Localized Carbon Deposition Enables Non-Invasive Trimming of Photonic Integrated Circuits

Localized Carbon Deposition Enables Non-Invasive Trimming of Photonic Integrated Circuits | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Localized Carbon Deposition Enables Non-Invasive Trimming of Photonic Integrated Circuits Wolfram Pernice, Rongyang Xu, Zhongyu Tang, Liam McRae, Akhil Varri, and 12 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7410735/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted You are reading this latest preprint version Abstract Photonic integrated circuits (PICs), widely used in optical communications and computing, require precise post-fabrication trimming due to their high sensitivity to fabrication imperfections. Focused ion beam (FIB) carbon deposition offers a promising solution, combining high spatial precision with minimal invasiveness and inherently broad material compatibility. Here, we demonstrate this technique for the first time to enable non-volatile post-fabrication trimming of PICs. To validate this approach, we use asymmetric directional couplers as representative fabrication-sensitive components. Structural characterizations verify the non-invasive nature of the localized carbon deposition, and device measurements show discrete transmission tuning levels of 1.46–16.1 dB with a trimming-induced insertion loss of only 0.3 dB. Furthermore, the optical response remains stable over two months following a brief initial settling phase. These results establish FIB carbon deposition as a robust and broadly applicable PIC trimming technique, empowering optical communications, computing, and beyond to meet rising data demands. Physical sciences/Optics and photonics/Optical materials and structures Physical sciences/Nanoscience and technology/Nanoscale materials/Synthesis and processing Physical sciences/Optics and photonics/Optical physics/Nanophotonics and plasmonics Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Introduction A photonic integrated circuit (PIC) is a microchip that brings together multiple photonic components with different functions. Due to the inherent properties of light, such as high speed, low propagation loss, and high parallelism, PICs have been widely explored in applications such as optical communications 1 – 4 , quantum computing 5 – 7 , and neuromorphic computing 8 – 10 . The outstanding performance of PICs in these applications relies on the proper operation of precisely engineered photonic structures, which are often vulnerable to fabrication imperfections. The accumulation of optical response deviations with increasing device count has thus made fabrication imperfections a key obstacle to the further development of PICs. Post-fabrication trimming provides an effective approach to correct such imperfections, offering precise, localized, and flexible control over photonic device performance 11 – 13 . Various post-fabrication trimming techniques have been proposed to address fabrication imperfections, including micro-heaters 14 , 15 , electron beam exposure of polymer claddings 16 – 18 , and focused ion beam processing 15 , 19 . The trimming method based on micro-heaters usually requires continuous heating to maintain the desired tuning states, resulting in significant power consumption and can cause thermal crosstalk. Polymer cladding trimming is a simple and cost-effective method, but it generally cannot guarantee long-term stability. Focused ion beam (FIB) techniques have been widely used in nanofabrication due to their high spatial resolution and mask-free characteristics 20 . In recent years, silicon ion implantation using FIB has emerged as a promising technique for post-fabrication trimming, offering permanent refractive index changes with low optical loss and excellent stability 19 , 21 . However, this approach faces challenges in achieving delicate fine-tuning. The strong refractive index contrast between the amorphous silicon-rich regions and the silicon nitride (SiN) waveguide leads to drastic changes in the field distribution of the guided mode, making nanometer-scale tuning highly exacting and sometimes unattainable. In addition, the high-energy ion beam can cause structural damage, restricting its applicability primarily to platforms such as silicon and SiN. FIB deposition of amorphous carbon as a non-invasive trimming technology has great potential for achieving platform-independent fine-tuning. This technique utilizes a focused ion beam to locally decompose precursor gas molecules, enabling precise deposition of diamond-like amorphous carbon 22 – 24 . Over the past few decades, the technique has been mainly used for the fabrication of 3D nanostructures 20 , 25 , 26 , though diamond-like amorphous carbon can also be used in anti-reflection coating 27 – 29 , protective coatings 30 , 31 , and as hard masks for etching 32 , 33 . However, its application in post-fabrication trimming of PICs has not yet been explored. Given that FIB is a mature technology with equipment widely available in research facilities and advanced manufacturing laboratories, this presents a valuable opportunity for precise post-fabrication trimming in PICs. To validate the trimming capability of FIB carbon deposition, we investigated asymmetric directional couplers (A-DCs) as a representative device. A-DCs are widely used for on-chip mode conversion 34 – 36 and rely on precise phase matching between selected modes, which occurs when their effective refractive indices ( n eff ) are equal at a given wavelength. Because of this strong dependence on precise phase matching, A-DCs are sensitive to fabrication imperfections, making them an ideal platform for evaluating post-fabrication trimming techniques. In this study, we demonstrate FIB carbon deposition as a non-invasive technique with great stability that can be used for high-resolution post-fabrication trimming of PICs. To validate its effectiveness, we employ A-DCs as a sensitive test structure, where trimming allows precise control of device responses. We also quantify the trimming-induced insertion loss, accessible tuning range, and long-term stability, confirming the method’s flexibility and robustness for post-fabrication trimming in PICs. Results To visualize the trimming process, consider an A-DC composed of a narrow waveguide (width w 0 ) placed adjacent to a wider waveguide (width w 1 ), separated by a gap g . Both waveguides share the same height h and sit atop a buried oxide (BOX) layer on a silicon substrate. Trimming begins when a carbon-containing precursor gas is directed toward the narrow waveguide through a gas nozzle, as shown in Fig. 1 a. The Ga + ion beam induces local decomposition of the precursor gas molecules when exposed to the waveguide, resulting in the accumulation of carbon in the irradiation region with nanometer scale precision (see Supplementary S1 for the actual FIB sample stage diagram). In this study, we chose to deposit carbon on the narrow waveguide to control the n eff of the TE 0 mode for demonstration purposes. Carbon can also be deposited on the wide waveguide to control higher-order modes 21 . Complementing the schematic, a fabricated A-DC based on the TE 2 mode (see Supplementary S2 for fabrication process) is shown in Fig. 1 b. Light is input via a grating coupler and is equally split by a multimode interferometer. Half of the input light is directed to the output grating coupler (indicated by the white dashed box) and measured by a photodetector. The resulting transmission spectrum is used as the reference for normalization. The first A-DC converts the remaining light from the TE 0 mode to the TE 2 mode. After conversion, the light propagates along the bus waveguide in the TE 2 mode and is converted back to the TE 0 mode by the second A-DC before reaching the output grating coupler. For simplicity, we assume in this study that the measured insertion loss is dominated by the mode conversions. Building on the device structure, we simulated the mode conversion behavior of A-DCs based on the TE 1 and TE 2 modes, using parameters of w 0 = 1200 nm, g = 300 nm, and h = 335 nm. By adjusting the bus waveguide width, an insertion loss of 0.055 dB for the TE 1 mode and 0.045 dB for the TE 2 mode is achieved for the A-DC at 1550 nm, as shown in Fig. 1 c and 1 f. The insets show that the TE 0 mode light in the narrow waveguide is converted into the higher-order mode light in the wide waveguide. Figure 1 d and 1 g exhibit the transmission of the A-DCs at a wavelength of 1550 nm under different w 1 values, where the highest transmission appears at w 1 = 2550 nm and 3900 nm, respectively. These structures were then fabricated for verification. Experimentally, there was only minimal deviation from the expected waveguide widths, mainly owing to the shrinkage of resist during the hot plate baking process. We observe the measured insertion losses to be 0.4 dB and 0.5 dB with waveguide widths of 2475 nm and 3750 nm for TE 1 and TE 2 modes, in that order. In addition, the observed insertion losses of 2.7 dB and 7.4 dB for the TE 1 and TE 2 modes at w 1 = 2550 nm and 3900 nm, respectively, further highlight the necessity of a wide tuning range for mode control. In the experiment, the w 1 value that achieved the highest transmission was slightly lower than the simulation results for the two modes. This is because resist shrinkage has a greater influence on the n eff of the TE 0 mode in the narrow waveguide, which can be compensated by depositing carbon. Figure 2 a shows the n eff of the TE 0 mode can be flexibly controlled by changing the carbon geometry. The refractive index n and extinction coefficient k of the carbon in our simulation models are 2.48 and 0.019, respectively 37 . Due to its small geometric dimensions and moderate refractive index, the deposited carbon has a negligible effect on the field distribution of the TE 0 mode, as shown in Fig. 2 b. To study the effect of carbon on the optical response of A-DCs, we reduce w 0 from 1200 nm to 1150 nm while keeping w 1 unchanged, thereby breaking the phase-matching conditions. As shown in Fig. 2 c and 2 d, after placing the carbon model on the narrow waveguide, the insertion loss decreases from approximately 5 dB to 0.24 dB (TE 1 mode) and 0.33 dB (TE 2 mode). The trimming-induced insertion losses are approximately 0.2 dB for the TE 1 -mode A-DCs and 0.24 dB for the TE 2 -mode A-DCs, which are slightly higher than those induced by silicon ion implantation 19 or polymer-based trimming 18 , as the deposited carbon is not perfectly lossless at 1550 nm. These values, however, remain lower than those typically reported for germanium ion implantation (~ 3 dB for 100 µm 38 ). For trimming applications, a small k has only a minimal impact on the overall insertion loss. As shown in Fig. 2 e, when the k value gradually increases from 0 to 0.1, the insertion loss of the TE 1 -mode A-DC increases from 0.035 dB to 1.08 dB. This can be generalized as an approximation: for every 0.01 increase in the k value here, the insertion loss increases by approximately 0.1 dB. Since the output of the bar port is almost the same in Fig. 2 f, the increased insertion loss is mainly caused by the increase in light absorption. The TE 2 mode-based A-DC also exhibits the same trend, but with an insertion loss that increases by approximately 0.13 dB for every 0.01 increase in the k value. This is mainly owed to the crossover length of the TE 2 mode,110 