Nucleation and Growth of CVD Si Thin Films: AFM, SE and Tem Analysis

  • PDF / 1,375,090 Bytes
  • 6 Pages / 414.72 x 648 pts Page_size
  • 4 Downloads / 195 Views

DOWNLOAD

REPORT


1020 cm- 3 . Deposition times at 550'C ranged from 10 to 320 seconds corresponding to nominal thicknesses from 0.38 to 12.0 nm, based on thick film measurements. All the films were deposited on 100 nm of thermal oxide grown on silicon wafers. The starting oxide surfaces were characterized using AFM and had an RMS roughness of 0.2 nm. The AFM analysis was done in air using a tapping mode of operation [10,11]. Three regions on each sample were scanned and the data analysis was taken from lgtm x ltm areas yielding Zr, the maximum peak-to-valley height, and RMS, the standard deviation of the vertical deflections. An SEM image of the AFM tip reveals an approximately spherical shape with a diameter of 35 nm. The actual contact area will be smaller than the measured diameter. Single line scans were used to measure typical nuclei diameters and the spaces between particles. Statistical analysis of these AFM images yields average grain sizes and number of grains per square micron. Variable angle spectroscopic ellipsometry (SE) measurements yield A and T parameters at wavelengths from 250 to 800 nm. The resulting data was modeled using two-layers combined with an effective medium model [8] that includes amorphous Si and voids. The thickness and optical parameters were adjusted to minimize the errors using a regression analysis [5]. Cross-section TEM micrographs were made by thinning the samples to electron transparency. The nuclei dimensions were determined and compared with the AFM results. SE MEASUREMENTS The optical measurements of these very thin silicon films were fit by a two layer model that is indicated by the sketch in Fig. 1. The top layer consisted of a mixture of amorphous silicon (aSi) and voids (i.e. nuclei), and the optical constants of this layer were calculated from the Bruggemann effective medium approximation [5,8]. The bottom layer was a continuous amorphous silicon film. The analysis of the SE data yielded the fraction of a-Si and voids in the top layer as well as the thicknesses of both layers. The results of these measurements are shown in Fig. 1. The effective thickness of the top layer (circles) is between 1 and 5 nm. The percentage voids in this top layer is shown by the square symbols. The void fraction starts at 97% for the shortest times and then decreases to about 65-72% for times from 40 to 160 seconds. At the longest deposition time the percentage of voids decreases to 15%. The bottom layer is not required to fit the data for the first three deposition times. For times beyond 80 seconds, the bottom, continuous a-Si layer becomes thicker and significantly improves the fit to the optical data. The triangles plot the total thickness which is the sum of the top and bottom layer values. 20. 100 S---ill-r-'_..5-

80

S60

0P,

...

-

-....

JQ

% Voids

-- Sum

--.... " Top Layer ....... 0 Bottom Layer

40~ 20

Oxide 0 200

!

Silicon

I

Time (s) Fig. 1 The layer thickness and percentage voids versus deposition time as determined by SE along with a sketch of the 2-layer model SE model. 78

EPOXY