Microchimica Acta (2024) 191:3 https://doi.org/10.1007/s00604-023-06029-x ORIGINAL PAPER Microflow injection analysis based on modular 3D platforms and colorimetric detection for Fe(III) monitoring in a wide concentration range David Ricart1 · Antonio David Dorado1 · Conxita Lao‑Luque1 · Mireia Baeza2 Received: 23 May 2023 / Accepted: 2 October 2023 © The Author(s) 2023 Abstract A modular microflow injection analysis (microFIA) system for the determination of Fe(III) in a bioleaching reactor has been designed, developed and validated. The different modules of the analyzer (mixer, diluter, disperser and detector) were 3D-printed. Fe(III) quantification is due by measuring the color intensity of the chelate formed between Fe(III) and salicylic acid at 525 nm. The device has been designed to dilute, disperse and detect high Fe(III) concentrations in the form of an inexpensive multi-step photometric flow cell that uses an light-emitting diode (LED) as a light source and an light-dependent resistor (LDR) as a light intensity detector. This microFIA system has been shown to be suitable for automatic and continuous determination of Fe(III) in the operation of a bioreactor for the oxidation of Fe(II). The device has a good repeatability (less than 5% of coefficient of variation in the whole range of concentrations) and accuracy of around 100%. The analyzer features an exceptional wide linear range, between 25 and 6000 mg·L−1. The device was successfully applied to the determination of Fe(III) in real samples. The obtained results proved that the method is applicable for accurate, precise, rapid, and low-cost colorimetric analysis and didn’t show significant differences with a conventional UV–Vis method. Keywords Fe(III) · 3D-printed microfluidic platform · FIA · Bioleaching · e-waste · Circular economy Introduction ceramics, and metals. Currently, pyrometallurgical and hydrometallurgical methods are the primary processes uti- In recent decades, the combination of rapid technological lized for the recovery of these metals from e-waste. None- advances and obsolescence has increased e-waste by 3–5% theless, these methods entail substantial economic costs per year according to the World Health Organization (WHO) and have a significant environmental impact due to their [1]. Consequently, it has become increasingly urgent to find use of high temperatures and harsh chemicals. Recently, methods for recycling e-waste not only for environmental other recycling technologies that are more environmentally reasons but also because e-waste typically contains valuable sustainable have been under investigation. Among these is metals such as copper, silver, cobalt, lithium, gold, nickel, the bioleaching process, which is based on the activities among other and so, their recovery for new uses is highly of certain microorganisms that generate a leaching agent profitable. responsible for metal extraction. Bioleaching is currently However, e-waste recycling is difficult due to its com- employed in the mining industry for low-grade ores [2], such plex structure and heterogeneous blend of organic materials, as in bio-heap processes for commercial applications [3, 4], but it may also be effective in the industrial-scale recovery of metals from e-waste [5, 6]. * Mireia Baeza Acidithiobacillus ferrooxidans has been employed for the mariadelmar.baeza@uab.cat bioleaching of copper from electronic waste via the oxida- 1 tion capacity of Fe(III) at acidic conditions. The bacteria Universitat Politècnica de Catalunya, Avinguda de les Bases de Manresa 61‑73, 08240 Manresa, Spain possess the ability to re-oxidize Fe(II) into Fe(III) following 2 the oxidation and solubilization of the copper content in the GENOCOV Research Group, Department of Chemistry, Faculty of Science, Edifici C-Nord, Universitat Autònoma de matrix, thus enabling a cyclic process to be established [7]. Barcelona, Carrer dels Til·Lers, 08193 Bellaterra, Spain The copper bioleaching process is typically executed in two 13 Vol.:(0123456789) 3 Page 2 of 13 Microchim Acta (2024) 191:3 stages. In the initial stage, the oxidation process transpires Most of these problems and limitations can be overcome within a bioreactor where microorganisms oxidize Fe(II) to by flow injection analysis (FIA). The FIA technique has Fe(III). Following this, the Fe(III) solution, without the pres- the capability to monitor various parameters, including Fe ence of bacteria, is introduced into the reactor containing ions, in different matrices [14, 28–31]. FIA methods provide the e-waste where Cu(II) extraction occurs. Monitoring the real-time and in situ analysis, allowing for automation and concentrations of Fe(II) or Fe(III) is crucial in both stages high sampling throughput. However, there is currently a lack for controlling and optimizing the process. Thereby, a fast of published research regarding the monitoring of Fe(III) and accurate technique for measuring these ions would prove concentrations in a bioreactor using FIA. Additionally, the beneficial in researching, managing, and overseeing these linear range response of these methods is quite narrow, so processes. large dilutions must be made when analyzing high concen- Various instrumental techniques have been proposed to tration samples, which is often a source of error for Fe(III) analyze Fe(III) in water samples, such as atomic absorp- concentration. The linear range found in similar works is tion spectrometry [8], inductively coupled plasma-mass 0.05–4.0 mg·L−1 [32], 1–40 mg·L−1 [29], 0.2–5 mg·L−1 spectrometry [9], inductively coupled plasma-optical emis- [33], 0.021–2.4 mg·L −1 [34], 40–350 mg·L −1 [35] and sion spectrometry [10], fluorescence spectroscopy [11, 12], 110–560 mg·L−1 [36]. as well as cathodic or anodic stripping voltammetry [13] Accordingly, the objective of this study is to design, and chromatography [14]. However, colorimetric analysis develop, and validate a reliable and adaptable microFIA of iron ions is considered a particularly significant, attrac- system, fully automated and integrated, utilizing 3D print- tive, and straightforward method. A myriad of chelating ing technology for on-site monitoring a wide range of Fe(III) reagents of Fe(III) have been proposed in the literature. concentrations in a bioreactor. The detection of Fe(III) is One of the most commonly employed colorimetric reagent achieved through the development of a coloured complex is 1,10-phenanthroline [15, 16]. To determine ferric iron, between the iron ion and salicylic acid which is specific for it is necessary to reduce it to Fe(II) using reducing agents Fe(III) at acidic pH [37] that absorbs at 525 nm. The device like hydroxylamine hydrochloride. The use of a phenanth- will be designed using a CAD program, featuring narrow roline derivative, bathophenanthroline, has also been sug- microchannels that require minimal volumes of both samples gested for the determination of ferrous iron [17]. A new and reagents. These features will enable cost savings and approach for concurrently measuring ferrous and ferric iron frequent, on-site, automatic and reliable determination of concentration in aqueous samples was presented by Kara- Fe(III) in a broad range of concentrations. manev [18]. This method involves determining both ferric and total iron simultaneously at distinct light wavelengths, thereby substantially