2.4.3. GeSn Ohmic Contact

Among the summarized GeSn contact works is that of Henry. H. Radamson et al., who proposed a novel method to improve the thermal stability of the Ni–GeSn contact. It is well-known that carbon stabilize NiSiGe materials, so after GeSn growth, they implanted C into GeSn. In Figure 15, we can see that the NiGeSn film with C was more uniform than the NiGeSn film without C. Characterization results indicated that the presence of C not only led to the improved thermal stability but also tended to change the preferred orientation of NiGeSn [187]. A comparison work with different contact metals (10 nm of Ni, Ti, and Pt) [199] showed that Ni–GeSn was the most promising candidate due to its low sheet resistance and low formation temperature (below 400 ◦C). Moreover, Pt–GeSn showed better behavior in terms of thermal stability compared to Ni–GeSn and Ti–GeSn. Because Sn loss occurs during B-doped GeSn CVD growth, it is still challenging to create low contact resistivity p-type GeSn contacts with high Sn contents, a challenge that is particularly critical for GeSn lasers and GeSn TFETs [207,208].

**Figure 15.** TEM images for Ni–GeSn interface (**a**) annealing at 400 ◦C without C and (**b**) annealing at 400 ◦C with C. Reproduced from [187], IOP Publishing, open access, 2015.

### **3. Research Progress for GeSn Detectors**

*3.1. GeSn Photoconductive Detector*

Photoconductive detector, which can also be defined as metal–semiconductor–metal (MSM) detector, is regarded as the simplest structure to achieve detection. In this type of structure, two Schottky junctions are designed and the total layer structure does not require any doping. Therefore, it can only work at a high bias voltage due to the existence of high contact resistance. However, the capacitance of a photoconductive detector is quite low, which is helpful for high-speed detection. Based on the photoconductive structure, researchers have put great effort into GeSn photoconductive detectors (Figure 16). Table 4 shows the reported performance levels of GeSn photoconductive detectors grown by CVD technology.

**Figure 16.** Cross-sectional schematic of a device structure for a GeSn photoconductive detector.

As previously reported, IMEC mastered low-cost and commercially available cuttingedge GeSn growth technology in 2011 (Ge2H6/GeH<sup>4</sup> precursor combination) [151]. Subsequently, they further grew a GeSn/Ge MQWs structure, and they also fabricated a photoconductive detector [58]. In 2014, Benjamin, R. Conley et al. reported the temperaturedependent spectral responses and detectivity of GeSn photoconductors with Sn contents ranging from 0.9 to 7% [59]. For a GeSn photoconductor with 7.0% Sn, a maximum wavelength response of 2100 nm was achieved. Experimental results showed that lowtemperature responsivity was two orders of magnitude higher than room-temperature responsivity at 1550 nm, and the maximum specific detectivity was 1 <sup>×</sup> <sup>10</sup><sup>9</sup> cm·Hz1/2/W

at 77 K. In the same year, Benjamin, R. Conley et al. further extended the spectral response using a GeSn layer with 10% Sn [60]. The room-and low-temperature (77 K) wavelength cutoffs for the GeSn detector were found to be 2400 and 2200 nm, respectively. Maximum peak responsivity was observed as 1.63 A/W at 77 K due to photoconductive gain. More importantly, the specific detectivity was increased by about five times compared to the previously reported result (a GeSn photoconductor with 7.0% Sn), indicating that the material quality of the GeSn layer with 10% Sn was greatly improved (Figure 17).

**Table 4.** Summary of reported GeSn photoconductive detectors in terms of Sn content, GeSn thickness, device structure, wavelength cutoff, and responsivity.


**Figure 17.** Specific detectivity for a Ge0.9Sn0.1 photoconductive detector at temperatures of 77, 160, 220, and 300 K. Reproduced from [60], OSA Publishing, open access, 2014.

In 2019, Huong Tran et al. reported a GeSn photoconductor with high Sn contents (the maximum Sn contents of the top GeSn layer were 12.5%, 15.9%, 15.7%, 17.9%, 20%, and 22.3%) [63]. As the Sn content increased, the cutoff wavelength shifted toward longer wavelength due to the bandgap shrinkage. From 77 to 300 K, the cutoff wavelengths were 3200–3650 nm for the GeSn photoconductor with 22.3% Sn. It is worth noting that this D\* value was superior to that of a PbSe detector at the given wavelength range and was comparable to that of a commercial extended-InGaAs detector (4 <sup>×</sup> <sup>10</sup><sup>10</sup> cm·Hz1/2 ·W−<sup>1</sup> ) at the same wavelength range (Figure 18). Even at 300 K, the passivated device showed better results D\* than the PbSe detector from 1500 to 2200 nm.

