An Autonomous Chemical Robot Discovers the Rules of Inorganic Coordination Chemistry without Prior Knowledge

Abstract

We present a chemical discovery robot for the efficient and reliable discovery of supramolecular architectures through the exploration of a huge reaction space exceeding ten billion combinations. The system was designed to search for areas of reactivity found through autonomous selection of the reagent types, amounts, and reaction conditions aiming for combinations that are reactive. The process consists of two parts where reagents are mixed together, choosing from one type of aldehyde, one amine and one azide (from a possible family of two amines, two aldehydes and four azides) with different volumes, ratios, reaction times, and temperatures, whereby the reagents are passed through a copper coil reactor. Next, either cobalt or iron is added, again from a large number of possible quantities. The reactivity was determined by evaluating differences in pH, UV-Vis, and mass spectra before and after the search was started. The algorithm was focused on the exploration of interesting regions, as defined by the outputs from the sensors, and this led to the discovery of a range of 1-benzyl-(1,2,3-triazol-4-yl)-N-alkyl-(2-pyridinemethanimine) ligands and new complexes: [Fe(L¹ )₂ ](ClO₄ )₂ (1); [Fe(L² )₂ ](ClO₄ )₂ (2); [Co₂ (L³ )₂ ](ClO₄ )₄ (3); [Fe₂ (L³ )₂ ](ClO₄ )₄ (4), which were crystallised and their structure confirmed by single-crystal X-ray diffraction determination, as well as a range of new supramolecular clusters discovered in solution using high-resolution mass spectrometry.

Keywords: algorithms; artificial intelligence; autonomous discovery robot; supramolecular chemistry.

Figure 1

A) Connection diagram with the prominent connections of the system highlighted. B) Photograph of the system with major features annotated.

Scheme 1

Small organic molecules used as chemical inputs for the synthesis of ligands, via tandem CuAAC and imine formation reactions, prior to complexation.

Figure 2

A diagram depicting the connections in our system as a network graph. Nodes depict either locations (in black) or valves (in blue), and the edges signify the physical connections between the nodes. Left: Two equivalent paths, in red and green, to move from node E to node G, the system will randomly choose one of them. Middle: The shortest path to move from node B to node E. Right: If the connections between nodes E and C, and C and B are not clean (shown in brown), for example, in response to the movement of material from B to E (see middle panel), the system chooses a new route. This is shown in orange, for movement from node G to node E, avoiding the unclean route from B to E.

Figure 3

Schematic of the exploration algorithm operating in 2D space. The first point is chosen at random (A). The measured exploration factor, α for Point 1 determines the radius away at which Point 2 is placed (panel B). The larger the value of α, the larger the radius to the next point. This stepwise exploration is continued in panels C and D. Note that, over time, the exploration is concentrated to areas of higher chemical interest (shown as red).

Figure 4

Crystal structures of the isolated compounds. Skeletal structures of ligands L^1–3; 1: [Fe(L¹)₂](ClO₄)₂; 2: [Fe(L²)₂](ClO₄)₂; 3: [Co₂(L³)₂](ClO₄)₄; 4: [Fe₂(L³)₂](ClO₄)₄.

Figure 5

ESI‐MS detection of different possible structures based on the helicate complex 4 (noted here as architecture A). Architectures B and E have the general formula M₂L₃ and would be required to adopt κ²‐coordination through the pyridyl and imine groups. Architectures C and D have the general formula M₂L₂, as does A, and would adopt the same κ³‐coordination motif as the helicates shown in Figure 4. See Table 1 for the calculated and measured peak data.