µm, which is longer than the 90 µm in the TE 1 mode. To explore the impact of carbon deposition at the nanoscale, electron energy loss spectroscopy (EELS) spectra were first used to determine the bandgaps of the SiN waveguide (5.25 eV) and the deposited carbon (3.9 eV), the latter corresponding to diamond-like carbon 39 . As a complementary technique, scanning transmission electron microscopy (STEM) imaging with energy-dispersive spectroscopy (EDS) analysis was used to examine the cross-section of the cut narrow waveguide and reveal how the carbon integrates with the underlying structure. Figure 3 a and 3 b show the elemental mapping of the waveguide after carbon deposition. In the elemental maps of carbon and gallium, we observe an increase in brightness in the local area above the waveguide, indicating that the deposited layer has a mixture of gallium and carbon 40 . The gallium originates from the Ga + ion beam used during the FIB-assisted carbon deposition. Importantly, gallium signal is confined to the deposited layer and absent from the underlying SiN waveguide. This is consistent with the use of a low probe current (20 pA) at 30 kV acceleration voltage, conditions that favor surface-assisted carbon deposition rather than gallium implantation. Under these parameters, the interaction volume of Ga + ions is minimal relative to the waveguide core, keeping the process firmly in the deposition regime. This is further supported by the line scan profile in Fig. 3 c, showing a localized increase in both carbon and gallium above the waveguide. According to the elemental maps, the carbon deposition is non-invasive and limited to the designated area, with no significant changes in elemental composition detected in the underlying SiN. Furthermore, the plasmon map in Fig. 3 d and the corresponding high-angle annular dark-field (HAADF) micrograph in Fig. 3 e together corroborate this conclusion: the plasmon map reveals negligible variations in the plasmon energy and intensity after carbon deposition, suggesting that the collective electronic response of the material remains unchanged. Combined with the HAADF micrograph, these observations confirm that the deposition is confined to this region and no significant structural alterations are detectable within the resolution limits of this method. Therefore, this method is expected to be compatible with different platforms without requiring any platform-specific modifications. Since the carbon is not completely lossless, we prepared simple Mach-Zehnder interferometers (MZIs) to evaluate the introduced insertion loss. Carbon was deposited on one arm of the MZIs with a fixed length of 20 µm (see Supplementary S3 for MZI images). The measured insertion loss is approximately 0.14 dB, comparable to 0.125 dB obtained from simulations. Compared to a standalone waveguide, the loss caused by trimming in A-DCs is approximately half. The supermodes formed in the coupling region distribute the optical power between the two waveguides, thereby reducing the proportion of the light interacting with the deposited carbon. Hence, the deposited carbon introduces only 0.3 dB loss to the TE 1 -mode A-DCs (90 µm trimmed length). If a shorter structure is used and the trimming length is reduced, the introduced loss becomes almost negligible. With both the elemental distribution and optical loss confirmed, attention turns to the trimmed region within the A-DC structures. A scanning electron microscope (SEM) image of the TE 1 -mode A-DC after carbon deposition is presented in Fig. 4a. A dimmer line can be observed on the narrow waveguide, formed by carbon deposition, which increases the n eff value of the TE 0 mode to meet the phase-matching conditions. The length of the deposited carbon is designed to match the crossover length of the A-DC, ensuring efficient mode conversion across the entire coupling region. After carbon deposition, the measured insertion loss of the A-DC is reduced from 1.81 dB to 0.36 dB, as shown in Fig. 4b. To corroborate the experimental results, we conducted simulations, and the results in Fig. 4c exhibit good agreement with measurement results, further validating the effectiveness of carbon deposition in controlling the mode conversion. In the SEM image of the TE 2 -mode A-DC after carbon deposition (Fig. 4d), we observe that the carbon is not perfectly centered in the narrow waveguide, however, according to our simulation results even when the center position offset reaches ± 100 nm, there is almost no influence on the n eff of the TE 0 mode. Figure 4 SEM image of A-DCs after carbon deposition and the corresponding transmission spectra. a SEM image of the trimmed A-DC for TE 1 mode. b Measured transmission spectra of the TE 1 -mode A-DC. c simulation results used to study the measurement results. In this model, w 0 = 1155 nm, w 1 = 2510 nm, h = 340 nm, and g = 300 nm. The major and minor axis lengths of the carbon structure are 120 nm and 60 nm, respectively. d SEM image of the TE 2 -mode device. e Measured transmission spectra of the TE 2 -mode A-DC. f Simulation results for studying the measured spectra. In this model, w 0 = 1155 nm, w 1 = 3870 nm, h = 340 nm, and g = 280 nm. The major and minor axis lengths are 100 nm and 60 nm, respectively Figure 4e shows a significant transmission improvement, where the insertion loss at approximately 1550 nm decreased from 4.5 dB to 0.82 dB after trimming. This result is in close agreement with the simulated data presented in Fig. 4f. Herein we discuss carbon deposition using TE 1 and TE 2 modes as examples, but it can be extended to higher-order modes, demonstrating excellent versatility. Carbon deposition also provides wide tuning capability, ensuring compatibility with a broad range of initial waveguide conditions, as shown in Fig. 5 a. Here, the initial insertion losses of the A-DCs to be trimmed are gradually increased from 1.8 dB to 17.8 dB. For the TE 1 -mode A-DCs at w 1 = 2525 nm (marked as 1), the improvement in the insertion loss is as fine as 1.46 dB, from 1.83 dB to 0.37 dB, confirming the feasibility of this method for fine-tuning. For the TE 2 -mode A-DCs at w 1 = 3950 nm (marked as 6), the improvement in the insertion loss can reach as high as 16.1 dB, from 17.8 dB to 1.7 dB. To evaluate the long-term stability of this non-volatile post-fabrication trimming method, we measured the transmission of the TE 1 -mode A-DC at different time points. As shown in Fig. 5 b, the transmission slightly decreases during the first two weeks and then stabilizes. The minor change in transmission observed within the first two weeks is likely attributed to stress relaxation or gradual structural stabilization of the deposited carbon, a process can be accelerated by thermal annealing. To verify this hypothesis, a device was prepared and annealed at 200 ℃, followed by measurement at different time intervals. The annealing temperature was selected because heating beyond approximately 300 ℃ causes the deposited carbon to transition from diamond-like carbon to graphite-like carbon 23 , 41 . The optical response of the device tended to stabilize after approximately 30 minutes of annealing treatment (see Supplementary 4). Based on the results, the chip containing the TE 2 -mode A-DC was placed on a hot plate at 200 ℃ for 30 minutes. As shown in Fig. 5 c, the transmission exhibited a slight decrease after heating, but it remains stable over the following weeks after the heat treatment, indicating great stability. Due to its mask-free nature, the FIB-based carbon deposition demonstrates exceptional flexibility in the post-fabrication trimming. This trimming method combined with A-DCs can be applied to photonic crossbar arrays 9 , 42 , 43 , which are commonly used to perform matrix-vector multiplication (MVM) operations to accelerate artificial intelligence tasks. As shown in Fig. 6 a, the A-DCs (indicated by a white dashed box) can be added before the output grating coupler to combine the output light of two small 9×2 crossbar arrays, which is equivalent to performing an addition operation. Figure 6 b shows an image of the A-DC in the very same crossbar configuration. For directional couplers based on the TE 0 mode, we observe that after combination, the output power of each input is reduced by 3 dB due to the 50:50 power splitting. Because the orthogonal modes (e.g., TE 0 and TE 1 modes) are independent, when using an A-DC based on mode-division multiplexing (MDM) technology for combination, the loss here can be approximated as the insertion loss, which is much lower than 3 dB after carbon deposition. Figure 6 c shows a comparison of transmitted light before and after carbon deposition. The insertion loss for the conversion from the TE 0 mode to TE 1 mode decreases from approximately 3 dB to 0.5 dB after trimming, ensuring efficient combination. Discussion We present a trimming method for PICs based on FIB carbon deposition, which is non-invasive, non-volatile, low-loss, and highly stable. As a representative case, we demonstrate its application in post-fabrication tuning of A-DCs, achieving insertion losses as low as 0.36 dB after trimming and showcasing discrete trimming levels ranging from ~ 1.5 dB to 16.1 dB. These results validate the method’s effectiveness on a fabrication-sensitive component, while its combined advantages render it broadly applicable well beyond A-DCs. For a trimming length of ~ 100 µm, the carbon-induced loss is only ~ 0.3 dB, demonstrating the minimal optical penalty associated with this technique. The trimmed devices also maintain stable performance over a period exceeding two months, confirming the long-term reliability of the method. Compared with thermal tuning, this approach consumes no continuous power and requires no local heaters, enabling more compact and densely integrated devices. It also circumvents the limitations of ion implantation: while this technique offers high precision and excellent long-term stability, its most common form, Ge ion implantation, often results in significant absorption loss and waveguide core damage; in contrast, our method incurs only minimal insertion loss and fully preserves core integrity. These features make the method well suited for a wide range of PIC applications demanding precise optical control, such as MDM components for expanding processing capacity and large-scale photonic accelerators requiring low-loss signal combination. Crucially, elemental mapping and EELS plasmon map confirm its non-invasive nature, underpinning broad compatibility across diverse material platforms. Although Ga ions are detected in the STEM EDS analysis, our EELS measurements show that the deposited carbon exhibits a bandgap of 3.9 eV, well within the accepted range for diamond-like amorphous carbon. This in turn, indicates that Ga content remains below levels that would compromise optical performance, as further supported by the transmission spectra of photonic devices after carbon deposition. Building on these advantages, FIB carbon deposition offers a mask-free solution for post-fabrication tuning in both current silicon-based platforms and emerging material systems. As such, this method provides a practical and forward-compatible pathway toward large-scale, energy-efficient, and yield-enhancing photonic integration across a wide spectrum of applications from high-density interconnects to next-generation optical computing systems. Methods Focused Ion Beam-Induced Carbon Deposition on