Room temperature lasing from electrically pumped n-type doped Ge edge emitting devices has been observed. The edge emitter is formed by cleaving Si-Ge waveguide heterodiodes, providing optical feedback through a Fabry-Perot resonator. The electroluminescence spectra of the devices showed optical bleaching and intensity gain for wavelengths between 1660 nm and 1700 nm. This fits the theoretically predicted behavior for the n-type Ge material system. With further pulsed electrical injection of 500 kA/cm2 it was possible to reach the lasing threshold for such edge emitters. Different lengths and widths of devices have been investigated in order to maintain best gain-absorption ratios.
© 2015 Optical Society of America
Group IV photonics is a promising approach to overcome emerging problems of the electrical wiring technology . In recent years, great effort has been made in this field by combining Ge and Si . But newest publications still deal with the fact that there is no integrated light source available and external laser light sources are used to investigate propagation losses, modulation rates and the responsivity of the optical interconnects .
The Si-based laser light source has been investigated intensively in the last ten years . Various groups demonstrated the electroluminescence (EL) of the direct transition of Ge in light emitting diodes (LED) on Si substrates [5–9]. However, the successful proof of lasing in Ge by the Kimerling group is of particular importance. This was achieved at first by optical pumping [10–12] and later even by electrical pumping at room temperature  and can be seen as the proof of principle for the Ge based laser. However, there are contradicting opinions on this. The Sigg group investigated the optical gain of the direct Ge transition and concluded that a Ge layer cannot act as a gain medium .
In this paper we investigated the emission properties of Ge LED on Si structures which were embedded in a Fabry-Perot resonator by etching a Ge waveguide and forming mirror facets by cleaving. The injected current density was varied within a wide range up to 510 kA/cm2. Two different layer structures with well-defined strains and doping levels were selected to investigate the lasing threshold and the role of the parameters strain and doping.
2. Epitaxial growth and device fabrication
The LED have been deposited on p--type doped Si (100) substrates by using a solid source molecular beam epitaxy (MBE) system. The layer sequence is grown with one epitaxial run with a low temperature process to achieve very sharp doping profiles. The MBE system is equipped with a Si electron beam evaporator and a Ge effusion cell. As n and p-type dopant sources, Sb and B effusion cells are used, respectively.
Two samples are fabricated in a quasi-planar technology to investigate optical intensity enhancement by a tensile strained intrinsic (sample 1) and a fully relaxed n-type doped Ge (sample 2). Schematic cross sections of the MBE layer stacks are shown in Fig. 1. The epitaxial growth starts in both samples with a 400 nm thick and p++-doped (1020 cm−3) Si buried layer (BL). This technique allows contacting the BL from the top side which is necessary for monolithic integration. For the lattice accommodation between Ge and Si a special virtual substrate (VS) technique is used. A p++-doped (1020 cm−3) strain-relaxed Ge layer with a thickness of 50 nm is deposited at a growth temperature of 330 °C. An annealing step at 850 °C considerably reduces the initial threading dislocation density in the VS. After this procedure, the following Ge film grows fully relaxed on the VS. The BL heterojunction contact has a thickness of 450 nm in both samples. After the BL, the epitaxial growth sequences are different for the samples. The optical active layer of sample 1 consists of a 250 nm thick intrinsic region (see Fig. 1 left). After the growth of this layer a tensile strain was induced by a high temperature annealing step at 750 °C. The top contact is finally realized as n++-doped Ge/Si heterojunction contact (1020 cm−3) .
For sample 2 (see Fig. 1 right) the optical active layer consists of a 250 nm thick n+-doped region (3·1019 cm−3) grown at 160 °C. A high doping concentration in the lower 1019 cm−3 range is considered to improve the electroluminescence intensity due to an indirect valley state filling effect in Ge . The investigation of even higher n-type doping beyond 4∙1019 cm−3 shows massive degeneration in the optical output power instead . This could be due to enhanced carrier-carrier interaction like Auger recombination and less electrical activation of the dopants. The n+-doped layer is not annealed to avoid dopant interdiffusion. The top contact is finally realized as a 200 nm thick n++-doped (1020 cm−3) Si contact.
The LED are realized as double mesa structure. The first mesa process defines the dimensions of the diode; the second one isolates the diodes from each other. The mesas are passivated by a 300 nm thick layer made from SiO2 and fabricated by a low temperature plasma-enhanced chemical vapor deposition process. The contact metallization is realized by sputtered Al.
