All-optical nonlinear activation functions realized on phase-change photonic integrated circuits with microheaters

Jiyuan Jiang; Bingxin Ding; Shiyu Li; Xin Zhang; Haihua Wang; Jie Wu; Xiaoyan Liu; Zhou Wang; Xiaojuan Lian; Wen Huang; Lei Wang

doi:10.1088/1674-4926/24090045

J. Semicond. > 2025, Volume 46 > Issue 2 > 022405

SPECIAL ISSUE ARTICLES

All-optical nonlinear activation functions realized on phase-change photonic integrated circuits with microheaters

Jiyuan Jiang¹, Bingxin Ding¹, Shiyu Li¹, Xin Zhang¹, Haihua Wang¹, Jie Wu¹, Xiaoyan Liu¹, Zhou Wang¹, Xiaojuan Lian^1,, Wen Huang^{2, 3,} and Lei Wang^{1, 4,}

+ Author Affiliations

1. School of Integrated Circuit Science and Engineering, Nanjing University of Posts and Telecommunications, Nanjing 210023, China
2. School of Science, Nanjing University of Posts and Telecommunications, Nanjing 210023, China
3. The State Key Laboratory of Silicon and Advanced Semiconductor for Materials, Zhejiang University, Hangzhou 310027, China
4. Nantong Institute of Nanjing University of Posts and Telecommunications, Nantong 226021, China

Author Bio:

Jiyuan Jiang got BS from Wuhan Textile University in 2022. Now he is a MS student at Nanjing University of Posts and Telecommunications under the Prof. Xiaojuan Lian. His research focuses on neuromorphic computing based on non-volatile devices.

Bingxin Ding got BS from Yancheng Teachers University in 2021. Now she is a MS student at Nanjing University of Posts and Telecommunications under the Prof. Lei Wang. Her research focuses on the fabrication, mechanisms, and neuromorphic applications of novel MBene-Based Memristors.

Xiaojuan Lian got her BS degree in electronic science and technology and the MS degree in physical electronics from Xidian University in 2008 and 2011 respectively. She received her PhD degree in electrical engineering from the Universitat Autònoma de Barcelona, Spain, in 2014. She is currently an associate professor at the College of Integrated Circuit Science and Engineering, Nanjing University of Posts and Telecommunications. Her research interests include memristive devices (RRAM, PCRAM and so on), information storage technology, and neuromorphic computing applications.

Wen Huang received his Ph.D. degree in the School of Physics and Astronomy at Uppsala University in 2017. He is now an associate professor in the School of Science at Nanjing University of Posts and Telecommunications. His current research interest is mainly centered on semiconductor based memristors and optoelectronic synaptic devices.

Lei Wang received the PhD degree in "Tbit/sq.in. scanning probe phase-change memory" from the University of Exeter, Exeter, U.K., in 2009. Between 2008 and 2011, he was employed as a Postdoctoral Research Fellow in the University of Exeter to work on a fellowship funded by European Commission. These works included the study of phase-change probe memory and phase-change memristor. Since 2020, he joined the Nanjing University of Posts and Telecommunications, Nanjing, P. R. China as a Professor, where he is engaged in the phase-change memories, phase-change neural networks, and other phase-change based optoelectronic devices and their potential applications.

Corresponding author: Xiaojuan Lian, xjlian@njupt.edu.cn; Wen Huang, wenhuang@njupt.edu.cn; Lei Wang, leiwang1980@njupt.edu.cn

DOI: 10.1088/1674-4926/24090045CSTR: 32376.14.1674-4926.24090045

Abstract: Photonic neural networks have garnered significant attention in recent years due to their ultra-high computational speed, broad bandwidth, and parallel processing capabilities. However, compared to conventional electronic nonlinear activation function (NAF), progress on efficient and easily implementable optical nonlinear activation function (ONAF) was barely reported. To address this issue, we proposed a programmable, low-loss ONAF device based on a silicon micro-ring resonator capped with the Antimony selenide (Sb₂Se₃) thin films, and with indium tin oxide (ITO) used as the microheater. Leveraging our self-developed phase-transformation kinetic and optical models, we successfully simulated the phase-transition behavior of Sb₂Se₃ and three different ONAFs—ELU, ReLU, and radial basis function (RBF) were achieved according to discernible optical responses of proposed devices under different phase-change extents. Classification results from the Fashion MNIST dataset demonstrated that these ONAFs can be considered as appropriate substitutes for traditional NAF. This indicated the bright prospect of the proposed device for nonlinear activation function in future photonic neural networks.