**Figure 18.** Specific detectivity for a GeSn photoconductive detector at the temperatures of (**a**) 77 K and (**b**) 300 K (the Sn contents for samples A–F were 12.5%, 15.9%, 15.7%, 17.9, 20.0%, and 22.3%, respectively). Reproduced with permission from [63], American Chemical Society, 2019.

To enable a comprehensive overview of the use of GeSn photoconductive materials for infrared detection applications, Figure 19 illustrates the Sn content vs. cutoff wavelength for reported GeSn photoconductive detectors. For GeSn with an Sn incorporation of 0.9–12.5%, the photoconductive detector wavelength coverage was found to range from 1800 to 2950 nm, indicating that GeSn with Sn contents of up to 12.5% or 13% is very promising for SWIR applications. For GeSn with an Sn incorporation of 15.9–22.3%, the photoconductive detector wavelength coverage was found to range from 3200 to 3650 nm, suggesting potential mid wavelength infrared (MWIR) applications. For wavelengths from 3650 to 5000 nm, no detectors have been reported. However, GeSn photoconductive detector performance is limited by current growth technology and Sn distribution uniformity in total layer structures, which causes a low responsivity (the responsivity values are listed in the table above).

**Figure 19.** Sn content vs. wavelength of a GeSn photoconductive detector, indicating that GeSn is a promising absorber in SWIR and MWIR detection applications.

### *3.2. GeSn PIN Detector*

The PIN detector is the most common and widely used detector type for Si-based optoelectronics applications. One side of a PIN detector device is for p-type doping, and the other side is for n-type doping; as such, the built-in electric field is able to locate the intrinsic region [18]. A typical cross-sectional schematic diagram of a GeSn PIN detector is shown in Figure 20, and the major device performance values for reported GeSn PIN detectors are summarized in Table 5.

**Figure 20.** Cross-sectional schematic of typical device structure for a GeSn detector.

**Table 5.** Summary of reported GeSn PIN detectors in terms of Sn content, GeSn thickness, device structure, wavelength cutoff, and responsivity.


In 2009, Jay Mathews et al. demonstrated the first GeSn photodetector with 2% Sn content; 350 nm Ge0.98Sn0.02 was directly grown on a B-doped Si (100) substrate in an UHVCVD system (the carrier concentration in the Si wafer was 4.3 <sup>×</sup> <sup>10</sup><sup>19</sup> cm−<sup>3</sup> ) [57]. Three cycles of post-growth annealing were carried out to decrease the TDDs in Ge0.98Sn0.02. Afterwards, n-doped Ge0.98Sn0.02 was further deposited, and its carrier concentration was found to be approximately 7.5 <sup>×</sup> <sup>10</sup><sup>19</sup> cm−<sup>3</sup> . Using the abovementioned layer structure, a circular GeSn photodetector was fabricated. To evaluate the quantum efficiency of the Ge0.98Sn0.02 photodetector, the circular mesa was continuously illuminated via a halogen source and 1270, 1300, 1550, and 1620 nm lasers. The Ge0.98Sn0.02 detector quantum efficiencies were higher than those in comparable pure Ge device designs processed at low temperatures (Figure 21). Additionally, the wavelength cutoff was extended to at least 1750 nm, which means that a GeSn photodetector with 2% Sn content can cover the entire telecommunication band.

**Figure 21.** Cross-sectional schematic of a GeSn photodetector and its quantum efficiency as a function of wavelength. Reproduced with permission from [57], AIP Publishing, 2009.

In 2018, Huong Tran et al. fabricated GeSn photodetectors with 700 nm thick GeSn layers using the p–Ge/p–Ge0.91Sn0.09/i–Ge0.89Sn0.11/n–Ge0.89Sn0.11/n–Ge layer structure (all layers were grown by RPCVD) [65]. In order to obtain detailed and accurate external reading of quantum efficiency, all GeSn photodetectors were illuminated with a 2000 nm laser. Room-temperature peak responsivity and external quantum efficiency were measured to be 0.32 A/W at 2000 nm and 20%, respectively. When the GeSn detector was illuminated by a 1550 nm laser, its external quantum efficiency reached up to 22%. Different from the previously reported thin film photoconductor, the thick film photoconductor showed an extended wavelength cutoff (2650 nm) due to the reduced strain relaxation and enhanced light absorption in the thick GeSn film. Nevertheless, the peak specific detectivity for the GeSn detector was compared to other commercial infrared detectors at a wavelength range from 1400 to 3000 nm, which showed that peak specific detectivity of the GeSn detector at 2000 nm was only one order of magnitude lower than that of the extended-InGaAs detector (Figure 22). To improve device performance, Xu S, et al. attempted to create a GeSn/Ge MQW detector [67,68], a GeSnOI detector [69], and a photon-trapping microstructure GeSn/Ge MQW detector [209].

