1. Introduction

Because of transparency in both the visible and infrared spectrum and capability to be deposited on almost any substrate using mature technologies such as low pressure chemical vapor deposition (LPCVD) or plasma enhanced chemical vapor deposition (PECVD), silicon nitride (Si₃N₄) has become one of the promising materials for integrated photonics applications, especially for three-dimensional (3D) integration of multiple photonic layers above the processed microelectronics [1, 2]. Propagation loss as low as 0.1 dB/cm and intrinsic quality factor of ring resonators as high as 3 × 10⁶ have been demonstrated on the Si₃N₄ platform [3]. Moreover, active functions such as parametric amplification [4] and broadband supercontiniuum generation [5] have been realized using the nonlinear properties of Si₃N₄. However, due to the relatively small refractive index contrast between the Si₃N₄ core and the oxide cladding (Δn ~0.5), the optical mode in Si₃N₄ waveguides is usually loosely confined and the minimum allowable bend radius is relatively large (e.g., >20 μm [3]), which results in large footprint for Si₃N₄-based photonic devices and low integration density for Si₃N₄-based photonic circuits.

On the other hand, deep subwavelength plasmonic waveguides such as metal-insulator-metal (MIM) can confine the optical mode in the nanometer scale and allow sharp bending, thus enabling to miniaturize the photonic devices to be comparable with the nanoelectronic devices and to increase the integration density significantly [6–8]. However, the high-confinement plasmonic waveguides suffer from large propagation loss due to the unavoidable metal loss [7]. A straightforward solution for this problem is to integrate both the high-confinement plasmonic waveguides and the conventional low-loss dielectric waveguides in the same chip, in which the plasmonic waveguides can be used to realize ultracompact functional photonic devices and the conventional dielectric waveguides are used for optical signal transmission over long distances [8]. For cost efficiency and ease of fabrication the integrated circuits are highly desired to be still CMOS compatible. In the silicon-on-insulator (SOI) platform, such CMOS-compatible photonic and plasmonic integrated circuits have been demonstrated based on horizontal Cu-insulator-Si-insulator-Cu nanoplasmonic waveguides [9–12] or vertical Cu-insulator-Si hybrid plasmonic waveguides [13, 14]. Copper is the metal of choice because it is widely used in CMOS backend processes and it offers much lower metal loss around 1550-nm telecom wavelengths than aluminum [15].

Here, CMOS-compatible plasmonics for integration in Si₃N₄-based photonic circuits is addressed. The plasmonic waveguide has a horizontal Cu-Si₃N₄-Cu or Cu-SiO₂-Si₃N₄-SiO₂-Cu structure. These two structures can be catalogued in general as the MIM waveguide. To date, MIM waveguides have been thoroughly investigated and many MIM-based photonic devices have been proposed theoretically [16–19]. However, only a few of them have been demonstrated experimentally [20–22], probably because the common-sense MIM waveguides have a layered metal-dielectric-metal structure using Au or Ag as the metal, thus they are not CMOS-compatible and are difficult to connect with the conventional low-loss dielectric waveguides. In contrast, the horizontal MIM-like waveguides developed in this paper are fully CMOS compatible and ease in fabrication, thus it provides an alternative way to realize the proposed MIM plasmonic devices cost-effectively.

2. Experiments

Fabrication was started from 200-mm-diameter silicon wafers. ~4-μm-thick SiO₂ was deposited by PECVD, followed by chemical mechanical polishing (CMP) to smooth the surface (Fig. 1(a)). Then, ~375-nm-thick Si₃N₄ was deposited by LPCVD. The deposition was carried out for two times (first ~200 nm, and then ~175 nm) to prevent stress-induced cracking (Fig. 1(b)). The wafers were patterned using standard ultraviolet (UV) lithography. ~326-nm Si₃N₄ was dry etched to leave ~49-nm Si₃N₄ slab layer using photoresistor as the mask (Fig. 1(c)). After removing the photoresistor, a thick SiO₂ layer was deposited by PECVD (Fig. 1(d)). SiO₂ windows were opened by lithography and dry etching SiO₂ down to the Si₃N₄ slab layer (Fig. 1(e)), which defined the plasmonic area. A thin dielectric layer can be deposited as an optional step to fabricate the so-called metal-multi-insulator-metal structures [16]. Attractive functionalities such as electro-optic modulation may be introduced if a functional dielectric such as VO₂ [23] or ITO [24] is deposited. Here, for simplification, a thin SiO₂ layer was deposited on some wafers by PECVD (Fig. 1(f)). Then, 150-nm-thick Cu was deposited by sputtering, followed by 1-μm-thick Cu electroplating (Fig. 1(g)). Low-temperature annealing at 200°C was carried out for half hour to improve the Cu film quality [15]. Then, Cu-CMP was carried out to remove Cu (as well as the deposited thin SiO₂ layer) outside the windows (Fig. 1(h)). Finally, a thick SiO₂ layer was deposited as a protective layer.