Key words: ONAF, Sb₂Se₃, microheater, photonic neural networks

References

[1]	Ikotun A M, Ezugwu A E, Abualigah L, et al. K-means clustering algorithms: A comprehensive review, variants analysis, and advances in the era of big data. Inf Sci, 2023, 622, 178 doi: 10.1016/j.ins.2022.11.139
[2]	Liu L K, Lu S D, Zhong R, et al. Computing systems for autonomous driving: State of the art and challenges. IEEE Internet Things J, 2021, 8, 6469 doi: 10.1109/JIOT.2020.3043716
[3]	Mohammed Sadeeq M, Abdulkareem N M, Zeebaree S R M, et al. IoT and cloud computing issues, challenges and opportunities: A review. Qubahan Acad J, 2021, 1, 1 doi: 10.48161/qaj.v1n2a36
[4]	Arikpo I I, Ogban F U, Eteng I E. Von Neumann architecture and modern computers. Glo Jnl Math Scie, 2008, 6, 97
[5]	Theis T N, Wong H S P. The end of Moore’s law: A new beginning for information technology. Comput Sci Eng, 2017, 19, 41 doi: 10.1109/MCSE.2017.29
[6]	Bengio Y, LeCun Y, Hinton G. Deep learning for AI. Commun ACM, 2021, 64, 58 doi: 10.1145/3448250
[7]	Shastri B J, Tait A N, Ferreira de Lima T, et al. Photonics for artificial intelligence and neuromorphic computing. Nat Photonics, 2021, 15, 102 doi: 10.1038/s41566-020-00754-y
[8]	LeCun Y, Bengio Y, Hinton G. Deep learning. Nature, 2015, 521, 436 doi: 10.1038/nature14539
[9]	Janiesch C, Zschech P, Heinrich K. Machine learning and deep learning. Electron Mark, 2021, 31, 685 doi: 10.1007/s12525-021-00475-2
[10]	Wetzstein G, Ozcan A, Gigan S, et al. Inference in artificial intelligence with deep optics and photonics. Nature, 2020, 588, 39 doi: 10.1038/s41586-020-2973-6
[11]	Dong B W, Brückerhoff-Plückelmann F, Meyer L, et al. Partial coherence enhances parallelized photonic computing. Nature, 2024, 632, 55 doi: 10.1038/s41586-024-07590-y
[12]	Woods D, Naughton T J. Photonic neural networks. Nat Phys, 2012, 8, 257 doi: 10.1038/nphys2283
[13]	De Marinis L, Cococcioni M, Castoldi P, et al. Photonic neural networks: A survey. IEEE Access, 2019, 7, 175827 doi: 10.1109/ACCESS.2019.2957245
[14]	Parhi R, Nowak R D. The role of neural network activation functions. IEEE Signal Process Lett, 2020, 27, 1779 doi: 10.1109/LSP.2020.3027517
[15]	Sharma S, Sharma S, Athaiya A. Activation functions in neural networks. Towards Data Sci, 2017, 6, 310
[16]	Feldmann J, Youngblood N, Karpov M, et al. Parallel convolutional processing using an integrated photonic tensor core. Nature, 2021, 589, 52 doi: 10.1038/s41586-020-03070-1
[17]	Zhou W, Dong B W, Farmakidis N, et al. In-memory photonic dot-product engine with electrically programmable weight banks. Nat Commun, 2023, 14, 2887 doi: 10.1038/s41467-023-38473-x
[18]	Lian X J, Jiang J Y, Tao Z H, et al. Electrically programmable phase-change nanophotonic circuits designed for edge detection application. IEEE Electron Device Lett, 2024, 45, 1101 doi: 10.1109/LED.2024.3406131
[19]	Li C Y, Qian G D, Cui Y J. Metal–organic frameworks for nonlinear optics and lasing. Inf Funct Mater, 2024, 1, 125 doi: 10.1002/ifm2.17
[20]	Miscuglio M, Mehrabian A, Hu Z B, et al. All-optical nonlinear activation function for photonic neural networks. Opt Mater Express, 2018, 8, 3851 doi: 10.1364/OME.8.003851
[21]	Feldmann J, Youngblood N, Wright C D, et al. All-optical spiking neurosynaptic networks with self-learning capabilities. Nature, 2019, 569, 208 doi: 10.1038/s41586-019-1157-8
