Introduction

The rapid development of e-motion, autonomous vehicles, IOTs, and 5G applications associated to green energy conversion recently lead to an underestimated growth of high frequency and power devices fabrication on 150 mm and 200 mm wafers of compounds (III–V, II–VI..) and Wide Band Gap (WBG) materials. GaAs, InGaAs, InP, HgCdTe, InAs, InSb, and GaSb are used for RF devices and optoelectronics, while SiC and GaN are more dedicated to high voltage and fast switching power devices. As the fabrication of these relatively new substrates is complicated, most of them can be found only on small diameter (100, 150, and recently 200 mm for SiC wafers) compared to mass volume production on Silicon which is mainly on 300 mm.

Most of these substrates are transparent, fragile, often exhibit large bow which provides a challenge for handling and wafer clamping systems in the current production of implanters. Those substrates also require new species for doping, for example, Aluminum for p-type and Nitrogen for n-type are commonly used for SiC, while Magnesium, Beryllium, and Calcium provide p-type and Silicon for n-type are utilized for GaN [1]. Furthermore, those substrates require specific implantations of non-doping species (H, He, Ar…) for the creation of insulating layers. In SiC, diffusion coefficient of dopants is very low and in GaN the thermal budget is limited to 850 °C to avoid GaN decomposition. Thus, the doping profile shape is adjusted using multi-energy steps and high-energy (about 1 MeV) implantations for the fabrication of deep edge terminations or “super junction” structures. High-temperature implants, about 600 °C, are required for SiC, while 400 °C to 1000 °C for GaN [2, 3] are necessary to minimize implantation defects and to allow a lowering of the thermal budget during the activation anneal.

To quickly address this fast-growing market, most of the leading implanter manufacturers have “downgraded” existing 300 mm implanters and added a heated wafer chuck as a specific required option. Even when optimized for the fabrication of advanced devices on 300 mm silicon wafers, these tools present several technical and cost drawbacks for the fabrication of WBG and compound devices on 100 to 200 mm substrates. Owing to these limitations, Ion Beam Services (IBS) has developed an optimized implantation platform for these applications, the FLEXion® ion implantation system (Fig. 1). The main characteristics of FLEXion® system are the following: Optimized 150 mm or 200 mm parallel beamline for reduced scanned surface to minimize implant time; 250 keV or 400 keV acceleration energy options allowing to reach 1.2 MeV using triple-charged ions; high-resolution mass spectrometer of 4700 keV. AMU to avoid beam contamination even using multicharged ions; optimized hardware for H2 and He implantations; dedicated handling for transparent and brittle wafers; high-temp chuck options; and either mechanical clamping for bowed wafers or electrostatic clamping (allowing throughputs of 55 wafers per hour in real implantation conditions at 600 °C).

Fig. 1
figure 1

Photos of the 400 keV FLEXion® implanter and its high voltage terminal

Full size image
Fig. 2
figure 2

a Ions species available on FLEXion® tool with highlights on the ions commonly implanted on WBG materials. b and c photo of the ECR ion source and example of associated Ar mass spectrum with up to 6 + charged Argon ions

Full size image

Ion sources and implanted species

In standard configuration FLEXion® tools are equipped with an Indirectly Heated Cathode (IHC) ion source. This source efficiently produces a wide range of ions (as shown in Fig. 2a) from gas precursors (high-pressure or low-pressure cylinders), high-temperature vaporizer, liquid precursors, or sputtering targets. As an example, for Aluminum implantation required for SiC doping, we use an AlN target associated with a novel and non-toxic gas mixture which optimize Al+ beam current without affecting source lifetime. Simple, double- and triple-charged ions can be generated allowing implantations up to 1.2 MeV using 400 kV acceleration voltage. However, with IHC ion sources the current of triple-charge ions is much lower (typically 100 times lower) than for single-charged ions making this strategy for high-energy implantations difficult owing to its poor throughput.

For higher current in multicharged ions, an Electron Cyclotron Resonant (ECR) ion source has been developed and is proposed as an option (Fig. 2b). This source is compatible with gas or solid precursor. Using Argon for example, we have demonstrated the generation of up to Ar6+ ions (Fig. 2c). With Aluminum, the current ratio between of Al+, Al2+, and Al3+ is, respectively, 1/1.4/ 0.4 (Arbitrary Units) making current levels for these 3 ions compatible for production requirements.