 

Fig. 1 Fabrication processes of horizontal Cu-dielectric-Si₃N₄-dielectric-Cu plasmonic waveguides using standard Si-CMOS backend processes.

Download Full Size | PDF

Figure 2(a) shows a microscope picture of one of the fabricated photonic devices. The Cu-covered plasmonic area is inserted in the conventional Si₃N₄ waveguides through linear taper couplers, as shown schematically in Fig. 2(b). Figure 2(c) is the cross-sectional transmission electron microscopy (XTEM) image of one of the fabricated plasmonic waveguides. A Si₃N₄ core above a thin Si₃N₄ slab is covered by ~10-nm SiO₂ layer almost conformally. Due to the imperfection of fabrication, the cross section of the Si₃N₄ core is not ideally rectangular, but looks like triangle-shaped with sidewall angle of ~83° at the bottom region and ~65° at the top region. The Si₃N₄ core shown in Fig. 2(c) has bottom width of ~90 nm and height (etch depth) of ~288 nm above the ~49-nm-thick Si₃N₄ slab. In fabrication, the bottom width of the Si₃N₄ core is tuned from ~110 nm to ~80 nm by changing the exposure dose during UV lithography. Due to the angled sidewall (assuming all Si₃N₄ core have similar sidewall angles), the Si₃N₄ core with the bottom width larger than ~110 nm is trapezoid-shaped with height of ~326 nm while that with the bottom width smaller than ~110 nm is triangle-shaped whose height decreases accordingly with decreasing the bottom width. Plasmonic waveguides without the thin SiO₂ interlayer (i.e., the Cu-Si₃N₄-Cu waveguides) were also fabricated.

 

Fig. 2 (a) Microscope picture of one of the fabricated devices; (b) Schematic layout of the horizontal Cu-dielectric-Si₃N₄-dielectric-Cu waveguide inserted in the conventional Si₃N₄ waveguide through taper couplers with length of L_C; and (c) XTEM image of one of the fabricated horizontal Cu-SiO₂-Si₃N₄-SiO₂-Cu plasmonic waveguides.

Download Full Size | PDF

The diced chips were measured using the fiber-waveguide-fiber measurement setup as the conventional dielectric waveguides [9–11]. Quasi-TE-polarized light (the electric field is parallel to the chip surface plane) is coupled into the input Si₃N₄ waveguide from a lensed polarization-maintaining (PM) single-mode fiber, transports through the waveguide, and then is coupled out to another fiber to be measured by a power meter and an optical spectrum analyzer (OSA). A semi-auto micrometer piezo-stage was used to adjust both input and output fibers to search the maximum output power.

Numerical simulation was performed using commercial software Lumerical [25]. The refractive indices of Si₃N₄ and SiO₂ are set to 2.0 and 1.444, respectively, and the complex index of Cu at 1550 nm is set to 0.181 + j11.05 [15, 26].

3. Results and discussion

3.1 Si₃N₄ rib waveguides

As the abovementioned, the plasmonic devices will be integrated in the low-loss dielectric waveguide circuits. Therefore, Si₃N₄ waveguides which were fabricated together with the plasmonic devices in the same chip are characterized first. Figure 3(a) shows the XTEM image of the fabricated Si₃N₄ waveguide, it has a rib structure with ~892-nm width, ~375-nm height, and ~326-nm etch depth, embedded in the thick SiO₂ cladding layer. The electrical field (|E_x|) distribution of the fundamental 1550-nm TE mode is depicted in Fig. 3(b), calculated using the eigen-mode expansion (EME) method [25]. The calculated effective modal index is 1.626 and the ratio of optical power contained in the Si₃N₄ core is 60%. The remaining 40% optical power is contained in the cladding SiO₂ layer. The propagation loss measured by the standard cutback method is plotted in Fig. 3(c) as a function of wavelength ranging from visible to infrared, using six available laser sources in our laboratory, which operate at different wavelengths. The propagation loss is smaller than ~1 dB/cm over a wide wavelength range, comparable to those reported in literature for single-mode Si₃N₄ channel waveguides [2, 3]. The loss increases in the C-band and the large loss (~10.5 dB/cm) at 420 nm may be attributed to Si-H and N-H bonds in the Si₃N₄ film. Waveguide ring resonators (WRRs) with radii ranging from 20 μm to 50 μm were fabricated. They exhibit typical resonant characteristics with extinction ratio of ~20-dB, loaded Q-value of ~1100, and group index of ~1.9 [27] (not shown here). These results indicate that the subsequent processes for the Cu-based plasmonic devices fabrication do not degrade the processed Si₃N₄ waveguides in the same chip.

