Affordable Automated Modules for Lab‐Scale High‐Throughput Synthesis of Inorganic Materials

Abstract

The development of sustainable advanced materials is increasingly driven by the need for sustainable, faster, scalable, and more efficient research workflows. Advancements in computational screening, high‐throughput experimentation, and artificial intelligence (AI) are accelerating progress in materials discovery. To fully leverage the benefits of these complementary approaches, the implementation of materials acceleration platforms (MAPs) and self‐driving laboratories (SDL) has emerged as a promising strategy. Here, we present the development of a semi‐automated station for the lab‐scale high‐throughput synthesis (HTS) of inorganic materials, as part of the Materials Acceleration and Innovation plaTform for ENergy Applications (MAITENA). The system integrates two in‐house‐designed liquid‐handling modules capable of performing sol‐gel, Pechini, solid‐state, and hydro/solvothermal syntheses. Each module enables the preparation of several dozen gram‐scale samples per week with high reproducibility and minimal manual intervention. The system's capabilities are demonstrated through three case studies involving Li‐ion battery materials. Results highlight the module's utilization for efficient screening of compositions and synthesis conditions to vary materials’ properties. This accessible and modular infrastructure offers a practical route to implementing high‐throughput strategies in inorganic materials research.

A semi‐automated high‐throughput synthesis station, developed within the MAITENA platform, features two modular units: one for solid‐state and wet‐chemistry routes (ceramic, sol‐gel, Pechini), and another for microwave‐assisted solvothermal synthesis. It enables reproducible workflows, phase diagram exploration, and morphology control, demonstrated on Li‐ion battery materials.

1. Introduction

Addressing the global challenges of climate change and critical materials scarcity requires accelerating the development of sustainable advanced materials. While traditional research approaches have historically been effective, they have become too slow and resource‐intensive to meet the urgent demands of strategic sectors such as energy, catalysis, or electronics.^{[ 1} , ² , ^{3 ]} In response, Materials Acceleration Platforms (MAPs) and Self‐Driving Laboratories (SDLs) have emerged as paradigm‐shifting solutions. These integrated research infrastructures combine automation, high‐throughput experimentation, and Artificial Intelligence (AI)‐driven decision‐making to enable faster and more efficient materials discovery.^{[ 3} , ⁴ , ⁵ , ⁶ , ⁷ , ⁸ , ⁹ , ^{10 ]}

Several different initiatives have been launched in recent years to accelerate materials development including the Materials Project (USA), BIG‐MAP (EU), the Accelerated Consortium (Canada), and DIADEM (France). High‐throughput theoretical approaches have been widely used to identify promising advanced materials. These approaches include leveraging Density Functional Theory (DFT) calculations to predict new materials through combinatorial compositions screenings,^{[ 11 ]} as well as mining materials databases to find structures with desirable properties using fast evaluation methods like electron density analyses^{[ 12} , ¹³ , ^{14 ]} and bond‐valence site energy calculations.^{[ 15} , ¹⁶ , ¹⁷ , ¹⁸ , ^{19 ]} Although computational methods are resource‐efficient alternatives in terms of time, materials, and infrastructure,^{[ 20 ]} they often generate a large number of candidates that require labour‐intensive experimental validation. To avoid any bottleneck at this latter stage, high‐throughput experimentation becomes essential.