**Figure 22.** Specific detectivity for a Ge0.89Sn0.11 photodetector at the temperatures of 77 and 300 K. Reproduced with permission from [65], AIP Publishing, 2018.

Figure 23 summarize the Sn content vs. cut-off wavelength for a reported GeSn PIN detector. For GeSn with an Sn incorporation of 2–11%, the PIN detector wavelength coverage was found to range from 1750 to 2650 nm, indicating that a GeSn PIN detector is very promising for SWIR applications. Due to the limitations of growth technology, PIN detectors at wavelengths from 2650 to 5000 nm have yet to be reported.

**Figure 23.** Sn content vs. cut-off wavelength of the GeSn PIN detector.

### **4. Research Progress for GeSn Lasers**

Since Si-based high-efficiency light sources comprise the technical bottleneck for Sibased monolithic optoelectronic integration, researchers have conducted extensive research into Ge and GeSn lasers. Ten years ago, the rapid development of the GeSn CVD growth technique enabled researchers from MIT to demonstrate optically injected and electrically injected Ge lasers at room temperature. The lasing thresholds of these laser devices were very high, which made it difficult to achieve efficient lasing. As a result, more attentions has been paid to the GeSn material due to its direct bandgap property. In this section, we review the latest research on GeSn lasers with different optical cavities, as well as their device performance.

### *4.1. Optically Injected GeSn Lasers*

### 4.1.1. Optically Injected GeSn Laser with FP Cavity

Based on the GeSn optical gain medium, the world's first optically injected FP cavity GeSn laser was demonstrated at a low temperature [70]. The typical threshold power densities of FP cavity GeSn lasers with cavity lengths of 1 mm, 500 µm, and 250 µm were maintained between 300 and 330 kW/cm<sup>2</sup> (Figure 24a). When the optically injected power density was above its threshold power density, the full width half maximum of the optical emission spectrum was dramatically reduced and the intensity was significantly increased; when the optical injection power density increased to 650 kW/cm<sup>2</sup> , the threshold curve tended to be flat (possibly due to a self-heating effect) (Figure 24a). When the optically injected power density increased to 1000 kW/cm<sup>2</sup> , the maximum lasing temperature for the GeSn laser with 12% Sn content was 90 K. Figure 24b shows high-resolution laser spectra that indicate the performance of a GeSn laser under multi-mode operation.

**Figure 24.** (**a**) Integrated PL intensity vs. excitation power density for a GeSn FP cavity laser with different cavity lengths; (**b**) high-resolution laser spectra for a GeSn laser with cavity lengths of 250 and 500 µm. Reproduced with permission from [70], Springer Nature, 2015.

In 2017, Joe Margetis et al. systematically studied the performance of optically injected GeSn lasers with different Sn contents [73]; the Sn contents of samples A–G were 7.3%, 9.9%, 11.4%, 14.4%, 15.9%, 16.6%, and 17.5%, respectively, and the maximum operation temperatures of samples A–G were 77, 110, 140, 160, 77, 140, and 180 K, respectively (Figure 25). Except for sample A (lower Sn content) and sample E (poor material quality), the samples could be lased at 140 K. It is worth noting that the maximum operation temperature of samples D and G were 160 and 180 K, respectively. The results showed

that the operating temperature of the optically injected GeSn laser was closely related to the Sn content of GeSn, and the GeSn lasers with higher Sn contents possessed higher operating temperature (except for sample F because of its poor material quality). Therefore, increasing the Sn content in GeSn can effectively increase the operating temperature of the laser device. From the theoretical point of view, the main factors that affect the performance of laser devices are material gain, active layer thickness, device surface roughness, and non-radiative recombination. Therefore, there are differences in the operating temperatures of GeSn laser devices with different Sn contents.

**Figure 25.** (**a**) GeSn laser spectra for samples A–G; (**b**) comparison of the PL and laser spectra of samples D and G. Reproduced with permission from [73], American Chemical Society, 2017.

Thanks to the discovery of the GeSn strain-relaxation-enhanced growth mechanism [88], researchers were able to increase the Sn content of GeSn to 22.3%. In this layer structure, the GeSn buffer layer is grown with a nominal recipe for 11% GeSn. When the thickness of the 11% GeSn layer reaches its critical thickness, internal strain in the GeSn layer gradually relaxes and more Sn atoms can be incorporated into the Ge lattice. Experimental results showed that the strain relaxation growth mechanism could lead to high-Sn-content GeSn alloys (higher than 22.3%). Later, Wei Dou et al. reported an optically injected bulk GeSn laser with an Sn content of up to 22.3% [75]; both 1064 and 1950 nm pulsed lasers were used for optical injection, and the maximum operating temperatures were 150 and 180 K, respectively (Figure 26).

