Silicon nitride waveguides with directly grown WS₂ for efficient second-harmonic generation

Abstract

Different functions can be directly realized by silicon (Si) in integrated electronic circuits. Although Si and silicon nitride (Si₃N₄) photonics have shown great potential in integrated optoelectronic devices, different functions, such as light generation, transparency for guided light, and light detection, cannot be simultaneously achieved only by Si or Si₃N₄. Second-order nonlinearity is another optical property they do not possess due to their centrosymmetric properties. Several kinds of 2D materials emerged recently and were transferred to specified photonic devices aimed at improving their nonlinear performance. However, the transferring methods are time-consuming, unable to achieve large-scale production, and will inevitably cause materials damage and introduce impurities at the interface. Herein, we demonstrate the direct growth of large-area homogeneous monolayer WS₂via a physical vapor deposition method onto Si₃N₄ waveguides. The WS₂ growth can be controlled mainly along the Si₃N₄ waveguides and the waveguides show an obvious enhancement of second-harmonic generation with the elongated WS₂ coverage. The direct growth of WS₂ endows Si₃N₄ integrated photonics with new nonlinear optical properties. As an alternative method of transferring 2D materials, the method we present here is compatible with large-scale integrated photonic fabrication, which lays the foundation for on-chip integrated optical fabrication and applications.

---

1. Introduction

Integrated Si and Si₃N₄ photonics have developed rapidly recently.^1–5 However, these materials do not possess an even-order optical nonlinearity due to their centrosymmetric properties, and the lowest nonlinear response is third-order nonlinearity, which limits the applications of Si or Si₃N₄ photonics in the nonlinear field. Second-harmonic generation (SHG) is a key topic in nonlinear optics and is a process during which twice the frequency of the incident light can radiate from the nonlinear optical medium.⁶ Some 2D materials, including monolayer transition metal dichalcogenides (TMDCs), possess high second-order nonlinear coefficients due to their broken inversion symmetry. The nonlinear properties of these 2D materials have also been widely studied in theory and application.^7–9 One of the great advantages of these 2D materials is that they can be integrated with various materials or structures,^10–15 especially with microfibers^16–19 and waveguides,^20–22 such that SHG enhancement or other enhanced nonlinear processes are realized and analyzed.

However, the integrations of 2D materials mentioned above are all realized by transferring methods which are not suitable for large scale production, and the inevitable introduction of impurities during the transferring process will affect the optical performance of these 2D materials. Researchers therefore began to investigate the integration of 2D materials by direct growth, aiming to improve the optical performance of the 2D materials. Large-area monolayer graphene²³ and TMDCs²⁴ can be grown directly and uniformly in microstructure fibers, and these kinds of microstructure fibers show excellent nonlinear characteristics. Growing monolayer TMDCs on a Si step was achieved and, based on this method, a FinFET was obtained.²⁵ However, presently to the best of our knowledge, the direct growth of TMDCs on Si₃N₄ waveguides has not been achieved.

As one kind of TMDC, monolayer WS₂ shows excellent optical properties²⁶ and possesses a high second-order nonlinearity.²⁷ For a certain pump power, light intensity can be greatly enhanced at the surface of the Si₃N₄ waveguide compared with that at a silica microfiber, and SHG should be more efficient in the Si₃N₄/WS₂ waveguide than that in the fiber/WS₂ structure (Fig. S1†), since the SHG is proportional to |E|². Hence, we propose directly growing monolayer WS₂ onto on-chip Si₃N₄ waveguides utilizing physical vapor deposition (PVD) method to realize effective harmonic frequency conversion in the Si₃N₄/WS₂ hybrid waveguide structure.