reducing analytical procedure errors. Materials and methods The technique relies on the colorimetric assessment of a red- coloured ferric-sulfosalicylate complex formed under acidic Chemicals and reagents conditions that absorbs at 500 nm. Following the addition of ammonia, which raises the pH, 5-sulfosalicylic acid (SSA) In this study, we utilized a polylactic acid (PLA) filament forms a yellow complex with all iron ions in the solution of 1.75 mm produced by LEON 3D (Valverde de la Virgen, it has highest light absorbance at 425 nm. More recently, Spain) in three different colours: black, white and transpar- novel chelating reagents have been used to analyze Fe(III) ent. We also used F ­ eNH4(SO4)2·12H2O (99%), salicylic by colorimetry such us deferiprone [19], benzimidazole [20], acid (99.5%), and ethanol (96%), all of which were obtained 4-mercaptophenol and mercaptoacetic acid [21], glycine from Scharlab (Barcelona, Spain). ­H2SO4 (95–98%) was [22], 8-hydroxyquinoline [23], gold nanoparticles [24, 25], purchased in Panreac (Castellar del Vallès, Spain). For the 2,5,7-triarylimidazopyridine [26] and ferrozine [27]. selectivity study, a solution of 1000 mg·L−1 of Cu(II) was However, a major drawback of many colorimetric meth- obtained from ­CuSO4·5H2O (Labkem, Barcelona, Spain), ods is that they are highly manual and non-automated, a 1000 mg·L−1 of Al(III) from A ­ l2(SO4)3·18H2O (Schar- requiring trained personnel, controlled laboratory environ- lab, Barcelona, Spain) and 1000 mg·L −1 of Fe(II) from ments, and in many cases, prior sample treatment, which ­FeSO4·7H2O (Chem-lab, Barcelona, Spain). Additionally, affect analysis time and quality. Furthermore, each meas- we utilized a red phenol dye from Scharlab (Barcelona, urement made using colorimetric methods requires a high Spain). All chemicals utilized in this study were of analyti- consumption of sample, reagents, and time, which is a sig- cal reagent grade. nificant limitation in monitoring bacterial activity inside the To prepare Fe(III) solutions, an aqueous acidic solu- bioreactor over time and limits the number of measurements. tion at pH = 1.8–2.0 was prepared by dissolving ­H2SO4 in These constraints render these methods unsuitable for on- deionized water. For FIA experiments, a stock solution of site monitoring. Fe(III) with a concentration of 6000 mg·L−1 was prepared. 13 Microchim Acta (2024) 191:3 Page 3 of 13 3 For the UV–Vis assays, a Fe(III) stock solution with a Experimental setup concentration of 375 mg·L−1 was prepared by dissolving ­FeNH4(SO4)2·12H2O in the acidic solution. It was criti- The scheme of the complete microflow injection analy- cal to maintain a pH value of 2.0 or below throughout the sis (microFIA) device is shown in Fig. 1A. The main parts experiment to prevent the precipitation of Fe(OH)3 inside are: (i) the injection system, (ii) the four 3D-printed modules the channels. assembled (diluter, mixer, disperser and detector) and (iii) The chelating reactant solution was prepared by dissolv- the acquisition data system. ing salicylic acid in ethanol at a concentration of 5% (w/v) Fluids flow (sample, carrier, reactant) was managed by for its use in UV–Vis experiments and a 0.180% (w/v) solu- two active components, peristaltic pumps and valves. A tion was prepared for FIA experiments using the aqueous GILSON Minipuls 3 R8, with an 8-channel head, (Middle- acidic solution. The acidic solution was obtained by add- ton, WI, USA) peristaltic pump equipped with PVC tubing ing ­H2SO4 (1:1) to any volume of deionized water, stirring with 1.52 mm internal diameter Tygon®tubing (Ismatec, continuously, until a pH of 1.8 for batch and pH of 2.0 for Swiss) was used to pump the sample, reagent and H ­ 2SO4 FIA was reached. Salicylic acid is known to form a coloured solution as diluent and carrier. For the sample injection, a purple complex with Fe(III), which exhibits an absorption six-way injection valve manufactured by Hamilton Injector peak at 525 nm. MVP (Cary, NC, USA) was utilized. A 1% solution of a red phenol dye in water was prepared The experimental operational parameters were fixed to check the tightness of the modules and the correct func- as follow. Flow rate of solutions was between 0.33 and tioning of the detector. All solutions were prepared with 0.41 mL/min. The carrier flow rate was 0.33 mL/min, the deionized water for Milli-Q system (Millipore, Billerica, diluent flow rate was 0.39 mL/min, the reagent flow rate MA, USA). was 0.36 mL/min, and the ethanol flow rate was 1.36 mL/ min. The injection volume was 25 µL, and the volume of the diluter, mixer and disperser was 95.1 m ­ m3, 2034 m­ m3 and Microfluidic 3D modules design, printing conditions 3 1524 ­mm , respectively. and fabrication The detection system was manufactured by assembling a 525 nm light-emitting diode (LED) green (OSGGD25112A, The FIA-designed system consisted of four 3D-printed Transfer Multisort Elektronik, Polonia) and a light- microfluidic modules: diluter, mixer, disperser, and detector. dependent resistor (LDR) (PGM5537, Transfer Multisort The CAD model design of the four modules was developed Elektronik, Polonia), to serve as a detector, connected by using AutoCAD (Autodesk). A Prusa i3MK3S + printer was Arduino UNO with PC via USB. The signal acquisition used for the manufacturing of all the pieces (Prusa Research software by LabVIEW was custom made to this detection Prague, Czech Republic). system. The PLA was chosen to print modular blocks because it To clean the detector, a 3-way solenoid valve 161T031 is not expensive, eco-friendly, and has mechanical charac- from NResearch (West Caldwell, NJ, USA) and a Dinko teristics suitable for the purpose of this work. Black, white D-21 V peristaltic pump (Dinko instruments, Barcelona, and transparent PLA were tested to find the best colour for Spain) was used. The solenoid valve was connected to an pieces. The models of the different pieces were exported interface Cool Drive Valve Drivers for 161 series/12 vdc., as STL files to CURA software where the printing settings 161D5X12 of NResearch (West Caldwell, NJ, USA) to carry were defined. The files were then exported as gcode files to ethanol inside for detector cleaning between subsequent SD targets and the codes were read by a 3D-printer. measurements. The printing conditions for the PRUSA printer involved The four 3D printing modules (diluter, mixer, disperser, a layer height of 0.1 mm, with the first layer set to a height and detector) of the microFIA system were joined with PVC of 0.2 mm. Line width 0.2 mm, top/bottom thickness 1 mm, tubing and polytetrafluoroethylene (PTFE) tubes (Tecator, wall line count 5, main speed 60 mm/s, 10 bottom and top Hohganes, Sweden) of variable length and 0.8 mm internal layers, initial printing temperature was 220 °C and thereafter diameter. 