**Figure 26.** Temperature-dependent lasing spectra for a bulk GeSn laser with an Sn content of up to 22.3%; the optical injection sources were (**a**) a 1064 nm pulsed laser and (**b**) a 1950 nm pulsed laser. Reproduced from [75], OSA Publishing, open access, 2018.

In 2019, Yiyin Zhou et al. researched optically injected GeSn lasers (an Sn content of 20%) with different waveguide widths [76]; 1064 and 1950 nm lasers were used for pulsed optical injection characterization (Figure 27). They concluded that the operation temperature for sample A was lower than those of the other samples (the laser operation temperatures under 1064 and 1950 nm pulsed injection were 120 and 140 K, respectively). Moreover, the threshold for sample A was relatively larger than those of the other samples (at 77 K, the thresholds under 1064 and 1950 nm optical pulsed injection were 516 and 132 kW/cm<sup>2</sup> , respectively). When the sample width was wider than 20 µm, the operation temperatures of the laser devices could be increased to 260 and 270 K under 1064 and 1950 nm optical pulsed injection, respectively. The reasons for this are as follows: (i) compared with the side wall surface recombination, free carrier absorption loss and nonradiative recombination were the dominant losses at higher temperatures; (ii) the stripeshaped optical injection light beam had a Gaussian distribution, which may have resulted in absorption occurring in the middle of a wider waveguide (less absorption at the edge of the waveguide); and (iii) the optical confinement factor for sample D was lower, which led to a higher threshold.


**Figure 27.** Summary of laser performance under 1064 and 1950 nm pulsed laser injection (the cavity widths for samples A, B, C, and D were 5 µm, 20 µm, 100 µm, and planar, respectively). Reproduced with permission from [76], ACS Publishing, 2019.

The simplest optical cavity is that of Fabry–Pérot, which consists of two parallel reflecting surfaces that allow coherent light to travel through the whole cavity. Due to the directness difference between GeSn alloys with different contents, we summarize the reported operation temperatures for GeSn with different Sn contents in Figure 28. Operation temperatures were found to increase with more Sn incorporation, indicating that operation temperature is closely related to the directness of GeSn. Different from narrow bulk devices, broad bulk devices (with a cavity width greater than 20 µm) possess higher operation temperatures, possibly due to the following two reasons: (1) they have higher optical gains, and (2) they are wider and thus have higher optical injection efficiencies. However, the operation temperature for a GeSn laser with 22.3% Sn incorporation was found to be the same as that of a GeSn laser with 17.5% Sn incorporation, which means that there were many point defects in the high-Sn-content GeSn layer. For clarification, we also summarize the devices performance for the published FP cavity optically pumped GeSn laser (Table 6).

**Figure 28.** Maximum operation temperature vs. Sn content for optical pumped FP cavity GeSn laser (under pulsed 1064 nm laser) [70,74–76].



### 4.1.2. Optically Injected GeSn Laser with WGM Cavity

In 2016, Daniela Stange et al. realized a self-suspending microdisk GeSn laser for the first time [74] (Figure 29). The laser spectrum is shown in Figure 30. It can be seen in the figure that the maximum working temperatures of samples A and B were 80 and 140 K, respectively. Compared with sample B, the lasing spectrum of sample A was blue-shifted

due to its higher content. Although the operation temperature for sample A was lower than that of sample B, the threshold for sample A was lower than that of sample B (the thresholds of samples A and B were 125 and 220 kW/cm<sup>2</sup> at 50 K, respectively).

**Figure 29.** (**a**) Process flow for GeSn microdisk; (**b**) SEM image of GeSn microdisk with an Sn content of 12.5% (diameter was 8 µm). Reproduced with permission from [74], American Chemical Society, 2016.

**Figure 30.** (**a**) Temperature-dependent lasing spectra for samples A and B (the Sn contents for samples A and B were 8.5% and 12.5%, respectively); (**b**,**c**) L–L curves for samples A and B, respectively. Reproduced with permission from [74], American Chemical Society, 2016.