2. Results and discussion

In this paper, the fundamental transverse electric (TE) mode is analyzed. In the Si₃N₄ waveguide, the fundamental TE mode has the lowest transmission loss studied in the work. As discussed by Vlasov et al.,²⁸ the TE mode exhibited much lower propagation loss than that for transverse magnetic (TM) mode for a SOI strip waveguide since the TM polarization mode results in an increased interaction with the top surface roughness. Also, a higher order TE mode will increase the interaction with the sidewall surface roughness, resulting in a larger propagating loss.²⁹ Based on the above analysis, we only focus on the fundamental TE mode for the Si₃N₄ waveguide in this paper. The SHG light component is also assumed to be the fundamental TE mode. When conducting the experiment, we adjusted the polarization of the fiber and the coupling between the fiber and the waveguide to make sure that the fundamental TE mode is propagated in the Si₃N₄ waveguide.

2.1 Growth and characterization of monolayer WS₂ on Si₃N₄ waveguides

In order to grow WS₂ directly on the Si₃N₄ waveguide, as shown in Fig. 1(a), firstly a Si₃N₄ film with a thickness of 0.6 μm was deposited on the SiO₂/Si wafer by means of low-pressure chemical vapor deposition (LPCVD). The required waveguide pattern (waveguides with a width of 2.0 μm and thickness of 0.6 μm) was defined by electron beam lithography (EBL), and then etched by reaction ion etching (RIE). The PVD setup for WS₂ growth is shown in Fig. 1(b). WS₂ powder (0.4 g) was placed at the upstream of the quartz tube with diameter of 1 inch, and the distance from the WS₂ powder to the prepared chip, which was placed in the center of the furnace, was about 7.5 cm. Firstly, 80 sccm of Ar was flowed into the tube for 30 min to ensure a stable and inert chemical reaction environment for the following crystal growth. After that, the 80 sccm of Ar flow was kept unchanged, and the furnace was heated from room temperature to 1160 °C over 50 min, maintained for 15 min for WS₂ growth, and cooled to room temperature naturally. The temperature of the furnace is a key factor for the high quality growth of WS₂. At 1160 °C, a triangle-shaped monolayer WS₂ was grown on the Si₃N₄ waveguide.

Fig. 1 Growth of monolayer WS₂ on the Si₃N₄ waveguide. (a) Schematic of the proposed Si₃N₄/WS₂ hybrid waveguide structure. (b) Schematic of the PVD setup used in the experiment.

The corresponding second harmonic (SH) intensity mapping for the hybrid waveguide in Fig. 2(a) is displayed in Fig. 2(b). When conducting the experiment for SHG mapping (Fig. 2(b)), the input laser was incident normally on the sample and focused on top of the Si₃N₄ waveguide covered by the grown monolayer WS₂. We used Witec Alpha 300R, equipped with a picosecond pulsed laser, and a 20× and a 100× objective lens (N.A. = 0.9) to focus the power on the Si₃N₄/WS₂ hybrid region. Using the confocal technique, we could collect the SHG signal from the focal area and see an obvious SHG signal for the Si₃N₄/WS₂ hybrid region. The PL intensities of WS₂ are the same for normal incidence. For the 20× objective lens, the resolution in the vertical direction is low, and thus the WS₂ on the waveguide region and the bare WS₂ region far away from waveguide have similar PL intensities. While for the 100× objective lens, the resolution in the vertical direction is relatively high, the WS₂ on the waveguide region is in the focal plane while the bare WS₂ region far away from waveguide is out of the focal plane, and thus the PL intensities from the bare WS₂ region are lower than those from the WS₂ on the Si₃N₄ waveguide. The scanning electron microscopy (SEM) image in Fig. 2(c) shows that the WS₂ can be grown well on the Si₃N₄ waveguide. Photoluminescence (PL) and Raman analyses were also performed for the WS₂ grown on top of the Si₃N₄ waveguide using a 532 nm laser, as displayed in Fig. 2(d) and (e). The peak value of the PL spectrum in Fig. 2(d) is located at about 630 nm, and the characteristic peaks of E_2g and A_1g of monolayer WS₂ can be clearly seen from the Raman spectrum shown in Fig. 2(e). Comprehensive analysis from Fig. 2(a)–(e) clearly indicates that monolayer WS₂ was directly grown on the Si₃N₄ waveguide, although there was a 0.6 μm step of the waveguide.

