (see ESI for full experimental details inc.Fig. S4 †).Using a commercial ISE sensor (Cole-Palmer, Cupric Combination ISE), flat-line indistinguishable sensor response was obtained for [CuSO 4 ] (aq) < 10 −6 mol dm −3 and ∼10 −4 mol dm −3 in H 2 SO 4(aq) and IL (aq) media, respectively.Above the respective lower [Cu 2+ ] detection limit, all studied calibration plots exhibit near ideal Nernstian potential dependence of 29.6 mV per [Cu 2+ ] decade (75 mM H 2 SO 4 = 26.9mV; 450 mM [C 4 Him]-[HSO 4 ] = 30.6mV; 450 mM NH 4 •HSO 4 = 31.7 mV).Thus, in strongly Cu-coordinating IL (aq) media lower cupric detection limits are deemed unsuitable for the present purpose of monitoring ambient IL (aq) leaching on timescales of <2 days.However, as we show below, electrodeposition combined with anodic stripping of copper (ASV) indeed enables real-time quantification of [Cu 2+ ] in solution, with sufficient sensitivity even in the presence of ILs.
In light of our aforementioned automation objectives, a powerful robotic electrochemical platform has been built (Fig. 1).For our ASV studies, the instrument is fitted with a Ptdisc working electrode, assembled into a glass fused probe construct (d WE = 1 mm, Pt CE, Ag/AgCl RE).The fabricated probe is docked at the labelled 'electrode mount'.Motorised probe positioning and potentiostat functions are programmatically controlled via USB 2.0 serial port connectivity.Further details can be found below and in ESI.† Initial testing of the platform setup included determining the geometric factor for several fabricated electrodes in certi- fied KCl (aq) conductivity standards (Sigma Aldrich), followed by accurate measurement of CuSO 4(aq) solution conductivities with <6% error (1-50 mmol dm −3 ). 22Automated data acquisition for various IL (aq) ASV calibration plots (see ESI -Fig.S8 †) provided some ASV specific platform validation, however more complex electrochemical study was desirable, as described next.
For our lixiviant systems of interest, adaptation to a simplified one pot ('1-vial') ASV procedure proved beneficial from numerous perspectives.Cupric electrodeposition and Cu (s) stripping can be performed back-to-back within the sample vial, whilst significantly reducing the standard deviation of ASV repeats leading to 30-70% reduction in fitting standard errors (Table 2 and ESI Fig. S8b †).Additionally, '1-vial' ASV requires significantly fewer probe positioning steps, thereby minimising the combined duration of probe motion to <7% of the overall automation cycle, in turn maximising sample throughput.Hereafter, ASV experiments have been performed with the aforementioned in situ simplifications, unless otherwise stated.
Having established that ASV is capable of monitoring [Cu 2+ ] with sufficient sensitivity in the presence of IL (aq) , we then moved on to demonstrate array-based monitoring of leaching performance, as a precursor to large-scale IL (aq) screening experiments.
Subsequently, a two-variable IL (aq) lixiviant screening assay was undertaken, as a proof-of-concept experiment towards larger arrays.25  Strictly speaking, ASV Cu (s) stripping charge calibration parameters apply only at one single [IL] (aq) .However, for the concentration range used in this experiment, we found that the variation is in fact relatively small (Table 2).For simplicity, we used a single set of calibration parameters for all samples, namely those obtained for 450 mM [NH 4 •HSO 4 ] (aq) -Table 2. This decision is justified by retaining a strong correlation between [Cu] measures (m = 1.02 ± 0.04; R 2 = 0.963see ESI Fig. S21 †).
