Page 79 - Read Online

Na et al. Soft Sci. 2026, 6, 25 Page 11 of 22

Figure 5. Challenges for zero-power optoelectronic synapses in future wearable neuromorphic systems.

large-area integration with mechanical softness. In this section, we discuss five key challenges [Figure 5],

including limited light utilization, insufficient adaptive learning capability, environmental instability, lack of
system-level integration, and mechanical incompatibility.

Moreover, we provide a forward-looking perspective on how these challenges may be addressed by drawing
on insights from prior advances in optoelectronics, wearable electronics, and soft material systems [Figure 6].
Rather than focusing on isolated device improvements, we outline possible technological directions for
overcoming key limitations, thereby offering a roadmap toward practical and scalable zero-power
neuromorphic wearable platforms. In the following sections, detailed strategies are discussed for each
challenge, highlighting how coordinated progress in materials design, device physics, and system-level
architecture could collectively enable practical, scalable, and truly zero-power neuromorphic wearable
platforms.

Limited light utilization
The performance of zero-power optoelectronic synapses is determined by not only how efficiently light is
absorbed and converted into electrical or photothermal signals, but also how effectively these light-induced
processes modulate and retain synaptic weights. In zero-power systems, where no external electrical bias is
available to assist charge separation or amplification, achieving sufficient photoresponsivity becomes even
more critical for sensing-in-memory operation . A higher responsivity further expands the dynamic range
[63]
of photocurrent and enables finer control over the synaptic weights, thereby increasing the number of
accessible conductance states and improving the precision of learning [64,65] . Conversely, insufficient
responsivity restricts synaptic modulation to a narrow multi-level synaptic weight, leading to coarse weight
updates and degraded learning accuracy. This limitation is particularly detrimental for neuromorphic
systems that require gradual and analog weight tuning to emulate biological learning processes. However, in
most current devices, the internal driving forces generated by built-in potentials or thermal gradients are
relatively weak, resulting in low photocurrent and limited weight modulation [42,66] . Consequently, strong
illumination is often required to trigger learning behavior, which contradicts the zero-power concept and
limits applicability under ambient lighting or typical visual perception conditions. Therefore, enhancing