LASER DIFFRACTION

This page discusses a quantitative study of fresnel diffraction of laser radiation incident on different obstacles and apertures. We have experimentally plotted the diffration pattern of a straight edge at various distances, and of wire obstacles of various widths and distances. We also made a qualitative study of slits, double slits, circular holes and difraction from a spherical body. We then compare our results with theoretically predicted diffraction patterns using computer simulation.


Experimental Details

The beam from a 15 mW HeNe gas laser was focussed by a microscope objective, and spatially filtered by a pinhole. The undisturbed pattern several meters from the pinhole is TEM00 and almost exactly gaussian. Various kinds of obstacles are placed between the detector and the pinhole, and a motorized scanning photodetector is swept across the detector plane. Its output is amplified and sent to a chart recorder.

The laser beam was aligned parallel to an optical track (Z-axis) and at the height of a travelling photodetector. A microscope objective was placed near the output of the laser and was collimated by using a small screen on the laser head to measure the weak reflection from the lens surfaces, by tilting and displacing the lens the 'diagnostic' spots could be centered. A pinhole was placed after the lens near the focal point, this produced a much brighter kind of spot on the laser head due to reflections from the metal substrate of the pinhole, the diameter of this spot is very sensitive to focal adjustments, the focal plane was precisely located by minimizing the size of this spot. The pinhole itself was centered by adjusting the XY controls until the pattern at the detector was a bright bessel pattern. Finally, very small adjustements of the focal control and the XY controls produced a correctly filtered gaussian beam. To reduce stray light the entire optical assembly is placed in an enclosure which has been painted flat black on most interior surfaces.

A slit of width 0.2 ± 0.1mm and height 2.2 ± 0.2mm covers the photodetector, spatial detail finer than these dimensions are smoothed out. The detector scans along the X-axis and positional accuracy is maintained by guiding the detector head along a machined screw using a synchronous motor. The output of the photodetector is amplified, then printed on a chart recorder. To increase the long term temporal stability of the laser beam intensity, we allowed the laser to warm up for at least 30 minutes. Even after this precaution, high frequency intensity oscillations of about 5 to 10 Hz where observed in the undisturbed laser beam, however the output of the chart recorder does not suffer from this noise and probably filters out these higher frequencies. The output was sensitive to vibrations caused by walking, so movement was restricted during the scans.

OBSERVATIONS

The beam is linearly polarized horizontally and temporally stable as expected from a brewster window Melles Griot 05-LHP-171. The gaussian beam pattern is uniform and bright over a large field of view at the detector we found no need to set the source to obstacle distance greater than one meter as suggested by previous experiments using a narrower bessel beam. In fact a source to obstacle distance as little as 80 mm is possible without losing any detail in the pattern diffracted at larger angles.

The wavefront curvature near the pinhole changes greatly over small distances, the diffraction pattern is highly sensitive to small obstacle displacements here. Combined with the uncertainty in the pinhole to obstacle distance, this can lead to some difficulty in matching the theoretical with the experimental pattern. In the fresnel approximation all objects are treated as purely 2 dimensional in the aperture plane, however near the pinhole the wire obstacle begins to display 3 dimensional effects such as specular reflection from the sides of the wire. This effect can seen when visually inspecting the light reflected of the wire at very large angles such as 90° to the optic axis. The observer must remain in the XZ plane at Y=0 to see the effect since from a geometrical optics standpoint this is the zone of ray reflection. There is also a faint, very low spatial frequency sinusoidal pattern in this region, the pattern moves with the wire; as expected this effect is not predicted by theory.

