Open Access
Add to BibSonomyAbstract
We describe the integration of optically pumped silicon nanocrystals (Si-ncs) embedded in SiO2 with low loss silicon nitride slab waveguides. An emission waveguide containing Si-ncs with a broad band emission centered at 850 nm, together with a low loss transmission silicon nitride waveguide forms a two section device. The waveguides are fabricated via the deposition of SiOx and silicon nitride using ECR-PECVD. Incorporation of hydrogen through annealing, while beneficial to emission from the Si-ncs, is found to increase material absorption in silicon nitride. This is reconciled by annealing at low temperature. This work shows clearly the potential for this material system as a means for the integration of optical emission and waveguiding using a wholly VLSI compatible processing technology. We further suggest that immediate applications exist in particular in the field of evanescent sensing.
©2007 Optical Society of America
1. Introduction
Si photonics attempts to merge optoelectronics and electronics onto a Si platform using one seamless process flow. Such integrated systems would thus enable the fabrication of high volume, low cost photonic chips for a variety of applications such as telecommunications, optical interconnects, and biological and chemical sensors [1, 2]. Whereas the development of passive elements such as waveguides, splitters and filters, and active devices such as modulators has found success, an efficient, reliable, electrically pumped, VLSI compatible optical source has yet to emerge. Recent and notable are the developments of Si Raman lasers [3] and hybrid evanescently coupled III-V lasers [4], although these technologies have yet to be proven commercially.
Here we discuss the integration of Si nanocrystals (Si-ncs) with optical waveguides. Si-ncs have shown promising photoluminescence (PL) and electroluminescence (EL) efficiencies, while efforts continue at improving and explaining the emission mechanisms and at obtaining stable EL, as reviewed in [2]. In comparison, there exist few reports of attempts to explore the integration potential of Si-ncs. Previous studies that integrate Si-ncs in waveguides almost exclusively use Si-ncs embedded in SiO2 as a waveguide core, while SiO2 forms the bottom cladding, and SiO2 or air is used as the top cladding. The material is often co-doped with erbium to achieve emission at 1.5 μm, the Si-ncs themselves emitting with a broad spectrum centered at ~850 nm. This waveguide design is ideal both for its ease of fabrication and for the long interaction length which aids the search for stimulated emission, as exampled by [5–7]. To date, optically pumped slab guides have shown edge emission from integrated Si-ncs [6–10] and from Er doped Si-ncs [11], to name a few. Ridge waveguides operating at ~850 nm have been demonstrated only as passive devices [12] - where by passive, we refer to the fact that the Si-ncs are not pumped, and waveguide operation is demonstrated with externally coupled lasers. Edge emission at ~850 nm arising from the Si-ncs has not been reported. However, ridge waveguides operating at 1.5 μm have shown Si-nc related edge emission [13], presumably because of the lower propagation loss in the 1.5 μm band compared to ~850 nm. In [14], a ring resonator employing a silicon nitride core is demonstrated at 1.5 μm but operates only as a passive device and Si-nc related edge emission is not reported. Gardner and Brongersma’s work is of particular relevance to our own study because it does not use the standard Si-nc core design while attempting to integrate Si-ncs. In this report, we show Si-nc emission at ~850 nm in slab structures using a Si-nc/silicon nitride /SiO2 stack similar to [14], as well as a two-sectioned design.
There has been debate recently on observed spectral filtering of PL in slab waveguides containing Si-ncs. For example, Ostatnicky et al [8] showed the difficulties in differentiating between guided and unguided edge emission. The problem arises because the material loss of the Si-nc core is very large compared to the SiO2 cladding. Radiation modes propagating in the low loss cladding do so with considerably less loss than the fundamental mode, hence edge emission may be dominated by radiation modes as opposed to guided modes. This is an important consideration for loss and stimulated emission measurements where it is often assumed that all edge emission originates from guided modes. Other reports, [9, 15] argue that spectral filtering can be described entirely by considering the light propagation as guided modes. In the current work, we do not observe spectral filtering of the edge spectra.