215 °C, plate temperature was 60 °C and a retraction dis- The analytical signal obtained from the system is shown tance of 1.6 mm. During the CAD design process, a distance in Fig. 1Ai. The height of the peak is proportional to Fe(III) of 4–5 mm was deemed sufficient between the channel and concentration. The total analysis time is 605 s (10.1 min). the external wall. In detail, the partial times are 0 s injection time, 85 s start- In order to test for leaks, several pieces were 3D-printed up time, 127 s peak maximum height, 295 s recovery time using transparent PLA. The red phenol dye was then passed of 95% of the baseline value and 310 s clean step (Fig. 1B). through the channels in these pieces to detect any potential The inlet/outlet 3D-printed modules were designed as leaks. described in a previous study [34]. Different sizes of inlet/ 13 3 Page 4 of 13 Microchim Acta (2024) 191:3 1. Fill de loop 3. Dilution 4. Mix with reagent 5. Dispersion 2. Injection Solenoid valve sample b c d g e Sample a Peristaltic pump h Peristaltic pump A) Salicylic acid Diluent and carrier f pH=2.0 pH=2.0 Ethanol H2SO4 7. Clean with ethanol 6. Detection i B) Fig. 1  (A) Scheme of the microFIA system to determine Fe(III). portional to concentration. (B) Analytical signal and sequence of time (a) Peristaltic pump for solutions, (b) six-way valve, (c) diluter, (d) diagram to determine Fe(III): injection time, start-up time, time at the mixer, (e) disperser, (f) detector, (g) 3-way solenoid valve, (h) peri- maximum signal, and recovery time of the baseline. The total analysis staltic pump for cleaning, (i) analytical signal, the peak height is pro- time is 10.1 min outlet systems were tested in order to ensure complete tight- layer of thermoplastic (g) to minimize stray light and ensure ness. Finally, the external diameter was 2.72 mm, the inter- optimal optical performance and to prevent vibration of the nal diameter was 1.88 mm and the diameter on the bottom diodes. was 1.02 mm but the last one will vary according to the The LED and LDR were mounted in their designated module. Due to these dimensions, the connecting tubes had holder within the detector (Fig. S2A and S2B, respectively) to be expanded to be able to fit them into the inlet and outlet whose shape was optimized for the components. To avoid of the platform to ensure no leaks. The isolated structure is water leaks in the contact between optical windows and shown in Fig. S1. flow cell, a sealant NURAL28 (Henckel Ibérica, Barcelona, Spain) (Fig. 2Ah) was used. In addition, a black thermo- 3D‑optical flow‑cell detector plastic from SALKI (Barcelona, Spain) was used to fix all components and avoid external light interference. Figure 2A shows a scheme of the 3D-optical flow cell detec- Regarding the flow cell, both ends were open to allow tor designed and fabricated in the work presented herein. light from the LED (Fig. S2A) to pass through it and reach Black PLA was chosen to isolate the optical cell and prevent the LDR (Fig. S2B). The optical windows (Fig. 2Ae) made interference from ambient light in the colorimetric meas- of glass (26 mm) closed the microchannel and a sealant was urement of the chelate. The key elements are (a) LED; (b) used to block water flow between the contact surface of the LDR; (c) the holder for the LED and LDR (Fig. S2A and window and the flow cell. During the printing process, it S2B, respectively); (d) the flow cell which is the channel was imperative to suspend the ends of the flow cell in the air; where light interacts with sample (e) transparent windows; however, supports were necessary to achieve a good print, (f) inlets and outlets of sample. Figure 2B is an image of the as depicted in Fig. S2C. Once the printing process was com- assembled 3D-printed photometric detector flow cell with a plete, the supports were removed. 13 Microchim Acta (2024) 191:3 Page 5 of 13 3 Fig. 2  A CAD design of the detector flow cell and its differ- ent components. B 3D-printed photometric detector flow cell assembled The detector converts light intensity into a proportional the diluter was intended to adjust the Fe(III) concentrations electrical signal (mV) and the signal was collected every until measurable operational range. The diluted sample was 0.1 s. The detector consisted of a green LED (525 nm) then propelled into the mixer module (Fig. 1Ad) where it and an LDR of 5 mm diameter, and the LED was pow- came into contact with the chelating reagent (0.180 mg sali- ered by 3 V and 20 mA. The LDR was integrated with cylic acid/100 mL) developing a purple color whose inten- an Arduino UNO for data acquisition purposes, whereby sity was proportional to Fe(III) concentration. After that, the LDR was connected to a voltage divider circuit. Spe- the mixture flowed through the disperser (Fig. 1Ae) and cifically, the LDR was linked to the second resistance finally entered the 3D-optical flow cell (Fig. 1Af), where of the circuit, while being directly connected to ground. absorbance was measured and transformed into electrical The first resistance in the voltage divider was 10 KΩ. signals (peak height, mV). Subsequently, the solution was Thus, the measured voltage was directly proportional to released from the system and discarded as waste. To clean the absorbed light, and therefore to the analyte concen- the detector after each measure, a 3-way isolated solenoid tration. Each 100 ms Arduino measured the voltage and valve (Fig. 1Ag), controlled by cooldrive, and a peristaltic sent it to the computer. The CAD design of the detector pump were used to pump ethanol (Fig. 1Ah). The sequence is illustrated in Fig. S3. you can see in Fig. 1B. The total analysis time for a sample is 10.1 min, which means that almost 6 samples can be ana- MicroFIA procedure for Fe(III) determination lyzed per hour. A software application was developed for the purpose of The analysis cycle worked as follows. A peristaltic pump controlling the system, utilizing LabVIEW 2021. This soft- (Fig. 1Aa) automatically propelled sample (or standard ware was designed to manage the operation of the 6-way valve, solution), the carrier ­(H2SO4 solution at pH = 2.0) and the facilitated through the implementation of serial communica- chelating reagent into the system. Automatic injection of the tion. Additionally, to enable communication with the Arduino sample (25 µL) into the diluter (Fig. 1Ac) was performed by UNO, a package named LINX was installed into LabVIEW. a six-way valve (Fig. 1Ab). As Fe(III) concentration of sam- The Arduino UNO was connected to a relay shield 3.0 for ples from the bio-reactors could easily reach 6000 mg·L−1, Arduino from seed studio, which was utilized to send signals 13 3 Page 6 of 13 Microchim Acta (2024) 191:3 to the pumps and a 3-way isolated solenoid valve. The details conditions were tested to ensure optimal system efficiency. of the software specially designed for the control of the ana- The primary focus of the optimization process was to obtain lyzer and the active elements of the system (pumps, valves, a wide response range of Fe(III) concentrations, improve the and data reading/acquisition) are detailed in the Electronic degree of mixing within the analyzer, and enhance detector Supplementary Information. stability. Spiked and real samples Diluter Several real samples were analyzed to validate the analyzer: medium from a bioleaching reactor, water from a nearby reser- The diluter served to adjust the sample concentration to voir (Parc de l’Agulla, Manresa, Spain), a mineral medium of a ensure compatibility with the operational range response of bioreactor without Fe(III) and a leachate obtained from an acid the micro flow injection analysis system. Given that sam- digestion of a battery cathode from a mobile phone. Water from ples sourced from the bioleaching reactor were anticipated to the reservoir was filtered using syringe filters 0.2 μm (Chemlab