SiN Carbon deposition on SiN substrates was performed using a ZEISS Crossbeam 540 FIB/SEM system equipped with a multi-channel Gas Injection System (GIS). Samples were mounted onto the stage and introduced into the chamber, where the stage was tilted to 54° to align the sample surface normal to the FIB column. Once thermal and vacuum equilibrium was achieved, a carbon precursor gas was introduced via the GIS. Using a FIB probe current of 20 pA at 30 kV, a carbon line of a specified length was deposited directly on top of the waveguides. The ion beam and GIS were activated simultaneously, and the deposition dose was calibrated to achieve a nominal carbon thickness of ~ 50 nm. STEM analysis of the trimmed SiN waveguides Post-trimming analysis of the SiN waveguides was performed using a Thermo Fisher Scientific FEI Themis 300 G3 Titan scanning/transmission electron microscope (S/TEM), operated at 300/60 kV. Samples were prepared in cross-sectional orientation using the FIB technique on a ZEISS Crossbeam 540 FIB/SEM system. The specimens were thinned to electron transparency for imaging. Elemental mapping was conducted in STEM mode using energy-dispersive X-ray spectroscopy (EDS). To investigate bandgap variations in the trimmed waveguides, electron energy loss spectroscopy (EELS) was performed in monochromated STEM mode at 60 kV. The analysis focused on the low-loss region of the EELS spectrum, which contains information about interband transitions and plasmon excitations that are sensitive to changes in the electronic structure and optical bandgap. Declarations Acknowledgements The authors gratefully acknowledge Prof. Dr. Rasmus R. Schröder for generously providing the experimental facilities, technical assistance, and invaluable support that were essential for this study. R. X. and Z. Z. gratefully acknowledge the Alexander-von-Humboldt Foundation for providing a postdoctoral fellowship. W. H. P. P. acknowledges support from the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under project numbers 390761711 and 390900948. The work was also supported by the European Union’s Horizon Projects (PHONICS, grant no. 101017237; HYBRAIN, grant no. 101046878; 2DNEURALVISION, grant no. 101119489). The manuscript was written through the contributions of all authors. All authors have approved the final version of the manuscript. Author contributions Conceptualization: R. X., S. T. and R. R. S., Simulation: R. X., L. M. and L. M., Sample fabrication: R. X., Z. T., A. V. and S. T., Data acquisition and analysis: R. X., Z. T., Z. Z, R. P., X. M., Q. Z and F. B. P., Visualization: R.X., J. B., J. R., and J. D., Writing: All authors, Supervision: W. H. P. P., S. T., R. R. S., and H. B. Competing interests. The authors declare no competing interests. Correspondence and requests for materials should be addressed to W. H. P. P. and S. T. References Dolphin JA et al (2023) A hybrid integrated quantum key distribution transceiver chip. npj Quantum Inf 9:84 Terrasanta G, Ziarko MW, Bergamasco N, Poot M, Poliak J (2025) Photonic Integrated Circuits for Optical Satellite Links: A Review of the Technology Status and Space Effects. Satell Commun Netw 43:210–228 Elshaari AW, Pernice W, Srinivasan K, Benson O (2020) Zwiller, V. Hybrid integrated quantum photonic circuits. Nat Photonics 14:285–298 Luo W et al (2023) Recent progress in quantum photonic chips for quantum communication and internet. Light Sci Appl 12:175 PsiQuantum team et al (2025) A manufacturable platform for photonic quantum computing. Nature 641:876–883 Pelucchi E et al (2021) The potential and global outlook of integrated photonics for quantum technologies. Nat Rev Phys 4:194–208 Lomonte E et al (2021) Single-photon detection and cryogenic reconfigurability in lithium niobate nanophotonic circuits. Nat Commun 12:6847 Ahmed SR et al (2025) Universal photonic artificial intelligence acceleration. Nature 640:368–374 Brückerhoff-Plückelmann F et al (2024) Probabilistic photonic computing with chaotic light. Nat Commun 15:10445 Xu R et al (2023) Hybrid photonic integrated circuits for neuromorphic computing [Invited]. Opt Mater Express 13:3553–3606 Atabaki AH, Eftekhar AA, Askari M, Adibi A (2013) Accurate post-fabrication trimming of ultra-compact resonators on silicon. Opt Express 21:14139 Spector S, Knecht JM, Juodawlkis PW (2016) Localized in situ cladding annealing for post-fabrication trimming of silicon photonic integrated circuits. Opt Express 24:5996 Fan T, Xia Z, Adibi A, Eftekhar AA (2018) Highly-uniform resonator-based visible spectrometer on a Si 3 N 4 platform with robust and accurate post-fabrication trimming. Opt Lett 43:4887 Liu X et al (2020) Highly efficient thermo-optic tunable micro-ring resonator based on an LNOI platform. Opt Lett 45:6318 Jayatilleka H et al (2021) Post-Fabrication Trimming of Silicon Photonic Ring Resonators at Wafer-Scale. J Lightwave Technol 39:5083–5088 Farmakidis N et al (2023) Scalable High-Precision Trimming of Photonic Resonances by Polymer Exposure to Energetic Beams. Nano Lett 23:4800–4806 Prorok S, Petrov AY, Eich M, Luo J, Jen A (2012) K.-Y. Trimming of high-Q-factor silicon ring resonators by electron beam bleaching. Opt Lett 37:3114 Tang Z et al (2025) Ultralow Loss Coupling Tuning of Photonic Accelerators. Adv Photonics Res 2500175. 10.1002/adpr.202500175 Varri A et al (2023) Scalable Non-Volatile Tuning of Photonic Computational Memories by Automated Silicon Ion Implantation. Adv Mater 2310596. 10.1002/adma.202310596 Li P et al (2021) Recent advances in focused ion beam nanofabrication for nanostructures and devices: fundamentals and applications. Nanoscale 13:1529–1565 Xu R et al (2024) Mode Conversion Trimming in Asymmetric Directional Couplers Enabled by Silicon Ion Implantation. Nano Lett 24:10813–10819 Aisenberg S, Chabot R (1971) Ion-Beam Deposition of Thin Films of Diamondlike Carbon. J Appl Phys 42:2953–2958 Kanda K et al (2009) Graphitization of thin films formed by focused-ion-beam chemical-vapor-deposition. Diam Relat Mater 18:490–492 Matsui S (2007) Focused-ion-beam deposition for 3-D nanostructure fabrication. Nucl Instrum Methods Phys Res Sect B 257:758–764 Matsui S et al (2000) Three-dimensional nanostructure fabrication by focused-ion-beam chemical vapor deposition. J Vac Sci Technol B 18 Fujita J et al (2002) Structure and Resonant Characteristics of Amorphous Carbon Pillars Grown by Focused-Ion-Beam-Induced Chemical Vapor Deposition. Jpn J Appl Phys 41:4423–4426 Choi WS, Kim K, Yi J, Hong B (2008) Diamond-like carbon protective anti-reflection coating for Si solar cell. Mater Lett 62:577–580 Da Silva DS et al (2011) Application of amorphous carbon based materials as antireflective coatings on crystalline silicon solar cells. J Appl Phys 110 Yang Q et al (2021) Research on the fabrication and anti-reflection performance of diamond-like carbon films. Diam Relat Mater 111:108184 Lu Y et al (2021) Pulsed laser deposition of the protective and Anti-reflective DLC film. Infrared Phys Technol 119:103949 Scendo M, Staszewska-Samson K (2019) Effect of Temperature on Anti-Corrosive Properties of Diamond-Like Carbon Coating on S355 Steel. Materials 12:1659 Wang C et al (2024) Lithium tantalate photonic integrated circuits for volume manufacturing. Nature 629:784–790 Li Z et al (2023) High density lithium niobate photonic integrated circuits. Nat Commun 14:4856 Luo L-W et al (2014) WDM-compatible mode-division multiplexing on a silicon chip. Nat Commun 5:3069 Yin R et al (2023) Integrated WDM-compatible optical mode division multiplexing neural network accelerator. Optica 10:1709–1718 Yang KY et al (2022) Multi-dimensional data transmission using inverse-designed silicon photonics and microcombs. Nat Commun 13:7862 Larruquert JI, Rodríguez-de Marcos LV, Méndez JA, Martin PJ, Bendavid A (2013) High reflectance ta-C coatings in the extreme ultraviolet. Opt Express 21:27537 Milosevic MM et al (2020) Ion Implantation of Germanium Into Silicon for Critical Coupling Control of Racetrack Resonators. J Lightwave Technol 38:1865–1873 Robertson J (2002) Diamond-like amorphous carbon. Mater Sci Engineering: R: Rep 37:129–281 Igaki J et al (2007) Elementary Analysis of Diamond-Like Carbon Film Formed by Focused-Ion-Beam Chemical Vapor Deposition. jjap 46, 8003 Tallant DR, Parmeter JE, Siegal MP, Simpson RL (1995) The thermal stability of diamond-like carbon. Diam Relat Mater 4:191–199 Feldmann J et al (2021) Parallel convolutional processing using an integrated photonic tensor core. Nature 589:52–58 Brückerhoff-Plückelmann F, Feldmann J, Wright CD, Bhaskaran H, Pernice W (2021) H. P. Chalcogenide phase-change devices for neuromorphic photonic computing. J Appl Phys 129:151103 Additional Declarations There is NO Competing Interest. Supplementary Files Supplementary.docx Supplementary materials Cite Share Download PDF Status: Under Review Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7410735","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":503343565,"identity":"662a187e-f1bc-4f62-a20e-628d55b0a2cc","order_by":0,"name":"Wolfram Pernice","email":"data:image/png;base64,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","orcid":"https://orcid.org/0000-0003-4569-4213","institution":"Heidelberg University","correspondingAuthor":true,"prefix":"","firstName":"Wolfram","middleName":"","lastName":"Pernice","suffix":""},{"id":503343566,"identity":"1cf40e53-79c6-4ed9-a009-838bdb54e26c","order_by":1,"name":"Rongyang Xu","email":"","orcid":"https://orcid.org/0000-0001-8914-4454","institution":"Heidelberg University","correspondingAuthor":false,"prefix":"","firstName":"Rongyang","middleName":"","lastName":"Xu","suffix":""},{"id":503343567,"identity":"d62d89b5-5b5d-4bde-baac-32a579ed86c8","order_by":2,"name":"Zhongyu Tang","email":"","orcid":"","institution":"Heidelberg University","correspondingAuthor":false,"prefix":"","firstName":"Zhongyu","middleName":"","lastName":"Tang","suffix":""},{"id":503343568,"identity":"7710d782-9648-488f-8c88-3a859de2a910","order_by":3,"name":"Liam McRae","email":"","orcid":"","institution":"Heidelberg University","correspondingAuthor":false,"prefix":"","firstName":"Liam","middleName":"","lastName":"McRae","suffix":""},{"id":503343569,"identity":"5033f765-5b71-4428-90ef-282c84293351","order_by":4,"name":"Akhil Varri","email":"","orcid":"https://orcid.org/0009-0008-5948-4181","institution":"University of Münster","correspondingAuthor":false,"prefix":"","firstName":"Akhil","middleName":"","lastName":"Varri","suffix":""},{"id":503343570,"identity":"97abf697-c902-404e-942a-1a3fd374fef3","order_by":5,"name":"Frank Brückerhoff-Plückelmann","email":"","orcid":"","institution":"University of Münster","correspondingAuthor":false,"prefix":"","firstName":"Frank","middleName":"","lastName":"Brückerhoff-Plückelmann","suffix":""},{"id":503343571,"identity":"e9fc305a-1533-44cc-bd7e-c775932eb6fd","order_by":6,"name":"Xinyu Ma","email":"","orcid":"","institution":"Heidelberg University","correspondingAuthor":false,"prefix":"","firstName":"Xinyu","middleName":"","lastName":"Ma","suffix":""},{"id":503343572,"identity":"878faf0c-3abb-44a1-a32a-8d233bbb8edc","order_by":7,"name":"Rasmus Bankwitz","email":"","orcid":"","institution":"Heidelberg University","correspondingAuthor":false,"prefix":"","firstName":"Rasmus","middleName":"","lastName":"Bankwitz","suffix":""},{"id":503343573,"identity":"8810c6d6-045a-4a73-b7de-ce76c4d71b07","order_by":8,"name":"Julius Römer","email":"","orcid":"","institution":"Heidelberg