The conventional Si-nc core waveguide has disadvantages for device integration. The low index contrast between core and cladding results in waveguides that require thick films, have wide modes, and have a large modal overlap with the core. As a result, the loss is typically >10 dB/cm [7, 12, 15, 16]. Such waveguides are required to be pumped to or beyond transparency.
Fig. 1. Schematic of a two-sectioned optically pumped Si-nc emitter integrated with a low loss silicon nitride waveguide. Two devices are fabricated. One has an emitting region length and transmitting region length of 7 mm and 26 mm respectively, and is 19 mm wide. The other has an emitting and transmitting region length of 16 mm each, and is 10 mm wide.
Here, we demonstrate the design, fabrication, and characterization of a two-sectioned optically pumped device. An emitting waveguide region containing Si-ncs is integrated with a low loss transmitting waveguide region, shown in Fig. 1. The device is optically pumped from above with a laser or light emitting diode (LED). Such a device may be useful for applications that require cheap, disposable, high volume components without the need for high efficiency or high optical power. Indeed, thin silicon nitride waveguides where the mode extends deep into the upper cladding are ideal for evanescent chemical and bio-sensors as demonstrated in [17, 18]. Waveguides formed using silicon nitride tend to exhibit low propagation loss in the visible regime; are VLSI compatible; and have a relatively high refractive index allowing for small waveguide dimensions. Here, we show emission at ~850 nm, however, the wavelength can be tuned over much of the visible and near IR through co-doping with rare earths [19].
The significant disadvantage of off-chip optical pumping is avoided through the use of LEDs. For instance, while pump lasers provide a high intensity and well confined beam, they are difficult to align in free space coupling. Our use of pump LEDs, which provide a lower intensity over a larger area, reduces the alignment requirements in our device design at the possible expense of overall pumping efficiency. LEDs have been considered by others to be a low cost substitute for lasers in pumping Si-nc devices [20, 21].
2. Design and principal of operation
2.1. Design considerations
Silicon nitride is chosen as the waveguide core due to its transparency in the optical region of interest, its high refractive index (~1.9) relative to SiO2 which permits small waveguide dimensions, and its VLSI compatibility. Although SOI (silicon-on-insulator) waveguides have a much higher refractive index (~3.7) and would thus allow fabrication to take place using smaller dimensions, silicon strongly absorbs 850 nm light. Our design allows control of the modal overlap (and hence the loss) with the Si-nc layer, and the ability to couple to low loss transmitting regions using a simple fabrication procedure. There is a large parameter space that yields single mode operation while simultaneously keeping the mode tail away from the absorbing Si substrate, and having only a small coupling loss between the two waveguide regions.
The core remains an uninterrupted plane for passive alignment and is placed below the Si-nc layer so as not to act as a diffusion barrier to H2, a known enhancer of light emission efficiency in Si-ncs [22]. The structures are formed via masked deposition as opposed to post-deposition wet etching, for which a high selectivity (~1:30) etchant is not readily available, while dry etching roughens the core and introduces significant excess loss.
The refractive indices of the films were measured using spectroscopic ellipsometry. The films show only a small dispersion in the band of interest. Using the ellipsometry measurements, mode intensity overlap integrals, effective indices, and beam propagation were simulated using code based on [23], as well as RSoft commercial BPM software. The results are shown in Table 1 for a wavelength of 850 nm. The coupling loss between the emission and transmission regions, assuming an infinitely abrupt transition, is simulated to be 2% and 4% for the TE and TM modes, respectively. Here, coupling loss is defined as the loss experienced by the fundamental mode of the emission region as it propagates through the transition point and into the transmission region. This loss is due to the spatial mode mismatch between the two regions. Three chip designs are studied in this paper. The first is shown in Fig. 1. The thickness of the silicon nitride core was selected to be 300 nm. This results in a single mode waveguide with a low coupling loss between the emission and transmission regions. The 100 nm thickness of the Si-nc layer overlaps most of the mode tail, yet is thin enough to be “lifted off” from the transmission region during fabrication. The waveguide properties of the emission and transmission regions at a wavelength of 850 nm are given in rows 1 and 2 of Table 1. In order to isolate the optical properties of the silicon nitride and Si-nc films, two additional waveguides are also designed. The first is a nitride core waveguide which is identical to the device in Fig. 1 but without the Si-nc film. The optical properties of this waveguide are therefore identical to row 2 in Table 1. The second is a Si-nc core waveguide, that employs a 530 nm thick Si-nc core with a 3 μm thick SiO2 bottom cladding. The Si-nc layer thickness is selected in order to achieve a waveguide above cutoff, with a mode tail that does not extend into the Si substrate. The properties of the Si-nc core waveguide are given in the third row.