exhibit high Fe(III) concentrations, potentially reaching up group, Terrassa, Spain). All these samples were analyzed after to 6000 mg·L−1, this step was crucial. The diameters of the adding a known concentration of Fe(III) except the medium inlets, outlets, and channels within the diluter were of utmost from a bioleaching reactor which contains Fe(III) itself. significance, as they played a pivotal role in facilitating the For accuracy studies, the samples from the bioleaching proper flow of the solvent and sample mixture. reactor were diluted with the acidic solution with dilution fac- To optimize the dilution process and minimize overpres- tors of 1/10, 1/5, 1/2 and 3/4. For each dilution, 2 aliquots sure on the diluter module, dilution was performed at three were taken, one of them was directly analyzed and in the other, distinct points. The main channel of the diluter comprised 1500 mg·L-1 Fe(III), were added and then analyzed. All the two serpentine sections, followed by a straight section. samples were analyzed in triplicate. The internal width of the main channel was progressively Results from microFIA were compared to those obtained increased by 0.5 mm in each successive section. The first with the standard conventional method using a spectrophotom- section measured 43 mm in length and 1 mm in width, the eter UV–Vis (PerkinElmer, Lambda25). second section measured 67 mm in length and 1.5 mm in width, while the final section measured 17 mm in length and Interference study 2 mm in width. This design helped to facilitate the mixing process and reduce the pressure on the diluter. Each inlet An interference study of Cu(II), Al(III) and Fe(II) has been was connected to a rectangular channel measuring 0.5 mm conducted in order to corroborate that there are no interfer- in width, 0.5 mm in height, and 4 mm in length. The optimal ences caused by these ions that are frequently present in our diameters of the inlet points were 1 mm, while the outlet samples. For this purpose, four aliquots of 22.5 mg·L−1 of point was 1.5 mm. This configuration was found to yield Fe(III) solutions were prepared. Three of them were spiked the best results. The CAD design of the mixer reactor is with Cu(II) reaching a final concentration of 500 mg·L−1 of illustrated in Fig. S4. Cu (solution prepared from ­CuSO4·5H2O), 500 mg·L -1 of Al(III) (from ­Al2(SO4)3·18H2O) and 100 mg·L−1 Fe(II) (from ­FeSO4·7H2O), respectively. The absorbances at 525 nm of Mixer Fe(III) solution and the spiked ones were measured by the salicylic colorimetric method at pH 2.0. Moreover, solutions In the mixer or reactor, the diluted sample underwent a reac- containing only 500 mg·L−1 of Cu, Al and 100 mg·L−1 Fe(II) tion with the chelating reagent, resulting in the formation of a have been also analyzed by the same colorimetry. In the Fe(II) purple chelate that exhibited absorbance at 525 nm. The inten- solution, 10% (w/v) hydroxylamine was added in order to pre- sity of the signal was directly proportional to the concentration vent the oxidation of Fe(II) into Fe(III). The assay was carried of Fe(III) present in the sample. The CAD design of the mixer out in triplicate. reactor is illustrated in Fig. S5. The reactor comprised three separate entrances, with one dedicated to the introduction of the diluted sample, and the remaining two reserved for the Results and discussion introduction of reagents, serving to increase the reagent ratio. The reactor channel was designed with a succession of linear Optimization of microFIA passive mixing elements segments and square turns, which served to improve mixing efficiency [38], culminating in the outlet channel. In order to optimize the performance of microFIA device The width and height of the channels within the micro- described in the experimental section, multiple operational fluidic system were critical to ensure efficient and reliable 13 Microchim Acta (2024) 191:3 Page 7 of 13 3 performance. Channels with a height of 1 mm were used the detector could be accounted for and eliminated, resulting to prevent overpressure and liquid leaks from occurring at in more accurate measurements. In this case, measuring the the connections. The sample microchannel had a width of baseline for 20 s before the chelate measurement helped to 0.5 mm, and connected to the reagent channel at a T-junc- stabilize the baseline and improve the repeatability of the tion, which had a width of 1.0 mm. A second reagent inlet, measurements. The analytical signal was then calculated as also with a width of 1.0 mm, joined the main channel, result- the difference between the signal of the chelate and the sig- ing in a final channel width of 2.0 mm. The large mixing nal of the baseline. This approach helped to minimize any zone within the mixer (as shown in Fig. S5) provided ample interference or noise in the measurement and ensure more time for the sample and reagent to react completely and accurate and reliable results. allowed for an initial dispersion of the chelate obtained to reduce peak height and prevent detector signal saturation. Optimization of operating conditions To ensure optimal mixing efficiency, the mixer featured 76 90-degree turns, with each turn measuring 24 mm in length After obtaining the baseline signal, the diluted sample was and 5.3 mm in width, which generated sufficient turbulence introduced into the system and the analyte was analyzed within the channel. by measuring the signal of the chelate formed between Fe(III) and salicylic acid. The intensity of the signal was Disperser proportional to the Fe(III) concentration in the sample and the signal was measured by the LDR in the 3D-optical The addition of the disperser module was a key improvement micro flow-cell and processed by the Arduino microcon- to the microFIA design, as it allowed for better dispersion troller. The output signal was then sent to a computer for of the chelate, reducing the peak height and avoiding the data analysis. detector signal saturation. The 3D-platform disperser was After that, the injection valve (Fig. 1Ab), in load position, designed to be placed after the mixing module and before was filled (25 µL) with the sample or standard solution and the 3D-optical microflow cell. It consisted of a rectangular introduced in the system by the carrier. The flow rate of dif- serpentine-shaped main channel with smaller rectangular ferent solutions was optimized after several assays and was serpentine segments to increase turbulence and improve che- finally established as follows: carrier 0.33 mL/min; diluent late dispersion. The inlet diameter was 1.5 mm, the outlet 1, 0.40 mL/min; diluent 2, 0.37 mL/min; diluent 3, 0.39 mL/ was 2 mm, and the channel dimensions were 2 mm width min; reagent 1, 0.41 mL/min; and reagent 2, 0.32 mL/min. and 1 mm height, all of which helped to minimize overpres- After each measurement, the detector was washed with sure in the output. Fig. S6 shows the design of the disperser ethanol that was introduced in the disperser with a second module in detail. pump and a 3-way-valve (Fig. 1Ag). The flow rate of etha- nol was 1.36 mL/min. Total analysis and cleaning time was Optimization of detector response 10.1 min (Fig. 1B). To avoid the formation of bubbles that would interfere in the measure, all solutions were previously Based on the results, it was found that the best color to print