Microchimica Acta (2023) 190:400 https://doi.org/10.1007/s00604-023-05975-w ORIGINAL PAPER Live synthesis of selective carbon dots as fluorescent probes for cobalt determination in water with an automatic microanalyzer Alex Pascual‑Esco1 · Pere Lleonart1 · Antonio Calvo‑López1 · Julián Alonso‑Chamarro1 · Mar Puyol1 Received: 13 June 2023 / Accepted: 28 August 2023 / Published online: 19 September 2023 © The Author(s) 2023 Abstract A new strategy integrating the straight synthesis of carbon dots (CDs) and their direct use for the determination of heavy metals by means of fluorescence quenching is presented. The proposal consists of a modular analyzer, which includes a low temperature co-fired ceramics (LTCC) microreactor for the synthesis of CDs and a cyclic olefin copolymer (COC) microflu- idic platform, which automatically performs a reverse flow injection analysis (rFIA) protocol for the determination of heavy metal ions in water by CD fluorescence quenching. As a proof of concept, nitrogen-doped CDs were synthesized from acrylic acid and ethylenediamine (ED) with quantum yields (QYs) of up to 44%, which are selective to cobalt. With the described system, we synthesized homogeneous CDs without the need for further purification and with the minimum consumption of reagents, and optimized fluorescence measurements can be performed with freshly obtained luminescent nanomaterials that have not undergone decomposition processes. They have an average hydrodynamic diameter of 4.2 ± 0.9 nm and maxi- mum excitation and emission wavelengths at 358 nm and 452 nm, respectively. The system allows the automatic dilution and buffering of the synthesized CDs and the sample prior to the determination of cobalt. The concentration of cobalt was determined with good sensitivity and a limit of detection of 7 μg·L−1 with a linear range of 0.02–1 mg·L−1 of ­Co2+. Spiked tap water and river water samples were analyzed, obtaining recovery from 98 to 104%. This demonstrates the potential of the equipment as an efficient on-site control system for heavy metal monitoring in water. Keywords Microreactor · Carbon dots · Fluorescence quenching · Microfluidics · Cobalt determination Introduction (ICP-MS), inductively coupled plasma optical emission spectrometry (ICP-OES), and atomic absorption spectrom- Environmental pollution is receiving more attention in the etry (AAS) [6, 7]. These techniques offer high analytical last decades. One of the most critical polluting agents are features in terms of sensitivity, accuracy, and precision but heavy metals, considering that they are not biodegradable show some disadvantages for on-site monitoring because and tend to accumulate in the ecosystem and different tissues of the complexity and dimensions of the instrumentation, of living beings [1]. Different heavy metals can be found nat- which hinders the possibility of carrying out measurements urally in the environment, but the rise of some human activi- outside a laboratory and the need for specialized personnel ties like mining, manufacturing, smelting, and the improper [4]. In this sense, the miniaturization of analytical proce- disposal of industrial wastes is contributing to an increase dures through microfluidic devices solves some of the men- in their negative environmental and health impact [2–5]. tioned problems, increasing the portability and enabling the The most common techniques used to analyze heavy met- on-site continuous monitoring of a wide variety of analytes, als include inductively coupled plasma mass spectrometry including heavy metals [5, 8, 9]. Currently, with the fast expansion of nanotechnology, different photoluminescent nanoparticles have been inves- * Mar Puyol tigated due to their potential applicability in sensing [10] mariadelmar.puyol@uab.cat and bioimaging [11], to list just a few fields. Their use as 1 Group of Sensors and Biosensors, Department fluorescent probes for heavy metals analysis has proven to of Chemistry, Faculty of Sciences, Universitat Autònoma lead to better sensitivity and selectivity, improved detec- de Barcelona, Carrer dels Til·lers s/n, Bellaterra, tion limits, and faster response time [3, 12]. One type of 08193 Cerdanyola del Vallès, Spain 13 Vol.:(0123456789) 400 Page 2 of 11 Microchim Acta (2023) 190:400 photoluminescent nanoparticles that have recently drawn the synthesis processes. As an example of the integration of great attention from the scientific community are carbon the optical probe synthesis and the water quality parameter dots (CDs). They are carbon nanoparticles with sizes below analysis in a single unit, the microreactor was optimized to 10 nm [13] and have different characteristics to highlight synthesize N-doped CDs, which were selective to cobalt(II), as optical probes like their distinctive photoluminescence from two precursors, acrylic acid and ethylenediamine (ED) (strong fluorescence with tunable emission) [14] and good [27]. Therefore, the analytical microsystem was optimized to biocompatibility [15]. Compared to other luminescent mol- determine cobalt(II) in water samples, taking advantage of ecules, CDs do not suffer from photobleaching, which is its fluorescence quenching effect on the CDs. Cobalt, which one of the major drawbacks of the formers, and compared at trace levels is an essential element in the human body to other luminescent nanomaterials such as quantum dots [28], is harmful at high concentrations, causing asthma, rhi- (QDs), CDs are water soluble directly from a one-pot syn- nitis, gastritis, and, in severe cases, cardiomyopathy [28, 29]. thesis [13], so it is not necessary to proceed with solubiliza- Although it is not a heavy metal that causes great concern, tion steps before its use. Due to their carbonaceous nature, its determination in water served us perfectly as a model to they present lower toxicity than QDs [16, 17], which is in validate our proposal. accordance with green chemistry. Finally, CDs have been