Table 1. Waveguide Mode Properties
2.2. Fabrication
Si-ncs are fabricated commonly via deposition of silicon rich oxide, followed by high temperature annealing [2]. In the present case, the substrate was CZ Si with 2.4 or 3 μm of thermally grown oxide. Simulations show that a thinner, 500 nm layer would have been sufficient to isolate the optical mode in the waveguide. Silicon nitride and SiOx were deposited using ECR-PECVD (details discussed elsewhere [24, 25]). The refractive indices at 850 nm are typically 1.9 and 1.6 for silicon nitride and SiOx respectively. The refractive index of SiOx films deposited in our chamber has been previously calibrated as a function of Si content using Rutherford backscattering. Here, we interpolate a Si content consistent with x=1.7. None of the annealing processes subsequently described significantly alter the thickness or refractive index of our films. The second SiOx deposition, made at a substrate temperature of 120 °C, is masked using either a glass cover slide, or photoresist. In the latter case the SiOx is selectively removed via a lift-off process. The two methods yield chips with identical losses in the masked region. The samples are cleaved, then annealed using one of three recipes:
- A: 1100 °C in N2 for 2 hrs
- B: 1100 °C in N2: 5% H2 for 2 hrs
- C: 1100 °C in N2, for 2 hrs followed by 500–600 °C in N2: 5% H2 for 2 hrs
In a previous study, identically prepared SiOx films (but with a slightly larger x=1.78 and annealed for 1 hour longer) were shown by glancing X-ray diffraction to contain Si-ncs with a mean diameter of 2 nm [24]. We note that the PL spectra reported in [24] and from this report have the same shape and peak wavelength. It is therefore concluded that the films here contain similarly sized Si-ncs.
2.3. Principle of operation
The samples were optically pumped from above at room temperature by a 405 nm diode laser or Nichia NSPB500S 470 nm LED. The pump light was directed at the emitting section of the waveguide. Spontaneous emission generated in the film couples to (1) surface radiation, (2) substrate radiation modes, and (3) the desired guided modes. The relative amount of light coupled to the guided mode compared to the total emission is the spontaneous emission factor. The guided light propagates along the emitting region, experiencing material loss. In this report we describe slab guiding only, thus the mode propagates in all horizontal directions. At the Si-nc step, a portion of the light is coupled to the mode of the transmitting region where it propagates with much lower absorption-mediated loss.
Optimization of the light power launched into the guided mode depends on the length of the optically pumped region, quantum efficiency of the generation of the visible signal, spontaneous emission factor, material loss in each layer, and the coupling coefficient between the two regions. Optimization is the subject of future work.
3. Results and discussion
3.1. Characterization of material loss
Two methods were used to measure optical loss in the emission band, and are shown in Fig. 2. Obtained results are consistent verifying the validity of each measurement technique.
- In the scattering detection (or streak) method [26], a rutile prism is used to couple an external 850 nm laser beam into the waveguide mode, forming a streak of surface scattered light. A 1-D fiber bundle array is scanned across the streak thus monitoring optical power as a function of propagation distance. In the present measurements the prism coupling angles agree with calculated values of the fundamental TE and TM modes.
- In the scanning excitation spot (SES) method [6], edge emission power is monitored as the pump spot is translated away from the emission facet. The pump laser is a 405 nm diode laser, modulated at 150 Hz with a 50% duty cycle square wave. The peak power is 300 μW, and is focused to a spot with 1/e2 dimensions of 54 μm × 125 μm. Focusing the beam with a cylindrical lens to a line 2 mm × 125 μm, aligned with the smaller dimension of the beam parallel to the scanning direction, was also employed, yielding identical loss values. In both the streak and SES measurements, light is detected with a Si PIN biased detector using lock-in detection.