[22]	Jha A, Huang C R, Prucnal P R. Reconfigurable all-optical nonlinear activation functions for neuromorphic photonics. Opt Lett, 2020, 45, 4819 doi: 10.1364/OL.398234
[23]	Rasamoelina A D, Adjailia F, Sinčák P. A review of activation function for artificial neural network. 2020 IEEE 18th World Symposium on Applied Machine Intelligence and Informatics (SAMI), 2020, 281 doi: 10.1109/SAMI48414.2020.9108717
[24]	Banerjee C, Mukherjee T, Pasiliao E Jr. An empirical study on generalizations of the ReLU activation function. Proceedings of the 2019 ACM Southeast Conference, 2019, 164 doi: 10.1145/3299815.3314450
[25]	Huang G B, Siew C K. Extreme learning machine: RBF network case. ICARCV 2004 8th Control, Automation, Robotics and Vision Conference, 2004, 1029 doi: 10.1109/ICARCV.2004.1468985
[26]	Wuttig M, Bhaskaran H, Taubner T. Phase-change materials for non-volatile photonic applications. Nat Photonics, 2017, 11, 465 doi: 10.1038/nphoton.2017.126
[27]	Fu Z L, Wang Z, Bienstman P, et al. Programmable low-power consumption all-optical nonlinear activation functions using a micro-ring resonator with phase-change materials. Opt Express, 2022, 30, 44943 doi: 10.1364/OE.476110
[28]	Kim H J, Sohn J W, Hong N N, et al. PCM-net: A refractive index database of chalcogenide phase change materials for tunable nanophotonic device modelling. J Phys Photonics, 2021, 3, 024008 doi: 10.1088/2515-7647/abeb55
[29]	Delaney M, Zeimpekis I, Lawson D, et al. A new family of ultralow loss reversible phase-change materials for photonic integrated circuits: Sb2S3 and Sb2Se3. Adv Funct Mater, 2020, 30, 2002447 doi: 10.1002/adfm.202002447
[30]	Van Vaerenbergh T, Fiers M, Mechet P, et al. Cascadable excitability in microrings. Opt Express, 2012, 20, 20292 doi: 10.1364/OE.20.020292
[31]	Wang Z, Li Q, Fu Z L, et al. Threshold plasticity of hybrid Si-VO₂ microring resonators. Optical Fiber Communication Conference (OFC), 2020, Th2A. 26
[32]	Lugnan A, García-Cuevas Carrillo S, Wright C D, et al. Rigorous dynamic model of a silicon ring resonator with phase change material for a neuromorphic node. Opt Express, 2022, 30, 25177 doi: 10.1364/OE.459364
[33]	Priem G, Dumon P, Bogaerts W, et al. Optical bistability and pulsating behaviour in silicon-on-insulator ring resonator structures. Opt Express, 2005, 13, 9623 doi: 10.1364/OPEX.13.009623
[34]	Barclay P, Srinivasan K, Painter O. Nonlinear response of silicon photonic crystal microresonators excited via an integrated waveguide and fiber taper. Opt Express, 2005, 13, 801 doi: 10.1364/OPEX.13.000801
[35]	Yang C J, Liang L, Qin L, et al. Advances in silicon-based, integrated tunable semiconductor lasers. Nanophotonics, 2023, 12, 197 doi: 10.1515/nanoph-2022-0699
[36]	Huang C R, Jha A, de Lima T F, et al. On-chip programmable nonlinear optical signal processor and its applications. IEEE J Sel Top Quantum Electron, 2021, 27, 6100211 doi: 10.1109/JSTQE.2020.2998073
[37]	Jha A, Huang C R, Peng H T, et al. Photonic spiking neural networks and graphene-on-silicon spiking neurons. J Light Technol, 2022, 40, 2901 doi: 10.1109/JLT.2022.3146157
[38]	Wu B, Li H K, Tong W Y, et al. Low-threshold all-optical nonlinear activation function based on a Ge/Si hybrid structure in a microring resonator. Opt Mater Express, 2022, 12, 970 doi: 10.1364/OME.447330

