Hollow waveguides with low intrinsic photoluminescence fabricated with Ta₂O₅ and SiO₂ films

Abstract

A type of integrated hollow core waveguide with low intrinsic photoluminescence fabricated with Ta₂O₅ and SiO₂ films is demonstrated. Hollow core waveguides made with a combination of plasma-enhanced chemical vapor deposition SiO₂ and sputtered Ta₂O₅ provide a nearly optimal structure for optofluidic biofluorescence measurements with low optical loss, high fabrication yield, and low background photoluminescence. Compared to earlier structures made using Si₃N₄, the photoluminescence background of Ta₂O₅ based hollow core waveguides is decreased by a factor of 10 and the signal-to-noise ratio for fluorescent nanobead detection is improved by a factor of 12.

Hollow core, integrated waveguides have gained attention for optical sensing applications in liquid media based on fluorescence and Raman detection.^{1, 2} Unlike conventional waveguides, many hollow waveguides do not rely on total internal reflection. Therefore, light can be guided in very low refractive index materials using an interference based method. Bragg fiber,³ photonic crystal fibers,⁴ omniguides,⁵ and Fresnel reflecting hollow channels⁶ qualify as low-loss hollow core waveguides, but their structures are not easily integrated on a planar chip. Antiresonant reflecting optical waveguides (ARROWs),⁷ based on multilayer optical interference have been introduced as a natural match to a planar substrate. ARROWs do not require periodicity to reach low loss propagation and can be integrated on silicon wafers using common semiconductor microfabrication processes.

Hollow core ARROWs have been demonstrated using either wafer bonding or sacrificial etching approaches. For the former, hollow core channels were fabricated by first dry etching deep trenches into a silicon wafer.^{8, 9} Dielectric layers, serving as claddings, were deposited on the etched wafer as well as a flat wafer and the two were then bonded. For a fully self-contained sensing platform as shown in Fig. 1a, sacrificial etching techniques are preferable because this method provides great flexibility for integration of hollow core ARROWs with solid core waveguides, fluid reservoirs, and electrical elements. A fabrication process combining thin-film deposition and sacrificial etching also provides a good way to create ARROWs small enough for efficient fluorescence and Raman spectroscopy,¹⁰ down to single particles.¹¹

Figure 1.

(a) ARROW based sensing platform. (b) SEM pictures of a hollow core waveguide fabricated with Ta₂O₅∕SiO₂. (c) SEM picture of a hollow core waveguide fabricated with Ta₂O₅∕SiO₂ on a self-aligned pedestal.

In order to build an optical sensing platform with high sensitivity, the signal-to-noise ratio (SNR) must be as high as possible. A significant source of noise when measuring fluorescence comes from the cladding materials of the waveguides. They can produce photoluminescence (PL) in the same wavelength range as the fluorescence or Raman signal generated by an analyte of interest located inside the waveguide. Because the background caused by the native PL of materials cannot be filtered out, subsequent SNR decreases.

Silicon dioxide (SiO₂) and silicon nitride (SiN) have been used to fabricate ARROWs in previous structures.^{1, 2} However, silicon nitride has relatively high native PL at wavelengths of interest for fluorescence sensing. In this Letter, tantalum oxide (Ta₂O₅) is introduced as a replacement for SiN as a cladding material. Ta₂O₅ films are desirable because they have a refractive index (∼2.30 for sputtered film) comparable with SiN, high melting temperatures, low absorption in the visible wavelength, good adhesion to SiO₂, and high resistance to acids.¹² The design, fabrication, and optical measurements of a Ta₂O₅ and SiO₂ ARROW are presented in this letter, with the goal of producing a waveguide with lower PL while maintaining low optical loss.