 

Fig. 3 (a) XTEM image of the fabricated Si₃N₄ rib waveguide; (b) Electric field |E_x| distribution of the fundamental 1550-nm TE mode in the waveguide, calculated using the EME method; and (c) Propagation loss versus wavelength, measured using six laser sources operating at different wavelengths.

Download Full Size | PDF

3.2 Plasmonic waveguides

Figure 4(a) plots spectra measured from a set of straight Cu-SiO₂-Si₃N₄-SiO₂-Cu waveguides with the same Si₃N₄ core but different lengths (L_Ps) using a broadband (1520–1620 nm) laser source, normalized by that measured from a reference Si₃N₄ waveguide without the plasmonic area. All spectra exhibit similar small ripples (while the spectra measured from the Si₃N₄ waveguides do not exhibit such ripples). We suspect that these ripples may arise from the weak refection at the interfaces between the Si₃N₄ waveguide and the Cu-covered plasmonic area rather than the unstable input source. Except these small ripples, the spectra are almost wavelength-independent in the 1520–1620-nm wavelength range. The output power depends on L_P almost linearly, as shown in Fig. 4(b), from which the propagation loss is extracted by linear fitting: ~0.37 dB/μm for the Cu-SiO₂-Si₃N₄-SiO₂-Cu waveguide and ~0.78 dB/μm for the Cu-Si₃N₄-Cu waveguide. The waveguides were also measured using other laser sources. Unlike the Si₃N₄ waveguides which keep low propagation loss over a wide wavelength range, the plasmonic waveguides exhibit larger propagation loss at shorter wavelengths. For example, the propagation loss of the above Cu-SiO₂-Si₃N₄-SiO₂-Cu waveguide is ~0.88 dB/μm at 1310 nm, ~3.2 dB/μm at 1060 nm, and >5 dB/μm at 853 nm. The larger propagation loss at the shorter wavelength can be attributed to the larger metal loss of Cu at the shorter wavelength [15, 26]. Since low propagation loss is essential for all plasmonic devices, the plasmonic devices reported in the following text were measured using the broadband (1520–1620 nm) laser source only.

 

Fig. 4 (a) Output spectra measured on a set of straight Cu-SiO₂-Si₃N₄-SiO₂-Cu plasmonic waveguides with the same Si₃N₄ core but different L_Ps, normalized by that measured on the reference Si₃N₄ waveguide without the plasmonic area; and (b) Output power (normalized by that of the reference Si₃N₄ waveguide) at 1550 nm versus L_P for Cu-SiO₂-Si₃N₄-SiO₂-Cu and Cu-Si₃N₄-Cu waveguides. Each data point is averaged from 3 identical waveguides and the standard deviation is presented as the error bar.

Download Full Size | PDF

Figures 5(a)-5(f) show electric field (|E_x|), magnetic field (|H_y|), and energy density distributions of the 1550-nm fundamental TE mode in the Cu-SiO₂-Si₃N₄-SiO₂-Cu and Cu-Si₃N₄-Cu waveguides with the Si₃N₄ core as shown in Fig. 2(c), calculated using the EME method. One sees that the optical mode is tightly confined in the core for both waveguides: the ratio of modal power in the core is ~87% for the Cu-SiO₂-Si₃N₄-SiO₂-Cu waveguide and ~90% for the Cu-Si₃N₄-Cu waveguide. The Cu-Si₃N₄-Cu waveguide confines the mode slightly tighter, namely provides slightly smaller mode size, than the corresponding Cu-SiO₂-Si₃N₄-SiO₂-Cu waveguide. A significant difference between these two kinds of waveguides is the energy density distribution: the energy density maximizes at the center of the core in the Cu-Si₃N₄-Cu waveguide while the energy density maximizes in the both sides of the core (i.e., the thin SiO₂ interlayer) in the Cu-SiO₂-Si₃N₄-SiO₂-Cu waveguide because the electric field in the thin SiO₂ interlayer is significantly enhanced due to the continuity of electric displacement normal to the Cu/SiO₂ and SiO₂/Si₃N₄ interfaces. This significant difference in the energy density distribution may lead to different properties of these two kinds of waveguides, as will be discussed below.

 

Fig. 5 (a) Electric field (|E_x|); (b) Magnetic field (|H_y|); and (c) Energy density distributions of the 1550-nm fundamental TE mode in the Cu-SiO₂-Si₃N₄-SiO₂-Cu plasmonic waveguide; (d)-(f) Figures for the corresponding Cu-Si₃N₄-Cu plasmonic waveguide; (g) Propagation loss and (h) Real part of the modal index (n_eff) versus the width and height of the Si₃N₄ core for waveguides having ideal rectangular Si₃N₄ core cross sections; (i) Propagation loss and (h) n_eff versus the bottom width of the Si₃N₄ core for waveguides having real core cross sections, which are trapezoid-shaped with ~326-nm height when width > ~110 nm and are triangle-shaped with a reduced height when width < ~110 nm, as shown schematically in the inset.