sonicated. the detector was black PLA as it provided the highest voltage difference between the red dye solution and water, indicating Validation of microanalyzer performance minimal external light interference. Additionally, the dimen- sions of the flow cell were optimized to ensure a good fit of In order to prove the microanalyzer’s suitability, reliability the LED and LDR orientation, resulting in improved detec- and robustness for the continuous monitoring of Fe(III) in tor performance. a bioleaching reactor, the microFIA system was validated. Table S1 shows that the average voltage difference The parameters determined for this purpose were precision between water and dye solution was lower for the transpar- (repeatability and reproducibility), limit of detection and ent and white flow-cells, indicating higher interference due limit of quantification, linear response range, accuracy and to external light. The black flow-cell showed the highest real sample applications. voltage difference, indicating the least interference due to external light. This was because the black PLA absorbed all Precision external light and only allowed the light from the flow-cell to reach the LDR. Therefore, the black flow-cell was chosen Repeatability The precision expressed as repeatability was as the optimal color for printing the detector. determined by analyzing standard solutions with differ- The baseline was the signal obtained when only the sol- ent concentrations of Fe(III) several times. The closeness vent or reagent flowed through the detector without any sam- between the multiple obtained results indicates the repeat- ple. By measuring the baseline, any fluctuation or drift in ability of the method. The precision study was performed 13 3 Page 8 of 13 Microchim Acta (2024) 191:3 in a wide range of Fe(III) concentrations (150, 1500, 3000, coefficients were significantly lower than the maximum 4500 and 6000 mg·L−1). Every concentration was analyzed acceptable value given by Horwitz for each concentra- in three different days in triplicate (n = 9). For each concen- tion. Thus, the response of microanalyzer in the range of tration, the mean, standard deviation and the experimental concentration between 150 and 6000 mg·L−1 of Fe(III) coefficient of variation (CV) were calculated. The experi- was repeatable. Figure 3 shows the peaks of the calibra- mental CV obtained was compared to Horwitz coefficient tion points (Fig. 3A), blank (Fig. 3B), and calibration ­(Cvh) [39]. The ­Cvh = ­2(1–0.5·log(C)), where c is the concentra- plot (Fig. 3C). tion of analyte in g/mL. When experimental CV < ­Cvh, it means the precision is acceptable. Table S2 shows analytical Reproducibility In this study, the reproducibility was studied results for each concentration, the average, standard devia- in terms of the variation of the sensitivity of the system over tion, and coefficient of variation as well as the coefficient time. Sensitivity is represented by the slope of the calibra- of Horwitz. tion curve. Thus, the calibration of the microFIA system on As Table S2 shows, the CV was smaller than 2% for successive days (n = 7) under the optimal operational condi- almost all concentrations of Fe(III), except for the small- tions was studied. With an average sensitivity of 0.000634 V est one of 150 mg·L −1. In any case, the experimental L/mg and a standard deviation of 0.000020 V L/mg, the Fig. 3  Calibration performed in the microFIA at optimal conditions. n=5 in triplicate). Operational conditions are injection volume 25 µL, (A) Analytical signal (λ = 525 nm) at different concentrations, (B) carrier flow rate 0.33 mL/min, diluent flow rate 0.39 mL/min, reagent blank and (C) calibration plot. The equation of the calibration curve flow rate 0.36 mL/min, and the ethanol flow rate 1.36 mL/min is Voltage = 0.00063 (± 0.00002) [Fe(III)]—0.12 (± 0.09) (r2 = 0.996; 13 Microchim Acta (2024) 191:3 Page 9 of 13 3 response of the microFIA in successive calibrations was Table 1  Accuracy results of samples from bioleaching reactor using reproducible (Table S3). An acceptable relative standard microFIA system deviation (RSD) of sensitivity of 3.1% (less than 5%) was Samples Spiked samples obtained. Sample Average SD Average SD Recovery (%) (mg·L-1) (mg·L-1) Limit of detection and limit of quantification Sample 1 636 8 2115 60 98.64% The limit of detection (LoD) and limit of quantifica- Sample 2 1296 19 2806 90 100.67% tion (LoQ) were determined by signal-to-noise approach, Sample 3 3157 11 4673 30 101.08% a method suitable for analytical procedures that exhibit Sample 4 4517 50 6012 8 99.46% baseline noise. The signal-to-noise ratio was carried out by blank analysis (Fig. 3B). To determine the LoD and LoQ, 278 measurements of baseline from blanks were concentrations between 500 and 6000 mg·L−1. A recovery performed. The mean and standard deviation values were of around 100% in the analyzed samples was obtained. The then calculated according to [40]: LoD = (X + 3*Sb)/S and wide range of concentrations in which it can be applied and LoQ = (X + 10*Sb)/S, where Sb is the standard deviation of the additional advantage of automatically performing the the blank mean and S is the slope of the calibration curve dilution of the sample are noteworthy. or sensitivity. From this method, the limit of detection was 11 mg·L−1 and the limit of quantification was 25 mg·L−1. Interference study Linear response range The determination of Fe(III) by colorimetry using salicylic acid as a complexing agent may experience some interfer- The linearity of an analytical method expresses the rela- ence. According to [41], some anions (arsenate, citrate, tionship between analyte concentration in the sample and dichromate, cyanide, thiocyanate, orto- and pyrophosphate, the analytical signal obtained (in this case, height peak in fluoride, iodide, tungstate) and metallic ions (Al(III), Cu(II), volts). The linear response range of an analytical method Ba(II), Be(II), Bi(III), Cr(VI), Co(II), Mn(II), Hg(I), Sn(II), is the interval between the upper (limit superior of quan- Sn(IV) and Th(IV)) can interfere the determination of Fe(III) tification) and LoQ. In this study, linearity of the designed with salicylic. In the samples for which the analyzer has analyzer was determined by a series of 3 analyses of 5 dif- been designed (bioleachates from PCBs), these chemical ferent Fe(III) standard solutions whose concentrations were species that could cause interference are not found, with the between 25 and 6000 mg·L−1. From the obtained results, exception of Al(III), Cu(II) and Fe(II). However, it should the calibration plot by the method of least squares was cal- be noted [41] that the interferences occur at a pH greater culated. The r2 was 0.996 provided that pH was under 2.0. than 2.5. According to some authors, these ions don’t cause The calibration plot is shown at Fig. 3C. This linearity range interferences in the determination of Fe(III) with salicylic is very wide, higher than those described in the literature acid at a very low pH = 1.8–2.0 that are precisely our condi- [29, 32–36]. tions [42, 43]. Also it has been demonstrated that ­Hg2+, ­Fe2+, ­Pb2+, Accuracy ­C d2+, ­Na +, ­K +, ­Z n 2+, ­C a 2+, ­M g 2+, ­S n 2+, ­NO 3−, ­NO 2−, ­CO32−, ­SCN−, oxalate, and citrates do not interfere with Accuracy, which indicates the degree of agreement between