Fig. 2. The top figures show schematics of the experimental setup for the streak measurement of the surface scattered light for a two-sectioned device, a), and for a silicon nitride core waveguide, b). a) An external 850 nm diode laser is prism coupled to the emission region of the waveguide. The surface scattered light from the mode is detected with a fiber bundle scanned to the right. b) Streak measurement of a silicon nitride core waveguide. The Si-nc core waveguide is also characterized in this configuration. The bottom figures show schematics of the experimental setup for SES measurements of the two-sectioned device taken from the right facet, c), and left facet, d). c) A 405 nm pump laser is scanned to the left across the emission region while edge emission is monitored from the right facet. d) SES measurement of the edge emission from the left facet. Here, the pump laser is scanned to the right starting at the facet. The Si-nc core waveguide is also characterized in this configuration.
The drawback of the streak method is that it is susceptible to relatively high noise levels since signal is dependent on scattering uniformity. In general SES provides a relatively noise-free measurement, however it can have an associated large system loss and uncertainty that depends on collection optics. Careful consideration of the collection geometry is required in SES as pointed out in references [6] and [7]. Our work uses a 25.4 mm focal length, 25.4 mm diameter lens to collect the edge emission. The lens is positioned 3 focal lengths from the facet in a low magnification configuration to avoid the detector acting as an aperture. The limiting aperture is the lens, and the collection efficiency is essentially constant over the short scanning distance. A colored glass filter is used to block pump light, and surface emission is blocked with a knife edge hovering over the facet. Unless stated otherwise, the edge emission is focused directly on the Si detector and is therefore a broad band measurement over the entire emission spectrum.
To separate the optical properties of the two materials, waveguides containing either silicon nitride or Si-nc cores (with upper claddings of air) were prepared on the thermally oxidized substrates. Figure 3 shows results of loss measurements for a SiO2/Si-nc/air waveguide. The Si-nc layer has an index of 1.62 at 850 nm and is 530 nm thick. The streak and SES methods agree to within 5 dB/cm. The small discrepancy is believed to originate in part from noise in the streak measurement, but also from systematic errors in the SES measurement. Although SES in general tends to over-estimate loss due to geometrical optic effects, SES is a broad band measurement while the streak method is monochromatic. As will be shown, there is some wavelength dependence in Si-nc loss. Taking into account the mode shape, loss originating from material absorption is given by
where α is the waveguide loss for the TE or TM mode, αi, Γi, ni are the material loss, mode overlap and refractive index in the ith layer, Neff is the effective index of the waveguide, and the sum is taken over all layers [27]. We calculate from Table 1 and Eq. (1) that the TM loss should be about 0.86 times the TE loss. The SES and streak method measures this ratio to be 0.93 and 0.82 respectively. Assuming the total loss is dominated by absorption in the Si-nc layer, an upper limit for the Si-nc layer’s material loss can be calculated. Eq. (1) relates the measured waveguide loss, α, to the Si-nc film’s material loss, αi=Si-nc. Inserting the streak measurements into Eq (1) yields a Si-nc loss of 38 dB/cm at 850 nm. This result is arrived at by taking the average calculated values of αi=Si-nc for the TE and TM polarizations (39.3 dB/cm and 37.4 dB/cm respectively). The two polarizations should result in the same value for the material loss, but a difference arises due to measurement uncertainty.
Fig. 3. Loss measurements of a SiO2/Si-nc/air waveguide made using the streak method with an 850 nm source (left) and the SES method (right). Both TE and TM polarizations are shown. The SES (streak) method measures linear fits of 29 (34) dB/cm for TE polarization and 27 (28) dB/cm for TM polarization.
Fig. 4. Streak measurements at 850 nm showing loss of identically deposited but differently annealed SiO2/silicon nitride/air waveguides. The silicon nitride layer is 300 nm thick, with a refractive index of 1.9. Only TE coupled results are shown since TM gives similar results. For clarity, 11 and 2 dB/cm slopes are also shown.