The lateral Ge waveguides are broken for mirror facet formation. With this technique resonators with different lengths can be produced. The cleaving process for all devices is done by sawing from the backside through the Si substrate. The wafer fragment is then inversely mounted on a thin plastic film. By applying pressure with small tweezers the wafer fragment cracks along the crystallographic (110) plane. The small wafer bars are then mounted on copper plates for better temperature conductivity. A cross sectional view scanning electron microscopy (SEM) image of the processed edge emitter structure is shown in Fig. 3 (left). The right picture in Fig. 3 shows an enlargement of the active device area. The SEM figures show a perfect mirror facet. High quality parallel interfaces are necessary for providing optical feedback in the edge emitter.
3. Electro optical characterization of vertical LED
The setup for the EL measurements consists of a probe station with a glass fiber, an Ando AQ6315A optical spectrum analyzer (OSA) and a semiconductor parameter analyzer. The measurements with the OSA are done with a 10 nm resolution and a sampling rate of 1 nm. Very high integration time under continuous electrical injection provides the most accurate optical output spectrum as a function of wavelength λ. The OSA dynamic range goes down to −100 db. It is possible to measure spectral intensity of a few pW/nm with this setup. For wavelengths higher than 1700 nm the sensitivity decreases due to detector cut-off. This results in a lower signal to noise ratio. A multimode glass fiber with a core diameter of 600 µm is used to ensure negligible transfer and coupling losses to the spectrometer.
The EL spectra are measured for different continuous injection currents and pulsed injection currents with 4% duty cycle and 150 μs pulse width. The pulsed injection results in a lower internal diode temperature.
Typical room temperature EL spectra as function of wavelength for both samples at different continuous injection current densities are shown in Fig. 4. The wavelength of the EL peak shifts to higher values with increasing injection currents because the Ge bandgap shrinks with higher device temperature.
From this dependence the direct band gap energy Eg,Γ can be extracted at room temperature . For sample 1 an Eg,Γ = 0.786 eV is calculated from the EL measurements. The direct bandgap energy shifts to a lower value by about 14 meV compared to unstrained Ge. A shift of 14 meV is corresponding to a tensile strain of 0.18% . For sample 2 a direct band gap of Eg,Γ = 0.767 eV is calculated. The bandgap shrinks in this sample because of the bandgap narrowing in highly doped semiconductors . A comparison of both samples shows that sample 2 has a higher EL efficiency at the same injection current density. The highest injection current densities in these devices are 5 kA/cm2.
A further increase of the injection current could be achieved by cooling the sample. The EL intensity of sample 2 at a temperature of 280 K is shown in Fig. 5 for different injection current densities.
The backside cooling allowed injection current densities up to 7.5 kA/cm2 in the LED. The comparison of room temperature and 280 K measurements shows decreased maximum EL intensity for the cooled sample. This behaviour is typical for indirect semiconductor materials due to less thermal carrier excitation.
A change in the characteristics of the typical LED spectrum is observed for current densities greater than 5 kA/cm2. A new peak is formed in the LED spectrum which is further shifted into the infrared. The new peak increases with increasing current density. This is indicating optical intensity gain from sample 2 for the spectral range between 1660 nm and 1700 nm. Experimental results are indicating large optical gain for 1·1019 cm−3 n-type doping in this spectral range . Our experiment is consistent with the shape of this gain spectrum with respect to higher doping induced infrared shift. The behaviour is comparable to III/V laser diodes in the subthreshold regime . This is contrary to sample 1 where a typical LED behaviour even for higher injection currents is observed.
4. Electro-optical characterization of edge emitters
To accomplish a further increase of the injection current density, fabrication of lateral edge emitters is suitable. The optical window and the electrical contacts are separated in this device concept. This enables easy scaling of the optical active layer and an increase of injection current density with the same current.
A typical EL intensity distribution as function of wavelength of sample 2 at 280 K is shown in Fig. 6(a). The edge emitters are 6.4 µm wide and the length is varied. The injection current density of 110 kA/cm2 is kept constant for the measurements.