In 2020, Anas Elbaz et al. reported a CW optically injected GeSn microdisk laser with a low Sn content for the first time [80]. Compared with high-Sn-content GeSn, low-Sncontent GeSn has fewer internal point defects and better material quality. After its growth, a low-Sn-content GeSn layer was transferred to an Si substrate with SiN and Al layers. Then, the Si substrate, Ge buffer layer, and defective GeSn layers are removed; only 40 nm, highquality, low-Sn-content GeSn was left. Finally, the transferred GeSn layer was patterned into independent GeSn/SiN microdisks supported by Al microdisk pillars (Figure 31). The lasing spectrum in Figure 31 shows the continuous wave light injection laser spectrum of a GeSn microdisk with a diameter of 7 µm at 25 K: below the threshold, a light emission spectrum with a wide half-width (red line) was obtained under an optical injection power of 0.5 mW; above the threshold, lasing emission characteristics were obvious under the optical injection power of 6.4 mW. Under the pulsed optical injection and CW light injection, the maximum operating temperatures of the laser device were 90 and 50 K, respectively.

**Figure 31.** Fabrication process for a GeSn microdisk laser with SiN<sup>x</sup> all-around. Reproduced with permission from [80], Springer Nature, 2020.

In 2020, Anas Elbaz et al. created an optically injected GeSn microdisk laser after proper defect management [81,82], indicating that the threshold was greatly reduced compared to that of a GeSn microdisk laser without defect management (the lasing threshold reduction was 1 order of magnitude higher compared to examples in the literature). They also found that the maximum lasing temperature for the optically injected GeSn microdisk laser, with Sn contents ranging from 7% to 10.5%, only weakly depended on Sn content. Apart from the directness of the GeSn active region, the experimental results indicated that nonradiative recombinations and point defects are the main obstacles for high-temperature lasing (Figure 32).

**Figure 32.** GeSn microdisk laser with removed defects under the disk. Reproduced with permission from [82], American Chemical Society, 2020.

The abovementioned GeSn microdisk laser results show that both pulsed and CW injection have been achieved (Table 7). Especially for CW lasing, this is the most direct evidence to verify that GeSn can withstand a CW injection test. To gain a better understanding of GeSn microdisk lasers, we summarize the operation temperatures for GeSn lasers with different Sn contents in Figure 33. For the pulsed injection, the operation temperature for the GeSn microdisk laser followed a similar trend to that of an FP cavity GeSn laser (the operation temperature increased with Sn content). However, the operation temperature for the heterostructure and quantum well GeSn laser was lower than that of bulk laser, suggesting that there is still room to improve the operation temperatures of heterostructure and quantum well lasers. For CW injection, it seems that operation temperature enhancement is not that sensitive to Sn content, though it brings efficient heat dissipation.


**Table 7.** Summary of the reported optically pumped WGM cavity GeSn lasers in terms of structure, Sn content, thickness, disk size, pumping laser, maximum operation temperature (Tmax), and threshold.

**Figure 33.** Maximum operation temperature vs. Sn content for an optically pumped WGM cavity GeSn laser (under pulsed 1064 nm laser).

4.1.3. Optically Injected GeSn Laser with Other Microcavities

In addition to those on FP cavity and microdisk cavity GeSn lasers, there have been publications on hexagonal photonic crystal (PC) and micro-bridge GeSn lasers. In 2018, Q.M. Thai et al. reported optically injected GeSn laser with 16% Sn content for the first time [72]. By introducing defects in the photonic crystal defect cavity (such as removing the central hole), the periodic structure around the photonic band gap were able to provide optical feedback to the microcavity. The experimental results showed that the maximum working temperature of the hexagonal photonic crystal GeSn laser was 60 K, and the threshold values at 15 and 60 K were 227 and 340 kW/cm<sup>2</sup> , respectively (Figure 34).

**Figure 34.** L–L curves for a photonic crystal GeSn laser with an Sn content of up to 16%. Reproduced with permission from [72], AIP Publishing, 2018.

In 2019, Jerémie Chrétien et al. explored a novel approach to create a direct bandgap GeSn material via strain redistribution, thereby controlling band structure and lasing wavelength [77]. Tensile-strained GeSn micro-bridge heterostructures were optically injected using pulsed 1064 and 2650 nm lasers (Figure 35), and the maximum operation temperature for the L = 75 µm micro-bridge structure laser was 273 K, which indicates that the operation temperature was very close to room temperature.

**Figure 35.** L–L curves for a micro-bridge GeSn laser with an Sn content of up to 16%. Reproduced with permission from [77], American Chemical Society, 2019.