Fig. 2 Characterization of monolayer WS₂ grown on the Si₃N₄ waveguide. (a) Optical microscopy image of a single WS₂ crystal grown on the Si₃N₄ waveguide. (b) The corresponding SH intensity mappings of the structure shown in (a). The Si₃N₄/WS₂ hybrid waveguide region is highlighted by white dashed lines. The image on the right is an enlarged view focused on the waveguide region marked with the dashed white rectangle in the left image. (c) SEM image of monolayer WS₂ grown on the Si₃N₄ waveguide. (d and e) The PL spectrum (d) and Raman spectrum (e) of the monolayer WS₂ grown on the Si₃N₄ waveguide.

2.2 Harmonic generation from monolayer WS₂

A femtosecond laser with a central wavelength of about 1570 nm was used to verify the nonlinear property of WS₂ and excite the SHG and third harmonic generation (THG) in a monolayer WS₂ grown on the SiO₂/Si substrate without the Si₃N₄ waveguide. The schematic of the excitation is shown in the inset in Fig. 3(a). A tapered optical lensed fiber guided the oblique incidence of the excitation light. Although the excitation light was focused to a small area about 4 μm diameter with a very high optical power density, the interaction of the light and the monolayer WS₂ was very weak for the oblique incidence. Consequently, the SHG and THG signals were very weak, as shown in Fig. 3(a).

Fig. 3 Harmonic generation from monolayer WS₂. (a) SHG and THG signals generated from the monolayer WS₂ on the SiO₂/Si substrate pumped by a femtosecond laser pulse with a center wavelength of 1570 nm. The illustration is a schematic of the oblique incidence on the monolayer WS₂. (b) PL images of a single triangular monolayer WS₂ crystal grown on the Si₃N₄ waveguide. The image on the right is an enlarged view focused on the monolayer WS₂ region marked with the dashed red rectangle on the left image. Green light emission is excited from the Si₃N₄ waveguide. (c) Enhanced SHG spectrum of the Si₃N₄/WS₂ hybrid waveguide structure (blue line) and bare Si₃N₄ waveguide (red line) under a laser pulse excitation centered at 1570 nm. For clarity, the SHG spectrum is upshifted vertically. The illustration is a schematic of the SHG when the edge coupling method is adopted. (d) Corresponding transmission spectrum of the Si₃N₄/WS₂ hybrid waveguide shown in (b). The data are normalized. (e) SHG intensity dependence on the excitation power in the Si₃N₄/WS₂ hybrid waveguide structure.

However, the interaction between the WS₂ and light can be greatly enhanced through evanescent field coupling.^30,31 Next, we coupled the femtosecond laser to the Si₃N₄ waveguide with one triangular WS₂ crystal grown on the top and both sides adopting an edge coupling method,³² as depicted in the inset in Fig. 3(c). The coupling efficiency of the femtosecond laser to the Si₃N₄ waveguide was much higher than that of the oblique incidence and thus the harmonic intensity will considerably increase. The photoluminescence (PL) image in Fig. 3(b) verifies the monolayer property of WS₂ grown on the Si₃N₄ waveguide, since multilayer WS₂ possesses an indirect bandgap and thus a very low emission efficiency. The Si₃N₄ does not possess second-order nonlinearity, and the result shows that there was no SHG signal (Fig. 3(c), red line). While for the Si₃N₄/WS₂ hybrid waveguide, the SHG signal with a center wavelength of about 790 nm was apparent (Fig. 3(c), blue line). However, there was no THG signal from the Si₃N₄/WS₂ waveguide, which is because the energy of the third harmonic signal (530 nm or so) exceeds the band gap of monolayer WS₂ (about 2 eV),²⁶ and the generated third harmonic signal was absorbed during transmitting in the Si₃N₄/WS₂ hybrid waveguide. However, for Fig. 3(a), a tapered optical lensed fiber guided the oblique incidence of the excitation light to the monolayer WS₂, the generated third harmonic signals did not transmit in the waveguide, but emitted directly out of the substrate. As it cannot be absorbed by WS₂, the THG signal generated in monolayer WS₂ on the SiO₂/Si substrate could be collected. Fig. 3(d) is the corresponding transmission spectrum of the femtosecond pulses. Subsequently, we evaluated the harmonic emission intensity dependence of the Si₃N₄/WS₂ waveguide on the excitation power by monitoring the transmission power from 40 to 250 μW, where the quadratic dependence on the excitation power shown in Fig. 3(e) reveals that the collected signal was obviously from the SHG emission.