The consistency between SES and streak measurements shows that the observed edge emission is dominated by spontaneous emission propagating in the waveguide mode, and not substrate or surface radiation modes as has been observed in studies with similar waveguides made on silica [8]. It is likely that the thin bottom cladding attenuates radiation modes through Fresnel loss. Other studies of slab waveguides having a Si-nc core show a large variety of losses. Work reported in references [7, 15, 16] shows losses in the emission band of 78, 8.7, and 176 dB/cm, respectively. This large range is likely due to actual differences in the optical properties of differently prepared Si-nc films, differences in the modal overlap with the layer, and possibly to measurement interpretation as discussed in [6, 7].
As SES is not possible for SiO2/silicon nitride/air waveguides, only the streak method was used to determine waveguide loss. Figure 4 shows results of measurements, taken at 850 nm, for 4 co-deposited SiO2/silicon nitride/air waveguides annealed under different conditions. Three of the waveguides, including those annealed at 600 °C in N2: 5%H2, are found to have a loss of 2 dB/cm. The waveguide annealed at 1100 °C in N2: 5%H2 is found to have a much higher loss of 11 dB/cm. The error in the least squares fit is 1.6 dB/cm. If all waveguide loss is assumed to arise from absorption and scattering in the silicon nitride layer, then we find an upper limit for the material loss of 2.2 dB/cm. The loss of the TM mode is calculated to be a factor 0.72 lower than that of the TE mode, which is too small to be resolved from the measurement noise. In addition, it is likely that some scattering loss is introduced through interface roughness.
Silicon nitride waveguides on Si have been studied for a number of years [28], but the majority of this work has concentrated on the propagation of light with a wavelength around 1.5μm and the closely related but more versatile SiON material system [29]. In these cases, N-H stretching vibrations are found to be the predominant source of loss, the harmonics extending to the near-IR. The presence of N-H bonds is shown to be tremendously reduced by annealing in ambient N2 gas at high temperatures >1000 °C [30]. There has been very little reported work for propagation of light with a wavelength of ~850 nm in silicon nitride waveguides, although a recent paper targeting VLSI compatible photonics for optical interconnects reports a 0.1 dB/cm loss for channel waveguides [31].
The optical loss associated with the silicon nitride slab waveguides of 2 dB/cm shown here may well be improved by varying our deposition parameters, although it is sufficiently low to be considered suitable for small propagation distances. Hydrogen is likely incorporating into the film, for the case of the high 1100 °C temperature anneal, dramatically increasing the loss to 11 dB/cm. However, there appears to be a processing window as high as at least 600 °C where H2 is not incorporated into the film. This is an important result because annealing in H2 at temperatures between 400–1200 °C is well known to improve the emission efficiency of Si- ncs in SiO2 by termination of parasitic dangling bonds [22], opening to us the process window of 400–600 °C for the production of low-loss silicon nitride waveguides while maximizing the emission of light from the optically pumped Si-ncs.
3.3. Characterization of the two- section device
Figure 5 shows results of SES and streak measurements of the two-sectioned device shown in Fig. 1. SES (total power) measures a loss in the emitting region of 7 dB/cm, while the streak method (TE at 850 nm) result agrees to within 1.2 dB/cm. Using the material loss values for Si-nc and silicon nitride,measured in the previous section, Eq. (1) predicts a waveguide loss of 4.7 and 2.0 dB/cm in the TE mode of the emission and transmission regions, respectively. This is within 1.1 dB/cm in agreement with the streak measurement.
The streak measurement in Fig. 5 (left) confirms coupling between the two regions of the chip, and the SES measurement in Fig. 5 (right) confirms the chip’s operation as an integrated optically pumped emitter. If we assume that the scattering cross-section in the two regions is equal then the coupling loss is smaller than the measurement noise.
Fig. 5. Streak measurement of a two sectioned chip (left). Light is prism coupled to the emitting region on the left of the chip and propagates to the right, as per Fig. 2 a). A dashed line shows the transition in the two section chip. SES measurement of an identically deposited and annealed chip (right). Light is collected from the right facet after propagation through a 26 mm transmission region as the laser is scanned to the left, as per Fig 2 c) (note the x-axis in the two graphs are opposite directions).