successfully used to selectively detect different heavy metal ions in water by easy methods of preparation and function- alization [18, 19]. CDs can be easily modified by element Experimental doping [1]. More specifically, nitrogen doping originates an increase in fluorescence intensity of CDs, caused by changes Reagents, materials, and preparation of precursors in the electronic structure, and also leads to the formation of active sites [2, 20], which are useful for further surface All reagents, namely acrylic acid (99%), ethylenediamine functionalization in order to provide them with selectivity. (99%), citric acid (99%), sodium citrate tribasic (99%), However, the main bottom-up synthetic methods of CDs quinine sulfate (90%), sulfuric acid (95%), and the metal (hydrothermal synthesis and microwave-assisted synthesis) salts Co(NO3)2 · ­6H2O, Hg(NO3)2 · ­H2O, Pb(NO3)2, ­FeCl3 · [1, 21] lack the required reproducibility to produce materials ­6H2O, Cu(NO3)2 · ­3H2O, Ni(NO3)2 · ­6H2O, Cd(NO3)2 · for analytical applications [21], which will directly impact ­4H2O, Zn(NO3)2 · ­6H2O, ­NaNO3, ­CaCl2 · ­2H2O, ­MgCl2 · the reliability of the final analytical result. ­6H2O, and Cr(NO3)3 · ­9H2O (> 98% for all the metal salts) In order to face this issue, microreactor technology arises were supplied from Merck Sigma-Aldrich (Barcelona, for synthesis process intensification [22–24]. It allows for Spain) (https://​www.​sigma​aldri​ch.​com/​ES/​en). enhancing reagents mixing and heat and mass transfer effi- Solutions of the precursors for the CD synthesis and ciency. In consequence, better control over the particle size 0.1 M citric/citrate (pH 4) buffer were prepared in MilliQ distribution and a reduction in the consumption of energy water. The solutions used for the CD characterization, selec- and reagents are achieved. From the different substrate mate- tivity, and cobalt ion determination were prepared in the rials for microreactor fabrication, low temperature co-fired mentioned buffer. ceramics (LTCC) technology has some advantages like the For the different syntheses of N-doped CDs, 1.6 mL capability to withstand harsh temperature and pressure con- of acrylic acid (99%, density 1.05 g·mL−1) was dissolved ditions, chemical inertness, and compatibility with screen in 10 mL of MilliQ water (2.33 M) and loaded in a glass printing technology, which enables the easy integration of syringe (Hamilton series Gastight 1000 TLL). Three dif- electrical components [11, 25, 26]. ferent concentrations of the nitrogen source precursor were Taking this context into account, our proposal was the prepared: 0.5 mL, 1.0 mL, and 1.5 mL of ED (99%, density development of a modular automatic analyzer consisting of 0.9 g·mL−1) were dissolved in 20 mL of MilliQ water in a cyclic olefin copolymer (COC) microfluidic system, which each case (0.37, 0.75, and 1.12 M, respectively). For each was optimized for the determination of heavy metals in synthesis process, two glass syringes were loaded with the water, and an LTCC microreactor for the direct synthesis of same nitrogen source precursor. CDs and their use as fluorescent probes, without the need of LTCC 951 green tapes with various thicknesses were any purification steps. The synthetic strategy used, an adap- supplied by DuPont Corporation (Wilmington, DE, USA) tation of a hydrothermal batch method, based on the use of a (https://​www.​dupont.​es/) and used to fabricate the microre- microreactor, takes profit of the advantages of a microfluidic actor: 254 μm thick DuPont 951PX green tapes and 114 μm continuous strategy that allows much better control of the thick DuPont 951PT green tapes. DuPont 5742 gold co- chemical variables of the reaction (minimizing temperature fireable conductor paste was used to print the gold resistor, gradients, increasing the speed of mixing, and controlling and DuPont 6141 silver co-fireable paste was used to print the reaction times), which improves the reproducibility of the contact pads. 13 Microchim Acta (2023) 190:400 Page 3 of 11 400 The microfluidic platform was fabricated with COC this, the total volume of the microreactor channel is approxi- sheets of diverse thicknesses and grades, which were pur- mately 100 μL. chased from TOPAS Advanced Polymers GmbH (Florence, The heating resistor, which occupies the same area as KY, USA) (https://​www.​topas.​com/): 400 µm Topas 5013 the microfluidic channel to reduce energy consumption, was COC and 25 µm Topas 8007 COC layers. screen-printed on the reverse of the fluidic inlets and out- let. Another LTCC green tape was laminated on top of the Fabrication of the low temperature co‑fired ceramics resistor, where the connecting pads were also screen-printed. (LTCC) and cyclic olefin copolymer (COC) microfluidic The device with all the layers laminated was then sintered in platform a CBCWF11/23P16 programmable box furnace (Carbolite Gero, Hope Valley, England) (https://​www.​carbo​lite-​gero.​ The LTCC microreactor was fabricated by a procedure pre- com/) following a two-step thermal profile, including a 1 h viously developed by our research group [11]. Briefly, the organic burnout at 350 °C and 1 h firing at 850 °C. Finally, different layers of the microreactor were designed with com- a PT100 temperature sensor (Innovative Sensor Technol- puter-aided design (CAD) software. The microfluidic chan- ogy, Ebnat-Kappel, Switzerland) (https://​www.​ist-​ag.​com/​ nels and other elements were cut in the LTCC green tapes en) was adhered to the microreactor using EPO-TEK H20E using a Nd:YAG Protolaser 200 (LPKF Laser and Electron- epoxy paste (Epoxy Technology, Billerica, MA, USA) ics, Garbsen, Germany) (https://​www.​lpkf.​com/​en/). Then, (https://​www.​epotek.​com/). The final device (Fig. 1b) has a the processed layers were aligned and thermolaminated with thickness of 3 mm and a diameter of 6 cm. a hydraulic press (Talleres Francisco Camps, Granollers, The COC microfluidic platform that can be seen in Spain) (http://​www.