Crucially, equivalent regions of darkened 'hotspot' lixiviant performance are highlighted in each panel of Fig. 4. Within the 25 sample array, 2 orders of magnitude difference in leaching performance are observed between the best performing (450 mmol dm −3 ; 600 mg) and poorest performing combinations (1800 mmol dm −3 ; 75 mg).Broad variation in leaching performance is also reflected by relative variances of 70-320%; a minimum of 35-fold larger than rel.σ 2 values for 10 equivalently leached [NH 4 •HSO 4 ] samples (cf.Table 1).Thus, we have established confidence limits for distinguishing lixiviant performance from intersample variability, which operate on different magnitude scales.
Notably, these results suggest a non-trivial optimal [NH 4 •HSO 4 ] (aq) in the vicinity of 450 mmol dm  ance enhancement may be achieved with a more comprehensive screening effort.
Proceeding to scale-up this 'hotspot' performance system, a two-neck round bottomed flask was used for a 120 mL scale, 6 day leaching study with automated [Cu] sensing.Freshly milled CuFeS 2(s) (3 g; 32 ≤ x ≤ 75 μm) was leached at room temperature in [NH 4 •HSO 4 ] (aq) (450 mmol dm −3 ; 120 mL; 40 mL g −1 ), while stirred at a constant rate of ∼120 rpm.Stir-ring was intermittently stopped (marked * -Fig.5a), providing extended periods of unstirred ASV for ease of calibration.A second equivalent experiment was conducted using 450 mmol   dm −3 [C 4 Him][HSO 4 ] (aq) .Our electrode system was pre-conditioned § and inserted as a static probe, with a programmed electrochemical schedule set to ascertain ASV response at 2 h intervals, for a total leach duration of 140 h.Beginning with 300 s, electrodeposition duration was adjusted to maintain Cu (s) stripping charges within a calibrated linear range (<1.5 mC).Calibration irregularities ( plateaus and high standard deviations) have previously been observed above 2 mC, the origins of which are unclear and are under investigation (see ESI -Fig.S9 †).3), producing a Cu : Fe extraction ratio of unity. 6,8espite using CuFeS 2(s) from the same batch and equal [IL] (aq) , comparative leaching in 450 mmol dm −3 [C 4 Him]-[HSO 4 ] (aq) ( □ / ■ , Fig. 5b) produces a parabolic, kinetically slow, Cu extraction profile.Interestingly, and in stark contrast to NH 4 •HSO 4 , an initial period with very little leaching is observed below 50 h, after which point, familiar electrochemical response and [Cu] tracking is resumed (see ESI -Fig.S30a †).This further exemplifies the value of our continuous automated approach to leached [Cu] monitoring and the leach-specific insights that can be extracted from reconstruction of a time-dependent extraction profile.Extracted [Fe] levels were found to be significantly higher than that of [Cu], averaging 132.6 ± 1.4% of corresponding [Cu] (see ESI -Fig.S30b †).
Aqua regia-based digestion of the milled, unleached CuFeS 2(s) starting material, confirmed the expected Cu : Fe metal ratio of unitysee below.Differing solution pH of 450 mmol dm −3 [NH 4 •HSO 4 ] (aq) (0.9 ± 0.05) and [C 4 Him][HSO 4 ] (aq) (1.2 ± 0.05) may go some way in explaining the difference in leaching per-formance.However, pH alone cannot explain the presence/ absence of induction periods or linear/parabolic Cu extraction behaviour for equivalent CuFeS 2(s) starting materialfurther detailed study is required.
Overall, our automated platform for data acquisition has proven effective in addressing several challenges existing within the field of acid-sulfate hydrometallurgy.Indication of promising IL (aq) systems amongst wide-ranging leaching performances within a modest-scale screening experiment has paved the way for large array screening of unstudied IL (aq) systems, which can utilise assessed sample-to-sample variability to define confidence limits.Furthermore, we have presented a new in situ approach to automated CuFeS 2(s) leach monitoring.The tool is applicable across diverse [IL] (aq) systems, minimising reliance on laborious ex situ ICP-AES sampling, and allows full reconstruction of Cu extraction profileswhere leaching dynamics can be clearly observed.There are thus significant prospects in employing this approach to even larger scale studies in pursuit of next generation Cu lixiviant systems.