The peaks in the EL spectra outline are evident, providing intensity gain for the edge emitters in the wavelength regime between 1650 nm and 1700 nm. For the shortest edge emitter with 35 µm, Fabry-Perot oscillations can be observed in the EL intensity spectrum. The edge emitter oscillates in the TE mode due to high damping of the TM mode induced by the Al contact pads . These oscillations are caused by feedback of the optical resonator. Equation (1) describes the proportionality of the mode number m in the resonator to the length l of the device at wavelength λ:
The effective model index neff = 3.82 for the TE modes can be extracted for a symmetric Si/Ge/Si waveguide . The wavelength distance Δλ between two consecutive modes m and m + 1 of the Fabry-Perot resonator can be calculated by Eq. (2):
For cavity lengths l of 100 µm and 35 µm at a wavelength λ of 1.66 µm from the EL peak, the free spectral range Δλ in the Fabry-Perot resonator is calculated as 3.6 nm and 10.4 nm, respectively. Resonances in the EL spectrum can be resolved better for shorter resonator lengths. This is shown in Fig. 6(b). The shortest 35 µm long resonator was measured with sub 5 nm resolution of the OSA for different injection currents. The free spectral range of the 35 µm long edge emitter fits with the physical dimensions of the resonator. The intensity of the Fabry-Perot enhanced EL spectrum increases with increasing injection current.
The relative magnitude of the oscillations in the EL spectrum ΔI pictures the change of the Fabry-Perot resonance intensity. It is normalized to the total EL intensity I at 1660 nm as shown in the inset in Fig. 6(b). This ratio ΔI/I increases with higher injection currents with a transparency threshold at a current density of around 70 kA/cm2. The increase of the transmittance indicates the reduction of optical absorption. This is an optical bleaching effect due to optical gain which counterbalances partly the modal and absorption losses. In order to electrically pump the edge emitter to the lasing threshold, the injected current density must be increased further to obtain a net gain (optical gain larger than the losses). The Kimerling group reported a lasing threshold of about 280 kA/cm2. For this purpose, edge emitters with 1.6 µm width and a length of 380 µm are fabricated and electrical pumping is performed by pulsed current injection with 4% duty cycle.
The EL measurements for these edge emitters at room temperature are shown in Fig. 7. The spectrum is measured with 20 samples per nanometer to provide maximum spectral quality. The lowest current density at 100 kA/cm2 shows the typical LED spectrum. For a current density of 300 kA/cm2 the outline of the gain spectrum is visible in the EL intensity. Further pumping to Jth = 510 kA/cm2 results in observation of narrow laser pulses in the wavelength range between 1675 nm and 1685 nm. The threshold current density matches the theoretical prediction . Due to the high injection current, it is just possible to pump slightly above the threshold. So the predicted multi wavelength pulsed laser operation is observed . Strong linewidth narrowing is indicating lasing. We expect single mode operation for higher pumping current densities .
The inset shows the integral EL intensity as function of injection current. Transparency threshold at a current density of around Jt = 120 kA/cm2 can be extracted for these edge emitters. Above the threshold for 500 kA/cm2 we observed a strong increase in integral EL intensity.
We investigated the edge emission from Ge LEDs on Si in a Fabry-Perot resonator for two epitaxial structures which differed with respect to two properties (tensile strain, n-doping) which were considered as essential [11, 16, 18, 20] for lasing from the indirect semiconductor Ge. Sample 1 was tensile strained (0.18%) Ge but with an undoped active region, whereas sample 2 had an unstrained but highly n-doped (3·1019 cm−3) active region. Sample 1 behaved like a LED without lasing up to the highest available current injection density (pulsed injection density of 510 kA/cm2). Sample 2 exhibited three regions: At low current densities the emission was comparable to that of a vertical LED. At higher current densities (around 100 kA/cm2, in detail depending on the Fabry-Perot dimensions) the stimulated emission leads to optical gain, optical bleaching by reduced net absorption and strong intensity increase. At even higher current densities (500 kA/cm2) we found the onset of lasing with small but intensive multimode peaks.
The observation confirms the principal validity of the earlier work of Kimerling’s group against the ambiguities fed by theoretical considerations and missing experimental confirmation. The higher threshold of lasing (500 kA/cm2 vs 280 kA/cm2) compared to the work of Kimerling’s group is probably due to the unstrained active region.
Further work will concentrate on the extraction of optical gain information from the transparency region below laser threshold, on the reduction of the laser threshold by different strain levels of the doped Ge and on an intensity increase from carrier confinement in Ge quantum well and quantum dot structures.
The work was supported by the DFG Deutsche Forschungsgemeinschaft (Grant no. OE 520/5-1).