### *4.2. Electrically Injected GeSn Lasers*

Different from optically injected GeSn lasers, electrically injected GeSn lasers are more suitable for practical applications. However, electrically injected GeSn lasers are more challenging to create due to the GeSn active gain medium having to overcome the extra metal absorption loss and more free carrier absorption (FCA) losses. Theoretical predication for the realm of possibility of electrically injected GeSn/SiGeSn lasers can be traced back to ten years ago, when Greg Sun et al. presented modelling and simulation results for an electrically injected SiGeSn/GeSn/SiGeSn double heterostructure laser with an Sn contents ranging from 6 to 12% [210]; they found that this type of laser requires cooling in the temperature range of 100–200 K after taking radiative, nonradiative, and Auger recombinations into consideration. Afterwards, Greg Sun et al. theoretically proposed that the lattice matched that of an Si0.1Ge0.75Sn0.15/Ge0.9Sn0.1/Si0.1Ge0.75Sn0.15 MQW laser [211], and they found that modal gain was very sensitive to the QW number in the active region and SiGeSn/GeSn/SiGeSn MQW could operate up to room temperature with a 2300 nm emission wavelength. For the SiGeSn/GeSn/SiGeSn MQW laser with 20 QWs, the optical confinement factor was calculated to be 0.74, and the modal gain was able to exceed 100/cm at a pumping current density of 3 kA/cm<sup>2</sup> , which was sufficient to attain roomtemperature lasing.

In 2020, Yiyin Zhou et al. reported the first electrically injected FP cavity GeSn/SiGeSn laser on Si with a lasing temperature of up to 100 K; its minimum threshold was approximately 598 A/cm<sup>2</sup> [89,90] (Figure 36). This work was regarded as an essential achievement for Si-based on-chip light source in the development of Si-based OEICs. Later, the effects of cap layer, cap layer thickness, and Sn content in the active region on the operating temperature, threshold, and emission wavelength were further systematically studied [89,90]. Experimental results showed that: (I) an SiGeSn cap had a better optical confinement effect than a GeSn cap; (II) the optical confinement factor was improved via changing the SiGeSn cap layer thickness; and (III) the use of a GeSn laser with an Sn content of up to 15% did not significantly improve device performance. **Figure 14.** SIMS result for the GeSn sample with an Sn content of up to 22.3% (the maximum Sn contents for regions I, II, and III were 11.9%, 15.5%, and 22.3%, respectively). Reproduced from [113], Springer Nature, open access, 2018.

*Nanomaterials* **2021**, *11*, x FOR PEER REVIEW 2 of 3

deposition (LPCVD) in 2018 [115,116].

[113], thus showing that compressive strain is the primary limiting factor for achieving greater Sn incorporation under an Sn oversaturation condition (Figure 14). In this research, the following growth strategy was proposed: (i) for first GeSn layer growth, they used a growth recipe of 9–12% Sn (the Sn content ranged from 8.8 to 11.9%); (ii) for second GeSn layer growth, they used the same growth recipe, and the SnCl4 flow fraction increased by ~8% compared to the first GeSn layer (the Sn content ranged from 12.5 to 16.5%); and (iii) for third GeSn layer growth, they used the same growth recipe, and the SnCl4 flow fraction increased by ~8% compared to the second GeSn layer. It should be noted that the grading rate of Sn incorporation was well-designed to suppress the growth breakdown. Inspired by the discovery of the SRE GeSn CVD growth mechanism, S. Assali et al. grew a high-quality GeSn layer with 15% Sn using low pressure chemical vapor

**Figure 36.** (**a**) Cross-sectional device structure for the first electrically injected FP cavity GeSn/SiGeSn laser; (**b**) calculated band structure and fundamental TE mode profile. Reproduced from [89], OSA Publishing, open access, 2020.

### **5. GeSn Transistors**

In addition to the rapid advancement of GeSn detectors and GeSn lasers grown by CVD technology, there have been some achievements in the field of GeSn transistors due to their mobility properties. In the hyper-scaling era, the quest for high-performance and low-power transistors is continuing and intensifying. One of the key technology enablers of these goals is that of channel materials with high carrier mobility and direct band gap structures [212,213]. GeSn films have emerged as the most promising candidate for next generation nano-electronic devices of computing due to their excellent properties, including ultrahigh hole mobility, band structures with direct and low band gaps, Si-based CMOS compatibility, and low thermal budget, all of which are of great importance for ultrahigh density devices and 3D integration in the hyper-scaling era. Anisotropy at the top of the GeSn valence band makes the effective mass of light hole rapidly decrease with increases of Sn content and the transport capacity rapidly increase. GeSn is a very promising channel material for the next generation pMOSFET, and its hole mobility is even higher than that of Ge. The hole mobility of Ge pMOSFET is increased by more than 10 times with respect to Si devices. In addition, compressive strain can improve the mobility of a GeSn channel by decreasing the effective mass of the hole carrier. GeSn is generally grown on Si substrates using Ge as the buffer layer, and GeSn subjects the Ge buffer layer to compressive strain since the Sn lattice constant is greater than that of Ge. As GeSn materials are compatible with Si-CMOS technology, a few research groups have studied GeSn-based transistors (Table 8 lists the reported transistors with CVD-grown GeSn layers).