High‐Throughput Synthesis (HTS) allows for an efficient production and testing of large libraries of materials. The concept was first pioneered in organic chemistry by Merrifield and Stewart in the 1960s for automated peptide synthesis,^{[ 21 ]} and has since been widely applied in combinatorial chemistry and materials science.^{[ 1} , ² , ²² , ²³ , ²⁴ , ²⁵ , ²⁶ , ²⁷ , ²⁸ , ^{29 ]} However, although HTS is now standard in fields such as drug development, its application for inorganic materials remains limited. In general, organic syntheses are more amenable to automation because they typically follow well‐characterized and rationalized reaction pathways and are usually performed in solutions under mild conditions. In contrast, inorganic syntheses involve a broader range of elements, frequently require the handling, mixing, and reacting of solid precursors, and often demand diverse reaction conditions including the use of autogenous pressure (e.g., hydro‐ and solvothermal routes) or high‐temperature treatments (e.g., solid‐state routes); all of which make automation considerably more challenging.^{[ 30} , ³¹ , ^{32 ]} When producing large sets of samples, high‐throughput characterizations coupled with efficient data analysis methodologies are also essential to prevent both workflow bottlenecks and misinterpretations (e.g., unreliable Rietveld analysis^{[ 33 ]} or overlooked materials disorder^{[ 34 ]}). Recent fully autonomous platforms, such as the A‐Lab proposed by Ceder et al., have drawn attention to these challenges, highlighting the importance of robust validation protocols in automated discovery workflows.^{[ 33} , ^{35 ]} High‐throughput characterization techniques can include automated sample preparation and measurements of X‐Ray Diffraction (XRD), electrochemical properties,^{[ 36} , ^{37 ]} X‐ray fluorescence (XRF) microscopy,^{[ 38 ]} Scanning Electron Microscopy (SEM),^{[ 39 ]} Raman microscopy,^{[ 40 ]} etc. As for automated data analysis, FullProfAPP^{[ 41} , ⁴² , ^{43 ]} was developed to carry out automatic Rietveld refinements on large powder diffraction datasets, PRISMA^{[ 44 ]} for processing chemical spectra, and Batalyse for processing electrochemical datasets.^{[ 45 ]} Moreover, the integration of AI, particularly machine learning models, is increasingly enabling real‐time data interpretation, anomaly detection, and predictive analysis, thereby enhancing decision‐making within autonomous workflows.^{[ 46} , ⁴⁷ , ^{48 ]}

Initial HTS approaches in inorganic materials relied on combinatorial screening of materials through thin‐film methods (e.g., ALD)^{[ 38} , ^{49 ]} or drop‐casting approaches.^{[ 50} , ^{51 ]} The emergence of more sophisticated data‐driven platforms has recently taken high‐throughput experimentation a step further. Examples of MAPs include AMANDA^{[ 23 ]} (perovskites and organic materials for photovoltaics), PANDA^{[ 22 ]} (electrodeposition of polymer films for electrochromic devices and optoelectronics), MEDUSA^{[ 29 ]} (solution‐phase synthesis and electrochemical screening of metal–ligand complexes), ASTRAL^{[ 52 ]} (metal oxides for intercalation batteries), MINERVA^{[ 24 ]} and AFION^{[ 53 ]} (nanomaterials for bioimaging, biomedical, and microelectronics). Commercial lab automation solutions offer advantages such as being ready‐to‐use and benefit from technical support, but they are often expensive, and tailored for specific tasks, with limited flexibility for modifications or upgrades, and involve relying on the manufacturer for supplies or replacements. They typically target either microgram‐scale (typically for pharmaceutical research) or large‐scale (industrial applications) productions. However, many material science research fields require intermediate quantities (hundreds of milligrams to grams), allowing all characterizations to be conducted on a single representative sample. This improves reliability, as well as provides results that better support scalability development. In these cases, in‐house‐developed solutions provide cost‐effective and customizable alternatives.^{[ 54 ]} Such systems have to integrate tailored hardware and software, and robust quality protocols (e.g., precision, calibration, validation) to ensure data accuracy and reliability.

To address the limitations of existing solutions, in this work we present the development of an in‐house station for semi‐automated high‐throughput inorganic synthesis, part of the Materials Acceleration and Innovation plaTform for ENergy Applications (MAITENA) at CIC energiGUNE. This station integrates two functional modules that enable precise handling and mixing of precursor solutions, and support various synthesis routes—including solid‐state, sol‐gel, Pechini, and hydro/solvothermal—while minimizing manual work. Each module can produce up to 12 samples per run. Typical isolated yields are in the range of 400 mg to 1 gram, depending on the reaction conditions, which enables comprehensive characterizations on single batches. They also offer precise microstructure control, which is key for tuning material properties. Herein, we describe the design and functionality of these modules and demonstrate their capabilities through three case studies.

