1、Laser Annealing of Power DevicesThe discrete IGBT (Insulated Gate Bipolar Transistor) is the most important power semiconductor device for power conversion and control in the medium power range with voltages 400 V. Next generation IGBTs require low thermal budget p-n junction formation for the back
2、side field stop and emitter layer. The requirement for the limited thermal budget is due to the fact that the front side metallization does not allow high temperature treatment for the activation of the field stop layer. Therefore, laser annealing experiments have been carried out using a frequency
3、doubled Yb:YAG laser at a wavelength of 515 nm with an energy density up to 4 J/cm2. In order to find out appropriate process conditions for a single step Boron and Phosphorous laser annealing process, parameters for ion implantation as well as energy density and pulse duration of the laser have bee
4、n varied. The doping profiles of Boron and Phosphorous were measured by Secondary Ion Mass Spectroscopy (SIMS) and Spreading Resistance Probe (SRP) in order to assess the dopant activation behaviour. The main interest was the activation of the Phosphorous doped field stop layer in a depth range of 1
5、 mum. A strong dependency of Phosphorous activation on implanted Boron dose was observed with dopant activation up to 70%.This paper appears in: Advanced Thermal Processing of Semiconductors, 2007. RTP 2007. 15th International Conference on , Issue Date: 2-5 Oct. 2007 , Written by: Friedrich, Detlef
6、; Bernt, Helmut; Hanssen, Henning; Oesterlin, Peter; Schmidt, Henning INTRODUCTIONIGBTs are used for most power switching applications in the fields of e.g. automotive, industry and consumer electronics. The main advantage of IGBTs is the excellent conduction behaviour combined with high robustness,
7、 so IGBTs are widely used for all kinds of motor control purposes. Discrete IGBTs are vertical transistors, means the current flow is directed from the front side of the device to the backside IGBTs have been continuously improved over the years with special focus on reduction of switching and condu
8、ction losses. Therefore, one of the main activities is the reduction of the substrate thickness down to a blocking voltage dependent minimal value which ensures the Off-state without voltage breakdown. For a 600 V IGBT the device physical limit for the substrate thickness will be reached with the ne
9、xt generations of trench field stop IGBTs having a substrate thickness below 50m . The realization of ultra thin IGBTs with low conduction losses is possible only by implementing a field stop layer from the back side of the wafer 2 3 as shown in figure 1. Fig. 1: Schematic cross section of a 600V tr
10、ench field stop IGBT having a Phosphorous field stop and Boron emitter implanted from the wafer backside View All | Next Since homogeneously doped FZ substrates are preferred to be used instead of more expensive epitaxial wafers the field stop and emitter layers have to be created by ion implantatio
11、n using Phosphorous and Boron, respectively with low thermal budget activation. The challenge in fabrication of ultra thin fieldstop IGBTs is the handling of substrates in the thickness range of 50m after back side grinding and the limited thermal budget for back side processing in the case the fron
12、t side has been finalized before. Here, the requirement for the limited thermal budget is due to the fact that the front side metallization of the IGBT does not allow high temperature treatment for the activation of the implanted species.Therefore, Yb: YAG laser annealing at a wavelength of 515nm wi
13、th maximum energy density up to 4J/cm2 and pulse durations in the range of 350600nsec has been investigated. Sufficient dopant activation of the implanted field stop layer down to a Si-depth of 1m is required which is in accordance with the penetration depths of a laser for 515nm wavelength. Due to
14、this excellent matching between the penetration depths and the depths of the field stop layer a laser wavelength of 515nm is preferential compared to shorter wavelengths. Since the minimum activated dose required for a field stop layer is about 31012cm2 the depths of the field stop layer can be in t
15、he 1m range or even below as long the dopant concentration is in the range of 1017cm3 . The implantation of both, field stop layer and emitter with subsequent laser annealing is preferred since in this case homogeneously doped float zone substrates can be used instead of substrates with epitaxial dr
16、ift zone layers.Activation of dopants with frequency doubled pulsed solid state lasers has been proposed and pursued since several years 4 5. Especially Kudo and Wakabayashi 5 investigated dopant activation as a function of laser pulse parameters in detail. They were able to show good activation rat