Fig. 1. (Color online) (a) A schematic of the 3D structure of the micro-ring resonator capped with Sb₂Se₃. (b) A cross-sectional diagram of the device. (c) Top view of the device, ITO microheater above Sb₂Se₃ has not shown. The simulation parameters are: width = 0.5 μm, height = 0.22 μm, gap = 0.3 μm, L_PCM = 0.5 μm, H_PCM = 20 nm, and H_ITO = 10 nm.

DownLoad: Full-Size Img PowerPoint

Fig. 2. (Color online) (a)−(c) Show the time-dependent variations of $ a $, $ \Delta T $, and $ N $ in the micro-ring resonator, respectively. (d) The relationship between the input and output power of the micro-ring resonator, and (e) is a magnified view of the output power from (d). (f) The time evolution of the $ \delta {\omega }_\rm{nl,hy} $.

DownLoad: Full-Size Img PowerPoint

Fig. 3. (Color online) (a) voltage distribution across the Sb₂Se₃ cell at the plateau stage of a 5 V pulse of 500 ns; (b) temperature distribution across the Sb₂Se₃ cell at the plateau stage of a 5 V pulse of 500 ns; (c) front view of temperature profile across the Sb₂Se₃ cell for (b); bottom images described the zoom-in of temperature profile inside the ITO layer in x−z direction (left) and of temperature profile inside the Sb₂Se₃ plane in x−y direction (right), respectively; different crystallization extents of the Sb₂Se₃ cell were afforded by a 500 ns pulse of (d) 5 V, (e) 5.3 V, (f) 5.5 V, (g) 5.8 V, (h) 6 V, and (i) 6.3 V, respectively.

DownLoad: Full-Size Img PowerPoint

Fig. 4. (Color online) Relationship between input and output power for 10, 20, and 30 nm thick Sb₂Se₃ in the crystalline (a), 10% crystallization (b), and 20% crystallization (c). Optical mode of the device with the Sb₂Se₃ in the amorphous state (d) and crystalline state (e). (f) Transmission of the device as a function of the crystallization fraction of the Sb₂Se₃ (inset: magnified view of the transmission near the 1548.911 nm wavelength); (g) ELU-like nonlinear activation behavior observed in the input−output power relationship when the Sb₂Se₃ is in the amorphous state; (h) ReLU-like nonlinear activation behavior observed when the Sb₂Se₃ is 10% crystallized; (i) RBF-like nonlinear activation behavior observed when the Sb₂Se₃ is 20% crystallized. The input and output power in Figs. (a)−(c) and (g)−(i) are normalized.

DownLoad: Full-Size Img PowerPoint

Fig. 5. (Color online) (a) Schematic diagram of the neural network structure used for fashion item image recognition. (b)−(d) Show the recognition accuracy and loss function variations during the training and testing of the neural network using the classical and optical ELU, ReLU, and RBF activation functions, respectively. (e) The confusion matrix illustrates the prediction results obtained with the optical ELU, ReLU, and RBF activation functions.

DownLoad: Full-Size Img PowerPoint

Table 1. Describe the characteristics of the micro-ring resonator, along with the parameters and values used in the simulations.