References and links
1. R. Soref, “Silicon photonics: A review of recent literature,” Silicon 2(1), 1–6 (2010). [CrossRef]
2. E. Kasper, “Prospects and challenges of silicon/germanium on-chip optoelectronics,” Front. Optoelectron. China 3(2), 143–152 (2010). [CrossRef]
3. P. Chaisakul, D. Marris-Morini, J. Frigerio, D. Chrastina, M. S. Rouifed, S. Cecchi, and L. Vivien, “Integrated germanium optical interconnects on silicon substrates,” Nat. Photonics 8(6), 482–488 (2014). [CrossRef]
4. D. Liang and J. E. Bowers, “Recent progress in lasers on silicon,” Nat. Photonics 4(8), 511–517 (2010). [CrossRef]
6. S. L. Cheng, J. Lu, G. Shambat, H. Y. Yu, K. Saraswat, J. Vuckovic, and Y. Nishi, “Room temperature 1.6 µm electroluminescence from Ge light emitting diode on Si substrate,” Opt. Express 17(12), 10019–10024 (2009). [CrossRef] [PubMed]
7. M. de Kersauson, R. Jakomin, M. El Kurdi, G. Beaudoin, N. Zerounian, F. Aniel, S. Sauvage, I. Sagnes, and P. Boucaud, “Direct and indirect band gap room temperature electroluminescence of Ge diodes,” J. Appl. Phys. 108(2), 023105 (2010). [CrossRef]
8. T. Arguirov, M. Kittler, M. Oehme, N. V. Abrosimov, E. Kasper, and J. Schulze, “Room temperature direct band-gap emission from an unstrained Ge p-i-n LED on Si,” Solid State Phenomena 178–179, 25–30 (2011). [CrossRef]
9. M. Oehme, M. Gollhofer, D. Widmann, M. Schmid, M. Kaschel, E. Kasper, and J. Schulze, “Direct bandgap narrowing in Ge LED’s on Si substrates,” Opt. Express 21(2), 2206–2211 (2013). [CrossRef] [PubMed]
11. M. de Kersauson, M. El Kurdi, S. David, X. Checoury, G. Fishman, S. Sauvage, R. Jakomin, G. Beaudoin, I. Sagnes, and P. Boucaud, “Optical gain in single tensile-strained germanium photonic wire,” Opt. Express 19(19), 17925–17934 (2011). [CrossRef] [PubMed]
13. R. E. Camacho-Aguilera, Y. Cai, N. Patel, J. T. Bessette, M. Romagnoli, L. C. Kimerling, and J. Michel, “An electrically pumped germanium laser,” Opt. Express 20(10), 11316–11320 (2012). [CrossRef] [PubMed]
14. L. Carroll, P. Friedli, S. Neuenschwander, H. Sigg, S. Cecchi, F. Isa, D. Chrastina, G. Isella, Y. Fedoryshyn, and J. Faist, “Direct-Gap gain and optical absorption in germanium correlated to the density of photoexcited carriers, doping, and strain,” Phys. Rev. Lett. 109(5), 057402 (2012). [CrossRef] [PubMed]
15. M. Oehme, M. Kaschel, J. Werner, O. Kirfel, E. Kasper, and J. Schulze, “Germanium on Silicon photodetectors with broad spectral range,” J. Electrochem. Soc. 157(2), H144–H148 (2010). [CrossRef]
16. X. Sun, J. Liu, L. C. Kimerling, and J. Michel, “Toward a germanium laser for integrated silicon photonics,” IEEE J. Sel. Top. Quantum Electron. 16(1), 124–131 (2010). [CrossRef]
17. M. Schmid, M. Oehme, M. Gollhofer, M. Kaschel, E. Kasper, and J. Schulze, “Electroluminescence of unstrained and tensile strained Ge-on-Si LEDs,” in Proceedings of IEEE Conference on Group IV Photonics (GFP, 2012), pp. 135–137. [CrossRef]
18. D. Sukhdeo, D. Nam, Z. Yuan, B. Dutt, and K. Saraswat, “Toward an efficient Germanium-on-Silicon laser: ultimate limits of tensile strain and n-type doping,” J. Opt. Soc. Am. 109, JTh2A (2013).
19. R. N. Hall, “Coherent light emission from p-n junctions,” Solid-State Electron. 6(5), 405–408 (1963). [CrossRef]
20. E. Kasper, M. Oehme, J. Werner, T. Aguirov, and M. Kittler, “Direct bandgap luminescence from Ge on Si pin diodes,” Front. Optoelectron. 5(3), 256–260 (2012). [CrossRef]