Automated Arc Melting System

• Automated arc melting system is a cyber-physical synthesis cell integrating arc discharge with automated weighing, robotic transfer, and digital traceability.
• The platform leverages ROS2 for modular orchestration of heterogeneous devices, ensuring precise process control, reproducibility, and scalability.
• In situ diagnostics using high-speed two-color pyrometry enable real-time control of arc parameters and prediction of molten-electrode ablation rates.

Searching arXiv for the cited papers and closely related work on automated arc melting systems. An automated arc melting system is an integrated experimental platform in which arc-discharge melting is coupled to automated weighing, transfer, atmosphere control, sample manipulation, monitoring, and digital logging for the synthesis of bulk intermetallic compounds and related materials. In the literature, the concept spans both UHV-compatible arc furnaces integrated with an argon glovebox and fully automated platforms orchestrated by Robot Operating System 2 (ROS2), as well as in situ diagnostic subsystems that estimate molten-anode ablation rates from high-speed pyrometry (Bauer et al., 2016, Wang et al., 15 Sep 2025, Musikhin et al., 12 Jun 2025).

1. System composition and functional scope

The most explicit realization of the concept is a fully automated arc melting platform for bulk intermetallic compound synthesis controlled by ROS2. Its major hardware components include an automated electric balance with dispensing (Mettler-Toledo Quantos), raw materials storage holding up to 20 dosing heads, a 6-axis collaborative robotic arm (Denso-Wave Cobotta) with a custom 3D-printed gripper, electric sliders in 1-axis, 2-axis, and 3-axis configurations, a custom small-volume water-cooled vacuum/arc melting chamber with an arc-discharge tungsten tip, a programmable logic controller (Mitsubishi, FX5 series), an arc power supply (Panasonic, YC-300BZ3), and a secondary PC for logging and digital laboratory notebook integration (Wang et al., 15 Sep 2025).

A broader materials-synthesis preparation chain places the arc melting furnace alongside an argon glovebox, an inductively heated horizontal cold boat furnace, an inductively heated rod casting furnace, an optically heated floating-zone furnace, a resistively heated annealing furnace, and an inductively heated annealing furnace. Within that chain, the arc melting furnace is a custom, all-metal sealed setup intended for the synthesis of high-purity intermetallic compounds, particularly when air-sensitive elements are involved (Bauer et al., 2016).

Component	Function
Automated electric balance with dispensing	Precise, contamination-resistant raw material weighing
6-axis collaborative robotic arm	Manipulation and transfer tasks
Electric sliders	Robot positioning, sample transfer, and arc guide control
Vacuum/arc melting chamber	Arc melting under controlled atmosphere
PLC	Low-level control of pumps, valves, sensors, and flippers
Arc power supply	Controlled arc discharge for melting
Secondary PC and recording	Logging and digital laboratory notebook integration

The functional scope therefore extends beyond the arc itself. It includes raw-material handling, inert transfer, coordinated actuation across heterogeneous hardware, and complete digitization of the synthesis record. This suggests that an automated arc melting system is best understood as a laboratory-scale cyber-physical synthesis cell rather than as a single furnace.

2. Vacuum compatibility, inert-atmosphere transfer, and furnace mechanics

A central design principle is the realization of a high-purity growth environment. In the UHV-compatible implementation, all furnaces are all-metal sealed, bakeable, and may be pumped to ultra-high vacuum; base pressures down to $\sim 10^{-8}\,\mathrm{mbar}$ are reached within a day, and operation proceeds after backfilling with up to 1.2 bar of 6N argon passed through gas purifiers (Bauer et al., 2016).

Integration with an argon glovebox is accomplished by a load-lock arrangement. The arc melting furnace is connected via lock (C) on the glovebox, comprising an ISO-K100 flange and a manually tightened lid. A CF DN100 plate valve at the back of the arc furnace, together with a short U-profile bellows for length adaptation, provides a direct, vacuum-tight and inert gas-compatible passage for transferring elements and samples. After aligning argon partial pressures, the plate valve is opened and samples are loaded from the glovebox using a custom shovel; entry and manipulation can also be done using a wobble stick for precise positioning into the copper molds (Bauer et al., 2016).

The mechanical realization includes a water-cooled copper base plate tightly bolted to a CF DN160 flange, a lanthanum-doped tungsten electrode attached to a water-cooled vertically and horizontally movable feedthrough with bellows, and electrical insulation realized using Stycast 2762 and a PEEK sleeve. Quick access is provided by a CF DN100 port with a clamping chain. Front and top windows permit visual control and attachment of a pyrometer for temperature monitoring. The bottom flange can be lowered and pivoted for cleaning, loading, or replacing crucible and electrode parts, while the furnace can remain docked to the glovebox during melting (Bauer et al., 2016).

These design choices are directly tied to purity and safety. The load-lock permits synthesis starting with air-sensitive elements while handling the constituents exclusively in an inert gas atmosphere. All-metal sealing and bakeability are described as an important prerequisite for handling compounds with high vapor pressure under high-purity argon atmosphere. The exclusion of organic seals such as Viton and the use of water-cooling on critical parts minimize contamination risks (Bauer et al., 2016).