CALIBRATION

The chart recorder X-axis scale was calibrated by dividing the distance the photodetector head travels during a certain amount of time to obtain the scan rate, then dividing the chart rate by the scan rate.
   Distance = 290 ± 0.5 mm
   Time     = 14min 31.06sec ± 0.1 sec = 871.7 ± 0.1 sec

   Error    = SQR ( (0.5/290)^2 + (0.1/871.7)^2 ) = 0.001728
   Scan Rate= Distance/Time = 0.3327 (1 ± 0.0017) mm/sec
                            = 0.3327 ± 0.0006 mm/sec

   Chart Rate 1 = 100 mm/min= 1.667 mm/sec  (As indicated on control knob)
   Chart Rate 2 = 250 mm/min= 4.167 mm/sec  (As indicated on control knob)

   Chart Recorder Scale   =  Chart Rate/Scan Rate
   Chart Recorder Scale 1 =  5.0095 ( 1 ± 0.0017 ) =  5.010 ± 0.009
   Chart Recorder Scale 2 = 12.5237 ( 1 ± 0.0017 ) = 12.52  ± 0.02
Depending on which chart speed used, positional units along the chart recorder X-axis are magnified by this factor.

GAUSSIAN OPTICS

We used plot 1, at 100 mm/min to study the undisturbed laser pattern

               Chart reading           True           Intensity
             Position   Intensity      Position
              Xc  dXc   Y    dY         X     dX      I        dI
              mm        mm             mm             Normalized

Zero         ---        15.0 ± 1        ----          ------
Peak           2 ± 4   148.0 ± 1        0.0 ± 5       1.000 ± 0.017
Minimum     -404 ± 1    16.0 ± 1      -81.0 ± 5       0.008 ± 0.001
Maximum     +418 ± 1    15.0 ± 1       83.0 ± 5       0.000 ± 0.000
Gauss waist -265 ± 1    33.0 ± 1      -53.3 ± 5       0.135 ± 0.015
Gauss waist +257 ± 1    33.0 ± 1       50.9 ± 5       0.135 ± 0.015

X  =  ( Xc ± dXc - 2 ± 4 ) / ( 5.010 ± 0.009 )        mm
dX =  X * SQR[ ((dXc+4)/Xc)^2 + (0.009/5.010)^2 ]     mm

I  = ( Y ± dY - 15.0 ± 1) / ( 148.0 ± 1 - 15.0 ± 1)   Normalized
dI = I * SQR [ (2/133.0)^2 + ( (dY+1)/(Y-15) )^2 ]    Normalized
We measured at the detector the positions of the gaussian beam waists at 13.5 percent peak intensity and from an average of both positive and negative waists we can compute the beam waist at the focal point using the equations of gaussian optics: see ref.3

  W =  (53.3 ± 5mm + 50.9 ± 5 mm) / 2 = 52.1 ± 5/SQR(2)  mm
    = 52.1 ± 4.0 mm           Average Beam waist at detector
  S =  337 ± 5mm              Position of Source on optical track scale
  P = 1809 ± 1mm + 85 ± 5mm   Position of detector on optical track scale
  Z = P-S = 1557 ± 11 mm      Source to detector distance
  Wavelength = 632.82 nm      Helium Neon laser

 W0 = Wavelength*Z/(PI*W)     Beam waist at focal plane of lens
 W0 = 6.0198( 1 ± SQR [ (11/1557)^2 + (4/52.1)^2 ])   microns
    = 6.0 ± 0.5 microns

3*W0 = 18 ± 2 microns         Focal diameter containg 99 percent of beam power
  P  = 30 ± 10 microns        Measured pinhole width
  P  = 25 microns             Indicated by label on pinhole substrate
The laser spot at the focal point is smaller than the pinhole by a reasonable margin, leading to the clean gaussian beam observed at the detector plane. If the pinhole were any smaller or not placed in the focal plane, the truncated gaussian would produce a pattern similar to a bessel function outline.

The Intensity profile of the gaussian beam is

 I(R) = I0*EXP[-2*R^2/W^2]
 R^2  = X^2 + Y^2
If the detector is not at Y=0 when it scans the X-axis across the gaussian pattern, I(R) will not change shape but just be multiplied by the constant EXP[-2*Y^2/W^2] as long as Y does not change. Thus Y-axis centering is not critical.