Parameter	Value	Parameter	Value
$ {\tau }_{\mathrm{t}\mathrm{h}} $	65 ns	$ {V}_{\mathrm{F}\mathrm{C}\mathrm{A}} $	2.36 × 10⁻¹⁸ m³
$ {\tau }_{\mathrm{f}\mathrm{c}} $	5.3 ns	$ {V}_{\mathrm{T}\mathrm{P}\mathrm{A}} $	2.59 × 10⁻¹⁸ m³
$ {\rho }_{\mathrm{S}\mathrm{i}} $	2.33 × 10³ kg·m⁻³	$ {\sigma }_{\mathrm{S}\mathrm{i}} $	10⁻²¹ m²
$ {c}_{\mathrm{p},\mathrm{S}\mathrm{i}} $	700 J·kg⁻¹·K⁻¹	$ \mathrm{d}{n}_{\mathrm{S}\mathrm{i}}/\mathrm{d}T $	1.86 × 10⁻⁴ K⁻¹
$ {V}_{\mathrm{t}\mathrm{h}} $	3.19 × 10⁻¹⁸ m³	$ \mathrm{d}{n}_{\mathrm{S}\mathrm{i}}/\mathrm{d}N $	−1.73 × 10⁻²⁷ m³
$ {\beta }_{\mathrm{S}\mathrm{i}} $	8.4 × 10⁻¹² m·W⁻¹	$ {\gamma }_{\mathrm{c}\mathrm{o}\mathrm{u}\mathrm{p}} $	9.7561 × 10⁻⁹ s⁻¹
$ c $	299 792 458 m·s⁻¹	$ {\alpha }_{\mathrm{S}\mathrm{i},\mathrm{r}\mathrm{i}\mathrm{n}\mathrm{g}} $	16 m⁻¹
$ \hslash $	1.0545718 × 10⁻³⁴ J·s	$ {n}_{\mathrm{g},\mathrm{S}\mathrm{i}} $	3.476

DownLoad: CSV

Table 2. The comparisons with other ONAF components.

Reference	Programmability	Excitation	Energy consumption (nJ)	Scalability
[22]	Yes	Electrical	−	Difficult
[27]	Yes	Optical	1.748	Difficult
[36]	Yes	Electrical	−	Difficult
[37]	No	−	−	Difficult
[38]	No	−	−	Easy
Our work	Yes	Electrical	1.05	Easy