In this test device, only a thin, 100 nm Si-nc layer was deposited, and the difference in loss between the two regions is only ~4dB/cm. A thicker Si-nc film would add light generation at the cost of increasing the loss in the emitting region, making the two-sectioned design more useful.
Figure 6 shows edge emission PL from the two-sectioned chip. A lens with an f number of 1.97 directs edge emission to a grating spectrometer with a silicon PIN detector using lock-in detection. For comparison, emission from both facets is shown. A colored glass filter is used to block any pump light which may give 2nd order diffraction signals. A knife edge at the facet blocks surface emission. In Fig. 6 (left), PL is collected from the left facet. Each curve is obtained with the pump light at 3 mm increasing increments from the facet. Surface emission taken at normal incidence is also shown for comparison.
In Fig. 6 (left), strong Fabry-Perot modulation is apparent in the surface emission. The edge emission is featureless and has a shape similar to surface emission from a thin 100nm Si-nc film on Si, where no Fabry-Perot modulation is expected due to the lack of the underlaying silicon-nitride/SiO2 stack. Polarization-resolved spectra (not shown) reveal the same featureless shape for the edge emission. The edge emission is seen to shift to red wavelengths as the excitation spot is moved away from the emission facet. This indicates the presence of a wavelength dependent loss which increases in the blue end of the spectrum. We note that this wavelength-dependent loss is entirely due to the presence of the Si-nc layer. Indeed, PL excited at the right-most point of the emitting region and collected out the right facet (after propagating 16 mm through the Si-nc free region, and shown in the right of Fig. 6) has the same shape and peak as the bluest curve in Fig. 6 (left). In this pumping configuration, a calibrated power meter butt-coupled to the facet measures 2.4 nW of broad band edge emission when pumped with 460 μW, focused to a spot .
The right panel of Fig. 6 shows LED-pumped edge emission. The Nichia LED, model # NSPB500S, is specified to emit light peaked at 470 nm with a spectral width of 30 nm. The far-field beam from the LED has a full-width-at-half maximum angle cone of ~14°. At 100 mA bias, 50 % duty cycle, the peak power emitted by the LED is 15.8 mW, as measured with a large area butt-coupled calibrated power meter. Whereas the laser is directed with mirrors and lenses, the LED is suspended approximately 1 mm from the emitting surface, and demonstrates the ease of alignment. The curve has a slightly different shape owing in part to the use of a different glass filter, and a different pumping intensity.
Fig. 6. Edge and surface emission spectra from a two-sectioned waveguide. (Left) Edge spectra collected out the left facet, in likeness to Fig 2 d). Each curve is from emission excited 3 mm further from the collection facet. The dot-dashed curve shows surface emission for comparison. The dashed curve shows surface emission from a 100 nm thick single layer Si-nc film. All curves are normalized to the spectral peak. (Right) Edge spectra of PL excited at the right most part of the emission region and collected out the right facet after propagation through a 16 mm transmitting region, in likness to Fig 2 c). The thin black curve is from laser excitation, while the thicker blue curve is from LED excitation.
4. Conclusions
We demonstrate an optically pumped Si-nc emitter integrated with low loss silicon nitride waveguides. Edge emission resulting from PL coupled to the waveguide mode is confirmed. Waveguide loss is dominated by material absorption in the Si-nc layer, measured to be 38 dB/cm at 850 nm. Our deposited silicon nitride is found to be a compatible material exhibiting only 2 dB/cm loss, even after annealing in the presence of H2 at a temperature as high as 600 °C. A two-sectioned design realized with a planar process is demonstrated to couple an emission region with a low loss transmission region.
Acknowledgments
The authors would like to thank Dr O.H.Y. Zalloum for his valuable advice with regard to the measurement of photoluminescence and to H. (Charlie) Zhang for his efforts depositing films. This work is sponsored by the Natural Sciences and Engineering Research Council of Canada, the Canadian Institute for Photonic Innovation and the Ontario Photonics Consortium.