Fe(III)-salicylic under high acidic conditions [44]. the experimental data and the true value, can be expressed Regarding Fe(II) interference, according to [37], as recovery (%) of an added standard in the sample. Here, 100 mg·L−1 of Fe(II) also does not interfere in the deter- recovery was determined by spiking a sample from a mination of Fe(III) with salicylic acid as a chelant under bioleaching reactor containing different Fe(III) concen- acidic conditions. In fact, to determine Fe(II) and Fe(III) trations (150 to 6000 mg·L -1) with a standard solution of simultaneously using the sulfosalicylic acid, Fe(III) is first 1500 mg·L−1 Fe(III). All samples were analyzed 3 times determined under acidic conditions, and then the medium is before and after being spiked. made alkaline, and under these alkaline conditions, total iron The recovery percentage was calculated as % R = (|Y (Fe(II) + Fe (III)) is then determined [18]. -X|/1500)·100, being X the Fe(III) concentration before spik- In the present study, the potential interference of Cu(II), ing and Y after standard addition of 1500 mg·L−1. Results Al(III), and Fe(II) ions in the determination of Fe(III) was for the microFIA analyzer are shown in Table 1. investigated. The results are displayed in Fig. 4. As it can be As can be seen in Table 1, the microFIA method seen, Cu(II), Al(III) and Fe(II) did not produce interference developed in the present work is accurate in the range in the determination of Fe(III) under our working conditions 13 3 Page 10 of 13 Microchim Acta (2024) 191:3 over other automatic analysis systems are the wide linear response range and the minimization of interferences it pre- sents. The low LoD (11 mg·L−1) obtained by our system is also surprising comparable to that of other microFIA sys- tems [35] (20 mg·L−1) which have a much narrower work- ing range (40–305 mg·L−1) and more interferences such as Cu(II). Kim et al. [11] present a very sensitive microfluidic system based on fluorescence measurements with a very small application range (0–12 mg·L−1) but with interference from Al(III). While other FIA systems [36] have a wider dynamic range (110–560 mg·L−1), they also have a higher LoD (110 mg·L−1) and a high selectivity. It is important to highlight the high accuracy (99.9%) and precision (3.1%) of this analyzer comparable to those of other more sensitive Fig. 4  Interference study. Comparative absorbance (λ = 525 nm) for microFIA systems [32, 33, 35]. On the other hand, the high Fe(III) and mixtures of Fe(III) and potentially interfering ions analysis frequency of classical flow systems such as FIA [45] and SIA [47] between 25 and 102 ­h−1, respectively, is not feasible for our system. The high degree of dilution (pH = 1.8–2.0), even at concentrations of 500 mg·L−1 for required to extend the linear response range causes the inte- Cu(II) and Al(III), and 100 mg·L−1 for Fe(II). On the other gration of multiple mixing and dilution stages that reduce hand, the absorbances shown in the Table S4, indicate that the analysis frequency to 6 ­h−1. the absorbance of these potential interferent ions using sali- cylic acid as a chelant is negligible under acidic conditions. Real sample applications The results of the characterization study are presented in Table 2, which summarizes all the quality parameters of The microFIA system was evaluated for its applicability by the method, including accuracy and precision (repeatability analyzing several real samples with increasing matrix com- and reproducibility). The sensitivity of the method is suit- plexity in a final assessment. The first sample was water able for application in bioreactors and allows the wide linear from the Agulla reservoir in Manresa, Spain, which has response range from 25 to 6000 mg·L−1 Fe(III). On the other a complex composition and is rich in organic matter. The hand, the reproducibility of the method, determined as the second sample was the growth medium typically used for relative standard deviation of the sensitivities on different Acidithiobacillus ferrooxidans, the microorganism used operation days (RSD = 3.1%), ensures the proper response in bioleaching process. The third sample was a leachate of the analyzer for process monitoring. obtained from acid leaching of the black mass from ion-Li In Table S5, the characteristics of other analytical meth- mobile batteries, while the last sample was a bioleachate ods based on different reagents and automation techniques from the bioleaching reactor, which was the only sample such as FIA [36, 45, 46], SIA [19, 47] and microsystems [11, containing naturally occurring Fe(III). 32, 33, 35] are shown. The main advantages of our analyzer A total of 12 real and spiked samples were analyzed by the developed microFIA system in triplicate for several days. Table 2  Principal parameters of quality of the automatic microFIA To compare the results, they were also analyzed by a UV–vis system developed for determination of Fe(III) conventional method. Table 3 presents the analytical results obtained from both methods. To evaluate the method statisti- Parameter Value cally, a paired-test and least-square linear regression were Sensitivity (V L ­mg−1) 0.00063 ± 0.00002 used, and no significant differences were observed at a 95% Detection limit (mg·L−1) 11 confidence level. In the paired test: ­tcal = 0.91 < ­ttab = 2.07, Quantification limit (mg·L−1) 25 while in the least square linear regression the slope and inter- Linear range (mg·L−1) 25–6000 cept were 1.02 ± 0.02 and 21 ± 43, respectively (r2 = 0.996; Repeatability (CV) (n = 9) 0.69–4.92% n = 12; 95% confidence level). Reproducibility (RSD) (n = 7) 3.1% To showcase the microFIA analyzer’s capability to con- Selectivity At pH acids not tinuously and automatically monitor Fe(III) concentration interference was in an in situ bioreactor, the system was integrated with a observed reactor that was oxidizing copper from electrical cables by Sample throughput (­ h−1) 6 converting Fe(III) to Fe(II). The microFIA analyzer was able Accuracy (RV%) 99.9 ± 1.1 13 Microchim Acta (2024) 191:3 Page 11 of 13 3 Table 3  Results from the microFIA system of the different samples Conclusions compared with the results from UV–Vis method FIA UV–Vis The microfluidic system designed and manufactured in this [Fe(III)] added Average SD Average SD work has proven to be a highly suitable instrument for con- (mg·L−1) (mg·L−1) (mg·L−1) tinuously and automatically determining the Fe(III) concen- tration in a bioreactor in a wide concentration range, from Bioleaching reactor - 636 8 713.8 1.1 25 to 6000 mg·L−1. - 3157 11 2962 70 The system was successfully validated in terms of preci- - 2115 60 2120 5 sion (repeatability and reproducibility), limit of detection Water from lake in 300 377 6 249 4 and limit of quantification, linear response range, accuracy, “Parc de l’agulla” 600 608 8 518.7 0.6 and real sample applications. Repeatability was less than 5% 1000 915 24 834.1 1.6 coefficient of variation throughout the concentration range, Mineral medium 1050 1042 11 924.0 0.6 and accuracy was around 100%. 2000 1941 60 1930 4 The effectiveness of the system was also evaluated for its 4050 4289 40 4211 13 applicability by analyzing four real samples with different Leachate from 500 428 6 402 4 matrix complexities and no significant statistical differences dissolved mobile 2500 2267 90 2312 24 batteries in acid were found compared to a conventional UV–Vis method. 