**Table 8.** Summary of reported transistors with GeSn layers grown by CVD technology in terms of institution, transistor type, Sn content, subthreshold swing (SS), Ion/Ioff ratio, and VDS.

> Tunnel-field-effect transistors (TFETs) features subthreshold swings (SS) below 60 mV/decade at room temperature, which also enable a decreased power supply without discounting the off-current. Although Si-TFETs have been reported with SS below 60 mV/decade at low current, band-to-band tunneling (BTBT) is limited by its indirect bandgap property and low SS at high current. Therefore, researchers have investigated GeSn with high Sn contents (12% and 15% Sn incorporation; Figure 37) to create highperformance GeSn TFETs [217]. A higher Sn content enhances device performance, but the subthreshold swing is affected by the increased leakage level. For ultrasmall supply voltages, the device structure should be optimized to improve device characteristics. Using Ge/GeSn heterostructure pTFETs led to the improvements of the BTBT rate. Thus, higher on-current and lower off-current were achieved simultaneously. Christian et al. reported the fabrication and characterization of Ge/GeSn pTFETs (Figure 38), and they recorded a low accumulation capacitance of 3 µF/cm<sup>2</sup> [99]. Moreover, their room-temperature (RT) current–voltage characteristics showed that the Ge/GeSn pTFETs with the 11% Sn content had the highest BTBT current (Figure 39).

**Figure 37.** HR–XRD curves for GeSn samples with (**a**) 12% and (**b**) 15% Sn incorporation. Reproduced with permission from [217], IEEE, 2017.

To suppress the short channel effects (SCEs) of multi-gate transistors, Dianlei et al. investigated the p-FinFETs with a CVD-grown GeSn channel [93]. For GeSn p-FinFETs grown on GeSnOI substrates with 8% Sn incorporation (Figure 40), compressive strain and hole mobility were found to be <sup>−</sup>0.9% and 208 cm2/V·s, respectively. Record low SS of 79 mV/decade for GeSn p-FETs were also achieved.

from [89], OSA Publishing, open access, 2020.

**Figure 38.** Process flow for Ge/GeSn vertical heterojunction pTFETs. Reproduced with permission from [99], IEEE, 2017. **Figure 38.** Process flow for Ge/GeSn vertical heterojunction pTFETs. Reproduced with permission from [99], IEEE, 2017.

**Figure 36.** (**a**) Cross-sectional device structure for the first electrically injected FP cavity GeSn/SiGeSn laser; (**b**) calculated band structure and fundamental TE mode profile. Reproduced

**Figure 39.** (**a**) RT current–voltage characteristics for GeSn p-i-n diode; (**b**) extracted BTBT current vs. electric field. Reproduced with permission from [99], IEEE, 2017.

**Figure 40.** 3D diagram and highlights of the first GeSn FinFETs grown on a GeSnOI substrate. Reproduced with permission from [93], IEEE, 2018. **Figure 40.** 3D diagram and highlights of the first GeSn FinFETs grown on a GeSnOI substrate. Reproduced with permission from [93], IEEE, 2018.

Compared with FinFETs, gate-all-around (GAA) FETs hold better electrostatic control, which can reduce the SCEs for the gate-length scaling. With down-scaling came the proposition of a vertically stacked Si channel for GAAFETs in order to improve drive current [218,219]. Yu-Shiang Huang et al. systematically investigated the strain response, LF noise, and temperature-dependence properties of vertically stacked GeSn nanowire pGAAFETs [214] (Figure 41). Their experimental results showed that: (I) Ion = 1850 µA/µm was improved with higher Sn incorporation; (II) the 6.3% extra enhancement of Ion was observed due to the uniaxial compressive strain that occurred when using wafer bending; and (III) the SS for one-nanowire and stacked two-nanowire GAAFETs were 84 and 88 mV/dec, respectively. To further improve the drive current for GAAFETs at a given footprint (Figure 42), vertically stacked 3-GeSn nanosheet pGAAFETs were studied and the Ion was increased 1975 µA/µm at VDS = −1 V.

**Figure 42.** (**a**) Process flow for vertically stacked 3-GeSn nanosheet pGAAFETs; (**b**) top view after the fin formation; (**c**) RMS value for as-grown GeSn; (**d**) SEM image of stacked 3-GeSn nanosheets. Reproduced with permission from [91], IEEE, 2018.