​talle​resca​mp.​com/). Fluidics comprises Fig. 1b is 30 mm wide, 50 mm high, and 2 mm deep and three inlets for the introduction of the synthesis precursors, has two inlets, a two-dimensional meander micromixer (0.8 a spiral-shaped microfluidic channel, and an outlet. The mm wide and 1 mm deep), an optical flow cell (4.5 mm cross-section dimensions of the channel are 525 μm width diameter and 1 mm deep), and an outlet. It was fabricated and 290 μm height, while the total length is 630 mm. With with a multilayered approach in the same way as the LTCC Fig. 1  (a) Experimental setup used, including the synthesis of the with a custom-built aluminum connector) and electrical connections, CDs (part 1), their automated dilution (part 2), and the determination and the custom-made miniaturized optical detection system with an of ­Co2+ (part 3). The fluidic management equipment is also depicted. inserted microfluidic platform (b) Detailed image including the microreactor with fluidics (secured 13 400 Page 4 of 11 Microchim Acta (2023) 190:400 microreactor was. The process was also developed in our Emission spectra were recorded by mixing in a cuvette research group [8, 30], and the first step was the CAD design 0.5 mL of CD dispersion at the optimized dilution factor (to of the layers. Then, the different layers were micromachined have an approximate absorption of 0.05 a.u. and avoid possi- on a Protomat S63 Computer Numerical Control (CNC) ble self-absorption effects) and 2.5 mL of citric/citrate buffer micromilling machine (LPKF Laser and Electronics Garb- as reference emission value and with 2.5 mL of buffered sen, Germany). As for the LTCC microreactor, when micro- solutions containing the mentioned metallic ions. fluidics were integrated into the COC substrate, the differ- ent layers were thermolaminated with the hydraulic press to Experimental setup obtain the final device. The complete system setup (Fig. 1a) is computer-controlled Carbon dots characterization and selectivity test and consists of three main parts: (1) an automatic system for the synthesis of CDs (containing a fluidic management system and a temperature-controlled microreactor), (2) an Absorption spectra of the synthesized CDs were registered interface (including the automated dilution of the CDs), and with a UV-3101PC UV-Vis-NIR double beam spectropho- (3) the determination system of C ­ o2+ (containing a fluidic tometer (Shimadzu, Kyoto, Japan) (https://​www.​shima​dzu.​ management system and the COC microfluidic platform com/), and fluorescence excitation and emission spectra integrated into a miniaturized optical detection system). were obtained with a Fluorolog FL3-11 spectrofluorometer Part 1): For the synthesis of the CDs, three syringes were (Horiba Jobin Yvon, Longjumeau, France) (https://​www.​ filled with the precursors. The syringes were mounted on horiba.​com/​int/​scien​tific/). QY values were determined three NE-500 OEM syringe pumps (New Era Pump Systems by applying the optically dilute measurement method [31] Inc., Farmingdale, NY, USA) (https://​www.​newer​ainst​rumen​ using quinine sulfate in 0.1 M sulfuric acid as a reference ts.​com/) and connected to the microreactor using 0.8 mm as follows: internal diameter Teflon tubing (Tecnyfluor, Barcelona, Spain) (https://​www.​tecny​fluor.​com/). The connections were secured Ix Ast 𝜂x2 with FPM75 O-rings (Epidor, Barcelona, Spain) (https://​ 𝜑x = 𝜑st × × × Ist Ax 𝜂st2 epidor-​srt.​com/) and a custom-built aluminum connector. To improve mixing, acrylic acid (the carbon source) was where φ is the photoluminescence QY, I is the integrated introduced through the central inlet, while ED (the nitrogen area of the corrected emission spectrum, A is the absorb- source) was introduced through the other two inlets, as can ance, and η is the refractive index of the solvent. The subin- be seen in Fig. 1a. The pressure inside the microreactor was dexes x and st refer to the sample and the reference standard, regulated by a back-pressure regulator connected to the outlet. respectively. Considering that both the standard and sample A pressure of 17 bars was applied to all the syntheses, and were dissolved in water, the last term of the equation was three different working temperatures were tested (150 °C, ignored. 170 °C, and 190 °C). The synthesis precursors were pumped Fourier transform infrared (FTIR) spectra of the CDs at a flow rate of 3.33 μL·min−1 for each inlet (total flow rate were acquired with a Tensor 27 FTIR spectrophotometer of 10 μL·min−1), and considering that the internal volume of (Bruker, Billerica, MA, USA) (https://​www.​bruker.​com/​ the microreactor is about 100 μL, the residence time of the en.​html). High-resolution transmission electron micros- reagents is of approximately 10 min. The temperature was copy (HR-TEM) images were collected using a Tecnai G2 controlled by means of a proportional-integral-derivative F20 HR(S)TEM (Field Electron and Ion Company, Hills- (PID) system implemented on a PIC18F4431 microcontroller boro, OR, USA) (https://​w ww.​fei.​c om), and dynamic (Microchip Technology Inc., Chandler, AZ, USA) (https://​ light scattering (DLS) measurements were carried out in www.​micro​chip.​com/), receiving the sensor input [32]. a Zetasizer Nano ZS (Malvern Panalytical, London, Eng- Part 2): CDs from the synthesis were automatically diluted land) (https://​www.​malve​r npan​alyti​cal.​com/​en) to check and buffered at pH 4 with the help of a 161T031 three-way morphology and particle size distribution of the CDs. solenoid valve (NResearch, West Caldwell, NJ, USA) (https://​ To test the selectivity of the CDs, their emission spec- www.​nrese​arch.​com/). They were diluted 100 times with tra in the presence of different heavy metal ion solutions 0.1 M citric/citrate buffer; 0.8 mm internal diameter Teflon ­(Cd2+, ­Co2+, ­Cu2+, ­Cr3+, ­Fe3+, ­Hg2+, ­Ni2+, ­Pb2+, ­Zn2+) at a tubing was used for fluidic connections. A FlowTest automated concentration of 10 mg·L−1 and other metal ions normally controller (BioTray, Villeurbanne, France) (https://​www.