3500 3444 40 3340 6 The use of the 3D technology for manufacturing the proposed device allows the creation of different pieces in a quickly, precise, and independent way. The technique is to measure the decreasing Fe(III) concentration in real time, highly adaptable and permits making changes on the fly in with measurements programmed every 20 min. an easy and cost-effective manner. To validate the microFIA analyzer’s suitability for this Taken together, the results demonstrate several advan- purpose, the results obtained were compared against those tages of combining miniaturization and continuous flow acquired through batch measurements using UV–Vis analy- techniques in a symbiotic association to achieve on-line sis. The results of this comparison are presented in Fig. 5. monitoring of high concentrations of Fe(III) in complex The results illustrate that the microFIA system is capable sample matrices. These advantages include lower cost due of on-line Fe(III) monitoring in real-time, without the need to reduced sample and reagent volumes, less maintenance, for any previous sample pretreatment, with programmed lower waste disposal, and lower personnel costs because the measurements taken automatically every 20 min. Fur- system is suitable for unattended operation over extended thermore, the system was able to perform this monitoring periods of time. smoothly and demonstrated no significant differences when The FIA technique is an open approach to automate anal- compared to the UV–vis method. ysis processes. Combining FIA with 3D-printed platforms enables the adaptation of analyses based on the formation of coloured products. The determination of other metals or ions of interest by the formation of coloured complexes can be implemented by modifying the wavelength of the LED in the flow cell and some characteristics of the modules. This study provides a novel strategy for future work, making it easier to extend to other target analytes based on colorimet- ric reactions. Supplementary Information The online version contains supplemen- tary material available at https://d​ oi.o​ rg/1​ 0.1​ 007/s​ 00604-0​ 23-0​ 6029-x. Funding Open Access Funding provided by Universitat Autonoma de Barcelona. Project BIOMETAL PID2020-117520RA-I00 funded by MCIN/AEI/10.13039/501100011033, BIOCOLI ACE034/21/000044 funded by ACCIÓ. The authors gratefully acknowledge the UPC for the financial support of a predoctoral grant FPU-UPC, with the col- laboration of Banco de Santander. Fig. 5  Comparison between online measurement using new microFIA system (n = 1) and batch determination UV–Vis method (n = 3). Exper- Data availability The data that support the findings of this study are imental uncertainty is determined for error interpolation available from the corresponding author upon reasonable request. 13 3 Page 12 of 13 Microchim Acta (2024) 191:3 Declarations chips. Sensors Actuators, B Chem 224:404–412. https://​doi.​org/​ 10.​1016/j.​snb.​2015.​10.​058 Conflict of interest The authors declare no competing interests. 12. Desai ML, Basu H, Saha S et al (2019) Investigation of silicon doping into carbon dots for improved fluorescence properties for Open Access This article is licensed under a Creative Commons Attri- selective detection of Fe3+ ion. Opt Mater (Amst) 96:109374. bution 4.0 International License, which permits use, sharing, adapta- https://​doi.​org/​10.​1016/j.​optmat.​2019.​109374 tion, distribution and reproduction in any medium or format, as long 13. Abualhaija MM, van den Berg CMG (2014) Chemical speciation as you give appropriate credit to the original author(s) and the source, of iron in seawater using catalytic cathodic stripping voltamme- provide a link to the Creative Commons licence, and indicate if changes try with ligand competition against salicylaldoxime. Mar Chem were made. The images or other third party material in this article are 164:60–74. https://​doi.​org/​10.​1016/j.​march​em.​2014.​06.​005 included in the article’s Creative Commons licence, unless indicated 14. Obata H, Karatani H, Nakayama E (1993) Automated determi- otherwise in a credit line to the material. If material is not included in nation of iron in seawater by chelating resin concentration and the article’s Creative Commons licence and your intended use is not chemiluminescence detection. Anal Chem 65:1524–1528. https://​ permitted by statutory regulation or exceeds the permitted use, you will doi.​org/​10.​1021/​ac000​59a007 need to obtain permission directly from the copyright holder. To view a 15. Harvey AE, Smart JA, Amis ES (1955) Simultaneous spectropho- copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. tometric determination of iron(II) and total iron with 1,10-phen- anthroline. Anal Chem 27:27 16. Schilt AA (1969) Analytical applications of 1, 10-phenanthroline and related compounds. Elsevier Science & Technology, Oxford 17. McMahon JW (1969) An acid-free bathophenanthroline method References for measuring dissolved ferrous iron in lake water. Water Res 3:743–748. https://​doi.​org/​10.​1016/​0043-​1354(69)​90038-4 18. Karamanev D, Nikolov L, Mamatarkova V (2002) Rapid simul- 1. Laurmaa V, Kers J, Tall K et al (2011) Mechanical recycling of taneous quantitative determination of ferric and ferrous ions in electronic wastes for materials recovery. In: Recycling of elec- drainage. Miner Eng - Min ENG 15:341–346. https://​doi.​org/​10.​ tronic waste II. John Wiley & Sons, Ltd, pp 1–10 1016/​S0892-​6875(02)​00026-2 2. Dorado AD, Solé M, Lao C et al (2012) Effect of pH and Fe(III) 19. Pragourpun K, Sakee U, Fernandez C, Kruanetr S (2015) Defer- ions on chalcopyrite bioleaching by an adapted consortium from iprone, a non-toxic reagent for determination of iron in samples biogas sweetening. Miner Eng 39:36–38. https://​doi.​org/​10.​ via sequential injection analysis. Spectrochim Acta - Part A Mol 1016/j.​mineng.​2012.​06.​009 Biomol Spectrosc 142:110–117. https://​doi.​org/​10.​1016/j.​saa.​ 3. Dorado Castaño AD, Gamisans Noguera X, Solé Sardans M et al 2015.​01.​081 (2019) Method for the biological recovery of metals in electric 20. Kim YS, Lee JJ, Lee SY et al (2016) A highly sensitive benzimi- and electronic waste (Spain, WO2019/206755A1). Universitat dazole-based chemosensor for the colorimetric detection of Fe(ii) Politècnica de Catalunya. https://​paten​ts.​google.​com/​patent/​ and Fe(iii) and the fluorometric detection of Zn(ii) in aqueous WO201​92067​55A1/​en media. RSC Adv 6:61505–61515. https://d​ oi.o​ rg/1​ 0.1​ 039/C​ 6RA1​ 4. Brierley CL (1999) Bacterial succession in bioheap leaching. 0086D Process Metall 9:91–97. https://​doi.​org/​10.​1016/​S1572-​4409(99)​ 21. Li Z, Ti M, Yang K et al (2019) Colorimetric detection of iron (II) 80008-X using 4-mercaptobenzoic acid and 3,3′,5,5′-tetramethylbenzidine 5. Benzal E, Solé M, Lao C et al (2020) Elemental copper recov- modified silver nanoparticles. Microchem J 151:104254. https://​ ery from e-wastes mediated with a two-step bioleaching process. doi.​org/​10.​1016/j.​microc.​2019.​104254 Waste Biomass Valorization 11:5457–5465. https://​doi.​org/​10.​ 22. Bhagat B, Jadeja V, Sharma P et al (2020) Case study on the use of 1007/​s12649-​020-​01040-2 image analysis for the simple and inexpensive colorimetric detec- 6. Benzal E, Cano A, Solé M et al (2020) Copper recovery from tion of Fe(III) in water. Anal Methods 12:4509–4516. https://​doi.