University","correspondingAuthor":false,"prefix":"","firstName":"Julius","middleName":"","lastName":"Römer","suffix":""},{"id":503343574,"identity":"38659ccb-b516-4ffa-9817-774ac91caa53","order_by":9,"name":"Ravi Pradip","email":"","orcid":"","institution":"Heidelberg University","correspondingAuthor":false,"prefix":"","firstName":"Ravi","middleName":"","lastName":"Pradip","suffix":""},{"id":503343575,"identity":"87c7c85a-045d-4d9b-8b42-24e32d752310","order_by":10,"name":"Qinlin Zhang","email":"","orcid":"","institution":"Heidelberg University","correspondingAuthor":false,"prefix":"","firstName":"Qinlin","middleName":"","lastName":"Zhang","suffix":""},{"id":503343576,"identity":"24aa6dc5-5354-465f-b18d-33dd7da42408","order_by":11,"name":"Lennart Meyer","email":"","orcid":"","institution":"Heidelberg University","correspondingAuthor":false,"prefix":"","firstName":"Lennart","middleName":"","lastName":"Meyer","suffix":""},{"id":503343577,"identity":"1af7a3d9-d8b0-453f-b499-dd4dca4b246c","order_by":12,"name":"Zhe Zhao","email":"","orcid":"https://orcid.org/0000-0002-1819-5518","institution":"University of Muenster","correspondingAuthor":false,"prefix":"","firstName":"Zhe","middleName":"","lastName":"Zhao","suffix":""},{"id":503343578,"identity":"04ef0189-6b97-4ed3-8ebc-c7bc5bc69f61","order_by":13,"name":"Jelle Dijkstra","email":"","orcid":"","institution":"Heidelberg University","correspondingAuthor":false,"prefix":"","firstName":"Jelle","middleName":"","lastName":"Dijkstra","suffix":""},{"id":503343579,"identity":"df3d6010-d23e-4f53-9db7-28683c6e9992","order_by":14,"name":"Harish Bhaskaran","email":"","orcid":"https://orcid.org/0000-0003-0774-8110","institution":"University of Oxford","correspondingAuthor":false,"prefix":"","firstName":"Harish","middleName":"","lastName":"Bhaskaran","suffix":""},{"id":503343580,"identity":"ca23c29b-a321-4fd2-aa5b-e3426c520400","order_by":15,"name":"Rasmus Schröder","email":"","orcid":"","institution":"Heidelberg University","correspondingAuthor":false,"prefix":"","firstName":"Rasmus","middleName":"","lastName":"Schröder","suffix":""},{"id":503343581,"identity":"93f43100-f3d1-4291-8a41-0588b27fc64d","order_by":16,"name":"Shabnam Taheriniya","email":"","orcid":"","institution":"Heidelberg University","correspondingAuthor":false,"prefix":"","firstName":"Shabnam","middleName":"","lastName":"Taheriniya","suffix":""}],"badges":[],"createdAt":"2025-08-19 17:06:13","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7410735/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7410735/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":89622266,"identity":"e5c70ed4-90fb-4f53-861c-9b8d5b1845e2","added_by":"auto","created_at":"2025-08-22 04:24:59","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":409159,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eSchematic diagram of a trimmed A-DC using carbon deposition, and the transmission response of the untrimmed A-DCs.\u003c/strong\u003e \u003cstrong\u003ea\u003c/strong\u003e Schematic of a trimmed A-DC by carbon deposition. \u003cstrong\u003eb\u003c/strong\u003e Images of a fabricated A-DC for mode conversion between TE\u003csub\u003e0\u003c/sub\u003e and TE\u003csub\u003e2\u003c/sub\u003e modes. \u003cstrong\u003ec\u003c/strong\u003e Simulated transmission spectrum and intensity distribution of the TE\u003csub\u003e1\u003c/sub\u003e-mode A-DC. \u003cstrong\u003ed, e\u003c/strong\u003e Simulated and measured transmission of the TE\u003csub\u003e1\u003c/sub\u003e-mode A-DCs with increasing \u003cem\u003ew\u003c/em\u003e\u003csub\u003e1\u003c/sub\u003e. The difference between the simulation results and the measurement results is due to fabrication errors in the width of the waveguides. \u003cstrong\u003ef\u003c/strong\u003e Simulated transmission spectrum and intensity distribution of the TE\u003csub\u003e2\u003c/sub\u003e-mode A-DC. \u003cstrong\u003eg,\u003c/strong\u003e \u003cstrong\u003eh\u003c/strong\u003e Simulated and measured transmission of the TE\u003csub\u003e2\u003c/sub\u003e-mode A-DCs with increasing \u003cem\u003ew\u003c/em\u003e\u003csub\u003e1\u003c/sub\u003e.\u003c/p\u003e","description":"","filename":"image1.png","url":"https://assets-eu.researchsquare.com/files/rs-7410735/v1/34fa1fe5663d6af1af6ca03c.png"},{"id":89622268,"identity":"ecb4aa1a-33a8-4826-9441-89d40c4ef667","added_by":"auto","created_at":"2025-08-22 04:24:59","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":228829,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eSimulation results of asymmetric directional couplers before and after carbon deposition. a\u003c/strong\u003e Effective refractive index of the TE\u003csub\u003e0\u003c/sub\u003e mode varies with the major axis length of carbon. When \u003cem\u003ew\u003c/em\u003e\u003csub\u003e0\u003c/sub\u003e = 1150 nm and major axis length = 160 nm, the \u003cem\u003en\u003c/em\u003e\u003csub\u003eeff \u003c/sub\u003eof the TE\u003csub\u003e0\u003c/sub\u003e mode is equal to the \u003cem\u003en\u003c/em\u003e\u003csub\u003eeff\u003c/sub\u003e at \u003cem\u003ew\u003c/em\u003e\u003csub\u003e0\u003c/sub\u003e=1200 nm. \u003cstrong\u003eb\u003c/strong\u003e Intensity distribution of a 1200 nm-wide waveguide, an 1150 nm-wide waveguide, and an 1150 nm-wide waveguide with carbon deposition. The carbon is elliptical with major axis and minor axis lengths of 160 nm and 60 nm. The \u003cem\u003en\u003c/em\u003e and \u003cem\u003ek\u003c/em\u003e of the deposited carbon are 2.48 and 0.019\u003csup\u003e37\u003c/sup\u003e, respectively. Transmission spectra of A-DC with 1150 nm-wide waveguide before and after depositing carbon for \u003cstrong\u003ec\u003c/strong\u003e TE\u003csub\u003e1\u003c/sub\u003e mode and \u003cstrong\u003ed\u003c/strong\u003e TE\u003csub\u003e2\u003c/sub\u003e mode. \u003cstrong\u003ee\u003c/strong\u003e and \u003cstrong\u003ef\u003c/strong\u003e When the \u003cem\u003ek\u003c/em\u003e of the deposited carbon is assumed to vary from 0 to 0.1, the transmission spectra of the A-DC based on the TE\u003csub\u003e1\u003c/sub\u003e mode. \u003cstrong\u003eg\u003c/strong\u003e and \u003cstrong\u003eh\u003c/strong\u003e Assuming that the \u003cem\u003ek\u003c/em\u003e of the deposited carbon varies from 0 to 0.1, the transmission spectra of the A-DC based on the TE\u003csub\u003e2\u003c/sub\u003e mode.\u003c/p\u003e","description":"","filename":"image2.png","url":"https://assets-eu.researchsquare.com/files/rs-7410735/v1/0f3a1057064ca4b491f75821.png"},{"id":89622823,"identity":"c322b7c2-efc2-48ff-bbca-2742b8ccd22b","added_by":"auto","created_at":"2025-08-22 04:40:59","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":468307,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eTransmission electron microscope study of the deposited carbon. a\u003c/strong\u003e Cross-section of the cut waveguide with carbon deposition. \u003cstrong\u003eb\u003c/strong\u003e STEM EDS elemental mapping of the waveguide. \u003cstrong\u003ec\u003c/strong\u003e Line scan profile of the deposited carbon indicated using a red dashed line in the cross-section image. \u003cstrong\u003ed\u003c/strong\u003e, \u003cstrong\u003ee\u003c/strong\u003e Plasmon map and HAADF micrograph of the cross-sectional region near the deposited carbon.\u003c/p\u003e","description":"","filename":"image3.png","url":"https://assets-eu.researchsquare.com/files/rs-7410735/v1/c4f986230d17ee26be1f8a36.png"},{"id":89622270,"identity":"72afb533-bd5d-4158-bb58-f83265fb67bc","added_by":"auto","created_at":"2025-08-22 04:24:59","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":283980,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eSEM image of A-DCs after carbon deposition and the corresponding transmission spectra. a\u003c/strong\u003e SEM image of the trimmed A-DC for TE\u003csub\u003e1\u003c/sub\u003e mode. \u003cstrong\u003eb\u003c/strong\u003e Measured transmission spectra of the TE\u003csub\u003e1\u003c/sub\u003e-mode A-DC. \u003cstrong\u003ec\u003c/strong\u003e simulation results used to study the measurement results. In this model, \u003cem\u003ew\u003c/em\u003e\u003csub\u003e0\u003c/sub\u003e = 1155 nm, \u003cem\u003ew\u003c/em\u003e\u003csub\u003e1\u003c/sub\u003e = 2510 nm, \u003cem\u003eh\u003c/em\u003e = 340 nm, and \u003cem\u003eg\u003c/em\u003e = 300 nm. The major and minor axis lengths of the carbon structure are 120 nm and 60 nm, respectively. \u003cstrong\u003ed\u003c/strong\u003e SEM image of the TE\u003csub\u003e2\u003c/sub\u003e-mode device. \u003cstrong\u003ee\u003c/strong\u003e Measured transmission spectra of the TE\u003csub\u003e2\u003c/sub\u003e-mode A-DC. \u003cstrong\u003ef\u003c/strong\u003e Simulation results for studying the measured spectra. In this model, \u003cem\u003ew\u003c/em\u003e\u003csub\u003e0\u003c/sub\u003e = 1155 nm, \u003cem\u003ew\u003c/em\u003e\u003csub\u003e1\u003c/sub\u003e = 3870 nm, \u003cem\u003eh\u003c/em\u003e = 340 nm, and \u003cem\u003eg\u003c/em\u003e = 280 nm. The major and minor axis lengths are 100 nm and 60 nm, respectively\u003c/p\u003e","description":"","filename":"image4.png","url":"https://assets-eu.researchsquare.com/files/rs-7410735/v1/5daec0e535ed0090678120f1.png"},{"id":89622269,"identity":"3f473cd8-f93a-4c96-8783-a68eebcf2742","added_by":"auto","created_at":"2025-08-22 04:24:59","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":181523,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eTransmission of A-DCs at 1550 nm with different initial insertion losses and stability of the optical response of the trimmed A-DCs over time.\u003c/strong\u003e \u003cstrong\u003ea\u003c/strong\u003e Devices 1–3 are TE­\u003csub\u003e1\u003c/sub\u003e­-mode A-DCs with \u003cem\u003ew\u003c/em\u003e­\u003csub\u003e1\u003c/sub\u003e = 2525 nm, 2550 nm, and 2575 nm, respectively. Device 4–6 are TE\u003csub\u003e2\u003c/sub\u003e-mode A-DCs with \u003cem\u003ew\u003c/em\u003e\u003csub\u003e1\u003c/sub\u003e = 3900 nm, 3925 nm, 3950 nm, respectively. The initial insertion losses of these devices vary from 1.8 dB to 17.8 dB. A single carbon deposition can improve insertion loss by up to 16.1 dB.\u003cstrong\u003e b\u003c/strong\u003e Transmission at 1550 nm for the TE\u003csub\u003e1\u003c/sub\u003e-mode A-DC over time. The decrease in transmission stops after the second week, and the transmission remains stable for at least 6 weeks thereafter. The final insertion loss is approximately 0.75 dB. \u003cstrong\u003ec\u003c/strong\u003e Transmission of the A-DC based on the TE\u003csub\u003e2\u003c/sub\u003e mode at 1550 nm over time. The device was annealed on a hot plate at 200℃ for 30 minutes. After annealing, the transmission decreases slightly and remains stable for at least 4 weeks thereafter. All the samples in this study were stored in ambient air without any special treatment.\u003c/p\u003e","description":"","filename":"image5.png","url":"https://assets-eu.researchsquare.com/files/rs-7410735/v1/fecb68bd0a1fc63906ded612.png"},{"id":89622286,"identity":"738c5f4e-e843-4ac0-8ca0-cf68e9a616ee","added_by":"auto","created_at":"2025-08-22 04:24:59","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":706077,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eFlexible post-fabrication trimming using carbon deposition in a photonic crossbar array. a\u003c/strong\u003e An image of the fabricated photonic crossbar array based on the wavelength-division multiplexing and MDM techniques. \u003cstrong\u003eb\u003c/strong\u003e Schematic of the application of MDM in the photonic crossbar array. \u003cstrong\u003ec\u003c/strong\u003e Sum of the output powers from the two constituent arrays before and after carbon deposition. Since the TE\u003csub\u003e0\u003c/sub\u003e mode light from array 1 experiences almost no loss when passing through the A-DC, its output power here is used as a normalized reference.