​biotr​ present in water (­ Na+, ­Mg2+, ­Ca2+) at a concentration of ay.​fr) controlled the operation of the solenoid valve. It was 1000 mg·L−1 were acquired per triplicate. programmed through the dedicated CosDesigner software. 13 Microchim Acta (2023) 190:400 Page 5 of 11 400 Part 3): For the determination of C ­ o2+, a reverse flow literature, the CD quenching mechanism of C ­ o2+ can be injection analysis (rFIA) strategy was applied [30]. The associated with static quenching [27]. diluted CDs were sequentially injected into the buffer (blank) Calibration plots were obtained by injecting per triplicate and different C ­ o2+-containing solutions. The microfluidic 100-time diluted CDs in standard solutions with different setup includes a pre-buffering step of samples that is auto- concentrations of cobalt, namely 0.01 mg·L−1, 0.05 mg·L−1, matically performed by using an in-line T connector mixer, 0.1 mg·L−1, 0.5 mg·L−1, and 1 mg·L−1. This CD dilution through which the sample and the buffer are introduced in a ensures a good signal-to-noise ratio and avoids the inner 1:1 ratio. The flow management is performed with a Gilson filter effect (the absorbance is less than 0.04 at the maxi- Minipuls 2 peristaltic pump (Middleton, WI, USA) (https://​ mum excitation wavelength of 365 nm). The repeatability of es.​gilson.​com/), 0.8 mm internal diameter Teflon tubing, the measurement, calculated as relative standard deviation and 1.14 mm internal diameter Tygon tubing (Ismatec, (RSD), was checked by performing ten injections of CDs Wertheim, Germany) (https://​heido​lph-​instr​uments.​com/​es/​ into a solution containing 0.05 mg·L−1 of C ­ o2+. The limit start). An MVP six-port injection valve (Hamilton Company, of detection (LOD) and limit of quantification (LOQ) were Bonaduz, Switzerland) (https://​www.​hamil​tonco​mpany.​ calculated as three times and ten times the standard devia- com/) was used to inject the CDs, and the connections with tion of the blank signal (citric/citrate buffer) divided by the the microfluidic platform are secured with FPM75 O-rings. slope of the calibration plot ­(KSV), respectively. The microfluidic platform is introduced in a custom-made To assess the practical applicability of the synthesized miniaturized optical detection system previously reported [8, CDs to determine C ­ o2+, spiked tap water and river water 30]. Briefly, it contains an LED emitting at 365 nm, a band- (Besòs river, Spain) samples were evaluated. The spiking pass filter, and a PIN photodetector integrated into a printed process was done to undiluted samples prior to their analy- circuit board (PCB). The insertion of the platform is based sis, and the concentration range was chosen, taking into con- on a “lock-and-key” concept [8] that allows a reproducible sideration the maximum admissible limit of cobalt in drink- positioning of the device with respect to the LED and the ing water set at 0.1 mg·L−1 by the US EPA. Therefore, four photodetector. The signal is obtained with a data acquisition spiked concentrations of C ­ o2+ (0.10 mg·L−1, 0.25 mg·L−1, card. Some parameters affecting signal-to-noise ratio were 0.50 mg·L−1, and 0.75 mg·L−1) were tested. Recovery rates previously optimized [30]. To summarize, for this specific and % RSD were calculated for three measurements, con- work, the following conditions were applied: CD injection sidering these concentrations as the true value because the volume of 500 µL, flow rate of 1.5 mL·min−1, signal ampli- concentration of C ­ o2+ in tap and river water samples was fication of 10, and integration time of 0.1 s. under the LOD (1 µg·L−1) determined by ICP-OES. With this method, a continuous photoluminescent signal was established, with a maximum value when CDs were injected into a solution without cobalt, and quenched val- ues when the solution contained cobalt. The fluorescence Results and discussion intensity, which is obtained as the peak height, was corre- lated with the C ­ o2+ concentration using the Stern–Volmer Synthesis of carbon dots optimization equation [33]. The modification of the reaction conditions such as tem- perature, pressure, reaction time, and molar fraction of the Analytical characterization for ­Co2+ determination precursors allows modulating QYs because different pro- in water portions of the two fractions of CDs (crystalline and amor- phous) can be obtained, but these do not affect significantly The relative fluorescence intensity can be plotted against the the maximum excitation and emission wavelengths. The quencher concentration (in this case, cobalt), according to effect of the concentration of the nitrogen precursor and the Stern–Volmer equation. the temperature over the resulting CDs were evaluated by F0 comparing the corresponding QYs; 17 bars were the pres- = 1 + KSV × [Q] sure chosen for all the syntheses due to channel occlusion F observed at lower pressures [11]. where F0 is the fluorescence intensity of the CDs in the As can be seen in Fig. 2, the temperature of the synthesis absence of the quencher, F is the fluorescence intensity of had an important effect on the QY of the CDs obtained, the CDs in the presence of the quenching species Q ­(Co2+), showing a maximum at 170 °C. Performing the reaction at and KSV is the Stern–Volmer quenching constant, which higher temperatures can increase the carbonization, obtain- indicates the sensitivity of the method. According to the ing a carbon core-based product, which is more photostable 13 400 Page 6 of 11 Microchim Acta (2023) 190:400 the solution containing 0.75 M of ED, which means that an excess of ED in solution was affecting the final CDs QY. Therefore, the optimal synthetic conditions to obtain the highest QY (44%) were using a solution containing 0.75 M of the nitrogen precursor and performing the synthesis