2. Results and Discussion

2.1. Description of the Modules

The station described herein consists of two distinct in‐house‐designed automated liquid‐handling modules (referred to as Module‐I and Module‐II), each capable of synthesizing 12 samples simultaneously. Module‐I is dedicated to the preparation of reactant mixtures for wetsyntheses, including sol‐gel and Pechini methods, while Module‐II focuses on solution‐based syntheses, such as hydro‐ and solvothermal syntheses. The two modules are shown in Figure 1.

Figure 1.

Automated liquid‐handling modules designed to handle and mix precursor solutions at suitable stoichiometries for inorganic syntheses: a) Module‐I for solid‐state and wetsyntheses (e.g., sol‐gel synthesis) and b) Module‐II for solution‐type syntheses (e.g., hydro‐ and solvothermal syntheses). The main components of a) Module‐I (A‐F) and b) Module‐II (A‐H) are indicated.

Both modules are based on a robotic system that automates liquid handling, dispensing, and mixing. This approach minimizes manual intervention, reduces human error, and contributes to the overall process optimization. Instead of directly using solid reactants, liquid‐handling automation enhances operational ease, allowing for efficient and precise handling of the materials. The operation of the liquid‐handling robot involves dispensing the reactants as uniform solutions, where the solubilities of the reactants determine the concentration range used for the reactions. In both modules, a series of pumps are used to dispense the reactant solutions (A in Figures 1a,b). Atlas scientific programmable pumps were first selected as they enable accurate dispensing of the quantities required for the chemical systems under study, and they are also easy to calibrate (vide infra). However, these pumps can be easily substituted in the system by other types of pumps according to the needs. To mix the reactants, a different approach was adopted for each of the two modules (Figure SI1). In Module‐I (Figure 1a), the reactant solutions are combined and stirred in a round flask mixer (B in Figure 1a) before being introduced into the crucibles (C in Figures 1a, and SI1a). This approach offers the possibility to add one reactant mixture to another one that has previously been pumped into the crucibles. The mixer is automatically cleaned three times with distilled water between each mixture. In Module‐II (Figure 1b), all solutions are separately pumped into the final reactors (B in Figures 1b and SI1b). This avoids obstruction of the system in case a viscous solution is formed after mixing the reactants, generates less waste, and results faster compared to Module‐I. The mixing in the reactors is done either with classical magnetic stirring or using a vertical stirrer (C in Figure 1b) to allow high‐speed agitation when viscous solutions are formed.

To dispense the solutions into the different reactors, two options were again considered: (i) a multi‐axis robotic arm and (ii) a two‐axes (x‐y) Computer Numerical Control (CNC) table. While a robotic arm provides greater movement flexibility, it also introduces more complexity in terms of programming, calibration, and maintenance. In contrast, the CNC table (D in Figures 1a, b) proved to be a more accessible and robust alternative, less costly and technically demanding. It is adaptable to the available space dimensions and proved to be more precise and reliable over time (design schemes are shown in the GitHub repository that can be found in the Supporting Information). In the case of Module‐II, a tube distributor was coupled to the CNC table to avoid contamination between tubes (E in Figure 1b). A pH meter (Lab Grade pH Probe Part # ENV‐40‐pH from Atlas Scientific) was also included on the mobile head of the CNC table (F in Figure 1b), allowing the automated measurement of the pH of the solution at any moment of the process. Lastly, the modules include a multi‐position heating and mixing plate (E and G in Figures 1a, b, respectively), placed below the CNC table, which allows stirring and heating the solutions for conducting reactions and/or concentrating the resultant mixtures.

In the case of solid‐state and wet syntheses, the solvent fraction is entirely evaporated prior to the annealing step, for which the reactors containing the mixture of reactants are transferred to the muffle or tubular furnaces using quartz trays (Figure SI2). These customized trays also ensure consistent placement of the reactors inside the furnace, which allows tracking the reactor position in the furnace and eventually corroborate it with the final material's features (see Section 4c). Both types of furnaces, used to perform syntheses under air and controlled atmospheres (i.e., Ar/H₂, Ar, O₂), respectively, are equipped with programmable temperature controllers. These controllers play a role in maintaining precise control over the heating and cooling ramps during the synthesis processes, ensuring optimal reproducibility of the experiments by facilitating comparison and validation of results.