DownLoad: CSV

[1]	Ikotun A M, Ezugwu A E, Abualigah L, et al. K-means clustering algorithms: A comprehensive review, variants analysis, and advances in the era of big data. Inf Sci, 2023, 622, 178 doi: 10.1016/j.ins.2022.11.139
[2]	Liu L K, Lu S D, Zhong R, et al. Computing systems for autonomous driving: State of the art and challenges. IEEE Internet Things J, 2021, 8, 6469 doi: 10.1109/JIOT.2020.3043716
[3]	Mohammed Sadeeq M, Abdulkareem N M, Zeebaree S R M, et al. IoT and cloud computing issues, challenges and opportunities: A review. Qubahan Acad J, 2021, 1, 1 doi: 10.48161/qaj.v1n2a36
[4]	Arikpo I I, Ogban F U, Eteng I E. Von Neumann architecture and modern computers. Glo Jnl Math Scie, 2008, 6, 97
[5]	Theis T N, Wong H S P. The end of Moore’s law: A new beginning for information technology. Comput Sci Eng, 2017, 19, 41 doi: 10.1109/MCSE.2017.29
[6]	Bengio Y, LeCun Y, Hinton G. Deep learning for AI. Commun ACM, 2021, 64, 58 doi: 10.1145/3448250
[7]	Shastri B J, Tait A N, Ferreira de Lima T, et al. Photonics for artificial intelligence and neuromorphic computing. Nat Photonics, 2021, 15, 102 doi: 10.1038/s41566-020-00754-y
[8]	LeCun Y, Bengio Y, Hinton G. Deep learning. Nature, 2015, 521, 436 doi: 10.1038/nature14539
[9]	Janiesch C, Zschech P, Heinrich K. Machine learning and deep learning. Electron Mark, 2021, 31, 685 doi: 10.1007/s12525-021-00475-2
[10]	Wetzstein G, Ozcan A, Gigan S, et al. Inference in artificial intelligence with deep optics and photonics. Nature, 2020, 588, 39 doi: 10.1038/s41586-020-2973-6
[11]	Dong B W, Brückerhoff-Plückelmann F, Meyer L, et al. Partial coherence enhances parallelized photonic computing. Nature, 2024, 632, 55 doi: 10.1038/s41586-024-07590-y
[12]	Woods D, Naughton T J. Photonic neural networks. Nat Phys, 2012, 8, 257 doi: 10.1038/nphys2283
[13]	De Marinis L, Cococcioni M, Castoldi P, et al. Photonic neural networks: A survey. IEEE Access, 2019, 7, 175827 doi: 10.1109/ACCESS.2019.2957245
[14]	Parhi R, Nowak R D. The role of neural network activation functions. IEEE Signal Process Lett, 2020, 27, 1779 doi: 10.1109/LSP.2020.3027517
[15]	Sharma S, Sharma S, Athaiya A. Activation functions in neural networks. Towards Data Sci, 2017, 6, 310
[16]	Feldmann J, Youngblood N, Karpov M, et al. Parallel convolutional processing using an integrated photonic tensor core. Nature, 2021, 589, 52 doi: 10.1038/s41586-020-03070-1
[17]	Zhou W, Dong B W, Farmakidis N, et al. In-memory photonic dot-product engine with electrically programmable weight banks. Nat Commun, 2023, 14, 2887 doi: 10.1038/s41467-023-38473-x
[18]	Lian X J, Jiang J Y, Tao Z H, et al. Electrically programmable phase-change nanophotonic circuits designed for edge detection application. IEEE Electron Device Lett, 2024, 45, 1101 doi: 10.1109/LED.2024.3406131
[19]	Li C Y, Qian G D, Cui Y J. Metal–organic frameworks for nonlinear optics and lasing. Inf Funct Mater, 2024, 1, 125 doi: 10.1002/ifm2.17
[20]	Miscuglio M, Mehrabian A, Hu Z B, et al. All-optical nonlinear activation function for photonic neural networks. Opt Mater Express, 2018, 8, 3851 doi: 10.1364/OME.8.003851
[21]	Feldmann J, Youngblood N, Wright C D, et al. All-optical spiking neurosynaptic networks with self-learning capabilities. Nature, 2019, 569, 208 doi: 10.1038/s41586-019-1157-8
[22]	Jha A, Huang C R, Prucnal P R. Reconfigurable all-optical nonlinear activation functions for neuromorphic photonics. Opt Lett, 2020, 45, 4819 doi: 10.1364/OL.398234
[23]	Rasamoelina A D, Adjailia F, Sinčák P. A review of activation function for artificial neural network. 2020 IEEE 18th World Symposium on Applied Machine Intelligence and Informatics (SAMI), 2020, 281 doi: 10.1109/SAMI48414.2020.9108717
[24]	Banerjee C, Mukherjee T, Pasiliao E Jr. An empirical study on generalizations of the ReLU activation function. Proceedings of the 2019 ACM Southeast Conference, 2019, 164 doi: 10.1145/3299815.3314450
[25]	Huang G B, Siew C K. Extreme learning machine: RBF network case. ICARCV 2004 8th Control, Automation, Robotics and Vision Conference, 2004, 1029 doi: 10.1109/ICARCV.2004.1468985
[26]	Wuttig M, Bhaskaran H, Taubner T. Phase-change materials for non-volatile photonic applications. Nat Photonics, 2017, 11, 465 doi: 10.1038/nphoton.2017.126
[27]	Fu Z L, Wang Z, Bienstman P, et al. Programmable low-power consumption all-optical nonlinear activation functions using a micro-ring resonator with phase-change materials. Opt Express, 2022, 30, 44943 doi: 10.1364/OE.476110
[28]	Kim H J, Sohn J W, Hong N N, et al. PCM-net: A refractive index database of chalcogenide phase change materials for tunable nanophotonic device modelling. J Phys Photonics, 2021, 3, 024008 doi: 10.1088/2515-7647/abeb55
[29]	Delaney M, Zeimpekis I, Lawson D, et al. A new family of ultralow loss reversible phase-change materials for photonic integrated circuits: Sb2S3 and Sb2Se3. Adv Funct Mater, 2020, 30, 2002447 doi: 10.1002/adfm.202002447
[30]	Van Vaerenbergh T, Fiers M, Mechet P, et al. Cascadable excitability in microrings. Opt Express, 2012, 20, 20292 doi: 10.1364/OE.20.020292
[31]	Wang Z, Li Q, Fu Z L, et al. Threshold plasticity of hybrid Si-VO₂ microring resonators. Optical Fiber Communication Conference (OFC), 2020, Th2A. 26
[32]	Lugnan A, García-Cuevas Carrillo S, Wright C D, et al. Rigorous dynamic model of a silicon ring resonator with phase change material for a neuromorphic node. Opt Express, 2022, 30, 25177 doi: 10.1364/OE.459364
[33]	Priem G, Dumon P, Bogaerts W, et al. Optical bistability and pulsating behaviour in silicon-on-insulator ring resonator structures. Opt Express, 2005, 13, 9623 doi: 10.1364/OPEX.13.009623
[34]	Barclay P, Srinivasan K, Painter O. Nonlinear response of silicon photonic crystal microresonators excited via an integrated waveguide and fiber taper. Opt Express, 2005, 13, 801 doi: 10.1364/OPEX.13.000801
[35]	Yang C J, Liang L, Qin L, et al. Advances in silicon-based, integrated tunable semiconductor lasers. Nanophotonics, 2023, 12, 197 doi: 10.1515/nanoph-2022-0699
[36]	Huang C R, Jha A, de Lima T F, et al. On-chip programmable nonlinear optical signal processor and its applications. IEEE J Sel Top Quantum Electron, 2021, 27, 6100211 doi: 10.1109/JSTQE.2020.2998073
[37]	Jha A, Huang C R, Peng H T, et al. Photonic spiking neural networks and graphene-on-silicon spiking neurons. J Light Technol, 2022, 40, 2901 doi: 10.1109/JLT.2022.3146157
[38]	Wu B, Li H K, Tong W Y, et al. Low-threshold all-optical nonlinear activation function based on a Ge/Si hybrid structure in a microring resonator. Opt Mater Express, 2022, 12, 970 doi: 10.1364/OME.447330