Download Full Size | PDF

The plasmonic waveguide properties (both the propagation loss and the real part of the effective modal index, n_eff) depend on the Si₃N₄ core size (both the width and height). As shown in Fig. 2(c), the real Si₃N₄ core is not an ideal rectangle but has a less than 90° sidewall angle due to the imperfect fabrication. When the bottom width of the Si₃N₄ core is smaller than ~110 nm, the top width reduces to zero and the core becomes triangular-shaped whose height decreases accordingly with decreasing the bottom width if assuming that the sidewall angle keeps the same. To distinguish the effects of the width and height of the core on the waveguide properties, Figs. 5(g) and 5(h) plot propagation loss and n_eff for plasmonic waveguides having ideal rectangular Si₃N₄ cores, calculated using the EME method. In one case, the core height (above the 49-nm-thick Si₃N₄ slab) keeps 326 nm while the core width varies from 60 to 160 nm. In the other case, the core width keeps 100 nm while the core height varies from 140 nm to 320 nm. The interlayer SiO₂ thickness keeps 10 nm. Firstly, the propagation loss increases monotonously with decreasing both the width and height, reflecting the fact the smaller mode size results in the larger propagation loss. Secondly, n_eff increases monotonously with decreasing the width, as the usual MIM plasmonic waveguides, while it decreases monotonously with decreasing the height. This is because with the height of the core decreasing, the influence of the Si₃N₄ slab becomes larger and the mode deviates from TE, thus resulting in n_eff increasing. Thirdly, the Cu-SiO₂-Si₃N₄-SiO₂-Cu waveguides have lower propagation loss and smaller n_eff than the corresponding Cu-Si₃N₄-Cu waveguides evening scaled to the same total size of the core, which can be attributed to the different optical energy distribution induced by the step index modulation in the core.

Figure 5(i) shows propagation loss for both Cu-SiO₂-Si₃N₄-SiO₂-Cu and Cu-Si₃N₄-Cu plasmonic waveguides having the real Si₃N₄ core cross sections. The experimental data (represented by the symbols) are measured using the cutback method and the theoretical data (represented by the solid curves) are calculated using the EME method. The experimental propagation losses of the Cu-SiO₂-Si₃N₄-SiO₂-Cu waveguides agree well with the theoretical values, while the Cu-Si₃N₄-Cu waveguides exhibit larger experimental propagation losses than the theoretical values. One reason may be the dry-etching induced sidewall roughness of the Si₃N₄ core. The roughness induced scattering loss may be suppressed by the thin SiO₂ layer deposited above the Si₃N₄ core due to the step index modification, thus resulting in its minor influence in the Cu-SiO₂-Si₃N₄-SiO₂-Cu waveguides. The other possible reason may be the different Cu filling/adhesion conditions on SiO₂ and Si₃N₄ surfaces [12, 20], which may lead to different Cu film quality in these two kinds of waveguides. It is expected that the propagation loss of the Cu-Si₃N₄-Cu waveguide may be further improved (i.e., approaches to the theoretical value) after fabrication optimization. Figure 5(j) plots n_eff versus the bottom width of the Si₃N₄ core for two kinds of waveguides, it increases with increasing the width first, reaches a maximum at the width of ~110 nm, and then decreases with further increasing the width. This is because in the range of the width less than ~110 nm the core is triangle-shaped and the waveguide property is dominated by the core height, while in the range of the width larger than ~110 nm, the core is trapezoid-shaped with height of ~326 nm and the waveguide property is dominated by the core width. Moreover, one can see that in the range of the width larger than ~110 nm, the waveguides having the real Si₃N₄ core exhibit similar properties as those having the ideal rectangular Si₃N₄ core if the width of the real trapezoid core is scaled to the effective width, which is approximately the middle width of the trapezoid core. It indicates that the angled sidewall does not affect the waveguide’s properties significantly if the core height is not reduced (i.e., the bottom width of the real Si₃N₄ core is large enough, ~110 nm here). Therefore, the Si₃N₄ core is approximated to a 351-nm (height) × 100-nm (width) rectangle in the following simulations for simplification.