In the case of hydro‐ and solvothermal syntheses, once the solution mixtures are prepared in the borosilicate reactors, they are transferred to the microwave oven (CEM Discover 2.0) equipped with a robotic sample handler. This oven enables precise control of the temperature and pressure during the reaction, thanks to an infrared temperature sensor (iWave) and through the automated sealing and venting of the autogenous gases in the reactor. The utilization of microwave heating offers significant advantages over classical hydrothermal synthesis methods with conventional heating furnaces.^{[ 55} , ^{56 ]} One of them is the reduction in synthesis times that enhances the overall efficiency of the process enabling a fast screening of chemical spaces and synthesis conditions.

To ensure safe operation, the synthesis modules are placed on a containment tray equipped with a perforated upper shelf that directs any spills into a lower compartment. An internal ventilation system within the cabinet removes vapors during solvent evaporation. Electronic components are located outside the tray and kept under dry conditions.

2.2. Workflow

The entire hardware system described above is controlled using a microcontroller (F and H in Figures 1a, b, respectively). Here, an Arduino Mega board was chosen; Arduino being an open‐source platform based on free hardware and software that allows to develop electronic devices and projects in a user‐friendly mode.^{[ 57 ]} The Arduino Mega board is based on the ATmega2560 microcontroller, which is a high‐performance 8‐bit AVR microcontroller. It operates at a clock speed of 16 MHz and has 256 KB of flash memory for storing program code.

The experimental workflow is orchestrated by an in‐house code uploaded onto the Arduino microcontroller. Figure 2 shows examples of experimental workflows for sol‐gel syntheses and hydrothermal syntheses, outlining the sequence of actions necessary to carry out successful synthesis processes using Module‐I and Module‐II, respectively. An example of code for preparing a set of twelve samples with the sol‐gel method can be found in the GitHub repository (Supplementary Information). Working in Arduino code instead of other languages such as G‐code, typically used when working with CNC tables,^{[ 58 ]} was selected to enable the use of this kind of microcontroller and to ensure a flexible control over the entire experimental workflow including pumps and sensors.

Figure 2.

Examples of program workflows for producing N samples by a) sol‐gel and b) hydrothermal syntheses using the automated Modules I and II, respectively.

To conduct sol‐gel syntheses, Module‐I is typically used to produce a series of 12 samples consecutively (Figure 2a). First, the desired quantities of reactants for sample N = 1 are all pumped into the mixer from the different precursor solutions. Then, they are stirred in the mixer for a specific duration. Subsequently, the CNC table moves to align with the first reactor, into which the reactant mixture is transferred from the mixer. Following this, the dispensing system undergoes a series of washing steps with distilled water (DW). Afterwards, the system is ready to start the process again for the second sample. In a typical series of sample preparations, this process is repeated 12 times for 12 reactors that are continuously stirred. Once all samples are prepared, they are left to stir overnight. Then, the solutions are evaporated by heating at 80 °C until a solid mixture is obtained. The Arduino microcontroller manages the liquid handling steps, whereas heating and stirring are independently configured before each experiment. Once the evaporation stage is complete, the reactors are transferred into a quartz tray, which is directly placed in a muffle or tubular furnace to carry out the annealing step. After the thermal treatment, 300–400 mg of each sample are typically recovered.

To perform syntheses in solution (hydro‐ or solvothermal‐synthesis), Module‐II is designed to accommodate up to 15 samples simultaneously. The sample preparation consists of directly pumping the different solutions into the individual reactors (30‐mL borosilicate tubes) through the dispenser head of the CNC table. During the preparation process, an automated mixing step can be included using a vertical mixer that homogenizes viscous solutions through vigorous agitation (C in Figure 1b). The pH of the solution can also be measured, at any step of the process, using the electrode attached to the dispenser head of the CNC table (D in Figure 1b). Once the procedure finishes, the reactors are transferred to the microwave equipment (CEM Discover 2.0 microwave) equipped with a robotic sample handler to complete the syntheses in an autonomous manner, wherein different parameters such as temperature, pressure, and time can be varied for each sample. After the microwave treatments, the samples are washed and centrifuged together to collect up to 900 mg for each sample and perform the required characterizations.

