High-efficiency silicon-compatible photodetectors based on Ge quantum dots

S. Cosentino, Pei Liu, Son T. Le, S. Lee, D. Paine, A. Zaslavsky, D. Pacificia)

School of Engineering, Brown University, Providence, RI02912, USA

S. Mirabella, M. Miritello, I. Crupi, A. Terrasi

MATIS-IMM-CNR and Dipartimento di Fisica ed Astronomia, Università di Catania, Catania I-95123, ITALY

Supplementary Information

By using a calibrated silicon reference cell, we determined the spectral distribution of incident photons per unit time,, on our sample upon illumination with a nearly monochromatic light source. Our monochromator and lamp set-up is characterized by a spectral resolution of ~ 3 nanometers. In particular, by measuring the net light current (at V = 0 V) and the external quantum efficiency (EQE) of the reference cell with our EQE/IQE measurement system (see Fig. S1), we determined by using the formula:

In order to calculate the spectral responsivity of our a-Ge QDs photodetectors we took into account the spectrum of our light source, by using the formula:

where the quantity at the numerator is the photocurrent under illumination at a given voltage. Therefore, measures the ability of collecting generated carriers under illumination at a given voltage.In Fig. S2 we show a comparison between the spectral responsivity of our Ge QDs MIS photodetector (at V = –2 V) and the calibrated silicon reference cell – these data have also been added to Fig. 2 in the main paper.

To get insight into the role of Ge QDs in the light absorption and to clarify the mechanism of carrier transport in our Ge QDs photodetector, we calculated the spectral internal quantum efficiency (IQE), defined as the number of photo-carriers collected at the device terminals per absorbed photon at a given voltage, defined as follows:

where R is the reflectance of our device. This quantity can be easily related to the photogeneration of carriers occurring under light exposure.

The volume fraction of the QDs in the SiO2 layer calculated using the Maxwell-Garnett approximation and fit to the reflectance measurements, is around 36% as pointed out in our paper; this leads to a number density of QDs around 2.5x1019 cm-3. Considering the Ge dose measured by RBS and the size extracted by TEM analysis, a value of Ge QDs density of ~ 2x1019 cm-3was found, in reasonable agreement. The doping level of the n-type Si substrate was measured using the four-point method, giving the resistivity of 0.0075.cm,corresponding to substrate doping of ~ 8x1018 cm-3. Given the measured doping level, the threshold voltage for inversion in our MIS structure is calculated to be 115 V, much higher than the typical bias values used in our experiments.

Knowing the incident photon flux and absorption in the QD layer and silicon substrate, we calculated the number of excited QDs in the SiO2 layer and the generated carriers in the Si substrate within one diffusion length (about 4.5 μm), assuming all absorbed light generates excited carriers, without considering recombination loss. We further assume each QD only absorbs one photon in the process. Our results are shown in the Fig. S3. The actual number of photons absorbed is determined by both the absorption fraction in QDs and incident photon numbers at that particular wavelength.It is clear that a large number of carriers is generated in the Si substrate over a wide spectral range, while the excitation of QDs is mostly lower than the photogeneration in the Si substrate except at the small wavelengths.The IQE characteristic curve we have measured, where the maximum value peaks around 700nm, indicates that both QD and substrate absorption play an important role. At this point, the detailed mechanism requires further study, but a likelymechanism is that holes generated in the QDs and in the Si substrate within a diffusion length of the interface get trapped at the QD interface states. The resulting positive charge in the insulator enhances the electron injection from the IZO, giving rise to a higher current and IQE.

Finally, in Fig. S4, we show IQE versus voltage at different wavelengths, where the incident photon number at each wavelengthis specified in the previous discussion, see Fig. S1. As the magnitude of the reverse biasis increased, the measured IQE increases correspondingly. In particular, the IQE is largest at 700 nm, where the maximum value is reached at –10 V. We should point out that IQE below 100% are also observed, for example, at low reverse voltages for most wavelengths used, or at higher bias voltages for wavelength shorter than 450nm or longer than 1000 nm. Overall, our device works best in the 500–1000 nm wavelength range.

Fig. S1: Top panel: short-circuit current of the silicon reference cell at different wavelengths; middle panel: comparison between the quantum efficiency curves (external, EQE, and internal, IQE) of the silicon reference cell obtained from the EQE/IQE measurement system (dashed curve represents the EQE of the calibrated silicon reference cell); bottom panel: incident photon number at different wavelengths.

Fig. S2: Comparison between the spectral responsivity of the Ge QDs MIS photodetector (at V= –2 V) and of the silicon reference cell (data added to Fig. 2 of revised paper).

Fig. S3: Top panel: fraction of absorbed light intensity in QDs layer and Silicon layer (within the minority carrier diffusion length of 4.5 µm); bottom panel: number of excited QDs in SiO2 layer (assuming single-photon absorption per QD) and photogenerated carriers in Si substrate.

Fig. S4. Comparison of IQE vs. voltage at different wavelengths.