Design of Intelligent Autocatalytic Reaction Networks
Dmitrii Kriukov is a PhD student in the Department of Molecular Nanofabrication. (Co)Promotors are prof.dr.ir. J. Huskens and dr.ing. S.Y. Wong from the Faculty of Science & Technology.
Chemical reaction networks (CRNs) which contain autocatalytic reactions offer a new frontier for creating next-generation intelligent chemical systems. The nonlinear nature of autocatalytic CRNs enables them to mimic complex, natural decision-making processes ubiquitous in biology on a molecular level. By leveraging autocatalysis under out-of-equilibrium conditions, this work sought to develop design principles for programmable, self-regulating chemical systems that can perform logic operations and signal processing, exhibit temporal and persistent memory, and adapt to environmental changes. This thesis explored and interpreted autocatalytic reaction networks and flow-driven out-of-equilibrium conditions, highlighting their potential as building blocks for future autonomous intelligent systems.
Outlook
Chapter 1 introduced the foundational concepts and motivation behind exploring autocatalytic CRNs. This chapter established the theoretical basis for how autocatalysis, particularly under conditions away from equilibrium, enables chemical systems to operate autonomously like analog computational systems — displaying intelligent functions akin to memory retention, environmental adaptation, and signal processing. By delving into examples from both synthetic chemistry, biology, and physics, it laid out the essential attributes of autocatalysis that make it suitable for constructing systems capable of self-organization and complex response behaviors without needing continuous external input. Furthermore, an overview of the design principles that guide this work was performed, introducing the concept of open systems maintained by a constant influx of chemical resources. The continuous input appeared as a key for creating dynamic, persistent states that respond fluidly to changing conditions, paving the way for so-called emergent behavior and self-organization. The discussion emphasized the implications of designing such intelligent chemical systems, setting a clear trajectory for how autocatalytic networks, when effectively harnessed, could be deployed in applications requiring real-time adaptability and self-regulation. The chapter thus serves as both a theoretical foundation and a call to rethink the potential of chemical networks as practical tools for achieving functional autonomy in chemical systems.
Chapter 2 investigated how varying flowrate impacts the adaptability of trypsinogen autocatalysis in a continuous stirred-tank reactor (CSTR), specifically under conditions of periodic perturbation using linear inhibition sweeps of soybean trypsin inhibitor. The change of flowrate, and thus change of residence time, of the reaction mixture in the CSTR affects the response of an autocatalytic CRN to periodic inhibitory perturbations. The setup allowed for an exploration of the interactions between the frequency of periodic perturbations, the flowrate, and history-dependent switching capabilities within a bistable autocatalytic CRN. The results demonstrated that adjusting flowrate influences the positive feedback articulation, which is essential for the system's stability and adaptation, revealing conditions for controlled memory retention and history-dependent behavior. This chapter enhanced the fundamental understanding of how flowrate modulation can regulate the dynamics of autocatalytic CRNs, providing a basis for potential applications in neuromorphic information processing and self-regulated chemical circuits.
Chapter 3 examined the impact of adding metal ions that can accelerate trypsinogen autocatalysis in a CSTR, with the aim of diversifying the CRN’s behaviors and identifying new emergent intelligent functions. By exploiting ions that modulate the autocatalytic reaction rate in different ways, the study enabled kinetic tuning that transforms the CRN’s response, introducing versatile Boolean logic operations and programmable polynomial functions into the chemical out-of-equilibrium system. This chapter expanded on the role of catalytic modulation in enhancing the programmability of CRNs, showing that certain catalytic modifications allow for nonlinear and history-dependent behavior crucial for molecular information processing. These findings revealed that autocatalytic CRNs can be adapted to perform complex logical calculus, thus moving the fields of systems chemistry and natural computing towards applications in chemical computing and programmable molecular devices, demonstrating the potential for CRNs to execute tasks typically managed by electronic and mechanical systems.
Chapter 4 explored the use of a chlorite-tetrathionate autocatalytic reaction within a continuous tubular reactor, which allows for the self-sustained propagation of flow-supported autocatalytic wavefronts. The fast reaction kinetics and specific tubular geometry allow for the controlled propagation of chemical wavefronts that exhibit hysteresis and spatial patterning due to fluid motion effects, such as Poiseuille flow. Various factors — such as the geometry, the flowrate, and the concentration of reactants — were examined to control the properties of autocatalytic wavefronts, additionally revealing that selective pathfinding within the reactor by a liquid can be visualized by autocatalytic reactions. This chapter demonstrated how flow-supported chemical wavefronts can be modulated through reactor design, enabling CRNs to serve in applications requiring spatial information processing, such as in communication within microfluidic compartments and in spatially organized chemical sensing. This study underscores the versatility of CRNs in propagating chemical information through media and adapting to spatially heterogeneous environments, highlighting the potential for these systems in applications towards complex chemical dynamics in complex fluidic architectures.
Overall, the findings described in this work can be successfully incorporated into the rational design of new autocatalytic CRNs and fluidic platforms for them. This thesis brings society closer to exploiting fundamental properties of autocatalysis towards new sustainable technologies of computation and analog control over chemical systems inspired by living matter, which is much more effective for solving complex tasks [1] than contemporary semiconductor technology of computation.