In both cases, the final amount of the individual samples allows for the execution of all the required characterization techniques on a single sample, ensuring consistency and representativeness in the analysis.

2.3. Reproducibility Considerations

To ensure accuracy and reliability of the samples produced, a comprehensive process of calibrations, optimization, and reproducibility validation is conducted for every newly developed synthesis workflow.

For optimal sol‐gel syntheses, a careful study of the thermal distribution of the heating plate was conducted to ensure uniform evaporation conditions across all crucible positions. Initial results revealed significant temperature variations of up to 10 °C across the twelve positions, which was identified as being caused by a high variability in the contact resistances between the crucibles and the heating plate. This issue was solved by applying silicone oil or silicone grease at the bottom of the alumina reactors, improving thermal contact between surfaces (Figure SI3).

Regular calibration of the pumps and the furnaces is also performed to maintain precision and reliability. Typical solution volumes pumped range from 1.3 to 7.6 mL, with a tolerance limit of 0.1 mL established to minimize deviation in the sample compositions. To further ensure accuracy, the molarity of solutions is prepared between 0.3 and 1 M, depending on the reagent. Figure SI4 shows that all pumps operate consistently within the established tolerance range over time. The calibration results of the muffle and tubular furnaces are presented in Figures SI5, SI6, respectively. For each position, the average temperature and the standard deviation (based on three or more measurements) relative to each setpoint temperature are recorded as metadata for each sample.

2.4. Case Studies

This section presents three case studies demonstrating the capabilities of the automated modules for HTS and optimization of inorganic materials. The selected examples illustrate the reproducibility of the modules and their practical application to explore phase diagrams and microstructural tuning.

2.4.1. Reproducibility of the Sol‐Gel Synthesis of LiFe_0.5Mn_1.5O₄

LiFe_0.5Mn_1.5O₄ (LFMO) is a spinel material, which requires precise synthesis control to prevent the formation of Fe‐Li antisite defects. Twelve samples of LFMO were prepared with Module‐I using the sol‐gel method under identical conditions (see Supplementary Information). Figure 3 compares the XRD patterns of the 12 synthesized materials. The XRD results confirm the formation of a pure spinel phase of LFMO across all module positions, with no significant change in relative intensity or peak broadening. Notably, the (220) and (222) reflections, known to be sensitive to Fe‐Li defects, remain unaffected, demonstrating the robustness and reliability of the automated synthesis.

Figure 3.

Comparison of the XRD patterns of the LiMn_1.5Fe_0.5O₄ spinel samples synthesized in the 12 positions of the automated module (labeled from A1 to C4). Enlargements around the diffraction peaks (220) and (222) are shown on the upper part of the figure.

The results of the Rietveld refinements of the 12 XRD patterns are presented in Figure SI7, and Table SI4 to Table SI6. The refined unit cell parameters of the twelve samples range from 8.297(2) to 8.307(2) Å, with an average value of 8.301 Å. Despite the minor variations in annealing temperature observed across the different crucible positions in the furnace, the structural parameters remained consistent, demonstrating the reproducibility of the synthesis method.

2.4.2. Mapping of the Li‐Ni‐Mn Oxide Pseudo‐Ternary Phase Diagram through High‐Throughput Sol‐Gel Synthesis

Module‐I was used to explore a subset of the pseudo‐ternary phase diagram of Li‐Ni‐Mn oxides, which have been widely studied for their application in energy storage, catalysis, and electronics. As detailed in the Experimental Section, 48 Li‐Ni‐Mn oxide samples were synthesized via the sol‐gel process under identical synthetic conditions, varying solely the metal stoichiometries for each sample to illustrate the module's capabilities for combinatorial synthesis. The samples were characterized by Synchrotron XRD to identify and map the different phase domains of the phase diagram, as shown in Figure 4. For this analysis, the program FullProfAPP^{[ 41} , ⁴² , ^{43 ]} was used to perform the automated phase identification and pseudo‐Rietveld refinement of the 48 XRD patterns. The results of the refinements of a selection of 4 samples are gathered in Figure SI8, Table SI7, and Table SI8. The results reveal three distinct regions within the phase diagram:

1. Region 1 (in yellow in Figure 4), corresponding to nominal composition LiNi_0.5+xMn_1.5‐xO₄ (0 ≤ x ≤ 1). Around x = 0 (Figure 5, Point A), the spinel phase LiNi_0.5Mn_1.5O₄ predominates. A secondary phase increases as x rises. This phase, indexed as a rock salt, has been ubiquitously reported as a secondary phase in samples that deviate from the theoretical stoichiometry of LiNi_0.5Mn_1.5O₄.^{[ 59 ]} To account for the reflections appearing at 10–25° (Figure SI9), it is indexed with a cell parameter of 8.28 Å, corresponding to a 2a2b2c supercell of the cubic cell commonly reported for rock salt phases.^{[ 60 ]} These results illustrate that the spinel phase cannot accommodate additional Ni^II in substitution of Mn^IV, resulting in the formation of the rock salt phase as the Ni content in the sample increases. While both phases coexist throughout this region, the spinel diminishes at higher Ni concentrations.

2. Region 2 (in blue in Figure 4), with nominal composition Li_1+2xNi_0.5+xMn_1.5‐xO₄ (0 ≤ x ≤ 0.5). For x = 0, the LiNi_0.5Mn_1.5O₄ spinel is obtained (Figure 5, Point A). As the Ni and Li contents increase, a Li₂MnO₃ (C2/m) layered oxide phase appears, compensating for lithium excess, while the rock salt phase would balance the nickel excess, as observed in Region 1. The cell parameters of these phases remain unchanged over this range of compositions, suggesting the absence of solid solution domains for any of these phases. Toward the other end of this region, a layered oxide (R‐3 m) emerges at the expense of the Li‐rich phase, although pattern similarities complicate precise phase identification.

3. Region 3 (in green in Figure 4), with nominal composition Li_1+2xNi_0.5+xMn_1.5‐xO₄ (0.5 < x ≤ 1). A single‐phase solid‐solution domain of a layered oxide Li_aNi_0.5+bMn_0.5‐bO₂ (R‐3 m) is identified over this composition range. Unlike Regions 1 and 2, no rock salt phase is present in this region.

Figure 4.

On the left, a pseudo‐ternary phase diagram obtained for the Li‐Ni‐Mn oxide system where the black points represent the nominal compositions of the samples, and the colored lines indicate the three distinct phase regions identified. A, B, C, and D points correspond to selected samples, whose refined XRD patterns are shown in Figure SI‐8. On the right, the diffractograms obtained for the 48 synthesized samples. The arrows indicate the increasing amount of Ni and Li in the targeted composition, and the obtained phases are indicated as headings for each region.

Figure 5.

A) Evolution of the unit cell parameters, crystallite size, and strain values as the synthesis temperature and time of the LFP samples increase. Standard deviations are not clearly visible in the figure, as they are on the order of 10⁻⁴, rendering them too small to be discerned. B) Particle morphology and size evolution with temperature, from 160 to 200 °C, and synthesis time, from 5 to 30 minutes of the LFP samples (160 °C, a‐c; 180 °C, d‐f; 200 °C, g‐i) monitored by SEM.

This case study illustrates the module's capability to facilitate HTS to explore complex compositional screenings beyond this Li‐Ni‐Mn pseudo‐ternary phase diagram.