Conclusion
The setup of an automated electrochemical platform has been described, offering potentially high-throughput, low-volume screening capabilities, with proven applicability to IL (aq) systems and CuFeS 2(s) hydrometallurgy.
Screening has been characterized through parallel leaching of ten equivalent samples in 450 mmol dm reconstructing full extraction profiles for two IL (aq) systems, with high time resolution.Differentiation of the two 450 mmol dm −3 [IL] systems was straightforward, through clear differences in extraction rates, the shape of the extraction profiles (i.e.linear/parabolic).Additional, potentially mechanistically relevant features were uncovered, including a 50 h dormant period for CuFeS 2(s) leaching in [C 4 Him][HSO 4 ] (aq) .
Work to-date suggests that some promising IL (aq) lixiviant systems, such as NH 4 •HSO 4(aq) , may not suffer from the same surface passivation effects as conventional ferric-acid-sulfate media, 6,8 although longer duration studies with focus on other key variables (E h , constant T etc.) are required.Moving forward, we will employ the tools introduced herein, in an iterative approach to large scale IL (aq) screening and extended electrochemical monitoring of lead systems for up-scaled studies.A broad unexplored IL chemical space awaits.

Platform design
A commercial milling platform (Heiz CNC Technik High-Z S-400 T) provided the basis for platform development (Fig. 1).Four stepper motors (1600 step per rev, Nanotec) are wired appropriately to commercial driver boards (Easydriver) and digital output ports (DO) of a microprocessor board (ATmega328, Arduino UNO).The microcontroller is interfaced with a graphical programming package (VISA Instrument Control Palette, NI LabView) using USB-delivered custom-designed firmware.Fig. 1 shows an overhead scaled technical diagram of the platform, indicating the electrode probe mount and sample holder (204 vial wells).At first use, probe 3D positioning is zero-referenced at X (0,0,0), from which positive (referenced) coordinate changes define the current probe positioning (+x,+y,+z), as tracked by firmware coding.Fixed cartesian (x,y) vial locations are stored within the graphical programming suite and retrieved for motor operation as necessary.
Full potentiostatic functionality is accessed through a manufacturer designed dynamic link library (.dll -Compactstat, Ivium Technologies) interfaced with the graphical programming suite.All operations are sequenced back-to-back for custom automation design, with phase completion and triggering managed by monitoring appropriately constructed instrument status signals.
Calibration plots for unstirred ASV can be rapidly generated utilising the electrochemical platform and In an effort to quantify the detection limit for Cu stripping under the present conditions, we divide the standard error of the intercept in Table 2 by the sensitivity, and obtain values between 0.11 and 0.31 mM, depending on the solution medium.We take this is as an estimate for the minimum stripping charge that we can detect in the present experimental configuration.

Fig. 1
Fig.1To-scale (overhead) technical schematic of the automated robotic platform, marking key features.Overlying, is an exemplar twovariable lixiviant screening result, highlighting regions of enhanced and moderate Cu leaching from CuFeS 2(s) .

Fig. 4
Fig. 4 (a) ICP-AES determined [Cu] for an array of 25 samples after 264 h ambient leaching, with the [NH 4 •HSO 4 ] (aq) and CuFeS 2(s) mass indicated.(b) Equivalent [Cu] measurements made using ASV after the same leach duration (264 h).A failed reading within the dataset is marked X.

− 3 [
NH 4 •HSO 4 ] (aq) , revealing sample-to-sample variation of 0.7-2.1% (rel.σ 2 ).Subsequent screening within a modest 5 × 5 sample array returned lixiviant performances ranging over 2 orders of magnitude, at least 35-fold larger than measured sample-to-sample variability.The presence of an 'optimum' leaching performance at [NH 4 •HSO 4 ] (aq) = 450 mmol dm −3 was unexpected and is incompatible with purely pH dependent leach dynamics.This lixiviant composition was then further explored in up-scaled ambient leach experiments and displayed linear extraction dynamics over 6 days of continuous [Cu] sensing.This in situ electrochemical monitoring of leached [Cu] proved effective for § Prior to insertion, the electrode is pre-conditioned using five back-to-back electrodeposition/stripping cycles in 0.45 M IL (aq) containing 10 mM CuSO 4(aq) .
[CuSO 4 ] (aq) solutions of known concentration.Consequently, stirring was switched off intermittently during in situ Cu sensing experiments, providing sections of reference data for rapid production of electrochemical [Cu] measurements.ICP-AES sampling provided regular [Cu] and [Fe] reference points for comparison of ASV [Cu] measurements.

Table 2
Cu (s) stripping calibration parameters, in a range of leachate mimetic acidic media.Conventional 2-vial ASV is simplified to a 1-vial procedure using back-to-back electrodeposition and stripping cycles within the cupric analyte-containing leachate/standard solution