\u003c/p\u003e","description":"","filename":"image6.png","url":"https://assets-eu.researchsquare.com/files/rs-7410735/v1/e6436f3d745b0c43fd32bc08.png"},{"id":89622897,"identity":"a292040c-3685-4b77-96f7-99b113038647","added_by":"auto","created_at":"2025-08-22 04:49:00","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":3691614,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7410735/v1/d60f79a9-b775-41c0-8eb3-831bda80a3f3.pdf"},{"id":89622446,"identity":"f9377cae-b744-4320-bd27-b9c6c83da778","added_by":"auto","created_at":"2025-08-22 04:32:59","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":1273339,"visible":true,"origin":"","legend":"Supplementary materials","description":"","filename":"Supplementary.docx","url":"https://assets-eu.researchsquare.com/files/rs-7410735/v1/79abecbd99ec936c6f502744.docx"}],"financialInterests":"There is \u003cb\u003eNO\u003c/b\u003e Competing Interest.","formattedTitle":"Localized Carbon Deposition Enables Non-Invasive Trimming of Photonic Integrated Circuits","fulltext":[{"header":"Introduction","content":"\u003cp\u003eA photonic integrated circuit (PIC) is a microchip that brings together multiple photonic components with different functions. Due to the inherent properties of light, such as high speed, low propagation loss, and high parallelism, PICs have been widely explored in applications such as optical communications\u003csup\u003e\u003cspan additionalcitationids=\"CR2 CR3\" citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e\u003c/sup\u003e, quantum computing\u003csup\u003e\u003cspan additionalcitationids=\"CR6\" citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e\u003c/sup\u003e, and neuromorphic computing\u003csup\u003e\u003cspan additionalcitationids=\"CR9\" citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e\u003c/sup\u003e. The outstanding performance of PICs in these applications relies on the proper operation of precisely engineered photonic structures, which are often vulnerable to fabrication imperfections. The accumulation of optical response deviations with increasing device count has thus made fabrication imperfections a key obstacle to the further development of PICs. Post-fabrication trimming provides an effective approach to correct such imperfections, offering precise, localized, and flexible control over photonic device performance\u003csup\u003e\u003cspan additionalcitationids=\"CR12\" citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eVarious post-fabrication trimming techniques have been proposed to address fabrication imperfections, including micro-heaters\u003csup\u003e\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e,\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e\u003c/sup\u003e, electron beam exposure of polymer claddings\u003csup\u003e\u003cspan additionalcitationids=\"CR17\" citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e\u003c/sup\u003e, and focused ion beam processing\u003csup\u003e\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e,\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e\u003c/sup\u003e. The trimming method based on micro-heaters usually requires continuous heating to maintain the desired tuning states, resulting in significant power consumption and can cause thermal crosstalk. Polymer cladding trimming is a simple and cost-effective method, but it generally cannot guarantee long-term stability. Focused ion beam (FIB) techniques have been widely used in nanofabrication due to their high spatial resolution and mask-free characteristics\u003csup\u003e\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e\u003c/sup\u003e. In recent years, silicon ion implantation using FIB has emerged as a promising technique for post-fabrication trimming, offering permanent refractive index changes with low optical loss and excellent stability\u003csup\u003e\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e,\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e\u003c/sup\u003e. However, this approach faces challenges in achieving delicate fine-tuning. The strong refractive index contrast between the amorphous silicon-rich regions and the silicon nitride (SiN) waveguide leads to drastic changes in the field distribution of the guided mode, making nanometer-scale tuning highly exacting and sometimes unattainable. In addition, the high-energy ion beam can cause structural damage, restricting its applicability primarily to platforms such as silicon and SiN.\u003c/p\u003e\u003cp\u003eFIB deposition of amorphous carbon as a non-invasive trimming technology has great potential for achieving platform-independent fine-tuning. This technique utilizes a focused ion beam to locally decompose precursor gas molecules, enabling precise deposition of diamond-like amorphous carbon\u003csup\u003e\u003cspan additionalcitationids=\"CR23\" citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e\u003c/sup\u003e. Over the past few decades, the technique has been mainly used for the fabrication of 3D nanostructures\u003csup\u003e\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e,\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e,\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e\u003c/sup\u003e, though diamond-like amorphous carbon can also be used in anti-reflection coating\u003csup\u003e\u003cspan additionalcitationids=\"CR28\" citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e\u003c/sup\u003e, protective coatings\u003csup\u003e\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e,\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e\u003c/sup\u003e, and as hard masks for etching\u003csup\u003e\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e,\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e\u003c/sup\u003e. However, its application in post-fabrication trimming of PICs has not yet been explored. Given that FIB is a mature technology with equipment widely available in research facilities and advanced manufacturing laboratories, this presents a valuable opportunity for precise post-fabrication trimming in PICs.\u003c/p\u003e\u003cp\u003eTo validate the trimming capability of FIB carbon deposition, we investigated asymmetric directional couplers (A-DCs) as a representative device. A-DCs are widely used for on-chip mode conversion\u003csup\u003e\u003cspan additionalcitationids=\"CR35\" citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e\u003c/sup\u003e and rely on precise phase matching between selected modes, which occurs when their effective refractive indices (\u003cem\u003en\u003c/em\u003e\u003csub\u003eeff\u003c/sub\u003e) are equal at a given wavelength. Because of this strong dependence on precise phase matching, A-DCs are sensitive to fabrication imperfections, making them an ideal platform for evaluating post-fabrication trimming techniques.\u003c/p\u003e\u003cp\u003eIn this study, we demonstrate FIB carbon deposition as a non-invasive technique with great stability that can be used for high-resolution post-fabrication trimming of PICs. To validate its effectiveness, we employ A-DCs as a sensitive test structure, where trimming allows precise control of device responses. We also quantify the trimming-induced insertion loss, accessible tuning range, and long-term stability, confirming the method\u0026rsquo;s flexibility and robustness for post-fabrication trimming in PICs.\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003eTo visualize the trimming process, consider an A-DC composed of a narrow waveguide (width \u003cem\u003ew\u003c/em\u003e\u003csub\u003e0\u003c/sub\u003e) placed adjacent to a wider waveguide (width \u003cem\u003ew\u003c/em\u003e\u003csub\u003e1\u003c/sub\u003e), separated by a gap \u003cem\u003eg\u003c/em\u003e. Both waveguides share the same height \u003cem\u003eh\u003c/em\u003e and sit atop a buried oxide (BOX) layer on a silicon substrate. Trimming begins when a carbon-containing precursor gas is directed toward the narrow waveguide through a gas nozzle, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ea. The Ga\u003csup\u003e+\u003c/sup\u003e ion beam induces local decomposition of the precursor gas molecules when exposed to the waveguide, resulting in the accumulation of carbon in the irradiation region with nanometer scale precision (see Supplementary S1 for the actual FIB sample stage diagram). In this study, we chose to deposit carbon on the narrow waveguide to control the \u003cem\u003en\u003c/em\u003e\u003csub\u003eeff\u003c/sub\u003e of the TE\u003csub\u003e0\u003c/sub\u003e mode for demonstration purposes. Carbon can also be deposited on the wide waveguide to control higher-order modes\u003csup\u003e\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eComplementing the schematic, a fabricated A-DC based on the TE\u003csub\u003e2\u003c/sub\u003e mode (see Supplementary S2 for fabrication process) is shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eb. Light is input via a grating coupler and is equally split by a multimode interferometer. Half of the input light is directed to the output grating coupler (indicated by the white dashed box) and measured by a photodetector. The resulting transmission spectrum is used as the reference for normalization. The first A-DC converts the remaining light from the TE\u003csub\u003e0\u003c/sub\u003e mode to the TE\u003csub\u003e2\u003c/sub\u003e mode. After conversion, the light propagates along the bus waveguide in the TE\u003csub\u003e2\u003c/sub\u003e mode and is converted back to the TE\u003csub\u003e0\u003c/sub\u003e mode by the second A-DC before reaching the output grating coupler. For simplicity, we assume in this study that the measured insertion loss is dominated by the mode conversions.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eBuilding on the device structure, we simulated the mode conversion behavior of A-DCs based on the TE\u003csub\u003e1\u003c/sub\u003e and TE\u003csub\u003e2\u003c/sub\u003e modes, using parameters of \u003cem\u003ew\u003c/em\u003e\u003csub\u003e0\u003c/sub\u003e\u0026thinsp;=\u0026thinsp;1200 nm, \u003cem\u003eg\u003c/em\u003e\u0026thinsp;=\u0026thinsp;300 nm, and \u003cem\u003eh\u003c/em\u003e\u0026thinsp;=\u0026thinsp;335 nm. By adjusting the bus waveguide width, an insertion loss of 0.055 dB for the TE\u003csub\u003e1\u003c/sub\u003e mode and 0.045 dB for the TE\u003csub\u003e2\u003c/sub\u003e mode is achieved for the A-DC at 1550 nm, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ec and \u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ef. The insets show that the TE\u003csub\u003e0\u003c/sub\u003e mode light in the narrow waveguide is converted into the higher-order mode light in the wide waveguide. Figure\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ed and \u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eg exhibit the transmission of the A-DCs at a wavelength of 1550 nm under different \u003cem\u003ew\u003c/em\u003e\u003csub\u003e1\u003c/sub\u003e values, where the highest transmission appears at \u003cem\u003ew\u003c/em\u003e\u003csub\u003e1\u003c/sub\u003e\u0026thinsp;=\u0026thinsp;2550 nm and 3900 nm, respectively. These structures were then fabricated for verification. Experimentally, there was only minimal deviation from the expected waveguide widths, mainly owing to the shrinkage of resist during the hot plate baking process. We observe the measured insertion losses to be 0.4 dB and 0.5 dB with waveguide widths of 2475 nm and 3750 nm for TE\u003csub\u003e1\u003c/sub\u003e and TE\u003csub\u003e2\u003c/sub\u003e modes, in that order. In addition, the observed insertion losses of 2.7 dB and 7.4 dB for the TE\u003csub\u003e1\u003c/sub\u003e and TE\u003csub\u003e2\u003c/sub\u003e modes at \u003cem\u003ew\u003c/em\u003e\u003csub\u003e1\u003c/sub\u003e\u0026thinsp;=\u0026thinsp;2550 nm and 3900 nm, respectively, further highlight the necessity of a wide tuning range for mode control.