at 170 °C and 17 bars. Characterization of carbon dots and selectivity tests To follow the optical performance of the synthesized CDs, fluorescence emission and excitation spectra were examined (Fig. 3a). CDs showed wide excitation and emission bands with maximum intensities at 358 nm and 452 nm, respec- tively, obtaining a Stokes shift of approximately 100 nm. Fig. 2  Comparison of the QYs of the CDs under different tempera- An FTIR spectrum (Fig. 3b) was taken to character- tures and different concentrations of nitrogen precursor (ED). All the syntheses were performed at 17 bars and with a solution of 2.33 M of ize the functional groups present on the surface of the carbon precursor (acrylic acid) CDs. The following peaks can be observed, which can be assigned to different functional groups: the broad peak in the 3500–2500 ­cm−1 area comes from stretching vibrations of but has a lower QY [34]. Additionally, the effect of the O-H groups. Bands at 3170 ­cm−1 and 3010 ­cm−1 correspond precursor concentration was also noticeable. When the with N-H of an amide group and with C-H stretching vibra- concentration of the ED solution was 0.37 M, lower QY tions, respectively. The peak at 1646 ­cm−1 can be assigned values were obtained, indicating that the reaction was to the stretching vibration of C = O, the peak at 1550 ­cm−1 to incomplete, while when the concentration of the ED solu- the C = C stretching vibration, the peak at 1388 ­cm−1 belongs tion was 1.12 M, the QY was lower than that obtained with to the bending vibration of N-H, the one at 1278 ­cm−1 is Fig. 3  (a) Photoluminescence spectra of the synthesized CDs diluted 100 times to have an approximate absorption of 0.04 a.u; (b) FTIR spec- trum of the undiluted CDs; (c) HR-TEM image of the undiluted CDs; and (d) DLS measurement of the undiluted CDs 13 Microchim Acta (2023) 190:400 Page 7 of 11 400 from the asymmetric stretch of C-N, and, finally, the peak groups of the surface of the CDs (-COOH, -OH, -CONH-) at 1044 ­cm−1 is attributed to C-O aromatic stretching [35]. formed non-luminescent complexes, which can explain the Results indicated that nitrogen functionalization was observed decrease in fluorescence intensity [29, 33]. successful and that carboxyl, hydroxyl, amino, and amide bonds exist on the surface of the CDs, which can act as Analytical performance anchor sites to adsorb more ­Co2+ on the CDs’ surface to enhance the analysis sensitivity [27]. Different concentrations of cobalt were used to evaluate the The morphology and size distribution of the CDs were analytical performance of the system. Figure 5a shows the studied through HR-TEM images and DLS measurements transient fluorescence intensity signal of the microsystem by (Fig. 3c, d, respectively). Results show that the CDs have injecting the 100-time diluted CD dispersions into 0.1 M citric/ a quasi-spherical shape and an average hydrodynamic citrate buffer (blank) and solutions of different concentrations diameter of 4.2 ± 0.9 nm. However, it was very difficult of ­Co2+. The obtained calibration plot (Fig. 5b) shows a good to obtain high-quality contrast TEM images due to the linear correlation (R2 = 0.996) in the range of 0.02 to 1 mg·L−1 formation of two fractions of CDs, one with a crystalline of ­Co2+. The Stern–Volmer quenching constant, which is structure (carbon core particles) that can be identified equivalent to the sensitivity, obtained is 0.31 ± 0.01 L·mg−1. thanks to the observation of the diffraction planes and The repeatability of the system was calculated as the % another one that is amorphous and suggests the forma- RSD, obtaining a value of 1.3%, thus indicating good repeata- tion of polymer dots. These polymer dots have molecular bility. Additionally, the LOD and LOQ were calculated, obtain- fluorophore moieties embedded, causing the enhancement ing the values of 7 μg·L−1 and 21 μg·L−1 of ­Co2+, respectively. of the QY and making it more difficult to obtain TEM The maximum admissible limit of cobalt in drinking water is images [34]. Both fractions determine the optical char- not mentioned by the World Health Organization (WHO) nor acteristics of the final nanomaterial to be employed as an the European Union (EU). As a reference value, the United optical probe. In this sense, the use of the microreactor States Environmental Protection Agency (US EPA) has fixed is of special importance because it allows better control the maximum admissible limit at 0.1 mg·L−1 [5]. of the synthetic conditions (temperature, pressure and The analytical features of other reported CD-based opti- precursors mixing) than batch methods. This assures to cal methods are summarized in Table 1. As it can be seen, obtain not only more homogeneous CDs but also a high these methods reach slightly lower LODs and wider linear reproducibility between the different syntheses [36]. ranges. However, the method presented in this work allows Selectivity is an important issue to assess the applica- obtaining CDs with the highest QY, which ensures a strong tion of synthesized CDs as fluorescent probes. To evaluate signal even in case of possible reagent decomposition and selectivity, emission spectra of the CDs in the presence of as optical probes in the developed analyzer; they procure an different metal ion solutions ­(Ca2+, ­Cd2+, ­Co2+, ­Cu2+, ­Cr3+, LOD that is well below the reference value for ­Co2+ set by ­Fe3+, ­Hg2+, ­Mg2+, ­Na+, ­Ni2+, ­Pb2+, ­Zn2+) were recorded. the US EPA for drinking water. The automated synthesis of As can be seen in Fig. 4, the fluorescence intensity of CDs the CDs in the LTCC microreactor and its direct use for the was only quenched by C ­ o2+ among all the other metal ions determination of C ­ o2+ without any purification saves time studied, suggesting that the synthesized CDs are selective to and manual intervention. Compared to the reported manual ­Co2+. The interactions between ­Co2+ ions and the functional methods based on batch fluorescence