2.3 Growth of large-area monolayer WS₂ on Si₃N₄ waveguides for further enhanced SHG

As mentioned above, the quality, layers and size of the grown WS₂ can be controlled by the temperature. Fig. 4(a)–(c) show the growth results of monolayer WS₂ on the Si₃N₄ waveguides under different temperatures. When we increased the growing temperature from 1160 °C to 1170 °C to 1176 °C, the covered area of monolayer WS₂ on the Si₃N₄ waveguides increased gradually, from a single triangular region (Fig. 3(b)) to several triangular regions (Fig. 4(a)), and then a continuous monolayer WS₂ film (Fig. 4(b) and (c)). Fig. 4(b) shows that at 1176 °C, most of the WS₂ can be grown along the Si₃N₄ waveguide. By etching SiO₂ at both sides of the Si₃N₄ waveguide (for the etched structure, see Fig. S2†), a large-area and continuous monolayer of WS₂ can be grown on the Si₃N₄ waveguide, as shown in Fig. 4(c). Also, with the increasing light–matter interaction length, SHG can be enhanced by ∼6–7 times (red line, marked as “waveguide 1” in Fig. 4(a)), and ∼150 times (blue line, marked as “waveguide 2” in Fig. 4(c)), as depicted in Fig. 4(d). It is shown that a higher coverage ratio of WS₂ corresponds to more small peaks in the SHG spectrum. This is the result of the interference of light: when WS₂ is grown on the Si₃N₄ waveguides, the refraction index of the waveguide changes. A higher coverage ratio of WS₂ corresponds to more small interference peaks. Fig. 4(e) is the corresponding transmission spectrum of the femtosecond pulses. As shown in Fig. 4(e), the transmission spectrum from 1500 nm to 1630 nm also shows interference patterns, which is consistent with the interference patterns from the SHG spectrum. This indicates that the growth of WS₂ on the Si₃N₄ waveguide is inhomogeneous. The interference can be avoided if WS₂ is grown homogeneously onto the Si₃N₄ waveguides.

Fig. 4 Growth of a large-area of monolayer WS₂ on Si₃N₄ waveguides for an enhanced SHG. (a–c) Dependence of the WS₂ covering area on the PVD-growing temperatures of 1170 °C (a), and 1176 °C (b and c). An oblique LED illumination was used in (b) to make the waveguides clearer. The image on the right of (c) is an enlarged view of the Si₃N₄ waveguide covered by monolayer WS₂ marked with the dashed white rectangle in the left image. (d and e) Enhanced SHG (d) and transmission (e) spectra of the Si₃N₄/WS₂ hybrid waveguides under a laser pulse excitation centered at about 1570 nm grown at different temperatures shown in (a) (marked as waveguide 1) and (c) (marked as waveguide 2), respectively. Both spectra are shifted vertically for clarity and the data of (e) are normalized.

Besides, the orientation of the WS₂ crystalline will affect the SHG efficiency, as discussed by Chen et al. for monolayer MoSe₂ on a silicon waveguide.²¹ For our research in this paper, the crystalline orientation of monolayer WS₂ directly grown onto Si₃N₄ waveguides is arbitrary and the orientation will undoubtedly affect the SHG intensity. For a certain sample, by adjusting the polarization of the input light, we can obtain the maximum SHG intensity. For the grown large-area WS₂, the collected SHG signal is the consequence of the average effect of the crystalline orientation.