\u003c/p\u003e\u003cp\u003eIn the experiment, the \u003cem\u003ew\u003c/em\u003e\u003csub\u003e1\u003c/sub\u003e value that achieved the highest transmission was slightly lower than the simulation results for the two modes. This is because resist shrinkage has a greater influence on the \u003cem\u003en\u003c/em\u003e\u003csub\u003eeff\u003c/sub\u003e of the TE\u003csub\u003e0\u003c/sub\u003e mode in the narrow waveguide, which can be compensated by depositing carbon. Figure\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ea shows the \u003cem\u003en\u003c/em\u003e\u003csub\u003eeff\u003c/sub\u003e of the TE\u003csub\u003e0\u003c/sub\u003e mode can be flexibly controlled by changing the carbon geometry. The refractive index \u003cem\u003en\u003c/em\u003e and extinction coefficient \u003cem\u003ek\u003c/em\u003e of the carbon in our simulation models are 2.48 and 0.019, respectively\u003csup\u003e\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e\u003c/sup\u003e. Due to its small geometric dimensions and moderate refractive index, the deposited carbon has a negligible effect on the field distribution of the TE\u003csub\u003e0\u003c/sub\u003e mode, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eb. To study the effect of carbon on the optical response of A-DCs, we reduce \u003cem\u003ew\u003c/em\u003e\u003csub\u003e0\u003c/sub\u003e from 1200 nm to 1150 nm while keeping \u003cem\u003ew\u003c/em\u003e\u003csub\u003e1\u003c/sub\u003e unchanged, thereby breaking the phase-matching conditions. As shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ec and \u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ed, after placing the carbon model on the narrow waveguide, the insertion loss decreases from approximately 5 dB to 0.24 dB (TE\u003csub\u003e1\u003c/sub\u003e mode) and 0.33 dB (TE\u003csub\u003e2\u003c/sub\u003e mode).\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eThe trimming-induced insertion losses are approximately 0.2 dB for the TE\u003csub\u003e1\u003c/sub\u003e-mode A-DCs and 0.24 dB for the TE\u003csub\u003e2\u003c/sub\u003e-mode A-DCs, which are slightly higher than those induced by silicon ion implantation\u003csup\u003e\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e\u003c/sup\u003e or polymer-based trimming\u003csup\u003e\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e\u003c/sup\u003e, as the deposited carbon is not perfectly lossless at 1550 nm. These values, however, remain lower than those typically reported for germanium ion implantation (~\u0026thinsp;3 dB for 100 \u0026micro;m\u003csup\u003e\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e\u003c/sup\u003e). For trimming applications, a small \u003cem\u003ek\u003c/em\u003e has only a minimal impact on the overall insertion loss. As shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ee, when the \u003cem\u003ek\u003c/em\u003e value gradually increases from 0 to 0.1, the insertion loss of the TE\u003csub\u003e1\u003c/sub\u003e-mode A-DC increases from 0.035 dB to 1.08 dB. This can be generalized as an approximation: for every 0.01 increase in the \u003cem\u003ek\u003c/em\u003e value here, the insertion loss increases by approximately 0.1 dB. Since the output of the bar port is almost the same in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ef, the increased insertion loss is mainly caused by the increase in light absorption. The TE\u003csub\u003e2\u003c/sub\u003e mode-based A-DC also exhibits the same trend, but with an insertion loss that increases by approximately 0.13 dB for every 0.01 increase in the \u003cem\u003ek\u003c/em\u003e value. This is mainly owed to the crossover length of the TE\u003csub\u003e2\u003c/sub\u003e mode,110 \u0026micro;m, which is longer than the 90 \u0026micro;m in the TE\u003csub\u003e1\u003c/sub\u003e mode.\u003c/p\u003e\u003cp\u003eTo explore the impact of carbon deposition at the nanoscale, electron energy loss spectroscopy (EELS) spectra were first used to determine the bandgaps of the SiN waveguide (5.25 eV) and the deposited carbon (3.9 eV), the latter corresponding to diamond-like carbon\u003csup\u003e\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e\u003c/sup\u003e. As a complementary technique, scanning transmission electron microscopy (STEM) imaging with energy-dispersive spectroscopy (EDS) analysis was used to examine the cross-section of the cut narrow waveguide and reveal how the carbon integrates with the underlying structure. Figure\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ea and \u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eb show the elemental mapping of the waveguide after carbon deposition. In the elemental maps of carbon and gallium, we observe an increase in brightness in the local area above the waveguide, indicating that the deposited layer has a mixture of gallium and carbon\u003csup\u003e\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e\u003c/sup\u003e. The gallium originates from the Ga\u003csup\u003e+\u003c/sup\u003e ion beam used during the FIB-assisted carbon deposition. Importantly, gallium signal is confined to the deposited layer and absent from the underlying SiN waveguide. This is consistent with the use of a low probe current (20 pA) at 30 kV acceleration voltage, conditions that favor surface-assisted carbon deposition rather than gallium implantation. Under these parameters, the interaction volume of Ga\u003csup\u003e+\u003c/sup\u003e ions is minimal relative to the waveguide core, keeping the process firmly in the deposition regime.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eThis is further supported by the line scan profile in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ec, showing a localized increase in both carbon and gallium above the waveguide. According to the elemental maps, the carbon deposition is non-invasive and limited to the designated area, with no significant changes in elemental composition detected in the underlying SiN. Furthermore, the plasmon map in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ed and the corresponding high-angle annular dark-field (HAADF) micrograph in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ee together corroborate this conclusion: the plasmon map reveals negligible variations in the plasmon energy and intensity after carbon deposition, suggesting that the collective electronic response of the material remains unchanged. Combined with the HAADF micrograph, these observations confirm that the deposition is confined to this region and no significant structural alterations are detectable within the resolution limits of this method. Therefore, this method is expected to be compatible with different platforms without requiring any platform-specific modifications.\u003c/p\u003e\u003cp\u003eSince the carbon is not completely lossless, we prepared simple Mach-Zehnder interferometers (MZIs) to evaluate the introduced insertion loss. Carbon was deposited on one arm of the MZIs with a fixed length of 20 \u0026micro;m (see Supplementary S3 for MZI images). The measured insertion loss is approximately 0.14 dB, comparable to 0.125 dB obtained from simulations. Compared to a standalone waveguide, the loss caused by trimming in A-DCs is approximately half. The supermodes formed in the coupling region distribute the optical power between the two waveguides, thereby reducing the proportion of the light interacting with the deposited carbon. Hence, the deposited carbon introduces only 0.3 dB loss to the TE\u003csub\u003e1\u003c/sub\u003e-mode A-DCs (90 \u0026micro;m trimmed length). If a shorter structure is used and the trimming length is reduced, the introduced loss becomes almost negligible.\u003c/p\u003e\u003cp\u003eWith both the elemental distribution and optical loss confirmed, attention turns to the trimmed region within the A-DC structures. A scanning electron microscope (SEM) image of the TE\u003csub\u003e1\u003c/sub\u003e-mode A-DC after carbon deposition is presented in Fig.\u0026nbsp;4a. A dimmer line can be observed on the narrow waveguide, formed by carbon deposition, which increases the \u003cem\u003en\u003c/em\u003e\u003csub\u003eeff\u003c/sub\u003e value of the TE\u003csub\u003e0\u003c/sub\u003e mode to meet the phase-matching conditions. The length of the deposited carbon is designed to match the crossover length of the A-DC, ensuring efficient mode conversion across the entire coupling region. After carbon deposition, the measured insertion loss of the A-DC is reduced from 1.81 dB to 0.36 dB, as shown in Fig.\u0026nbsp;4b. To corroborate the experimental results, we conducted simulations, and the results in Fig.\u0026nbsp;4c exhibit good agreement with measurement results, further validating the effectiveness of carbon deposition in controlling the mode conversion. In the SEM image of the TE\u003csub\u003e2\u003c/sub\u003e-mode A-DC after carbon deposition (Fig.\u0026nbsp;4d), we observe that the carbon is not perfectly centered in the narrow waveguide, however, according to our simulation results even when the center position offset reaches\u0026thinsp;\u0026plusmn;\u0026thinsp;100 nm, there is almost no influence on the \u003cem\u003en\u003c/em\u003e\u003csub\u003eeff\u003c/sub\u003e of the TE\u003csub\u003e0\u003c/sub\u003e mode.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003e\u003cb\u003eFigure\u0026nbsp;4 SEM image of A-DCs after carbon deposition and the corresponding transmission spectra. a\u003c/b\u003e SEM image of the trimmed A-DC for TE\u003csub\u003e1\u003c/sub\u003e mode. \u003cb\u003eb\u003c/b\u003e Measured transmission spectra of the TE\u003csub\u003e1\u003c/sub\u003e-mode A-DC. \u003cb\u003ec\u003c/b\u003e simulation results used to study the measurement results. In this model, \u003cem\u003ew\u003c/em\u003e\u003csub\u003e0\u003c/sub\u003e\u0026thinsp;=\u0026thinsp;1155 nm, \u003cem\u003ew\u003c/em\u003e\u003csub\u003e1\u003c/sub\u003e\u0026thinsp;=\u0026thinsp;2510 nm, \u003cem\u003eh\u003c/em\u003e\u0026thinsp;=\u0026thinsp;340 nm, and \u003cem\u003eg\u003c/em\u003e\u0026thinsp;=\u0026thinsp;300 nm. The major and minor axis lengths of the carbon structure are 120 nm and 60 nm, respectively. \u003cb\u003ed\u003c/b\u003e SEM image of the TE\u003csub\u003e2\u003c/sub\u003e-mode device. \u003cb\u003ee\u003c/b\u003e Measured transmission spectra of the TE\u003csub\u003e2\u003c/sub\u003e-mode A-DC. \u003cb\u003ef\u003c/b\u003e Simulation results for studying the measured spectra. In this model, \u003cem\u003ew\u003c/em\u003e\u003csub\u003e0\u003c/sub\u003e\u0026thinsp;=\u0026thinsp;1155 nm, \u003cem\u003ew\u003c/em\u003e\u003csub\u003e1\u003c/sub\u003e\u0026thinsp;=\u0026thinsp;3870 nm, \u003cem\u003eh\u003c/em\u003e\u0026thinsp;=\u0026thinsp;340 nm, and \u003cem\u003eg\u003c/em\u003e\u0026thinsp;=\u0026thinsp;280 nm. The major and minor axis lengths are 100 nm and 60 nm, respectively\u003c/p\u003e\u003cp\u003eFigure\u0026nbsp;4e shows a significant transmission improvement, where the insertion loss at approximately 1550 nm decreased from 4.5 dB to 0.82 dB after trimming. This result is in close agreement with the simulated data presented in Fig.