measurements, the Fig. 4  Representative study showing the selectivity by the representation of the ratio between the fluorescence after the addition of various metal ions to the CDs and the fluores- cence of the CDs 13 400 Page 8 of 11 Microchim Acta (2023) 190:400 Fig. 5  (a) Fluorescence intensity signal of the microsystem (obtained in the absence of the quencher (F0)) and different standard solutions with an excitation source emitting at 365 nm, a CDs injection volume of the quencher (F). (b) Stern–Volmer plot of the quenching effect of of 500 µL, and a flow rate of 1.5 mL·min−1), where different peaks ­Co2+ on the fluorescence of the CDs appear when the CDs were injected into the buffer (intensity value Table 1  An overview of Precursors QY LOD Linear range Reference recently reported CD-based (%) (µg·L−1) (mg·L−1) optical methods for the determination of ­Co2+ ions Pyridoxal 5-phosphate and ethanediamine 15 3 0–3.5 [15] L-cysteine 27 2 0.06–2.9 [37] p-phenylenediamine and asparagine 16 1 0.02–3.8 [7] Carbopol 934 and diethylenetriamine 39 27 0–2.4 [29] Citric acid and L-cysteine - 5 0.005–5.9 [38] Acrylic acid and ethylenediamine 23 15 0.06–3.5 [27] Acrylic acid and ethylenediamine 44 7 0.02–1.0 This work proposed analytical method fulfills other analytical features be necessary to compact the whole instrumentation to make it such as automation and portability. more portable, robust, automatic, and autonomous. Long-term Regarding the Stern–Volmer quenching constant (KSV), it studies of the performance in continuous operation of the system has been doubled (0.31 ± 0.01 L·mg−1) compared with the should be carried out to detect possible instrumental instability synthetic method in batch (0.14 L·mg−1) [27] indicating an problems related to the fluidic microsystem or other interfer- improvement in the sensitivity. ence effects derived from analyzing complex polluted sample Taking into consideration that the response time of the matrices. Accuracy should also be assessed by comparison with CDs-Co2+ interaction to generate the analytical signal is a reference method for ­Co2+ determination in water samples. approximately 2 min, the overall sample throughput was calculated at 30 ­h−1. If, for monitoring purposes, an analyte was determined in the process solution to be monitored 4 Real sample analysis times a day and the analyzer was calibrated with 5 standard solutions and the blank, all of them per triplicate, once a The high selectivity and sensitivity of the synthesized CDs day, 30 measurements a day would be necessary to perform. toward ­Co2+ suggest that the current fluorescent sensor based This involves the synthesis of 150 µL of CDs a day (they are on CDs can be applied for measuring ­Co2+ in real water sam- diluted 100 times in the system before the injection). In this ples. To confirm that, spiked tap water and river water sam- situation, the microreactor would continuously perform the ples were analyzed. The results obtained are summarized in synthesis during 15 min. If other monitoring schedules were Table 2. The calculated recoveries are in the range between performed, the synthetic procedure would last differently. 98 and 104%, and RSD for three measurements is lower than The present modular prototype also has some limitations. 4% in all cases. This validates the application of the proposal To demonstrate its applicability for on-site monitoring, it would for online ­Co2+ determination in real water samples. 13 Microchim Acta (2023) 190:400 Page 9 of 11 400 ­ o2+ in spiked tap and river water samples Table 2  Determination of C management by using computer-controlled micropumps (n = 3) and microvalves to improve portability and robustness. To Sample [Co2+] [Co2+] found Recovery RSD demonstrate the applicability of the equipment in real on-site added (mg·L−1) (%) (%) monitoring and for validation purposes, it will be necessary (mg·L−1) to evaluate the effect of organic matter present in water sam- Tap water 0.10 0.104 ± 0.007 104 2.8 ples in the fluorescence quenching of the CDs, to perform 0.25 0.25 ± 0.02 101 2.5 long-term studies, compare results with a reference method, 0.50 0.50 ± 0.02 100 1.6 and analyze certified reference materials. 0.75 0.77 ± 0.05 102 2.5 River water 0.10 0.10 ± 0.01 102 3.8 Author contribution Alex Pascual-Esco: conceptualization, methodol- 0.25 0.24 ± 0.02 98 2.9 ogy, investigation, formal analysis, validation, and writing – original 0.50 0.49 ± 0.02 99 2.0 draft; Pere Lleonart: investigation and formal analysis; Antonio Calvo- 0.75 0.75 ± 0.04 100 2.0 López: conceptualization and methodology; Julián Alonso-Chamarro: resources, project administration, funding acquisition, and writing – review and editing; Mar Puyol: conceptualization, resources, supervi- sion, project administration, funding acquisition, writing – original draft, Conclusions and writing – review and editing. In this work, we describe a new analytical strategy for the Funding Open Access Funding provided by Universitat Autonoma de Barcelona. This work is part of the grant PRE2018-084834 funded by direct use of freshly synthesized nanoparticles by microreac- MCIN/AEI/https://doi.org/10.13039/501100011033and by the “ESF tor technology as optical probes in automatic microanalyz- Investing in your future.” This work has been financially supported ers. The complete system is modular and consists of three by the Spanish Ministry of Science and Innovation (MCIN) through main parts: a CD synthesis module with an LTCC microreac- project CTQ2017-85011-R and project PID2020-117216RB-I00, co- funded by FEDER (Fondo Europeo de Desarrollo Regional), and the tor, a dilution interface for the CDs obtained, and a fluores- Catalonia Government through project 2021SGR00124. cence detection module containing a COC microsystem. All of this is computer-controlled for the automatic determina- Data Availability Data will be made available on request. tion of heavy metals by fluorescence