3. Conclusions

To conclude, we have proposed a method to directly grow large-area monolayer WS₂ onto Si₃N₄ waveguides. The monolayer properties of the grown WS₂ were verified by means of SEM, and PL and Raman spectroscopy. By virtue of the growth of large-area and homogeneous WS₂, the enhanced SHG from the Si₃N₄/WS₂ hybrid waveguide was demonstrated. Our work helps to integrate 2D materials with Si₃N₄ photonics, enrich the optical properties of integrated Si₃N₄ photonics and pave the way for the large-area growth of 2D materials on integrated circuits.

Author contributions

N. L. and X. Y. contributed equally to this work. K. L., N. L. and Z. H. Z. conceived the study. X. Y. and N. L. conducted the experiment of WS₂ growth. X. Y. conducted the measurements of Raman spectra. F. C. and Y. B. Z. conducted the experiment of SH mapping. K. L. and N. L. conducted the main measurements. N. L. and J. P. X. drew the schematic diagrams in the paper. N. L. and K. L. led the writing of the paper. All authors discussed the results and commented on the manuscript.

Conflicts of interest

There are no conflicts to declare.

Notes and references

1. J. Leuthold, C. Koos and W. Freude, Nat. Photonics, 2010, 4, 535–544.
2. G. T. Reed, G. Mashanovich, F. Y. Gardes and D. J. Thomson, Nat. Photonics, 2010, 4, 518–526.
3. A. H. Atabaki, S. Moazeni, F. Pavanello, H. Gevorgyan, J. Notaros, L. Alloatti, M. T. Wade, C. Sun, S. A. Kruger, H. Meng, K. Al Qubaisi, I. Wang, B. Zhang, A. Khilo, C. V. Baiocco, M. A. Popovic, V. M. Stojanovic and R. J. Ram, Nature, 2018, 556, 349–354.
4. X. Li, N. Youngblood, C. Rios, Z. Cheng, C. D. Wright, W. H. P. Pernice and H. Bhaskaran, Optica, 2019, 6, 1–6.
5. J. Liu, E. Lucas, A. S. Raja, J. He, J. Riemensberger, R. N. Wang, M. Karpov, H. Guo, R. Bouchand and T. J. Kippenberg, Nat. Photonics, 2020, 14, 486–491.
6. G. P. Agrawal, J. Opt. Soc. Am. B, 2011, 28, A1–A10.
7. X. F. Liu, Q. B. Guo and J. R. Qiu, Adv. Mater., 2017, 29, 29.
8. X. L. Wen, Z. B. Gong and D. H. Li, InfoMat, 2019, 1, 317–337.
9. B. Guo, Q. L. Xiao, S. H. Wang and H. Zhang, Laser Photonics Rev., 2019, 13, 46.
10. A. Majumdar, C. M. Dodson, T. K. Fryett, A. Zhan, S. Buckley and D. Gerace, ACS Photonics, 2015, 2, 1160–1166.
11. T. K. Fryett, K. L. Seyler, J. J. Zheng, C. H. Liu, X. D. Xu and A. Majumdar, 2D Mater., 2017, 4, 015031.
12. X. T. Gan, C. Y. Zhao, S. Q. Hu, T. Wang, Y. Song, J. Li, Q. H. Zhao, W. Q. Jie and J. L. Zhao, Light: Sci. Appl., 2018, 7, 6.
13. J. W. Shi, W. Y. Liang, S. S. Raja, Y. G. Sang, X. Q. Zhang, C. A. Chen, Y. R. Wang, X. Y. Yang, Y. H. Lee, H. Ahn and S. Gwo, Laser Photonics Rev., 2018, 12, 7.