17、e. However, their data were for single implants only, either Boron implants for the emitter or Phosphorous implants for the field stop layer. In this work we demonstrate that both implants can be activated simultaneously in a single process step with sufficient activation rate. o INTRODUCTION o EXPE
18、RIMENTALo RESULTS o TEMPERATURE SIMULATIONS o CONCLUSION EXPERIMENTALTo explore the potential of pulsed laser annealing at 515 nm, test anneals of plain wafers with Boron and Phosphorous dopants were performed. For the sample preparation 6 inch Float Zone (FZ) wafers have been used with a n-type Pho
19、sphorous concentration of 1014cm2 . The ion implantation for the field stop layer and emitter was carried out with a medium current implanter (Varian E220). The emitter was created by Boron implantation in the range 51014cm2 up to 11016cm2 with an energy of 30 keV. For the field stop layer a dual Ph
20、osphorous implantation was performed with a dose of 21013cm2 at an energy of 200 keV and 41013cm2 with 400 keV. The laser energy density was varied between 2.6 and 4J/cm2 , and the pulse duration of the laser was adjusted between 350 and 600 ns.The test anneals were performed with an INNOVAVENT opti
21、cal annealing system VOLCANO which was operated with a laboratory ASAMA laser emitting pulsed laser radiation at 515 nm. This wavelength has a penetration into crystalline Si of roughly 1m , and thus allows to bring laser energy into the bulk material, in contrast to UV laser radiation which is abso
22、rbed in a10 nm thin surface layer. The laboratory VOLCANO system created a laser line on the wafer with a length of 1mm with a top-hat intensity profile (within +/3% variation) and a Gaussian profile width of 40 or 80m FWHM. Figure 2 shows typical beam profiles. Fig. 2a: Long axis homogenized beam p
23、rofile of the IN NOVAVENT VOLCANO optical system Previous | View All | Next Fig. 2b: Short axis Gaussian beam profile of the INNO VAVENT VOLCANO optical system Previous | View All | Next The ASAMA laser allows variation of the pulse duration by computer control. This unique feature made it very easy
24、 to perform test anneals with different pulse durations between 250 and 650 ns. Figure 3 shows two laser pulses with 300 and 646 ns. Energy density could be varied between 1J/cm2 up to 4J/cm2 using an optical attenuator. Fig. 3: Continuously variable pulse duration of the ASAMA laser. Two different
25、laser pulses with pulse duration 300 ns (upper trace) and 646 ns(lower trace)are shown as examples Previous | View All | Next More information about the VOLCANO annealing system and the ASAMA laser can be found elsewhere 6.For the laser annealing tests, single wafers were manually aligned and placed
26、 onto a motorized high resolution xy stage which was equipped with a vacuum chuck. The laser process shutter was synchronized to the movement of the stage. A computer control system allowed to illuminate parts of the wafer with different laser beam parameters. The laser line was scanned in the direc
27、tion of its small dimension with variable velocity (variation of pulse overlap). For all tests reported here the scan speed was chosen so that the laser pulses overlapped by 90% with reference to the FWHM of the line width. Figure 4 shows the typical overlap of pulses. Several of these lines, each 1
28、 mm wide, were stitched side by side to anneal larger areas. No indication could be found for differences of the annealing results in the stitching areas. Fig. 4: Pulseoverlap of the line scan process Previous | View All | Next All test anneals were performed at room temperature on air.For character
29、ization purposes the doping profiles have been analyzed by Secondary Ion Mass Spectroscopy (SIMS) and Spreading Resistance Probe (SRP). For SIMS analysis a Cameca IMS-4f system was used with Cs+ primary ions for Phosphorous at an energy of 12,5 keV. Instead, for Boron analysis O2+ ions were used at
30、the same energy.The spreading resistance measurements were carried out by using the SSM 150 system.Also, simulations for the ion implantation have been performed as a reference by using the commercial simulator Athena from Silvaco. o INTRODUCTION o EXPERIMENTAL o RESULTSo TEMPERATURE SIMULATIONS o C
31、ONCLUSION RESULTSIt was the aim of this investigation to verify whether a single step laser anneal for a Boron/Phosphorous p-n junction gives sufficient activated dose for the n-type field stop layer. For this purpose the as implanted Phosphorous and Boron profiles have been measured by SIMS and com
32、pared by simulations using the Silvaco Athena process simulator.Boron was implanted at a high dose of 11016cm2 with an energy of 30 keV. For Phosphorous a twofold implantation at 400 keV with double charged ions and 200 keV was being applied having dose values of 41013cm2 and 21013cm2 , respectively.As depicted in figure 5 a good correlation between the SIMS and simulation profiles is obvious, especially for the Boron profile. The Phosphorous tail of the SIMS measureme
copyright@ 2008-2022 冰豆网网站版权所有
经营许可证编号:鄂ICP备2022015515号-1