• PDF / 414,071 Bytes
• 6 Pages / 414.72 x 648 pts Page_size
• 87 Downloads / 219 Views

DOWNLOAD

REPORT

---

g=2.1

9.7 GHz g=1.90

Doping

0

10 8

57 84-

0

6-

g= .00

0 2-

g=2.0050 1 3000

a)

4000

" 1 b)

Magnetic Field (G)

, . . . .I

, , , . .

2

2

4

6

4

6

100 10 Doping Ratio (ppm)

2

Figure 1: a) Echo-detected field sweeps at 20 K of B-doped microcrystalline silicon. For better overview and to avoid overlap, the sharp line at 3470 G representing the DB was cut from the bottom spectra. b) Total intensity of the broad feature in a) as a function of doping ratio.

EXPERIMENT Microcrystalline silicon samples were deposited by PECVD on Al foil from mixtures of silane and hydrogen, adding phosphine or diborane (1 to 162 ppm) for doping. For more details see [4, 6]. All doping levels mentioned in the following represent the doping ratio of phosphine or diborane to silane in the gas phase. Two pulsed ESR spectrometers were used, a home-built and a commercial spectrometer ESP380E by Bruker. The field-sweep spectra were recorded with a twopulse-echo technique. Inversion recovery curves were measured within a time window of 256 /s with the Bruker spectrometer [ 1] and over the whole dynamic range of recovery up to 48 ms with the home-built spectrometer [5]. Recovery curves were recorded with the static magnetic field at positions corresponding to g=1.998 or g=2.0052, to obtain the T1 values of the CE and the DB, respectively. The T, -values were extracted by fitting the recovery curves with a stretched exponential, accounting for a distribution of relaxation times due to structural disorder in the material [1, 5]. The maximum excitation bandwidths for the inversion recovery pulse sequences used are 7 G for the home-built spectrometer and 24 G for the ESP380E. The excitation bandwidth can be reduced by increasing the length of the inversion pulse. RESULTS Field-sweep spectra of boron-doped microcrystalline silicon Figure 1 a) shows two-pulse-echo field-sweep spectra at 20 K scaled to same peak amplitude of the broad feature for different doping levels. There is a sharp line at 4000 G of about 10 G 758

Table 1: Line parameters obtained from deconvolution of the spectra in figure 1 line No 1 2 3

g-value 2.094-2.147 2.022-2.088 1.904- 1.918

FWHM (G) 220 -290 130- 180 25-35

resonance field at 9.7 GHz (G) 3230-3310 3320-3430 3615-3640

width; the intensity is independent of doping. The broad feature seems to consist of at least three contributions. Most easily recognized are the contributions at 3300 G for 161.7 ppm and at 3630 G for 1.1 ppm. For 3.4 ppm, one can see that the peak intensity is not at these extremal field values, but in-between. The shape of the spectra suggests that this is due to a third contribution and not due to an overlap of the two lines or a shift of the contribution at g=2. 100. Asuming that there are four distinct contributions to the overall line shape (including the DB resonance), we deconvoluted the spectra (not all shown in figure 1 a) ) with Gaussian line shapes to obtain the line parameters shown in table 1. The large uncertainty of the g-values reflects the broadness of the values a