The fabrication process for hollow core Ta₂O₅∕SiO₂ ARROWs is similar to the fabrication process used for SiN∕SiO₂ ARROWs.² To begin with, alternating SiO₂ and Ta₂O₅ layers are coated by plasma-enhanced chemical vapor deposition (PECVD) and sputtering respectively. Ta₂O₅ can also be deposited by PECVD,¹³ however, the process is more complicated than that used for silicon based films (i.e., SiN and SiO₂). SU-8 is then applied as a sacrificial material that forms a rectangular hollow core over the bottom layers. Six alternating dielectric layers are then deposited over the core, including a thick top oxide layer which is grown over the other five top layers to ensure mechanical strength. Substrate temperatures for both sputtering and PECVD are kept around 250 °C to ensure films are dense enough to survive an acid etching process for the removal of the SU8 core, which consisted of a piranha etch solution of sulfuric acid and hydrogen peroxide.¹⁴ Figure 1b shows a scanning electron microscope picture of a completed hollow ARROW with Ta₂O₅∕SiO₂ films with a core 12 μm wide and 5.0 μm high. Sputtered Ta₂O₅ and PECVD SiO₂ can also be used to make a structure as shown in Fig. 1c by depositing only one thick SiO₂ layer on the top of the SU8 core after etching into the silicon substrate to form a self aligned pedestal.¹⁵

To build an on-chip ARROW platform shown in Fig. 1a, solid core waveguides are created on the same substrate to propagate optical signals on and off the chip and into and out of the hollow waveguides. The structure consists of the layers employed to create hollow core ARROWs, including the thick SiO₂ layer used to increase the structural strength of the hollow waveguides. Ridges are then formed by plasma etching this topmost SiO₂ layer. Light is guided in this layer, confined by ARROW layers beneath and air on top and on the sides (the air-SiO₂ interfaces provide total internal reflection).

The optical loss of Ta₂O₅∕SiO₂ solid and liquid-core ARROWs was compared to SiN∕SiO₂ ARROWs. Solid core waveguides for both types of ARROWs were 12 μm wide and formed by etching a ridge 2 μm deep into the topmost SiO₂ cladding layer. The dielectric layers were (starting from the substrate—in nm): SiO₂∕Ta₂O₅∕SiO₂∕Ta₂O₅∕SiO₂∕Ta₂O₅∕SiO₂ (281∕93∕281∕93∕281∕93-5000). Using the cutback method, a loss coefficient of 0.9±0.1 cm⁻¹ was measured for Ta₂O₅∕SiO₂ ARROWs and 0.7±0.1 cm⁻¹ for SiN∕SiO₂ ARROWs. For liquid-core ARROWs, the loss coefficients were measured by the scattering loss imaging method^{16, 17} when the cores were filled with water. The core dimensions for both the Ta₂O₅∕SiO₂ and SiN∕SiO₂ hollow core waveguides were 12×5 μm². The dielectric layers for hollow core ARROW platform were: SiO₂∕Ta₂O₅∕SiO₂∕Ta₂O₅∕SiO₂∕Ta₂O₅ -core-Ta₂O₅∕SiO₂∕Ta₂O₅∕SiO₂∕Ta₂O₅∕SiO₂ (281∕93∕281∕93∕281∕93-5000-158∕281∕158∕281∕158∕5000). The hollow core loss was 2.7±0.2 cm⁻¹ for SiN∕SiO₂ samples and 3.1±0.2 cm⁻¹ for Ta₂O₅∕SiO₂ samples. These measurements clearly show that both types of waveguides have comparable losses and replacing the SiN layer with Ta₂O₅ does not add significantly to optical loss.

As mentioned above, it is preferable to use materials with low PL to fabricate waveguides in order to increase the sensitivity of fluorescence detection for ARROW platforms. Typically, the relevant PL range for cladding materials is from 660 to 690 nm for bimolecular detection on ARROWs with a pump beam of 633 nm. PL at other wavelengths can be removed with a band pass filter. To compare the PL from SiN based ARROWs with that from Ta₂O₅ based ARROWs, test films of 150 nm thickness were grown by sputtering and PECVD and measured by a spectrometer (Jobin, Yvon: LabRAM HR, λ_ex=633 nm, P∼5 mW, 5 s integration time). Since alternating PECVD and sputtered films are coated in the ARROW fabrication process, the PL of the sputtered Ta₂O₅ film was measured before and after being annealed in the PECVD vacuum chamber at 250 °C. Figure 2 shows that an as-grown SiN PECVD film has higher PL than a PECVD SiO₂ film or a Ta₂O₅ sputtered film. In particular, the PL intensity of Ta₂O₅ film drops substantially after annealing at 250 °C. This is fortunate for our end goal of producing a low PL waveguide since dense SiO₂ PECVD films must be grown over the Ta₂O₅ films at 250 °C. It has been suggested that the annealing in an oxygen-free environment helps to reduce the density of midgap trap states in Ta₂O₅ films which are responsible for the PL.¹⁸

Figure 2.