quenching. As a model system for the validation of the proposal, we synthesized Declarations highly efficient (high QYs) N-doped CDs from acrylic acid Conflict of interest The authors declare no competing interests. and ED to selectively determine ­Co2+ ions in water samples. To demonstrate the adequate performance of the sys- Open Access This article is licensed under a Creative Commons Attri- tem, ­C o 2+ was selectively detected among other heavy bution 4.0 International License, which permits use, sharing, adapta- tion, distribution and reproduction in any medium or format, as long metal and metal ions usually present in water with as you give appropriate credit to the original author(s) and the source, high sensitivity and an LOD and LOQ of 7 µg·L −1 and provide a link to the Creative Commons licence, and indicate if changes 21 µg·L−1, respectively, that are well below the reference were made. The images or other third party material in this article are value set by the US EPA for drinking water. The target included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in analyte was successfully determined in spiked tap water the article’s Creative Commons licence and your intended use is not and river water samples with good accuracy and precision, permitted by statutory regulation or exceeds the permitted use, you will thus demonstrating the viability to integrate the synthesis need to obtain permission directly from the copyright holder. To view a of the CDs prior to the analysis and their application as copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. optical probes for ­Co2+ determination in water. This rep- resents an advantage not only in terms of automation of the synthesis but also in terms of the use of a fresh opti- cal material that has not undergone any degradation prior References to analysis. The reduced dimensions of the microreactor 1. Mansuriya BD, Altintas Z (2021) Carbon dots: classification, and the microfluidic platform allow for minimizing the properties, synthesis, characterization, and applications in health amounts of reagents, which would favor less maintenance care - an updated review (2018–2021). Nanomaterials 11:2525. in a real application for heavy metals monitoring. https://​doi.​org/​10.​3390/​nano1​11025​25 The proposed system offers the possibility of expanding the 2. Qiu J, Zeng D, Lin Y, Ye W, Chen C, Xu Z, Hu G, Liu Y (2023) Carbon-polymer dot-based UV absorption and fluorescence per- number of analytes to determine by synthesizing other types of formances for heavy metal ion detection. Spectrochim Acta A selective CDs by alternating other precursors (containing N, S, P, 285:121913. https://​doi.​org/​10.​1016/j.​saa.​2022.​121913 etc.) and the reaction conditions (molar ratios and temperatures). 3. Salahinejad M, Sadjadi S, Abdouss M (2021) Investigating Further miniaturization and integration of the different fluorescence quenching of cysteine-functionalized carbon quantum dots by heavy metal ions: experimental and QSPR modular parts are currently being performed for fluidic 13 400 Page 10 of 11 Microchim Acta (2023) 190:400 studies. J Mol Liq 334:116067. https://​doi.​org/​10.​1016/j.​mol- 20. Hu G, Ge L, Li Y, Mukhtar M, Shen B, Yang D, Li J (2020) Car- liq.​2021.​116067 bon dots derived from flax straw for highly sensitive and selective 4. Hebbar A, Selvaraj R, Vinayagam R, Varadavenkatesan T, Kumar detections of cobalt, chromium, and ascorbic acid. J Colloid Interf Sci PS, Duc PA, Rangasamy G (2023) A critical review on the envi- 579:96–108. https://​doi.​org/​10.​1016/j.​jcis.​2020.​06.​034 ronmental applications of carbon dots. Chemosphere 313:137308. 21. Yu C, Zhuo C, Yundong W, Jianhong X (2022) Continuous syn- https://​doi.​org/​10.​1016/j.​chemo​sphere.​2022.​137308 thesis of N, S co-coped carbon dots for selective detection of 5. Lace A, Cleary J (2021) A review of microfluidic detection strate- CD(II) ions. J Photochem Photobiol A 429:113910. https://​doi.​ gies for heavy metals in water. Chemosensors 9:60. https://d​ oi.o​ rg/​ org/​10.​1016/j.​jphot​ochem.​2022.​113910 10.​3390/​chemo​senso​rs904​0060 22. Chen L, Wang CF, Liu C, Chen S (2022) Facile access to fabri- 6. Crawford SE, Kim KJ, Baltrus JP (2022) A portable fiber optic sensor cate carbon dots and perspective of large-scale applications. Small for the luminescent sensing of cobalt ions using carbon dots. J Mater 2206671:1–24. https://​doi.​org/​10.​1002/​smll.​20220​6671 Chem C 10:16506–16516. https://​doi.​org/​10.​1039/​d2tc0​2560d 23. Rao L, Tang Y, Li Z, Ding X, Liang G, Lu H, Yan C, Tang K, Yu B 7. Tian M, Liu Y, Wang Y, Zhang Y (2019) Facile synthesis of (2017) Efficient synthesis of highly fluorescent carbon dots by micro- yellow fluorescent carbon dots for highly sensitive sensing of reactor method and their application in Fe3+ ion detection. Mater Sci cobalt ions and biological imaging. Anal Methods 11:4077– Eng C 81:213–223. https://​doi.​org/​10.​1016/j.​msec.​2017.​07.​046 4083. https://​doi.​org/​10.​1039/​c9ay0​1244c 24. Tang Y, Rao L, Li Z, Lu H, Yan C, Yu S, Ding X, Yu B (2018) 8. Ymbern O, Berenguel-Alonso M, Calvo-López A, Gómez-de Pedro S, Rapid synthesis of highly photoluminescent nitrogen-doped car- Izquierdo D, Alonso-Chamarro J (2015) Versatile lock and key assem- bon quantum dots via a microreactor with foamy copper for the bly for optical measurements with microfluidic platforms and cartridges. detection of Hg2+ ions. Sens Actuators B Chem 258:637–647. Anal Chem 87:1503–1508. https://​doi.​org/​10.​1021/​ac504​255t https://​doi.​org/​10.​1016/j.​snb.​2017.​11.​140 9. Asif M, Tait RN, Berini P (2021) Hot embossing of microfluidics in 25. Golonka LJ, Zawada T, Radojewski J, Roguszczak H, Stefanow cyclic-olefin co-polymer using a wafer aligner-bonder. Microsyst M (2006) LTCC microfluidic system. Int J Appl Ceram Technol Technol 27:3899–3906. https://​doi.​org/​10.​1007/​s00542-​020-​05188-8 3:150–156. https://​doi.​org/​10.