Furthermore, a top–down approach was utilized to fabricate vertical heterojunction GeSn/Ge GAA nanowire pMOSFETs (Figure 43); with proper optimization, a record high <sup>I</sup>on/Ioff (3 <sup>×</sup> <sup>10</sup><sup>6</sup> ) was achieved [216].

**Figure 43.** (**a**) Fabrication process and (**b**) 3D schematic of single vertical heterojunction GeSn/Ge GAA nanowire pMOSFETs. Reproduced with permission from [220], Elsevier, 2020.

Similar to the n–Ge material, n–GeSn suffers from a large resistance in metal-n–GeSn contacts mainly due to a strong Fermi pinning effect. To improve the performance of GeSn n-FETs, Yen Chuang et al. researched GeSn n-FinFETs and n-Channel MOSFETs: n <sup>+</sup>–GeSn contact; in situ doped n+–GeSn was grown by CVD, and Ni was employed as the contact metal [94]. With the increasing Sn content and n-type doping level, contact resistivity reduced to 3.8 <sup>×</sup> <sup>10</sup>−<sup>8</sup> <sup>Ω</sup>/cm<sup>2</sup> , which may be attributed to the bandgap shrinkage of GeSn (8% Sn incorporation). With the optimized n+–GeSn contact, the highest drive current and best SS for GeSn n-FinFETs were 108 A/m and 138 mV/dec, respectively (8% Sn incorporation) [91]. To suppress the dopant diffusion for S/D carrier activation, microwave annealing (MWA) was proposed. For GeSn with 4.5% Sn incorporation, GeSn nMOSFETs were found to possess an electron mobility of 440 cm2/V·s, suggesting that CVD-grown GeSn and MWA technologies are very promising for GeSn CMOS applications. For higher electron mobility, a 0.46% tensile strain was introduced to Ge0.96Sn0.04; due to the introducing of tensile strain, the carrier population in the Γ valley was higher. Thus, the electron mobility of GeSn nMOSFETs was further improved to 698 cm2/V·s [215].

This discussion shows that pTFETs, pFin-FETs, pMOSFETs, nMOSFETs, and vertically stacked nanowire pGAAFETs with CVD-grown GeSn layers have been extensively studied; breaking the bottleneck the n-doped or p-doped GeSn CVD growth technology is one of the main routes forward for high-performance GeSn transistors. Uniformly stacked nanowires or nanosheets with low surface roughness are of great importance for 5 nm CMOS technology nodes and beyond. More importantly, It should be noted that Henry. H. Radamson et al. explored Ni–(GeSn)<sup>x</sup> contact formation [220]; the strain dependence, phase formation, and thermal stability of Ni–(GeSn)<sup>x</sup> were systematically investigated, and they found that an Sn-rich surface impeded the diffusion of Ni, thus paving the way for the optimization of high-performance nanowire pGAAFETs.

### **6. Conclusions and Outlooks**

In summary, the challenges and progress of GeSn CVD growth technology (including in situ doping technology and ohmic contact formation), GeSn lasers, GeSn detectors, and GeSn transistors were reviewed. Due to growth difficulties, such as the large lattice mismatch between GeSn and Si, the low solubility between Ge and Sn, and phase changes for Sn, more effort must be made in improving the quality of high-Sn-content GeSn materials, GeSn/SiGeSn heterostructures, and GeSn/SiGeSn QWs for high-performance electronic and optoelectronic devices, especially GeSn lasers and GeSn TFETs. Sn distribution uniformity and sharp GeSn/SiGeSn interfaces are the key issues in the development of room temperature, CW electrically pumped GeSn lasers. In addition, research on novel Si-based group IV materials, such as CSiGeSn and CSiGe [221–223], may pave the way for better strain compensation and lattice-mismatched laser structures.

**Author Contributions:** Conceptualization, Y.M., G.W. and H.H.R.; literature survey, Y.M., G.W., Z.K., X.Z., B.X., X.L., H.L., Y.D., B.L. and J.L.; formal analysis, Y.M., G.W., H.H.R., L.D. and J.Z.; project administration, H.H.R.; supervision, G.W. and H.H.R.; writing—original draft preparation, Y.M.; writing—review and editing, Y.M. and H.H.R. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was supported by the construction of a high-level innovation research institute from the Guangdong Greater Bay Area Institute of Integrated Circuit and System (Grant No. 2019B090909006) and the construction of new research and development institutions (Grant No. 2019B090904015), in part by the National Key Research and Development Program of China (Grant No. 2016YFA0301701), the Youth Innovation Promotion Association of CAS (Grant No. Y2020037), and the National Natural Science Foundation of China (Grant No. 92064002).

**Data Availability Statement:** The data presented in this study are available on request from the corresponding authors.

**Conflicts of Interest:** The authors declare no conflict of interest.

### **References**