\u0026nbsp;4f. Herein we discuss carbon deposition using TE\u003csub\u003e1\u003c/sub\u003e and TE\u003csub\u003e2\u003c/sub\u003e modes as examples, but it can be extended to higher-order modes, demonstrating excellent versatility.\u003c/p\u003e\u003cp\u003eCarbon deposition also provides wide tuning capability, ensuring compatibility with a broad range of initial waveguide conditions, as shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e5\u003c/span\u003ea. Here, the initial insertion losses of the A-DCs to be trimmed are gradually increased from 1.8 dB to 17.8 dB. For the TE\u003csub\u003e1\u003c/sub\u003e-mode A-DCs at \u003cem\u003ew\u003c/em\u003e\u003csub\u003e1\u003c/sub\u003e\u0026thinsp;=\u0026thinsp;2525 nm (marked as 1), the improvement in the insertion loss is as fine as 1.46 dB, from 1.83 dB to 0.37 dB, confirming the feasibility of this method for fine-tuning. For the TE\u003csub\u003e2\u003c/sub\u003e-mode A-DCs at \u003cem\u003ew\u003c/em\u003e\u003csub\u003e1\u003c/sub\u003e\u0026thinsp;=\u0026thinsp;3950 nm (marked as 6), the improvement in the insertion loss can reach as high as 16.1 dB, from 17.8 dB to 1.7 dB.\u003c/p\u003e\u003cp\u003eTo evaluate the long-term stability of this non-volatile post-fabrication trimming method, we measured the transmission of the TE\u003csub\u003e1\u003c/sub\u003e-mode A-DC at different time points. As shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e5\u003c/span\u003eb, the transmission slightly decreases during the first two weeks and then stabilizes. The minor change in transmission observed within the first two weeks is likely attributed to stress relaxation or gradual structural stabilization of the deposited carbon, a process can be accelerated by thermal annealing. To verify this hypothesis, a device was prepared and annealed at 200 ℃, followed by measurement at different time intervals. The annealing temperature was selected because heating beyond approximately 300 ℃ causes the deposited carbon to transition from diamond-like carbon to graphite-like carbon\u003csup\u003e\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e,\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e\u003c/sup\u003e. The optical response of the device tended to stabilize after approximately 30 minutes of annealing treatment (see Supplementary 4). Based on the results, the chip containing the TE\u003csub\u003e2\u003c/sub\u003e-mode A-DC was placed on a hot plate at 200 ℃ for 30 minutes. As shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e5\u003c/span\u003ec, the transmission exhibited a slight decrease after heating, but it remains stable over the following weeks after the heat treatment, indicating great stability.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eDue to its mask-free nature, the FIB-based carbon deposition demonstrates exceptional flexibility in the post-fabrication trimming. This trimming method combined with A-DCs can be applied to photonic crossbar arrays\u003csup\u003e\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e,\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e,\u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e\u003c/sup\u003e, which are commonly used to perform matrix-vector multiplication (MVM) operations to accelerate artificial intelligence tasks. As shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e6\u003c/span\u003ea, the A-DCs (indicated by a white dashed box) can be added before the output grating coupler to combine the output light of two small 9\u0026times;2 crossbar arrays, which is equivalent to performing an addition operation.\u003c/p\u003e\u003cp\u003eFigure\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e6\u003c/span\u003eb shows an image of the A-DC in the very same crossbar configuration. For directional couplers based on the TE\u003csub\u003e0\u003c/sub\u003e mode, we observe that after combination, the output power of each input is reduced by 3 dB due to the 50:50 power splitting. Because the orthogonal modes (e.g., TE\u003csub\u003e0\u003c/sub\u003e and TE\u003csub\u003e1\u003c/sub\u003e modes) are independent, when using an A-DC based on mode-division multiplexing (MDM) technology for combination, the loss here can be approximated as the insertion loss, which is much lower than 3 dB after carbon deposition. Figure\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e6\u003c/span\u003ec shows a comparison of transmitted light before and after carbon deposition. The insertion loss for the conversion from the TE\u003csub\u003e0\u003c/sub\u003e mode to TE\u003csub\u003e1\u003c/sub\u003e mode decreases from approximately 3 dB to 0.5 dB after trimming, ensuring efficient combination.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eWe present a trimming method for PICs based on FIB carbon deposition, which is non-invasive, non-volatile, low-loss, and highly stable. As a representative case, we demonstrate its application in post-fabrication tuning of A-DCs, achieving insertion losses as low as 0.36 dB after trimming and showcasing discrete trimming levels ranging from ~\u0026thinsp;1.5 dB to 16.1 dB. These results validate the method\u0026rsquo;s effectiveness on a fabrication-sensitive component, while its combined advantages render it broadly applicable well beyond A-DCs. For a trimming length of ~\u0026thinsp;100 \u0026micro;m, the carbon-induced loss is only\u0026thinsp;~\u0026thinsp;0.3 dB, demonstrating the minimal optical penalty associated with this technique. The trimmed devices also maintain stable performance over a period exceeding two months, confirming the long-term reliability of the method. Compared with thermal tuning, this approach consumes no continuous power and requires no local heaters, enabling more compact and densely integrated devices. It also circumvents the limitations of ion implantation: while this technique offers high precision and excellent long-term stability, its most common form, Ge ion implantation, often results in significant absorption loss and waveguide core damage; in contrast, our method incurs only minimal insertion loss and fully preserves core integrity. These features make the method well suited for a wide range of PIC applications demanding precise optical control, such as MDM components for expanding processing capacity and large-scale photonic accelerators requiring low-loss signal combination.\u003c/p\u003e\u003cp\u003eCrucially, elemental mapping and EELS plasmon map confirm its non-invasive nature, underpinning broad compatibility across diverse material platforms. Although Ga ions are detected in the STEM EDS analysis, our EELS measurements show that the deposited carbon exhibits a bandgap of 3.9 eV, well within the accepted range for diamond-like amorphous carbon. This in turn, indicates that Ga content remains below levels that would compromise optical performance, as further supported by the transmission spectra of photonic devices after carbon deposition. Building on these advantages, FIB carbon deposition offers a mask-free solution for post-fabrication tuning in both current silicon-based platforms and emerging material systems. As such, this method provides a practical and forward-compatible pathway toward large-scale, energy-efficient, and yield-enhancing photonic integration across a wide spectrum of applications from high-density interconnects to next-generation optical computing systems.\u003c/p\u003e"},{"header":"Methods","content":"\u003cdiv id=\"Sec5\" class=\"Section2\"\u003e\u003ch2\u003eFocused Ion Beam-Induced Carbon Deposition on SiN\u003c/h2\u003e\u003cp\u003eCarbon deposition on SiN substrates was performed using a ZEISS Crossbeam 540 FIB/SEM system equipped with a multi-channel Gas Injection System (GIS). Samples were mounted onto the stage and introduced into the chamber, where the stage was tilted to 54\u0026deg; to align the sample surface normal to the FIB column.\u003c/p\u003e\u003cp\u003eOnce thermal and vacuum equilibrium was achieved, a carbon precursor gas was introduced via the GIS. Using a FIB probe current of 20 pA at 30 kV, a carbon line of a specified length was deposited directly on top of the waveguides. The ion beam and GIS were activated simultaneously, and the deposition dose was calibrated to achieve a nominal carbon thickness of ~\u0026thinsp;50 nm.\u003c/p\u003e\u003c/div\u003e\n\u003ch3\u003eSTEM analysis of the trimmed SiN waveguides\u003c/h3\u003e\n\u003cp\u003ePost-trimming analysis of the SiN waveguides was performed using a Thermo Fisher Scientific FEI Themis 300 G3 Titan scanning/transmission electron microscope (S/TEM), operated at 300/60 kV. Samples were prepared in cross-sectional orientation using the FIB technique on a ZEISS Crossbeam 540 FIB/SEM system. The specimens were thinned to electron transparency for imaging. Elemental mapping was conducted in STEM mode using energy-dispersive X-ray spectroscopy (EDS). To investigate bandgap variations in the trimmed waveguides, electron energy loss spectroscopy (EELS) was performed in monochromated STEM mode at 60 kV. The analysis focused on the low-loss region of the EELS spectrum, which contains information about interband transitions and plasmon excitations that are sensitive to changes in the electronic structure and optical bandgap.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors gratefully acknowledge Prof. Dr. Rasmus R. Schr\u0026ouml;der for generously providing the experimental facilities, technical assistance, and invaluable support that were essential for this study. R. X. and Z. Z.\u0026nbsp;gratefully acknowledge the Alexander-von-Humboldt Foundation for providing a postdoctoral fellowship. W. H. P. P. acknowledges support from the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under project numbers 390761711 and 390900948. The work was also supported by the European Union\u0026rsquo;s Horizon Projects (PHONICS, grant no. 101017237; HYBRAIN, grant no. 101046878; 2DNEURALVISION, grant no. 101119489). The manuscript was written through the contributions of all authors. All authors have approved the final version of the manuscript.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eConceptualization: R. X., S. T. and R. R. S., Simulation: R. X., L. M. and L. M., Sample fabrication: R. X., Z. T., A. V. and S. T., Data acquisition and analysis: R. X., Z. T., Z. Z, R. P., X. M., Q. Z and F. B. P., Visualization: R.X., J. B., J. R., and J. D., Writing: All authors, Supervision: W. H. P. P., S. T., R. R. S., and H. B.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare no competing interests.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCorrespondence\u003c/strong\u003e and requests for materials should be addressed to W. H. P. P. and S. T.