​ PCBs by Acidithiobacillus ferrooxidans: toxicity of bioleached org/​10.​1039/​D0AY0​1212B metals on biological activity. Waste Biomass Valorization 23. Huyen VuT, Serradji N, Seydou M et al (2020) Electronic spec- 11:5483–5492. https://​doi.​org/​10.​1007/​s12649-​020-​01036-y troscopic characterization of the formation of iron(III) metal 7. Zhan Y, Yang M, Zhang S et al (2019) Iron and sulfur oxidation complexes: the 8-hydroxyquinoline as ligand case study. J Inorg pathways of Acidithiobacillus ferrooxidans. World J Microbiol Biochem 203:110864. https://​doi.​org/​10.​1016/j.​jinor​gbio.​2019.​ Biotechnol 35:60. https://​doi.​org/​10.​1007/​s11274-​019-​2632-y 110864 8. de Costa RC, C, Araújo AN, (2001) Determination of Fe(III) and 24. Koç ÖK, Üzer A, Apak R (2020) A colorimetric probe based on total Fe in wines by sequential injection analysis and flame atomic 4-mercaptophenol and thioglycolic acid-functionalized gold nano- absorption spectrometry. Anal Chim Acta 438:227–233. https://​ particles for determination of phytic acid and Fe(III) ions. Micro- doi.​org/​10.​1016/​S0003-​2670(01)​00845-5 chim Acta 187:586. https://​doi.​org/​10.​1007/​s00604-​020-​04478-2 9. Yan X-P, Hendry MJ, Kerrich R (2000) Speciation of dissolved 25. Ho TT-T, Dang C-H, Huynh TK-C et al (2021) In situ synthesis iron(III) and iron(II) in water by on-line coupling of flow injection of gold nanoparticles on novel nanocomposite lactose/alginate: separation and preconcentration with inductively coupled plasma recyclable catalysis and colorimetric detection of Fe(III). Carbo- mass spectrometry. Anal Chem 72:1879–1884. https://​doi.​org/​10.​ hydr Polym 251:116998. https://​doi.​org/​10.​1016/j.​carbp​ol.​2020.​ 1021/​ac990​9655 116998 10. Yokoi K, Kimura M, Itokawa Y (1991) Determination of nonheme 26. Sasan S, Chopra T, Gupta A et al (2022) Fluorescence “turn-off” iron using inductively coupled plasma-atomic emission spectrom- and colorimetric sensor for Fe2+, Fe3+, and Cu2+ions based on etry. Biol Trace Elem Res 31:265–279. https://​doi.​org/​10.​1007/​ a 2,5,7-triarylimidazopyridine scaffold. ACS Omega 7:11114– BF029​90196 11125. https://​doi.​org/​10.​1021/​acsom​ega.​1c071​93 11. Kim H, Rao BA, Jeong J et al (2016) A rhodamine scaffold immo- 27. Verschoor MJ, Molot LA (2013) A comparison of three colori- bilized onto mesoporous silica as a fluorescent probe for the detec- metric methods of ferrous and total reactive iron measurement in tion of Fe (III) and applications in bio-imaging and microfluidic freshwaters. Limnol Oceanogr Methods 11:113–125. https://​doi.​ org/​10.​4319/​lom.​2013.​11.​113 13 Microchim Acta (2024) 191:3 Page 13 of 13 3 28. Achterberg EP, Ussher S (2002) Real-time monitoring of picomo- rectangular meso (micro) channels with pressure-driven flows. lar concentrations of iron ( II ) in marine waters using automated Anal Bioanal Chem 393:1233–1243. https://​doi.​org/​10.​1007/​ flow instrumentation. Environ Sci Technol 36:4600–4607. https://​ s00216-​008-​2532-8 doi.​org/​10.​1021/​es020​045v 39. Horwitz W (1982) Evaluation of analytical methods used for regu- 29. Oungvises NY, Hanurak PT, Haida TC et al (2015) Double-sided lation of foods and drugs. Anal Chem 54:67–76. https://​doi.​org/​ microfluidic device for speciation analysis of iron in water sam- 10.​1021/​ac002​38a002 ples : towards greener analytical chemistry. Anal Sci 31:365–370 40. McNaught AD, Wilkinson A (1997) Compendium of chemical 30. Barzallo D, Danchana K, Cerdà V, Palacio E (2020) Design of terminology, 2nd edn. Blackwell Science, Oxford an automatic spectrophotometric system. Talanta 218:121163. 41. Mehlig JP (1938) Colorimetric determination of iron with sali- https://​doi.​org/​10.​1016/j.​talan​ta.​2020.​121163 cylic acid: a spectrophotometric study. Ind Eng Chem - Anal Ed 31. Saez J, Catalan-Carrio R, Owens RM et al (2021) Microfluidics 10:136–139. https://​doi.​org/​10.​1021/​ac501​19a007 and materials for smart water monitoring: a review. Anal Chim 42. Paipa C, Poblete E, Toral MI (2006) Simultaneous determina- Acta. https://​doi.​org/​10.​1016/j.​aca.​2021.​338392 tion of iron and copper in pregnant liquid solutions. Miner Eng 32. Kruanetr S, Liawruangrath S, Youngvises N (2007) A simple and 19:1465–1468. https://​doi.​org/​10.​1016/j.​mineng.​2006.​04.​004 green analytical method for determination of iron based on micro 43. Morales J (2018) Determinación de Fe III y/o Fe total en Aguas flow analysis. Talanta 73:46–53. https://​doi.​org/​10.​1016/j.​talan​ta.​ Potables y Naturales por Medio de Extracción en Fase Sólida. 2007.​02.​032 Final degree project, Universidad de Chile. https://​repos​itorio.​ 33. Prasertboonyai K, Arqueropanyo O, Liawraungrath B (2015) Min- uchile.​cl/​handle/​2250/​152933 iaturization of spectrophotometry based on micro flow analysis 44. Rajendraprasad N, Basavaiah K (2016) Modified spectropho- using norfloxacin as less-toxic reagent for iron determination. tometric methods for determination of iron(III) in leaves and Spectrochim ACTA PART A Mol Biomol Spectrosc 151:532– pharmaceuticals using salicylic acid. Indian J Adv Chem Sci 537. https://​doi.​org/​10.​1016/j.​saa.​2015.​07.​017 4:302–307 34. Calderilla C, Maya F, Cerdà V, Leal LO (2017) 3D printed device 45. Lunvongsa S, Tsuboi T, Motomizu S (2006) Sequential determina- including disk-based solid-phase extraction for the automated tion of trace amounts of iron and copper in water samples by flow speciation of iron using the multisyringe flow injection analysis injection analysis with catalytic spectrophotometric detection. technique. Talanta 175:463–469. https://​doi.​org/​10.​1016/j.​talan​ Anal Sci 22:169–172. https://​doi.​org/​10.​2116/​anals​ci.​22.​169 ta.​2017.​07.​028 46. Jaikrajang N, Kruanetr S, Harding DJ, Rattanakit P (2018) A 35. Ogawa K, Kaneta T (2016) Determination of iron ion in the water simple flow injection spectrophotometric procedure for iron(III) of a natural hot spring using microfluidic paper-based analytical determination using Phyllanthus emblica Linn. as a natural rea- devices. Anal Sci 32:31–34. https://d​ oi.o​ rg/1​ 0.​2116/​analsc​ i.​32.​31 gent. Spectrochim Acta - Part A Mol Biomol Spectrosc 204:726– 36. Petdum A, Waraeksiri N, Hanmeng O et al (2020) A new 734. https://​doi.​org/​10.​1016/j.​saa.​2018.​06.​109 water-soluble Fe3+ fluorescence sensor with a large Stokes 47. Mesquita RBR, Suárez R, Cerdà V et al (2013) Exploiting the use shift based on [5]helicene derivative: its application in flow of 3,4-HPO ligands as nontoxic reagents for the determination of injection analysis and biological systems. J Photochem Pho- iron in natural waters with a sequential injection approach. Talanta tobiol A Chem 401:112769. https://​d oi.​o rg/​1 0.​1 016/j.​j phot​ 108:38–45. https://​doi.​org/​10.​1016/j.​talan​ta.​2013.​02.​058 ochem.​2020.​112769 37. Ogawa K, Tobe N (1966) A spectrophotometric study of the com- Publisher's Note Springer Nature remains neutral with regard to plex formation between iron(III) and salicylic acid. Bull Chem jurisdictional claims in published maps and institutional affiliations. Soc Jpn 39:227–232. https://​doi.​org/​10.​1246/​bcsj.​39.​227 38. Baeza M, Montesinos JL, Alonso J, Bartrolí J (2009) Sim- ple modeling of the physical sample dispersion process in 13