3. Orchestration, control software, and modular integration

The automated platform uses ROS2 as the orchestration layer for heterogeneous experimental machines. Direct LAN or USB connections link the ROS2 host PC to hardware with built-in controllers, including the robotic arm, sliders, balance, and camera, whereas lower-level devices such as the power supply, pumps, valves, sensors, and flippers are abstracted by a PLC that is itself networked to the ROS2 host (Wang et al., 15 Sep 2025).

The ROS2 node structure is explicitly modular. The main node orchestrates the experiment’s process sequence; weighing controls the balance and dispensing system; cobotta manages robotic-arm actions; slider executes linear-stage motion; plc sends and receives PLC commands for low-level operations; and record handles experimental data and video logging. Sequences are encoded as text files such as Sequence.txt, with actions specified as nodename_action1_action2.... The main node cycles through the sequence by publishing commands, and nodes report status as action, standby, or error; the next step begins only when all nodes are ready. Multiple nodes can receive commands concurrently, enabling asynchronous execution and parallelization, for example when pumping and robot motion occur in parallel (Wang et al., 15 Sep 2025).

Users define the workflow either through Sequence.txt or through a Streamlit-based GUI that specifies parameters such as composition, weights, and arc trajectory. No code changes are required for routine use or when integrating new compatible hardware. Extension to new apparatus follows an explicit pattern: devices with LAN automation are wrapped in a new ROS2 node, whereas non-automated devices are first brought under PLC control and then exposed via ROS2. Process logs and video are stored locally and redundantly uploaded to the remote electronic lab notebook eLabFTW (Wang et al., 15 Sep 2025).

The architectural significance lies in the separation between process orchestration and device-specific control. This supports plug-and-play modular integration, future expansion of functions, and the simultaneous handling of multiple experimental apparatuses.

4. Operational workflow and execution of the melt cycle

The automated workflow begins with weighing. The robotic arm retrieves raw-material dosing heads from storage, brings them to the balance, and dispensing proceeds automatically until the desired quantities are met, with sequential processing for multi-element mixtures. Only coarse or fine chunks of approximately 1 mm are used, owing to powder-handling and contamination concerns. The loaded sample carrier is then conveyed through transfer stations and into the arc chamber by the robotic arm and sliders (Wang et al., 15 Sep 2025).

Arc melting is executed after chamber conditioning. The chamber is vacuum-purged with argon in four cycles requiring less than 3.5 minutes in total. Arc discharge is then applied, with the 3-axis slider controlling precise arc-tip positioning and trajectory. After each melt, a motorized flipper flips the sample for homogeneity. The entire arc process is video recorded. After completion, the sample is retrieved, weighed again, and returned to stock, and all process data—including raw weights, trajectories, sequences, and videos—are digitized and logged for reproducibility (Wang et al., 15 Sep 2025).

A related but non-robotic arc-melting workflow for thermoelectric Bi$_{2-x}$Sb$_x$Te$_3$ illustrates the underlying process physics. Stoichiometric mixtures of high-purity elemental Bi, Sb, and Te are weighed, thoroughly mixed, placed in an Edmund Bühler MAM-1 arc furnace under argon, rapidly melted in seconds, and immediately quenched by contact with a water-cooled copper crucible. Pressing under 70 bars is used to form test bars for physical measurements. The same report identifies atmosphere control, arc current, voltage, duration, electrode position, temperature sensing, timed quenching, sample loading and unloading, and parameter logging as readily automatable steps (Serrano-Sánchez et al., 2018).

Taken together, these workflows show that automation in arc melting is not confined to robotic transfer. It also includes recipe execution, chamber cycling, sample inversion for homogeneity, and full digital traceability of the thermal history and handling sequence.

5. In situ diagnostics and model-based control of molten electrodes

For DC arc systems with a meltable metal anode, real-time monitoring of the molten surface is a control problem because vaporization determines catalyst seed-particle formation in single-walled carbon nanotube synthesis. In situ monitoring has therefore been implemented with high-speed, 2D, 2-color pyrometry. The system uses a Phantom v7.3 high-speed camera with 512×256 pixels, 14-bit depth, and 106 $\mu$m/pixel resolution; wavelengths of 785 $\pm$ 3 nm and 890 $\pm$ 10 nm selected with Thorlabs bandpass filters; a beam splitter and filters arranged so that the optical axes are coincident; and a servo-driven mechanical shutter of approximately 100 ms placed between the electrodes to block specular reflections from cathode and plasma (Musikhin et al., 12 Jun 2025).

The measurement strategy is designed around three known error sources: reflections from molten steel surfaces, emissivity variations, and arc instabilities. The 2-color method minimizes error from uncertain or varying surface emissivity, while the shutter blocks lines of sight to reflective objects during measurement. High-speed imaging resolves rapid temperature and position changes associated with arc spot motion and Marangoni-driven flow in the molten pool. Angled views are also used when possible to reduce direct reflection (Musikhin et al., 12 Jun 2025).