References and links
1. G. T Reed and A. P. Knights, Silicon Photonics-An introduction (Wiley, 2004). [CrossRef]
2. D. J. Lockwood and L. Pavesi, “Silicon Fundamentals for Photonics Applications,” Top. Appl. Phys. 94, 1–50 (2004). [CrossRef]
3. H. Rong, R. Jones, A. Liu, O. Cohen, D. Hak, A. Fang, and M. Paniccia, “A Continuous-Wave Raman Silicon Laser,” Nature 433, 725 (2005). [CrossRef] [PubMed]
4. A. W. Fang, R. Jones, H. Park, O. Cohen, O. Raday, M. J. Paniccia, and J. E. Bowers, “Integrated AlGaInAs-silicon evanescent racetrack laser and photodetector,” Opt. Express 15, 2315–2322 (2007). [CrossRef] [PubMed]
5. L. Pavesi, L. Dal Negro, C. Mazzoleni, G. Franzo, and F. Priolo, “Optical gain in silicon nanocrystals,” Nature 408, 440–444 (2000). [CrossRef] [PubMed]
6. J. Valenta, I. Pelant, and J. Linnros, “Waveguiding effects in the measurement of optical gain in a layer of Si nanocrystals,” App. Phys. Lett. 81, 1396–1398 (2002). [CrossRef]
7. L. Dal Negro, P. Bettotti, M. Cazzanelli, D. Pacifici, and L. Pavesi, “Applicability conditions and experimental analysis of the variable stripe length method for gain measurements,” Opt. Commun. 229, 337–348 (2004). [CrossRef]
8. T. Ostatnicky, J. Valenta, I. Pelant, K. Luterova, R. G. Elliman, S. Cheylan, and B. Honerlage, “Photoluminescence from an active planar optical waveguide made of silicon nanocrystals: dominance of leaky substrate modes in dissipative structures,” Opt. Mater. 27, 781–786 (2005). [CrossRef]
9. L. Khriachtchev, D. Navarro-Urrios, L. Pavesi, C. J. Oton, N. E. Capuj, and S. Novikov, “Spectroscopy of silica layers containing Si nanocrystals: Experimental evidence of optical birefringence,” J. Appl. Phys. 101, 044310 (2007). [CrossRef]
10. R. T. Neal, M. D. C. Charlton, G. J. Parker, C. E. Finlayson, M. C. Netti, and J. J. Baumberg, “Ultrabroadband transmission measurements on waveguides of silicon-rich silicon dioxide,” Appl. Phys. Lett. 83, 4598–4600 (2003). [CrossRef]
11. R. G. Elliman, M. Forcales, A. R. Wilkinson, and N. J. Smith, “Waveguiding properties of Er-implanted silicon-rich oxides,” Nucl. Instrum. Methods Phys. Res. B 257, 11–14 (2007). [CrossRef]
12. P. Pellegrino, B. Garrido, C. Garcia, J. Arbiol, J. R. Morante, M. Melchiorri, N. Daldosso, L. Pavesi, E. Scheid, and G. Sarrabayrouse, “Low-loss rib waveguides containing Si nanocrystals embedded in SiO2,” J. Appl. Phys. 97, 074312 (2005). [CrossRef]
13. N. Daldosso, D. Navarro-Urrios, M. Melchiorri, L. Pavesi, F. Gourbilleau, M. Carrada, R. Rizk, C. Garcia, P. Pellegrino, B. Garrido, and L. Cognalto, “Absorption cross section and signal enhancement in Er-doped Si nanocluster rib-loaded waveguides,” Appl. Phys. Lett. 86, 261103 (2005). [CrossRef]
14. D. S. Gardner and M. L. Brongersma, “Microring and microdisk optical resonators using silicon nanocrystals and erbium prepared using silicon technology,” Opt. Mater. 27, 804–811 (2005). [CrossRef]
15. L. Khriachtchev, M. Rasanen, and S. Novikov, “Efficient wavelength-selectrive optical waveguideing in a silica layer containing Si nanocrystals,” Appl. Phys. Lett. 83, 3018–3020 (2003). [CrossRef]
16. J. Ruan, P. M. Fauchet, L. Dal Negro, M. Cazzanelli, and L. Pavesi, “Stimulated emission in nanocrystalline silicon superlattices,” Appl. Phys. Lett. 83, 5479–5481 (2003). [CrossRef]