3.3 Taper couplers

As the abovementioned, an effective coupling between the high-confinement plasmonic waveguide and the low-loss dielectric waveguide is essential for the photonic and plasmonic integrated circuits. Here, a linear taper coupler as shown schematically in Fig. 2(b) is used. The feasibility of such a liner taper is first investigated using the 3D finite-difference time-domain (FDTD) simulation. Figures 6(a) and (b) depicts the top and cross sectional view of the absolute value of Poynting vector along the in- and out-couplers, showing that the 1550-nm fundamental TE mode in the 1-μm-wide Si₃N₄ waveguide is effectively coupled in the Cu-SiO₂-Si₃N₄-SiO₂-Cu plasmonic waveguide through a 2-μm-long taper coupler, and effectively coupled out to the Si₃N₄ waveguide through another 2-μm-long taper coupler. For the Cu-SiO₂-Si₃N₄-SiO₂-Cu waveguide, the theoretical coupling loss is estimated to ~1.9 dB for in-coupling and ~1.3 dB for out-coupling by comparing powers monitored before and after couplers (i.e., position-a versus position-b, and position-c versus position-d, as indicated in Fig. 6(a), respectively). For the Cu-Si₃N₄-Cu waveguide, the theoretical coupling loss is ~2.5 dB for in-coupling and ~2.0 dB for out-coupling.

 

Fig. 6 (a) Top view; and (b) Cross sectional view of the absolute value of Poynting vector of the 1550-nm TE mode in- and out-coupling between the 1-μm-wide Si₃N₄ waveguide and the Cu-SiO₂-Si₃N₄-SiO₂-Cu plasmonic waveguide through 2-μm-long taper couplers; (c) Output spectra measured on Cu-SiO₂-Si₃N₄-SiO₂-Cu plasmonic waveguides with L_P = 15 nm and L_C ranging from 0 to 5 μm, normalized by that measured on the reference Si₃N₄ waveguide; (d) The measured coupling loss at 1550 nm versus L_C for Cu-SiO₂-Si₃N₄-SiO₂-Cu and Cu-Si₃N₄-Cu waveguides. Each data point is averaged from 4 identical waveguides and the standard deviation is presented as the error bar.

Download Full Size | PDF

Experimentally, the sum of the in- and out-coupling losses can be extracted from the y-intercept of the linear fitting lines in Fig. 4(b), but they cannot be distinguished each other. The average coupling loss (of the in-coupling and out-coupling losses) extracted from Fig. 4(b) is ~3.2 dB/facet for the Cu-SiO₂-Si₃N₄-SiO₂-Cu waveguide and ~4.2 dB/facet for the Cu-Si₃N₄-Cu waveguide. Plasmonic waveguides with different coupler length (L_C) (with the same L_P of 15 μm) were fabricated. Figure 6(c) plots output spectra measured on Cu-SiO₂-Si₃N₄-SiO₂-Cu plasmonic waveguides with L_C of 0, 0.3, 1, and 5 μm, respectively, normalized by that of the reference Si₃N₄ waveguide without the plasmonic area. The spectra depend on wavelength weakly except the small ripples. Figure 6(d) plots the coupling loss versus L_C for Cu-SiO₂-Si₃N₄-SiO₂-Cu and Cu-Si₃N₄-Cu waveguides. For both kinds of waveguides, the coupling loss initially decreases rapidly with increasing L_C, reaches a minimum at L_C of 1-2 μm, and then increases slowly with L_C further increasing. This is because the coupling loss includes the tapering loss (due to reflection and radiation) and the propagation loss through the taper coupler [8], the tapering loss decreases while the propagation loss increases with increasing L_C. Figure 6(d) shows that the optimal taper length is ~1-2 μm for both kinds of waveguides and the minimum coupling loss is ~2.8 dB/facet (i.e., ~52% coupling efficiency) for the Cu-SiO₂-Si₃N₄-SiO₂-Cu waveguides and ~4.2 dB/facet (i.e., ~38% coupling efficiency) for the Cu-Si₃N₄-Cu waveguides. The Cu-SiO₂-Si₃N₄-SiO₂-Cu waveguides provide higher coupling efficiency than the Cu-Si₃N₄-Cu counterparts, consistent with the theoretical prediction. However, the experimental coupling losses for both kinds of waveguides are larger than the theoretical values, which may be attributed to the different core cross section between the real devices and the simulations as well as the other imperfection fabrication.