14. J. W. Chen, K. Wang, H. Long, X. B. Han, H. B. Hu, W. W. Liu, B. Wang and P. X. Lu, Nano Lett., 2018, 18, 1344–1350.
15. Q. C. Yuan, L. Fang, H. L. Fang, J. T. Li, T. Wang, W. Q. Jie, J. L. Zhao and X. T. Gan, ACS Photonics, 2019, 6, 2252–2259.
16. P. G. Yan, A. J. Liu, Y. S. Chen, H. Chen, S. C. Ruan, C. Y. Guo, S. F. Chen, I. L. Li, H. P. Yang, J. G. Hu and G. Z. Cao, Opt. Mater. Express, 2015, 5, 479–489.
17. Y. Z. Wang, F. Zhang, X. Tang, X. Chen, Y. X. Chen, W. C. Huang, Z. M. Liang, L. M. Wu, Y. Q. Ge, Y. F. Song, J. Liu, D. Zhang, J. Q. Li and H. Zhang, Laser Photonics Rev., 2018, 12, 9.
18. J. H. Chen, J. Tan, G. X. Wu, X. J. Zhang, F. Xu and Y. Q. Lu, Light: Sci. Appl., 2019, 8, 8.
19. B. Q. Jiang, Z. Hao, Y. F. Ji, Y. G. Hou, R. X. Yi, D. Mao, X. T. Gan and J. L. Zhao, Light: Sci. Appl., 2020, 9, 8.
20. L. H. Liu, K. Xu, X. Wan, J. B. Xu, C. Y. Wong and H. K. Tsang, Photonics Res., 2015, 3, 206–209.
21. H. T. Chen, V. Corboliou, A. S. Solntsev, D. Y. Choi, M. A. Vincenti, D. de Ceglia, C. de Angelis, Y. R. Lu and D. N. Neshev, Light: Sci. Appl., 2017, 6, 7.
22. Y. J. Zhang, L. Tao, D. Yi, J. B. Xu and H. K. Tsang, J. Opt., 2020, 22, 5.
23. K. Chen, X. Zhou, X. Cheng, R. X. Qiao, Y. Cheng, C. Liu, Y. D. Xie, W. T. Yu, F. R. Yao, Z. P. Sun, F. Wang, K. H. Liu and Z. F. Liu, Nat. Photonics, 2019, 13, 754–759.
24. Y. G. Zuo, W. T. Yu, C. Liu, X. Cheng, R. X. Qiao, J. Liang, X. Zhou, J. H. Wang, M. H. Wu, Y. Zhao, P. Gao, S. W. Wu, Z. P. Sun, K. H. Liu, X. D. Bai and Z. F. Liu, Nat. Nanotechnol., 2020, 15, 987–988.
25. M. L. Chen, X. D. Sun, H. Liu, H. W. Wang, Q. B. Zhu, S. S. Wang, H. F. Du, B. J. Dong, J. Zhang, Y. Sun, S. Qiu, T. Alava, S. Liu, D. M. Sun and Z. Han, Nat. Commun., 2020, 11, 7.
26. H. L. Liu, C. C. Shen, S. H. Su, C. L. Hsu, M. Y. Li and L. J. Li, Appl. Phys. Lett., 2014, 105, 4.
27. C. Janisch, Y. X. Wang, D. Ma, N. Mehta, A. L. Elias, N. Perea-Lopez, M. Terrones, V. Crespi and Z. W. Liu, Sci. Rep., 2014, 4, 5.
28. Y. A. Vlasov and S. J. McNab, Opt. Express, 2004, 12, 1622–1631.
29. X. Ji, J. K. Jang, U. D. Dave, M. Corato-Zanarella, C. Joshi, A. L. Gaeta and M. Lipson, Laser Photonics Rev., 2021, 15, 2000353.
30. J. L. Kou, J. H. Chen, Y. Chen, F. Xu and Y. Q. Lu, Optica, 2014, 1, 307–310.
31. S. J. Koester and M. Li, IEEE J. Sel. Top. Quantum Electron., 2014, 20, 84–94.
32. V. R. Almeida, R. R. Panepucci and M. Lipson, Opt. Lett., 2003, 28, 1302–1304.

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: 10.1039/d1nr06216f

‡ These authors contributed equally to this work.

---