Search

GET CITATION

shu

Export: BibTex EndNote

Article Metrics

Article views: 830 Times PDF downloads: 155 Times Cited by: 0 Times

History

Received: 23 September 2024 Revised: 18 October 2024 Online: Accepted Manuscript: 06 November 2024Uncorrected proof: 02 January 2025Published: 15 February 2025

Article Navigation > Journal of Semiconductors > 2025 > 46(2): 022405

Jiyuan Jiang, Bingxin Ding, Shiyu Li, Xin Zhang, Haihua Wang, Jie Wu, Xiaoyan Liu, Zhou Wang, Xiaojuan Lian, Wen Huang, Lei Wang. All-optical nonlinear activation functions realized on phase-change photonic integrated circuits with microheaters[J]. Journal of Semiconductors, 2025, 46(2): 022405. doi: 10.1088/1674-4926/24090045 ****J Y Jiang, B X Ding, S Y Li, X Zhang, H H Wang, J Wu, X Y Liu, Z Wang, X J Lian, W Huang, and L Wang, All-optical nonlinear activation functions realized on phase-change photonic integrated circuits with microheaters[J]. J. Semicond., 2025, 46(2), 022405 doi: 10.1088/1674-4926/24090045

Citation:

J Y Jiang, B X Ding, S Y Li, X Zhang, H H Wang, J Wu, X Y Liu, Z Wang, X J Lian, W Huang, and L Wang, All-optical nonlinear activation functions realized on phase-change photonic integrated circuits with microheaters[J]. J. Semicond., 2025, 46(2), 022405 doi: 10.1088/1674-4926/24090045