The Si₃N₄-based plasmonic waveguides in this work exhibit relatively large coupling losses (both experimental and theoretical values), which may be attributed to the relatively small mode overlap between the Si₃N₄ waveguide and the plasmonic waveguide due to the loose mode confinement in the Si₃N₄ waveguide. Figures 6(a) and 6(b) show that the portion of optical power contained in the Si₃N₄ core (~60%) will be mostly tapered into the Cu-covered coupler while that contained in the SiO₂ cladding (~40%) will be mostly scattered and/or reflected. This relatively large scattering and/or reflection also cause the small ripples observed in the transmission spectra. For comparison, a Si channel waveguide can confine the mode mostly in the Si core, thus the scattering and/or reflection between the Si waveguide and the Cu-covered plasmonic area is smaller than that in the case of Si₃N₄ waveguides. As a result, a similar taper coupler between the Si channel waveguide and the horizontal Cu-SiO₂-Si-SiO₂-Cu plasmonic waveguide can provide coupling loss as low as ~0.1 dB/facet and their transmission spectra exhibit much smaller ripples [8, 9, 28]. Compared with the Cu-Si₃N₄-Cu counterparts, the Cu-SiO₂-Si₃N₄-SiO₂-Cu waveguides have a slightly looser mode confinement, thus they exhibit smaller coupling loss. Moreover, n_eff mismatch between the dielectric waveguide and the plasmonic waveguide is another source of the tapering loss. The Cu-SiO₂-Si₃N₄-SiO₂-Cu waveguides have n_eff more close to the Si₃N₄ waveguides than the Cu-Si₃N₄-Cu waveguides, as observed in Fig. 5(j), which also contributes their smaller coupling loss as compared with the Cu-Si₃N₄-Cu counterparts.

3.4 Ultracompact 90° bends

It has been theoretically verified that the high confinement MIM plasmonic waveguide allows sharp bending [18]. In this work, 90° bends with radii (R) of 0 and 0.5 μm were fabricated, as shown in Figs. 7(a) and 7(d), respectively. The total length of plasmonic route (i.e., from point-A to point-B, as indicated in the figures) is 3 μm and 3.8 μm, respectively. The coupler length keeps 2 μm. The top views of the absolute value of Poynting vector in the Cu-SiO₂-Si₃N₄-SiO₂-Cu and Cu-Si₃N₄-Cu bends with R = 0 and 0.5 μm are depicted in Figs. 7(b), 7(c), 7(e), and 7(f), respectively, obtained from the 3D FDTD simulations. The theoretical bending loss is estimated by comparing powers monitored 0.1-μm before and after the bend (i.e., point-a and point-b, as indicated in Figs. 7(b) and 7(e)). For Cu-SiO₂-Si₃N₄-SiO₂-Cu, the bending loss is 0.59 dB for R = 0 and is 0.68 dB for R = 0.5 μm. For Cu-Si₃N₄-Cu, the bending loss is 0.48 dB for R = 0 and 0.56 dB for R = 0.5 μm. The apparently larger bending loss for the 0.5-μm-R bends can be simply attributed to the propagation loss through the ~0.8-μm-long arc. After subtracting the propagation loss, the bending loss of the 0.5-μm-R bend is 0.30 dB for Cu-SiO₂-Si₃N₄-SiO₂-Cu and 0.12 dB for Cu-Si₃N₄-Cu. The Cu-Si₃N₄-Cu bends provide a slightly smaller theoretical bending loss than the Cu-SiO₂-Si₃N₄-SiO₂-Cu counterparts because of their tighter mode confinement. Experimentally, the bending loss is extracted by comparing the output spectrum measured on the bend with that measured on the corresponding 3-μm-long straight plasmonic waveguide. The measurement results are plotted in Fig. 7(g) for Cu-SiO₂-Si₃N₄-SiO₂-Cu and in Fig. 7(h) for Cu-Si₃N₄-Cu. One sees that the spectra are almost wavelength-independent except the small ripples. For Cu-Si₃N₄-Cu, the average bending loss is ~0.2 dB for R ~0 and ~0.7 dB for R = 0.5 μm, in good agreement with the theoretical results after taking the measurement error into account. On the other hand, the Cu-SiO₂-Si₃N₄-SiO₂-Cu bends exhibit a relatively large bending loss. The bending loss is ~3.2 dB for R ~0 and is ~1.9 dB for R = 0.5 μm. The larger experimental bending loss for Cu-SiO₂-Si₃N₄-SiO₂-Cu may be attributed to the possible imperfect fabrication, for example, the PECVD deposited SiO₂ interlayer may be thicker in the bend corner than in the other area.

 

Fig. 7 (a) SEM image of Si₃N₄ core of a 90° bend with R ~0; (b)-(c) The absolute value of Poynting vector in Cu-SiO₂-Si₃N₄-SiO₂-Cu and Cu-Si₃N₄-Cu 90° bends with R = 0, obtained from the 3D FDTD simulation; (d)-(f) Corresponding figures for 90° bends with R = 0.5 μm; (g) Experimental bending loss spectra measured on Cu-SiO₂-Si₃N₄-SiO₂-Cu bends after subtracting that measured on the corresponding 3-μm-long straight pasmonic waveguide; and (h) Experimental spectra for Cu-Si₃N₄-Cu bends.