2.4.3. Microwave‐Assisted Solvothermal Synthesis of LiFePO₄

The performance of olivine‐type LiFePO₄ (LFP), a state‐of‐the‐art cathode material for Li‐ion batteries, is known to be highly dependent on its microstructure and particle size, which critically influence lithium‐ion diffusion and charge transfer kinetics. Module‐II was employed to prepare the reactants mixture for the synthesis of nine distinct samples of LFP via microwave‐assisted solvothermal treatment, varying the synthesis parameters to modify the microstructure of the samples (Table SI9). All the synthesized samples show similar XRD patterns, with larger peak broadening when the samples are synthesized at lower temperatures with shorter times (Figure SI10b). The results of the Rietveld analysis of the XRD patterns confirmed that a pure triphylite phase of LiFePO₄ is obtained (Figure SI11, Table SI10, Table SI11, and Table SI12), while differences in the unit cell parameters, strains, and average crystallite sizes are observed, as reflected in Figure 5A. At the lowest synthesis temperature (160 °C), the unit cell volume of LFP increases with synthesis time due to the growth of the a and b cell parameters. In contrast, at higher temperatures (180 °C and 200 °C), all three cell parameters exhibit relatively stable values regardless of the reaction time. The strain shows a nearly linear decrease over time at all three temperatures. This trend suggests an increased homogeneity of the lattice parameters as the heating process advances, and could be attributed to crystal growth via the coalescence of adjacent particles and partial relaxation of surfaces. The apparent average crystallite sizes also increase with synthesis time at all three temperatures, ranging in size from 42 to 74 nm, indicating the progressive growth of the crystalline structure throughout the heating treatment. SEM indeed revealed a variety of particle sizes and morphologies obtained from slight modifications of the synthesis parameters (Figure 5B). At the shortest synthesis times (5 minutes) (Figure 5B‐a, d, and g), elongated pseudo‐hexagonal nanosheets are formed. At 160 °C (Figure 5B‐a, b, and c), particle thickness and size increase with synthesis time, transitioning from agglomerates to independent particles (100‐300 nm in length and 40–60 nm in thickness). At the highest temperature (200 °C) (Figure 5B‐g, h, and i), particles start thicker, forming larger semi‐welded structures within 15 minutes. From this point onwards, crystalline growth is triggered and, eventually, these structures end up sintering into much larger particles (500 nm to microns). At 180 °C (Figure 5B‐d, e, and f), particles evolve into unique hollow nanoprisms (Figure SI12).

This case study illustrates the capability of the modules to systematically prepare a series of samples for an automated screening of synthesis conditions, enabling an efficient assessment of their impact on material properties.

3. Conclusion

The HTS station, developed within the MAITENA Materials Acceleration Platform, integrates two in‐house‐designed modules that enable the precise and reproducible preparation of series of samples with a wide range of inorganic synthesis routes, including sol‐gel, Pechini, solid‐state, and hydro/solvothermal methods. The modular and flexible design, coupled with open‐source control and customizable hardware, provides a cost‐effective alternative to commercial automation systems, specifically adapted to meet the intermediate‐scale needs of materials science research.

The reproducibility and versatility of the modules were demonstrated through three case studies involving the synthesis of Li‐ion electrode materials: (i) the consistent synthesis of 12 LiFe_0.5Mn_1.5O₄ samples confirmed the high reproducibility of the automated process for sol‐gel synthesis; (ii) a combinatorial synthesis of 48 Li‐Ni‐Mn oxide compositions enabled the mapping of a pseudo‐ternary phase diagram, showcasing the system's potential for accelerated discovery; and (iii) the systematic variation of synthesis parameters in the microwave‐assisted hydrothermal synthesis of LiFePO₄ highlighted the system's capacity for microstructural tuning.

These modules enable an easy shift from traditional lab‐scale experimentation to high‐throughput workflows. They allow the production of material quantities suitable for thorough characterization and future scalable development. Traditionally, the manual preparation of inorganic materials typically limited researchers to just a handful to a couple dozen samples per week, depending on the complexity of the synthesis. In contrast, the automated modules presented in this work consistently deliver several dozen samples per module within the same timeframe; resulting in a 5‐ to 10‐fold increase in throughput while delivering consistent and reliable outcomes. We believe this accessible and adaptable solution can help move toward more efficient materials development workflows and ultimately accelerate innovation and breakthroughs in the energy sector and beyond.

Supporting Information

The authors have cited additional references within the Supporting Information.

Conflict of Interest

The authors declare no conflict of interest.

Supporting information

Supporting Information

Data Availability Statement

The data that support the findings of this study are openly available in Giithub and Zenodo at https://doi.org/10.5281/zenodo.15608412, reference number 15 608 412.