\u0026nbsp;\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eDolphin JA et al (2023) A hybrid integrated quantum key distribution transceiver chip. npj Quantum Inf 9:84\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eTerrasanta G, Ziarko MW, Bergamasco N, Poot M, Poliak J (2025) Photonic Integrated Circuits for Optical Satellite Links: A Review of the Technology Status and Space Effects. Satell Commun Netw 43:210\u0026ndash;228\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eElshaari AW, Pernice W, Srinivasan K, Benson O (2020) Zwiller, V. Hybrid integrated quantum photonic circuits. Nat Photonics 14:285\u0026ndash;298\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLuo W et al (2023) Recent progress in quantum photonic chips for quantum communication and internet. Light Sci Appl 12:175\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003ePsiQuantum team et al (2025) A manufacturable platform for photonic quantum computing. Nature 641:876\u0026ndash;883\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003ePelucchi E et al (2021) The potential and global outlook of integrated photonics for quantum technologies. Nat Rev Phys 4:194\u0026ndash;208\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLomonte E et al (2021) Single-photon detection and cryogenic reconfigurability in lithium niobate nanophotonic circuits. Nat Commun 12:6847\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAhmed SR et al (2025) Universal photonic artificial intelligence acceleration. Nature 640:368\u0026ndash;374\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBr\u0026uuml;ckerhoff-Pl\u0026uuml;ckelmann F et al (2024) Probabilistic photonic computing with chaotic light. Nat Commun 15:10445\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eXu R et al (2023) Hybrid photonic integrated circuits for neuromorphic computing [Invited]. Opt Mater Express 13:3553\u0026ndash;3606\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAtabaki AH, Eftekhar AA, Askari M, Adibi A (2013) Accurate post-fabrication trimming of ultra-compact resonators on silicon. Opt Express 21:14139\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSpector S, Knecht JM, Juodawlkis PW (2016) Localized in situ cladding annealing for post-fabrication trimming of silicon photonic integrated circuits. Opt Express 24:5996\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eFan T, Xia Z, Adibi A, Eftekhar AA (2018) Highly-uniform resonator-based visible spectrometer on a Si\u003csub\u003e3\u003c/sub\u003e N\u003csub\u003e4\u003c/sub\u003e platform with robust and accurate post-fabrication trimming. Opt Lett 43:4887\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLiu X et al (2020) Highly efficient thermo-optic tunable micro-ring resonator based on an LNOI platform. Opt Lett 45:6318\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eJayatilleka H et al (2021) Post-Fabrication Trimming of Silicon Photonic Ring Resonators at Wafer-Scale. J Lightwave Technol 39:5083\u0026ndash;5088\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eFarmakidis N et al (2023) Scalable High-Precision Trimming of Photonic Resonances by Polymer Exposure to Energetic Beams. Nano Lett 23:4800\u0026ndash;4806\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eProrok S, Petrov AY, Eich M, Luo J, Jen A (2012) K.-Y. Trimming of high-Q-factor silicon ring resonators by electron beam bleaching. Opt Lett 37:3114\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eTang Z et al (2025) Ultralow Loss Coupling Tuning of Photonic Accelerators. Adv Photonics Res 2500175. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1002/adpr.202500175\u003c/span\u003e\u003cspan address=\"10.1002/adpr.202500175\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eVarri A et al (2023) Scalable Non-Volatile Tuning of Photonic Computational Memories by Automated Silicon Ion Implantation. Adv Mater 2310596. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1002/adma.202310596\u003c/span\u003e\u003cspan address=\"10.1002/adma.202310596\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLi P et al (2021) Recent advances in focused ion beam nanofabrication for nanostructures and devices: fundamentals and applications. Nanoscale 13:1529\u0026ndash;1565\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eXu R et al (2024) Mode Conversion Trimming in Asymmetric Directional Couplers Enabled by Silicon Ion Implantation. Nano Lett 24:10813\u0026ndash;10819\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAisenberg S, Chabot R (1971) Ion-Beam Deposition of Thin Films of Diamondlike Carbon. J Appl Phys 42:2953\u0026ndash;2958\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKanda K et al (2009) Graphitization of thin films formed by focused-ion-beam chemical-vapor-deposition. Diam Relat Mater 18:490\u0026ndash;492\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMatsui S (2007) Focused-ion-beam deposition for 3-D nanostructure fabrication. Nucl Instrum Methods Phys Res Sect B 257:758\u0026ndash;764\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMatsui S et al (2000) Three-dimensional nanostructure fabrication by focused-ion-beam chemical vapor deposition. J Vac Sci Technol B 18\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eFujita J et al (2002) Structure and Resonant Characteristics of Amorphous Carbon Pillars Grown by Focused-Ion-Beam-Induced Chemical Vapor Deposition. Jpn J Appl Phys 41:4423\u0026ndash;4426\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eChoi WS, Kim K, Yi J, Hong B (2008) Diamond-like carbon protective anti-reflection coating for Si solar cell. Mater Lett 62:577\u0026ndash;580\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eDa Silva DS et al (2011) Application of amorphous carbon based materials as antireflective coatings on crystalline silicon solar cells. J Appl Phys 110\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eYang Q et al (2021) Research on the fabrication and anti-reflection performance of diamond-like carbon films. Diam Relat Mater 111:108184\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLu Y et al (2021) Pulsed laser deposition of the protective and Anti-reflective DLC film. Infrared Phys Technol 119:103949\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eScendo M, Staszewska-Samson K (2019) Effect of Temperature on Anti-Corrosive Properties of Diamond-Like Carbon Coating on S355 Steel. Materials 12:1659\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWang C et al (2024) Lithium tantalate photonic integrated circuits for volume manufacturing. Nature 629:784\u0026ndash;790\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLi Z et al (2023) High density lithium niobate photonic integrated circuits. Nat Commun 14:4856\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLuo L-W et al (2014) WDM-compatible mode-division multiplexing on a silicon chip. Nat Commun 5:3069\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eYin R et al (2023) Integrated WDM-compatible optical mode division multiplexing neural network accelerator. Optica 10:1709\u0026ndash;1718\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eYang KY et al (2022) Multi-dimensional data transmission using inverse-designed silicon photonics and microcombs. Nat Commun 13:7862\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLarruquert JI, Rodr\u0026iacute;guez-de Marcos LV, M\u0026eacute;ndez JA, Martin PJ, Bendavid A (2013) High reflectance ta-C coatings in the extreme ultraviolet. Opt Express 21:27537\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMilosevic MM et al (2020) Ion Implantation of Germanium Into Silicon for Critical Coupling Control of Racetrack Resonators. J Lightwave Technol 38:1865\u0026ndash;1873\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eRobertson J (2002) Diamond-like amorphous carbon. Mater Sci Engineering: R: Rep 37:129\u0026ndash;281\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eIgaki J et al (2007) Elementary Analysis of Diamond-Like Carbon Film Formed by Focused-Ion-Beam Chemical Vapor Deposition. \u003cem\u003ejjap\u003c/em\u003e 46, 8003\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eTallant DR, Parmeter JE, Siegal MP, Simpson RL (1995) The thermal stability of diamond-like carbon. Diam Relat Mater 4:191\u0026ndash;199\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eFeldmann J et al (2021) Parallel convolutional processing using an integrated photonic tensor core. Nature 589:52\u0026ndash;58\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBr\u0026uuml;ckerhoff-Pl\u0026uuml;ckelmann F, Feldmann J, Wright CD, Bhaskaran H, Pernice W (2021) H. P. Chalcogenide phase-change devices for neuromorphic photonic computing. J Appl Phys 129:151103\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"nature-portfolio","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"","title":"Nature Portfolio","twitterHandle":"","acdcEnabled":false,"dfaEnabled":false,"editorialSystem":"ejp","reportingPortfolio":"","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-7410735/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7410735/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cb\u003ePhotonic integrated circuits (PICs), widely used in optical communications and computing, require precise post-fabrication trimming due to their high sensitivity to fabrication imperfections. Focused ion beam (FIB) carbon deposition offers a promising solution, combining high spatial precision with minimal invasiveness and inherently broad material compatibility. Here, we demonstrate this technique for the first time to enable non-volatile post-fabrication trimming of PICs. To validate this approach, we use asymmetric directional couplers as representative fabrication-sensitive components. Structural characterizations verify the non-invasive nature of the localized carbon deposition, and device measurements show discrete transmission tuning levels of 1.46\u0026ndash;16.1 dB with a trimming-induced insertion loss of only 0.3 dB. Furthermore, the optical response remains stable over two months following a brief initial settling phase. These results establish FIB carbon deposition as a robust and broadly applicable PIC trimming technique, empowering optical communications, computing, and beyond to meet rising data demands.\u003c/b\u003e\u003c/p\u003e","manuscriptTitle":"Localized Carbon Deposition Enables Non-Invasive Trimming of Photonic Integrated Circuits","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-08-22 04:24:54","doi":"10.21203/rs.3.rs-7410735/v1","editorialEvents":[],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"nature-communications","isNatureJournal":true,"hasQc":false,"allowDirectSubmit":false,"externalIdentity":"NCOMMS","sideBox":"Learn more about [Nature Communications](http://www.nature.com/ncomms/)","snPcode":"","submissionUrl":"https://mts-ncomms.nature.com/","title":"Nature Communications","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"ejp","reportingPortfolio":"Nature Communications","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"4a83d46e-e30d-4055-adc3-488b916ca231","owner":[],"postedDate":"August 22nd, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[{"id":53472393,"name":"Physical sciences/Optics and photonics/Optical materials and structures"},{"id":53472394,"name":"Physical sciences/Nanoscience and technology/Nanoscale materials/Synthesis and processing"},{"id":53472395,"name":"Physical sciences/Optics and photonics/Optical physics/Nanophotonics and plasmonics"}],"tags":[],"updatedAt":"2026-04-29T08:26:30+00:00","versionOfRecord":[],"versionCreatedAt":"2025-08-22 04:24:54","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7410735","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7410735","identity":"rs-7410735","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}