Download Full Size | PDF

3.5 Ultracompact 1 × 2 and 1 × 4 power splitters

It has been theoretically verified that the MIM plasmonic waveguide supports ultracompact power splitters [18]. In this work, ultracompact 1 × 2 splitters were fabricated, as shown in Fig. 8(a). The opening angle is 90° and the length of each plasmonic route (i.e., from point-A to point-B or to point-B’, as indicated in Fig. 8(a)) is 3 μm. The simulation results are shown in Figs. 8(b) and 8(c) for Cu-SiO₂-Si₃N₄-SiO₂-Cu and Cu-Si₃N₄-Cu, respectively. The excess loss is estimated by comparing powers monitored 0.1-μm before and after the junction (i.e., point-a, -b, and -b’, as indicated in Fig. 8(b)). Due to the symmetrical structure, the power at point-b is the same as the power at point-b’. Normalized by the power at point-a, the power at point-b is −3.48 dB for Cu-SiO₂-Si₃N₄-SiO₂-Cu and is −3.69 dB for Cu-Si₃N₄-Cu. Subtracting the propagation loss over the corresponding 1.2-μm-long plasmonic waveguide, the excess loss is near zero for both kinds of splitters. Experimentally, each output port of these splitters is measured and compared with that of the corresponding 3-μm-long straight plasmonic waveguide. The measurement results are plotted in Figs. 8(d) and 8(e) for Cu-SiO₂-Si₃N₄-SiO₂-Cu and Cu-Si₃N₄-Cu, respectively. The spectra are almost wavelength-independent except the small ripples. The input light is split almost equally to each output port, as the simulation prediction. The average excess loss is ~0.1 dB for Cu-Si₃N₄-Cu, agrees very well with the theoretical value. For Cu-SiO₂-Si₃N₄-SiO₂-Cu, the average excess loss is ~2.1 dB, larger than the theoretical value. The larger experimental excess loss for Cu-SiO₂-Si₃N₄-SiO₂-Cu may be attributed to the same reason for their larger experimental bending loss, as described above.

 

Fig. 8 (a) SEM image of the Si₃N₄ core of 1 × 2 splitter; (b)-(c) The absolute value of Poynting vector in Cu-SiO₂-Si₃N₄-SiO₂-Cu and Cu-Si₃N₄-Cu splitters, obtained from the 3D FDTD simulation; (d) Spectra measured on two output ports of the Cu-SiO₂-Si₃N₄-SiO₂-Cu 1 × 2 splitter, normalized by that measured on the corresponding 3-μm-long straight plasmonic waveguide, and (e) Experimental spectra for the Cu-Si₃N₄-Cu 1 × 2 splitter.

Download Full Size | PDF

Ultracompact 1 × 4 splitters were also fabricated, as shown in Fig. 9(a). The opening angle is 90° and the length of each plasmonic route is 3 μm. The simulation results are shown in Figs. 9(b) and 9(c) for Cu-SiO₂-Si₃N₄-SiO₂-Cu and Cu-Si₃N₄-Cu, respectively. Powers at 0.1-μm before the junctions (i.e., point-a, -b, -b’, -c, and -c’, as indicated in Fig. 9(b)) are monitored. Due to the symmetrical structure, point-b has the same power as point-b’, and point-c has the same power as point-c’. Normalized by the power at point-a, the power at point-b is −6.25 dB and that at point-c is −6.04 dB for Cu-SiO₂-Si₃N₄-SiO₂-Cu, thus, this splitter has near zero average excess loss and 0.21-dB non-uniformity. For Cu-Si₃N₄-Cu, the power at point-b is −5.66 dB and that at point-c is −6.61 dB, normalized by the power at point-a, thus this splitter also has near zero average excess loss and slightly larger non-uniformity of 0.95-dB. Experimentally, each output port of these splitters is measured and compared with that of the corresponding 3-μm-long straight plasmonic waveguide. The measurement results are plotted in Figs. 9(d) and 9(e) for Cu-SiO₂-Si₃N₄-SiO₂-Cu and Cu-Si₃N₄-Cu, respectively. For Cu-Si₃N₄-Cu, the average output power delivered to the out-1 and out-4 ports is ~-7.2 dB and that delivered to the out-2 and out-3 ports is ~-6.2 dB, which corresponds to an average excess loss of ~0.7 dB and non-uniformity of ~1.0 dB, roughly agrees with the theoretical results after taking the measurement error into account. For Cu-SiO₂-Si₃N₄-SiO₂-Cu, the average output power delivered to the out-1 and out-4 ports is ~-7.7 dB and that delivered to the out-2 and out-3 ports is ~-6.2 dB, which corresponds to an average excess loss of ~1.0 dB and non-uniformity of ~1.5 dB, larger than the theoretical prediction. Again, these large experimental values for Cu-SiO₂-Si₃N₄-SiO₂-Cu may be attributed to the same reason described above.

