Photoluminescence from silicon dioxide photonic crystal cavities with embedded silicon nanocrystals

One-dimensional nanobeam photonic crystal cavities are fabricated in silicon dioxide with silicon nanocrystals. Quality factors of over $9\times {10}^3$ are found in experiment, matching theoretical predictions, with mode volumes of $1.5{\left(\lambda /n\right)}^3$. Photoluminescence from the cavity modes is observed in the visible wavelength range 600–820 nm. Studies of the lossy characteristics of the cavities are conducted at varying temperatures and pump powers. Free carrier absorption effects are found to be significant at pump powers as low as a few hundred nanowatts.

Figure 1

(Color online) (a) SEM picture of the fabricated beam structure. The $E_y$ component of the (b) first- (c) second-, and (d) third-order TE-like modes supported by the nanobeam with parameters $w=3.2a$, $h_x=0.5a$, $h_y=0.7w$, and thickness $d=0.7a$ are also shown.

Figure 2

(Color online) (a) $Q$ and (b) $V_m$ for the ${\text{TE}}_0$ mode for different beam widths $\left(w\right)$ and thicknesses $\left(d\right)$. We keep the same air hole design for the simulated cavities.

Figure 3

(Color online) (a) PL spectra from unpatterned film at 10 and 290 K, as well as a cavity spectrum at 290 K with the first two TE modes visible. (b) TEM image of representative Si-NCs. The scale bar represents 5 nm. (c) The PL spectrum of a representative fundamental cavity mode at 290 K and the fit to a Lorentzian with $Q=9000$. (d) The polarization angle dependence of the cavity mode at 290 K along with the angle dependence of PL from an unpatterned region. $0°$ corresponds to the $y$ direction of Fig. 1d.

Figure 4

(Color online) The $Q$’s of cavities at different wavelengths as a function of temperature.

Figure 5

(Color online) (a) The $Q$’s of the one set of cavities vs the cavity wavelengths, at 10 and 290 K. The PL from an unpatterned region of the sample is shown for reference. (b) The change in the cavity linewidth as the temperature is increased from 10 to 290 K.

Figure 6

(Color online) The pump power dependence of the integrated intensity for different cavities at (a) 290 K and (b) 10 K. The amplitude traces of each cavity are offset by factors of 10 to allow clear viewing. The pump power dependence of the change in cavity wavelength (with respect to the low pump power wavelength, ${\lambda }_0$) for different cavities at (c) 290 K and (d) 10 K. Power dependences were taken with both the CW diode laser and the pulsed frequency-doubled Ti:Sapphire. The pump power is measured in front of the objective.

Figure 7

(Color online) The linewidths of representative cavities as the pump power is changed at (a) 290 K and (b) 10 K. Both a CW diode laser and a frequency doubled mode-locked Ti:Sapphire laser are used as the pump source. The pump power is measured in front of the objective.

Figure 8

(Color online) (a) The Si-NC density distribution of this sample as a function of Si-NC radius. (b) Time-resolved measurements of the Si-NC rise and fall time. Fits to extended exponential distributions yield a rise time of $17\mu \text{s}$ and a fall time of $20\mu \text{s}$ for the sample at 290 K, and a rise time of $78\mu \text{s}$ and a fall time of $79\mu \text{s}$ for the sample at 10 K. (c) The excitation cross section and the lifetimes $\left({\tau }_f\right)$ of the Si-NCs as a function of emission wavelength, for both 290 and 10 K. (d) The calculated free carrier concentration as a function of pump intensity for 290 and 10 K.

Figure 9

(Color online) (a) The change in cavity linewidth as a function free carrier density for a representative cavity around 720 nm at 290 and 10 K. Linear fits to the data are also shown. (b) The free carrier absorption cross section $\left({\sigma }_{\text{FCA}}\right)$ obtained from linear fits such as those in part (a), as a function of wavelength, for different cavities throughout the PL spectrum of the Si-NCs. The inset shows the same data renormalized in a small $V_m$ setting. The dashed lines represent fits to a ${\lambda }^b$ model, where $b=1.3$ and 1.8 for the 290 K and 10 K data, respectively.