​1111/j.​1744-​7402.​2006.​02072.x 10. Sodzel D, Khranovskyy V, Beni V, Turner APF, Viter R, Eriksson 26. Lahti M, Kautio K, Karppinen M, Keränen K, Ollila J, Karioja MO, Holtz PO, Janot JM, Bechelany M, Balme S, Smyntyna V, Kole- P (2020) Review of LTCC technology for millimeter waves and sneva E, Dubovskaya L, Volotovski I, Ubelis A, Yakimova R (2015) photonics. Int J Electron Telecommu 66:361–367. https://​doi.​org/​ Continuous sensing of hydrogen peroxide and glucose via quenching 10.​24425/​ijet.​2020.​131886 of the UV and visible luminescence of ZnO nanoparticles. Microchim 27. Jing N, Tian M, Wang Y, Zhang Y (2019) Nitrogen-doped carbon Acta 182:1819–1826. https://​doi.​org/​10.​1007/​s00604-​015-​1493-9 dots synthesized from acrylic acid and ethylenediamine for simple 11. Berenguel-Alonso M, Ortiz-Gómez I, Fernández B, Couceiro P, Alonso- and selective determination of cobalt ions in aqueous media. J Chamarro J, Capitán-Vallvey LF, Salinas-Castillo A, Puyol M (2019) Lumin 206:169–175. https://d​ oi.o​ rg/1​ 0.1​ 016/j.j​ lumin.2​ 018.1​ 0.0​ 59 An LTCC monolithic microreactor for the synthesis of carbon dots with 28. Bano D, Kumar V, Chandra S, Singh VK, Mohan S, Singh DK, Talat photoluminescence imaging of the reaction progress. Sens Actuators B M, Hasan SH (2019) Synthesis of highly fluorescent nitrogen-rich Chem 296:126613. https://​doi.​org/​10.​1016/j.​snb.​2019.​05.​090 carbon quantum dots and their application for the turn-off detection 12. Chen T, Yin S, Wu J (2021) Nanomaterials meet microfluidics: improved of cobalt(II). Opt Mater 92:311–318. https://​doi.​org/​10.​1016/j.​optmat.​ analytical methods and high-throughput synthetic approaches. Trends 2019.​04.​045 Anal Chem 142:116309. https://​doi.​org/​10.​1016/j.​trac.​2021.​116309 29. Kong D, Yan F, Han Z, Xu J, Guo X, Chen L (2016) Cobalt(II) ions 13. Zu F, Yan F, Bai Z, Xu J, Wang Y, Huang Y, Zhou X (2017) detection using carbon dots as an sensitive and selective fluorescent The quenching of the fluorescence of carbon dots: a review on probe. RSC Adv 6:67481–67487. https://​doi.​org/​10.​1039/​c6ra1​2986b mechanisms and applications. Microchim Acta 184:1899–1914. 30. Pascual-Esco A, Alonso-Chamarro J, Puyol M (2022) Rapid warning https://​doi.​org/​10.​1007/​s00604-​017-​2318-9 microanalyzer for heavy metals monitoring in natural waters. Sens Actu- 14. Aswathy AO, Anju SM, Jayakrishna J, Vijila NS, Anjali Devi JS, ators B Chem 368:132180. https://​doi.​org/​10.​1016/j.​snb.​2022.​132180 Anjitha B, George S (2020) Investigation of heavy atom effect 31. Crosby GA, Demas JN (1971) The measurement of photolumi- on fluorescence of carbon dots: NCDs and S, N-CDs. J Fluoresc nescence quantum yields. A review. J Phys Chem 75:991–1024. 30:1337–1344. https://​doi.​org/​10.​1007/​s10895-​020-​02607-x https://​doi.​org/​10.​1021/​j1006​78a001 15. Liao S, Zhu F, Zhao X, Yang H, Chen X (2018) A reusable P, N-doped 32. Martínez-Cisneros CS, Gómez-de Pedro S, Puyol M, García-García carbon quantum dot fluorescent sensor for cobalt ion. Sens Actuators J, Alonso-Chamarro J (2012) Design, fabrication and characteriza- B Chem 260:156–164. https://​doi.​org/​10.​1016/j.​snb.​2017.​12.​206 tion of microreactors for high temperature syntheses. Chem Eng J 16. Huang SW, Lin YF, Li YX, Hu CC, Chiu TC (2019) Synthesis 211–212:432–441. https://​doi.​org/​10.​1016/j.​cej.​2012.​09.​101 of fluorescent carbon dots as selective and sensitive probes for 33. Song Y, Zhu S, Xiang S, Zhao X, Zhang J, Zhang H, Fu Y, Yang B cupric ions and cell imaging. Molecules 24:1785. https://​doi.​ (2014) Investigation into the fluorescence quenching behaviours and org/​10.​3390/​molec​ules2​40917​85 applications of carbon dots. Nanoscale 6:4676–4682. https://​doi.​org/​ 17. Zuo P, Lu X, Sun Z, Guo Y, He H (2016) A review on syntheses, 10.​1039/​c4nr0​0029c properties, characterization and bioanalytical applications of 34. Wang J, Zhang P, Huang C, Liu G, Leung KCF, Wang YXJ (2015) fluorescent carbon dots. Microchim Acta 183:519–542. https://​ High performance photoluminescent carbon dots for in vitro and doi.​org/​10.​1007/​s00604-​015-​1705-3 in vivo bioimaging: effect of nitrogen doping ratios. Langmuir 18. Naik V, Zantye P, Gunjal D, Gore A, Anbhule P, Kowshik M, 31:8063–8073. https://​doi.​org/​10.​1021/​acs.​langm​uir.​5b018​75 Sv B, Kolekar G (2019) Nitrogen-doped carbon dots via hydro- 35. Smith BC (1999) Infrared spectral interpretation: a systematic thermal synthesis: naked eye fluorescent sensor for dopamine approach. CRC Press LLC, Boca Raton. ISBN: 0-8493-2463-7 and used for multicolor cell imaging. ACS Appl Bio Mater 36. Gómez-de Pedro S, Puyol M, Izquierdo D, Salinas I, de la Fuente JM, 2:2069–2077. https://​doi.​org/​10.​1021/​acsabm.​9b001​01 Alonso-Chamarro J (2012) A ceramic microreactor for the synthesis of 19. Liu T, Zhang S, Fan X, Yang D, Wang M, Shao X, Wang S, Yue water soluble CdS and CdS/ZnS nanocrystals with on-line optical charac- Q (2022) Inner-filter effect induced fluorescence quenching of terization. Nanoscale 4:1328–1335. https://​doi.​org/​10.​1039/​c2nr1​1525e carbon dots for Cr(VI) detection with high sensitivity. J Fluoresc 37. Sun L, Liu Y, Wang Y, Xu J, Xiong Z, Zhao X, Xia Y (2021) 32:2343–2350. https://​doi.​org/​10.​1007/​s10895-​022-​03028-8 Nitrogen and sulfur Co-doped carbon dots as selective and visual 13 Microchim Acta (2023) 190:400 Page 11 of 11 400 sensors for monitoring cobalt ions. Opt Mater 112:110787. https://​ Publisher's Note Springer Nature remains neutral with regard to doi.​org/​10.​1016/j.​optmat.​2020.​110787 jurisdictional claims in published maps and institutional affiliations. 38. Chen Y, Shang P, Dong Y, Chi Y (2017) Regulating the over- lap between the absorption spectrum of metal ion-chromogenic agent and the emission spectrum of carbon-based dots to improve the sensing performance for metal ions. Sens Actuators B Chem 242:1210–1215. https://​doi.​org/​10.​1016/j.​snb.​2016.​09.​102 13