 

Fig. 9 (a) SEM image of the Si₃N₄ core of 1 × 4 splitter; (b)-(c) The absolute value of Poynting vector in Cu-SiO₂-Si₃N₄-SiO₂-Cu and Cu-Si₃N₄-Cu splitters, obtained from the 3D FDTD simulation; (d) Spectra measured on four output ports of the Cu-SiO₂-Si₃N₄-SiO₂-Cu 1 × 4 splitter, normalized by that measured on the corresponding 3-μm-long straight plasmonic waveguide, and (e) Experimental spectra for the Cu-Si₃N₄-Cu 1 × 4 splitter.

Download Full Size | PDF

3.6 Plasmonic waveguide ring resonators

Ultracompact MIM plasmonic waveguide ring resonators (WRRs) have been theoretically investigated [19]. For evanescently coupling between the bus and ring MIM plasmonic waveguides, the gap between them should be smaller than ~26 nm because the penetration depth of optical filed in the metal is around 26 nm. Such a small gap cannot be defined using the conventional UV lithography, but requires an expensive process such as electron beam lithography or focused ion beam lithography. To circumvent this problem, an aperture coupler has been proposed theoretically for MIM WRRs [29] and has been demonstrated experimentally for horizontal Cu-SiO₂-Si-SiO₂-Cu plasmonic WRRs [30]. In this work, ultracompact WRRs based on the Cu-SiO₂-Si₃N₄-SiO₂-Cu waveguide are fabricated, as shown in Fig. 10(a). The bus waveguide has L_P = 7 μm and L_C = 2 μm, and has the same cross sectional structure as the ring waveguide. The outer radius of the ring is 1 μm. The final dimension of the aperture coupler is determined by the nominal gap size in the mask (which ranges from 0 to 0.2 μm) as well as the exposure dose during UV lithography [30], which is very sensitive to the performance of the plasmonic WRRs. WRRs with optimal aperture size exhibit typical resonant properties.

 

Fig. 10 (a) SEM image of the Si₃N₄ core of a plasmonic waveguide ring resonator with radius of 1 μm; (b) Theoretical (obtained from 3D FDTD simulation) and experimental transmission spectra of a plasmonic WRR, normalized by that of the corresponding 7-μm-long straight plasmonic waveguide; (c) The absolute value of Poynting vector distribution in the off-state; and (d) The absolute value of Poynting vector distribution in the on-state.

Download Full Size | PDF

Figure 10(b) plots (the solid curve) a transmission spectrum in the range of 1520-1620 nm measured on a plasmonic WRR with normal gap of 0.15 μm, normalized by that measured on a corresponding 7-μm-long straight plasmonic waveguide. It exhibits one resonance at λ_r = 1564 nm with extinction ratio of ~18 dB, insertion loss of ~2 dB, and quality factor is ~84. The experimental spectrum agrees well with the theoretical spectrum obtained from the 3D FDTD simulation, which is represented by the dash curve in Fig. 9(b). The absolute value of Poynting vector distributions in the ring at the off-resonance state and on-resonance state are predicted in Figs. 10(c) and 10(d), respectively. One sees that light goes through the bus waveguide at the off-resonance state and is trapped in the ring at the on-resonance state, similar to the conventional WRRs. The attached movie (Media 1) shows the dynamic 1550-nm TE light propagating through this plasmonic WRR. The Cu-Si₃N₄-Cu WRRs also exhibit resonance property but with smaller extinction ratio and smaller quality factor (not shown here) due to its relatively large propagation loss as compared with the Cu-SiO₂-Si₃N₄-SiO₂-Cu counterparts.

4. Conclusion

Si₃N₄-based plasmonic waveguides and various passive devices including couplers, bends, power splitters, and ring resonators have been experimentally demonstrated for seamless integration in the Si₃N₄-based photonic circuits using Si-CMOS back0end-of-line processes. A thin SiO₂ interlayer between the Si₃N₄ core and the Cu cover can modify the devices properties significantly. Relatively low propagation loss of ~0.37 dB/μm and high coupling efficiency of ~52% have been measured on horizontal Cu-SiO₂-Si₃N₄-SiO₂-Cu waveguides and ultracompact ring resonators with 1-μm radius have been demonstrated with extinction ratio of ~18 dB and quality factor of ~84 around 1550-nm telecom wavelengths. On the other hand, the horizontal Cu-Si₃N₄-Cu waveguides supports sharp bending and splitting with near-zero loss. The devices reported in this work open the door to integrate dense photonic and plasmonic integrated circuits above the processed microelectronics.