17. R. G. Heideman and P. V. Lambeck, “Remote opto-chemical sensing with extreme sensitivity: design, fabrication and performance of a pigtailed integrated optical phas-modulated Mach-Zehnder interferometer system,” Sens. Actuators B-Chem. 61, 100–127 (1999). [CrossRef]
18. O. Hofmann, G. Voirin, P. Niedermann, and A. Manz, “Three-dimensional microfluidic confinement for efficient sample delivery to biosensor surfaces. Application to immunoassays on planar optical waveguides,” Anal. Chem. 74, 5243–5250 (2002). [CrossRef] [PubMed]
19. T. W. MacElwee, S. E. Hill, S. Campbell, D. Ducharme, B. B. Rioux, I. D. Calder, M. Flynn, J. Wojcik, S. Gujrathi, and P. Mascher, “Bright green visible electroluminescence from rare earth doped silicon rich SiOx,” in 2006 3rd IEEE International Conference on Group IV photonics (Institute of Electrical and Electronics Engineers, New York, 2006), pp.216–218. [CrossRef]
20. H. Lee, J. H. Shin, and N. Park, “Performance analysis of naocluster-Si sentized Er-doped waveguide amplifier using top-pumped 470 nm LED,” Opt. Express 13, 9881–9889 (2005). [CrossRef] [PubMed]
21. J. Lee, J. H. Shin, and N. Park, “Optical gain at 1.5 μm in nanocrystal Si-sensitized Er-doped silica saveguide using top-pumping 470 nm LEDs,” J. Lightwave Technol. 23, 19–25 (2005). [CrossRef]
22. A. R. Wilkinson and R. G. Elliman, “Maximixing light emission from silicon nanocrystals - The role of hydrogen,” Nucl. Instrum. Methods Phys. Res B 242, 303–306 (2006). [CrossRef]
23. J. Chilwell and I. Hodgkinson, “Thin-films field-transfer matrix theory of planar multilayer waveguides and reflection from prism-loaded waveguides,” J. Opt. Soc. Am. A 1, 742–753 (1984). [CrossRef]
24. D. Comedi, O. H. Y. Zalloum, E. A. Irving, J. Wojcik, and P. Mascher, “H-induced effects in luminescent silicon nanostructures obtained from plasma enhanced chemical vapor deposition grown SiyO1-y:H(y>1/3) thin films annealed in (Ar+5% H2),” J. Vac. Sci. Technol. A 24, 817–820 (2006). [CrossRef]
25. X. Tan, J. Wojcik, and P. Mascher, “Study of the optical properties of SiOxNy thin films by effective medium theories,” J. Vac. Sci. Technol. A 22, 1115–1119 (2004). [CrossRef]
26. H. Nishihara, M. Haruna, and T. Suhara, Optical Integrated Circuits (McGraw-Hill1985), Chapt. 8.
27. J. Haes, B. Demeulenaere, R. Baets, D. Lenstra, T. D. Visser, and H. Block, “Difference between TE and TM modal gain in amplifying waveguides: analysis and assessment of two perturbation approaches,” Opt. Quantum. Electron. 29, 263–273 (1997). [CrossRef]
28. W. Stutius and W. Streifer, “Silicon nitride films on silicon for optical waveguides,” Appl. Optics 16, 3218–3222 (1977). [CrossRef]
29. G. L. Bona, R. Germann, and B. J. Offrein, “SiON high-refractive-index waveguide and planar lighwave circuits,” IBM J. Res. & Dev. 47, 239–249 (2003). [CrossRef]
30. C. K. Wong, H. Wong, C. W. Kok, and M. Chan, “Silicon oxynitride prepared by chemical vapor deposition as optical waveguide materials,” J. Cryst. Growth 288, 171–175 (2006). [CrossRef]
31. N. Daldosso, M. Melchiorri, F. Riboli, M Girardini, G. Pucker, M. Crivellari, P. Bellutti, A. Lui, and L. Pavesi, “Comparison among various Si3N4 waveguide geometries grown within a CMOS fabrication pilot line,” J. Lightwave Technol. 22, 1734– 1740 (2004). [CrossRef]