Citation:

PDF( 9499 KB)

All-optical nonlinear activation functions realized on phase-change photonic integrated circuits with microheaters

DOI: 10.1088/1674-4926/24090045

CSTR: 32376.14.1674-4926.24090045

1.
School of Integrated Circuit Science and Engineering, Nanjing University of Posts and Telecommunications, Nanjing 210023, China
2.
School of Science, Nanjing University of Posts and Telecommunications, Nanjing 210023, China
3.
The State Key Laboratory of Silicon and Advanced Semiconductor for Materials, Zhejiang University, Hangzhou 310027, China
4.
Nantong Institute of Nanjing University of Posts and Telecommunications, Nantong 226021, China

More Information

Jiyuan Jiang got BS from Wuhan Textile University in 2022. Now he is a MS student at Nanjing University of Posts and Telecommunications under the Prof. Xiaojuan Lian. His research focuses on neuromorphic computing based on non-volatile devices
Bingxin Ding got BS from Yancheng Teachers University in 2021. Now she is a MS student at Nanjing University of Posts and Telecommunications under the Prof. Lei Wang. Her research focuses on the fabrication, mechanisms, and neuromorphic applications of novel MBene-Based Memristors
Xiaojuan Lian got her BS degree in electronic science and technology and the MS degree in physical electronics from Xidian University in 2008 and 2011 respectively. She received her PhD degree in electrical engineering from the Universitat Autònoma de Barcelona, Spain, in 2014. She is currently an associate professor at the College of Integrated Circuit Science and Engineering, Nanjing University of Posts and Telecommunications. Her research interests include memristive devices (RRAM, PCRAM and so on), information storage technology, and neuromorphic computing applications
Wen Huang received his Ph.D. degree in the School of Physics and Astronomy at Uppsala University in 2017. He is now an associate professor in the School of Science at Nanjing University of Posts and Telecommunications. His current research interest is mainly centered on semiconductor based memristors and optoelectronic synaptic devices
Lei Wang received the PhD degree in "Tbit/sq.in. scanning probe phase-change memory" from the University of Exeter, Exeter, U.K., in 2009. Between 2008 and 2011, he was employed as a Postdoctoral Research Fellow in the University of Exeter to work on a fellowship funded by European Commission. These works included the study of phase-change probe memory and phase-change memristor. Since 2020, he joined the Nanjing University of Posts and Telecommunications, Nanjing, P. R. China as a Professor, where he is engaged in the phase-change memories, phase-change neural networks, and other phase-change based optoelectronic devices and their potential applications
Corresponding author: xjlian@njupt.edu.cn; wenhuang@njupt.edu.cn; leiwang1980@njupt.edu.cn
Received Date: 2024-09-23
Revised Date: 2024-10-18

Available Online: 2024-11-06

Abstract

Abstract

Photonic neural networks have garnered significant attention in recent years due to their ultra-high computational speed, broad bandwidth, and parallel processing capabilities. However, compared to conventional electronic nonlinear activation function (NAF), progress on efficient and easily implementable optical nonlinear activation function (ONAF) was barely reported. To address this issue, we proposed a programmable, low-loss ONAF device based on a silicon micro-ring resonator capped with the Antimony selenide (Sb₂Se₃) thin films, and with indium tin oxide (ITO) used as the microheater. Leveraging our self-developed phase-transformation kinetic and optical models, we successfully simulated the phase-transition behavior of Sb₂Se₃ and three different ONAFs—ELU, ReLU, and radial basis function (RBF) were achieved according to discernible optical responses of proposed devices under different phase-change extents. Classification results from the Fashion MNIST dataset demonstrated that these ONAFs can be considered as appropriate substitutes for traditional NAF. This indicated the bright prospect of the proposed device for nonlinear activation function in future photonic neural networks.
- ONAF,
- Sb₂Se₃,
- microheater,
- photonic neural networks

FullText(HTML)

References(38)