PL of PECVD SiN and SiO₂ films (T=250 °C) and sputtered Ta₂O₅ (T=250 °C), including as deposited films and after annealing in a vacuum chamber at 250° (excitation wavelength 633 nm).

In order to explore and compare the detection sensitivity between SiN∕SiO₂ ARROW chips and Ta₂O₅∕SiO₂ ARROW chips, a previously described detection setup was employed¹¹ (Fig. 3). A 633 nm pump light of 3.3 mW emitted from a HeNe laser was coupled into the solid-core waveguides on an ARROW platform [Fig. 1a]. The fluorescence signals generated at the excitation volume were collected via an objective lens (Olympus, NA=0.85). Spurious excitation photons were removed by collecting the fluorescence perpendicularly to the excitation direction and by optical filters (Semrock, 640 nm long pass and 670±17 nm bandpass) before coupling into an avalanche photodiode detector (APD, Perkin Elmer, SPCM-AQR-14-FC). The data were finally analyzed by using a time-correlated single photon counting card and software (Picoquant, Timeharp 200).

Figure 3.

Optical fluorescence setup. OFC: optofluidic chip; F: filter; PC: polarization controller; MMF: multimode fiber; SPAPD: single-photon avalanche photodiode; TCSPC: time-correlated single-photon counting card; V: applied voltage between fluid reservoirs.

The PL intensity as a function of time from water filled (18MΩ-cm) SiN∕SiO₂ and Ta₂O₅∕SiO₂ ARROW chips are measured and compared in Fig. 4a. The liquid core channels were filled with water and the detector signal was recorded with laser off (t<10 s) and laser on (10 s<t<40 s). The experimental results show that the background noise baseline above the detector dark counts of liquid core ARROWs is reduced by a factor of 10 when SiN films are replaced by Ta₂O₅ films. Note that the total optical throughput from chip edge to edge (through both hollow and solid core waveguides) was approximately 10% for both types of ARROW chips.

Figure 4.

(a) Comparison of fluorescence background signals between SiN∕SiO₂ samples and Ta₂O₅∕SiO₂ samples. (b) Fluorescence signals from excited Tetraspeck nanoparticles in the excitation volume on SiN∕SiO₂ ARROWs. (c) Fluorescence signals from excited Tetraspeck nanoparcles in the excitation volume on Ta₂O₅∕SiO₂ ARROWs.

To characterize the SNR (detected signal divided by the standard deviation of the noise) of the devices, fluorescent nanoparticles (100 nm diameter, Tetraspeck, Invitrogen) were introduced into the chip via the reservoir [Fig. 1a, 10 μl]. The particle concentration (1.8×10¹⁰ particles∕ml) was chosen such that only one particle was detected in the device excitation volume (∼85 fl using 1∕e² beam diameters). The particles traveled down the channel using pressure driven flow and were optically excited at the orthogonal hollow core∕solid core interface. The particle fluorescence was then collected by the APD. The detected signals for SiN∕SiO₂ and Ta₂O₅∕SiO₂ devices are shown on Figs. 4b, 4c (over a 5 s span), respectively, where each spike corresponded to a detected nanoparticle. The device baseline and noise was clearly higher in Fig. 4b than in Fig. 4c. Over a 40 s timeframe, the number of detected particles for the Ta₂O₅∕SiO₂ sample was 948, with an average SNR of 126.7. The number of detected particles for the SiN∕SiO₂ ARROW sample was 53, with an average SNR of 10.3. In other words, the number of detected particles increased by a factor of 18 and the SNR improved by a factor of 12 by replacing SiN with Ta₂O₅ films.

In conclusion, Ta₂O₅ films were investigated as a replacement for SiN films for the construction of hollow core and solid core ARROWs. Ta₂O₅ based ARROWs were shown to have comparable optical losses to their SiN counterparts. Fluorescence measurements also show that the background noise between 660 and 690 nm has been reduced by a factor of 10 by replacing SiN with Ta₂O₅ films. The SNR of Ta₂O₅∕SiO₂ ARROWs was around 12 times higher than that of SiN∕SiO₂ ARROWs. This improved material system allows for more sensitive measurements, including pushing detection limits down to the level of single fluorophores.