Introduction Carbon in silicon crystal affects the performance of power devices for the hybrid cars [1]. Measurement of (substitutional) carbon concentration [C_s] is usually done by infrared absorption spectroscopy (IR). We have established the measurement procedures for the single crystal silicon (C-Si) and realized the detection limit of 1 x 10¹³ atoms cm^-3 and quantification limit of 1 x 10¹⁴ cm^-3 for C-Si at room temperature [2]. The key technologies are (1) “synthetic reference” made by electron irradiation [3, 4], (2) 3 point baseline between, for example, 595 and 615 cm^-1, (3) “spectrum processing” to cancel the interfering fractional virtual phonon absorption bands [5] and (4) IR machine calibration by measuring samples who's concentration was determined by SIMS. Fractional phonon bands were found to be corresponding to the inflection points of the finite difference of the phonon absorption band [2]. Generally, [C_s] measurement of semiconductor grade polycrystalline silicon (P-Si) is done after growing FZ crystal (poly-FZ method) because it is difficult to manage with phonon absorption in P-Si. Many polysilicon companies use the 2nd generation IR. As [C_s] has reduced, however, carbon contamination during FZ process has become relatively serious. Here we apply the 2nd generation IR to two applications, the measurement at low temperature (LT) and measurement of P-Si. Measurement of P-Si has been reported earlier [6, 7], but detailed information about the phonon absorption in P-Si is necessary.

Experimental The samples were both poly-FZ single crystal and polycrystalline semiconductor grade silicon. The typical thickness was 3 mm for the low temperature measurement and 2 mm for the room temperature measurement. Some of the polycrystalline samples were annealed. The sample surface was mirror finished. The nominally carbon-free reference was prepared by the electron irradiation. IR measurement was done either at room temperature or liquid nitrogen temperature at wavenumber resolution of 2 and 1 cm^-1, respectively. Peak position and width of virtual phonon fragments were obtained and compared to those in the spectra of C-Si measured at room temperature. They were cancelled in the spectra.

Results and discussion Low temperature measurement Fig. 1 shows the relation of the peak absorbances of poly-FZ samples measured at room temperature and low temperature. Linear relation was observed for [C_s] down to about 1 x 10¹⁴ cm^-3. In measurement at room temperature, the virtual inner phonon bands located between 595 and 615 cm^-1 overlap the carbon absorption peak at 605 cm^-1 and change the peak absorbance. At low temperature, the phonon peak absorbance is 1/2 and carbon peak height is 2 times high so that the phonon band interference is reduced. It is easier to cancel them for the low temperature measurement than RT. We have already confirmed such a relation down to about 1 x 10¹³ cm^-3. The problem is that the peak height is dependent on the machine and actual wavenumber resolution. Therefore, the calibration using the samples who’s [C_s] was measured by either CPAA [4] or SIMS [2] is indispensable. Also, nominally carbon-free reference fabricated by electron irradiation is indispensable [2].

Polycrystal measurement Basic information necessary for the measurement includes, full width at half maximum (FWHM), reference sample, baseline procedure (BL) and conversion coefficient k [8]. They were examined in detail and established for single crystal (C-Si), but not supplied for P-Si. Here these basic factors were examined. Virtual fractional phonon bands were examined by comparing to them in C-Si. Their location was basically at 591, 600, 602 and 610 cm^-1, corresponding to the maxima of finite difference also but a little different to that in C-Si. Measurement of polycrystal in 10¹⁴ cm^-3 range is reported performed for the first time.

[1] T. Sugiyama et al., Proc. 17^th Symp. Power Semiconductor Devices & ICs, 243 (2004). [2] N. Inoue, phys. stat. sol. c, (2016), DOI 10.1002/pssc.201600068. [3] N. Inoue et al., Proc. Forum Si Materials, 217 (JSPS, 2007). [4] N. Inou et. al., High Purity Silicon 2014, 19. [5] N. Inoue, M. Nakatsu, Solid State Phenomena 108-109, 621 (2005). [6] l. L. Hwang et al., J. Electrochem. Soc., 138, 576 (1991). [7] M. Porrini et al., Solid State Phenomena, 108-109, 591 (2016). [8] N. Inoue et al., Emerging Semiconductor Technology, STP 960, 365 (ASTM, 1987).