Temperature fields are converted into ablation rates through a diffusion-limited evaporation model derived from the Hertz-Knudsen equation and Fick’s laws. Images are segmented by isotherms, the surface area of each isothermal segment is determined from the measured projection, and the total ablation is obtained from

$I_{\rm total} = \sum_i A_i g_{\rm ablation}(T_i).$

For a steel anode, the condensation distance $d$ is treated as a fit parameter and was found to be 4.5 mm. In pure argon, calculated rates agreed well with mass-loss measurements, with $R^2 = 0.77$ for the model fit. Addition of methane at 2.4 wt.% raises the arc voltage needed to maintain the same current, increases power input, but reduces effective ablation rate from 10–40 mg/h in Ar to 2–8 mg/h in Ar/CH$_{2-x}$0 at similar apparent surface temperature (Musikhin et al., 12 Jun 2025).

These results define a path toward closed-loop automated control. Real-time, spatially and temporally resolved temperature maps permit ablation-rate prediction and therefore active adjustment of arc current, voltage, electrode gap, or gas composition. The same report identifies LabVIEW or a similar environment for synchronized data acquisition and actuation, automated isotherm and area analysis, calibration against blackbody sources, and protection of optics against deposition as practical design strategies for automated DC arc melters (Musikhin et al., 12 Jun 2025).

6. Materials synthesis outcomes and demonstrated capabilities

Automated arc melting has been demonstrated on the synthesis of the well-known superconductor V$_{2-x}$1Ge from stoichiometric vanadium and germanium. Phase formation was confirmed by XRD with no impurity peaks, and magnetic susceptibility showed full shielding close to the theoretical value $_{2-x}$2. Multiple samples were synthesized, and their physical characterization was reported as highly consistent, with improved reproducibility attributed to precise, automated process control. For recipes involving two to four components, asynchronous execution reduced total runtime relative to fully sequential execution (Wang et al., 15 Sep 2025).

Arc melting also functions as a rapid materials-processing route outside the fully automated context. For Sb-doped Bi$_{2-x}$3Te$_{2-x}$4, a straightforward and fast arc-melting procedure produced strongly textured, nanostructured polycrystalline pellets composed of stacked platelets with thickness below 50 nm and lateral size in the 200–1000 nm range. For Bi$_{2-x}$5Sb$_{2-x}$6Te$_{2-x}$7, the Seebeck coefficient reached 350 $_{2-x}$8V·K$_{2-x}$9 at 395 K, thermal conductivity was as low as 0.56 W·m$_x$0·K$_x$1 at 309 K, electrical resistivity was 28–35 $_x$2·m at 300 K, and the figure of merit exceeded 1.1 at 395 K (Serrano-Sánchez et al., 2018).

Within the broader UHV-compatible preparation chain, the operational aspects were illustrated through the single-crystal growth of the heavy-fermion compound CeNi$_x$3Ge$_x$4. The arc melting furnace was designed to interoperate with a rod casting furnace and an optical floating-zone furnace, so that melting can serve as a preparatory stage for downstream homogenization, rod fabrication, and single-crystal growth (Bauer et al., 2016).

These demonstrations indicate that automated arc melting systems are relevant to both exploratory bulk intermetallic synthesis and to tightly controlled upstream processing in multistage crystal-growth infrastructures.

7. Limitations, misconceptions, and development trajectory

The published systems also define clear technical constraints. Fine metal powders are not handled in the ROS2-based platform in order to avoid dispersion and clogging, so there are upper and lower particle-size limits for feedstock. Devices lacking automation interfaces are not plug-and-play and require initial setup through a PLC or custom microcontroller solutions. Physical access and laboratory layout may also limit reconfigurability (Wang et al., 15 Sep 2025).

In situ monitoring introduces a different set of constraints. Reflections can corrupt temperature measurements on reflective molten metals; arc and molten-pool instabilities create high-frequency, spatially non-uniform temperature fields; emissivity drift remains a residual error source even with the 2-color method; and harsh plasma environments necessitate protection of optics and reliable actuation of mechanical components. In hydrocarbon atmospheres, carburization can mask ablation loss so that gravimetric evaluation of the anode before and after a run becomes unfeasible; under those conditions, pyrometry-derived measurements become essential (Musikhin et al., 12 Jun 2025).

A common misconception is that automation primarily concerns robotic motion. The available literature shows instead that the decisive bottlenecks are atmosphere integrity, contamination control, repeatable arc placement, reliable sample flipping, and synchronized data handling. Another misconception is that arc melting is intrinsically a coarse, non-diagnostic method. The combination of UHV-compatible furnace construction, glovebox loading, high-speed pyrometry, and ROS2-based orchestration demonstrates that arc melting can be embedded in a high-purity, digitally traceable, and increasingly model-based synthesis environment (Bauer et al., 2016, Wang et al., 15 Sep 2025).

The development trajectory is therefore toward greater modularity, more complete closed-loop control, and tighter coupling between data-driven materials exploration and experimental validation. This suggests that future automated arc melting systems will increasingly be judged not only by melting capability, but also by interoperability, measurement fidelity, and the quality of the digital record produced during synthesis.