IBS also put a specific focus on new possible safety issues identified with precursors such as Beryllium or Hydrogen, both are required for GaN device fabrication. All our technical choices are safety oriented to allow the use of these species in a production environment. For example, high-pressure hydrogen cylinders have been replaced by a hydrogen generator using DI water from the fab and allowing safe and on the fly delivery of low-pressure hydrogen to the tool.

High-temperature implantation

For SiC and GaN, it is necessary to implant at high temperature to minimize implant defects and lower the thermal budget required for activation. IBS has developed 2 types of heating chucks for this purpose. For production purpose on wafers with an “acceptable” bow, the electrostatic clamping (e-chuck) (Fig. 3a) is the best solution to transfer the heat from the hot chuck to the wafer in a uniform and well reproducible way. On FLEXion® tools the proposed heating e-chucks allow the wafer to reach 650 °C with a temperature non-uniformity lower than 2.5% (measure with thermocouple wafers). The heating ramp up to 20 °C/min allows acceptable downtime when the chuck needs to be cooled down for maintenance purposes. However, compound wafers with EPI layers often have bows which are not compatible with electrostatic clamping technology. For this purpose, IBS also proposes a mechanical clamping heating chuck (Fig. 3b) with similar performance to the e-chuck..

Fig. 3
figure 3

a Photo of the 100 mm, 600 °C heating e-chuck. b Photo of the 200 mm 650 °C heating mechanical chuck

Full size image

In both cases real-time temperature control on the wafer was performed using a pyrometer.

Mass spectrometry and doping profiles

Most of the commercially available implanters integrate a mass spectrometer optimized for silicon device fabrication, i.e., its resolving power has been limited to M/ΔM < 100 to optimize costs and space constraints. While valid for standard dopants with single-charged ions, this approach finds its limit with the new implanted species required for SiC, GaN, and for compound substrates as new contamination risks linked to insufficient mass separation appear, especially when multicharged ions are used. As an example, Volker Häublein presented in IIT 2018 [3] the contamination of a Al2+ implantation by Ar3+ ions observed on a competitor tool. Here, argon was used to generate aluminum ions by sputtering in the ion source and the resolving power of the mass spectrometer magnet, corresponding to M/ΔM = 76, was not enough to separate the two peaks as illustrated in Fig. 4a. Using the mass spectrometer of FLEXion® with M/ΔM = 175, we can clearly distinguish the two peaks and avoid this contamination (Fig. 4b).

Fig. 4
figure 4

a Illustration of possible Ar3+ contamination in a Al2+ implantation with a low-resolution mass spectrometer from a competitor tool. b Ar3+ /Al2+ separation using the high-resolution mass spectrometer of FLEXion tool. c Multicharged Al implantation SIMS profiles in 4H-SiC. b Example of 600 °C chained Al implantation profiles in 4H-SiC (comparison with IBS simulator, SRIM and SIMS profiles measured at several wafer locations: center, north, south, east, west)

Full size image

The ability to achieve clean and predicable doping profiles using simple or multicharged Aluminum ions is illustrated in Fig. 4c and Fig. 4d. Figure 4c shows SIMS profiles of Aluminum in 4H-SiC realized using an acceleration energy of 180 keV and Al+, Al2+, and Al3+ ions. Figure 4d shows Al concentration profiles for a multi-energy implantation to create a flat concentration of 5E19/cm3 on a 200 nm wafer. In this case, we used our in-house simulator to establish the best number of implantations with respective energies and doses (4 implantations were forecasted at 40, 88,150, and 195 keV). Implantations were performed at 600 °C on a 150 mm 4H-SiC wafer. Al profile was measured by SIMS on 5 points on the wafer with comparison to IBS model and to SRIM. We can observe a very good correlation between the SIMS and IBS model. SRIM is efficient to predict the concentration of the plateau but underestimates the tail, it is also time consuming compared to the IBS model which computes the optimal implant conditions in a few seconds. This model is available as an option on FLEXion® software. Up to 10 chained implantations can be simulated and then realized on the tool with full autotune managing energy, charge, species change, with optimizing wafer transfers.

Conclusion

The FLEXion® ion implanter guarantees an optimized 150/200 mm implants for compounds and Wide Band Gap applications rather than a downgraded 300 mm tool. Thanks to FLEXion® optimized ions source and high-resolution mass spectrometer, it performs implant multicharged ions with production level currents and with no risk of contamination. FLEXion® systems cover a wide range of energies removing the need to purchase additional high-energy implanters to perform MeV range implants. FLEXion® handling solutions, ion species capabilities, and heating chuck options have specifically been developed to optimize a cost-effective solution for SiC, GaN, and compound material device fabrication on 150 and 200 mm substrates.