Updated: January 23, 2008
Raman cross section from silicon nanobumps



   Assume the incident laser beam is 1.0 mW with a square cross section of 1.0 μm. So the incident power density is (1 mW)/(1.0 μm)2 = 1.0×109 W/m2. Dividing by c, I get 3.3×100 J/m3. Since half the radiant energy is in the electric field this corresponds to an electric field of E = 6.1×105 V/m.
   Assume the temperature is 300 K. So the energy of a normal mode is kB T = (1.38×10-23 )(300K)= 4.14×10-21 J. The size of the oscillating object (a silicon "nanobump") is 10 nm × 10 nm × 10 nm, and has density 2329 kg/m3 So its mass is 2.329×10-21 kg. As it oscillates in a given single vibrational mode, half of its energy is kinetic energy, so the characteristic speed is
v = sqrt(2E/m) = sqrt(2(0.5)(4.14×10-21 J)/(2.329×10-21 kg) ), or 1.33 m/s.
Given the (longitudinal) speed of sound in silicon is 5840 m/s, the characteristic vibrational frequency of this object is around f = (5840 m/s)/(10 nm) = 5.8×1011 Hz. So that means that the amplitude of the vibration is on the order of d = (1.33 m/s)/(5.8×1011 Hz) = 2.3×10-12 m.
   Consider the electric dipole induced by the incident electric field. Silicon has a high dielectric constant, so the simplest approximation is to assume an induced surface charge density σ = εo E = 5.4×10-6 C/m2. This leads to an induced dipole moment of 5.4×10-30 C-m. This dipole moment is not constant, due to the thermal vibrations. Since thermal vibrations are 2.3×10-12 m, and assuming linear variation of dipole moment as the shape oscillates, the variation of the dipole moment should be approximately
(5.4×10-30 C-m) (2.3×10-12 m)/(10 nm) = 1.2×10-33 C-m.
It is this time varying part of the dipole moment that leads to the Raman scattering. The total radiated power from a dipole is:

P = μo p2 ω4 / (12 π c)

I assume the laser light is green light, with λ = 514.5 nm, so that ω = 3.7×1015 rad/s. In that case, the total power radiated from a single silicon "nanobump" is 3.6×10-20 W. I assume that a region of the sample of size 1 μm × 1 μm is imaged by a microscope into the spectrograph. The large "nanobumps" are approximately 10 nm in size and 50 nm apart. So the imaged region contains 4×102 such nanobumps. The total scattered power from all of them is 1.4×10-17 W. However, due to imaging optics aperature limitations, I would guess that only 5% of the light is received into the spectrograph. Of that, approximately 10% makes it to the CCD. This is 7.2×10-20 W. In other words, only 0.2 light photons are received per second. Assuming 50% quantum efficiency, there will be 0.1 electrons accumulated per second. So it is possible that the Raman scattered light is too faint to see.


Daniel Murray
Associate Professor
Math, Stats & Physics Unit
University of British Columbia - Okanagan
Kelowna, BC, Canada
daniel "dot" murray "at" ubc "dot" ca

For a list of related articles click here.

Hosted by www.Geocities.ws

1