We sampled the undisturbed gaussian into a computer and compared it to a theoretical gaussian and the deviations were mimimal. To compensate for the angular dependance of the frequency distribution of the undisturbed laser beam, we multiplied all theoretical computations by a gaussian outline of width 52.1 ± 4.0 mm.

DIFFRACTION PATTERNS

Circular Obstacles:

We mounted two ball bearings with superglue to microcope slides. The slides were placed in the beam with the side with the ball facing away from the pinhole, a tilt of about 45° to the optics axis was introduced to prevent internal reflections reaching the detector plane.

 d =  3.0 ± 0.5 mm                      Small spherical ball bearing
 d =  4.5 ± 0.5 mm                      Large spherical ball bearing
 P = 34 - 26 = 8 ± 1 cm                 Source to obstacle distance
 S = 181 + 30 + 26 - 34 = 203 ± 1 cm    Obstacle to screen distance
Width of geometrical shadow of small ball bearing was 8cm as measured on the screen which agrees with the theoretical value of : (d/P)*(S+P)= ( 3.0 ± 0.5 / 80 ± 10 ) * ( 203 ± 1 + 8 ± 1 ) = 7.9 ± 0.2 cm The boundary of the geometrical shadow of both spheres resembles the pattern of a straight edge. When observing the small sphere, a poisson spot of less than 0.5mm was observed and disappeared for P less than 5 cm. When we replaced it by the large sphere the poisson spot disappeared for P less than 10 cm. This lower cutoff for P in both cases could indicate that at that point the surface roughness of the spheres became comparable to the width of a fresnel zone at the diameter of the sphere, thus destroying the symmetry. As we move further away from the pinhole the width of fresnel zones increases, and the poisson spot becomes visible.

The theoretical width of the spot is

R = SQR(2)*SQR [ Wavelength*S*(S+P)/2P] /(4*SQR(N))
N = Last fresnel zone intercepted by sphere

THEORY

The validity of the fresnel diffraction method is based on the expansion of the exponential in the Fresnel Kirchhoff integral formula (ref7) which gives a quantitative criterion for wavefront curvature.

  let d1 = 0.080 m,  d2 = 1.469 m,  Width = 0.00011 m

  0.5*( 1/d1 + 1/d2 )*Width^2 < < Wavelength,  for Franhofer approximation.
                         76nm < < Wavelength,
For a wavelength of 632.82 nm the wavefronts are almost plane waves but do have a certain amount of curvature, thus the Fresnel diffraction formula which is correct to second order should be quite accurate.

SIMULATION

We compared various methods of computing the fresnel integrals on a personal computer ( TESTFRES.BAS) and chose the method of ref.4 as the most accurate and numerically stable over all possible input parameters.

In the program DIFFRACT.BAS we use this numerical me thod to simulate diffraction patterns and print them to laser printer and to the screen. The programs has options to include the gaussian angular intensity modulation or to simulate the actual 2 dimensional visual laser pattern on the video screen using 64 increasing levels of luminosity, the similarities between the computed an the actual pattern are remarquable.

The program ANIMATE.BAS to compute time animations has not yet been fully converted to IBM format, however a sample of 5 animations was recorded on VHS tape.

When observing the computer simulated diffraction pattern of a slit, we increased the slit width parameter at fixed source to obstacle distance of 80mm, a pattern resembling the classical franhoffer diffraction shrinks in extent and increases in intensity until the slit width is about d=0.05 mm (at P = 80 mm) then the pattern ceases to grow in intensity and begins to widen in synchronism with the geometrical shadow. When the width was increased beyond 1.4 mm the pattern began resembling the pattern of two straight edges at a similar source to obstacle distance.

As we move a straight edge further from the source or closer to the detector during a computer animation, the spatial frequency at the detector increases, but the pattern remains almost identical in shape; theory also predicts this effect.


Theoretical parameters used in the figures

Figure   Obstacle  Source to  Obstacle to  Width     Comment
number   type      obstacle   Detector      d
                    mm          mm          mm
 1       Wire       87.0       1469.0      0.110     Compare with Plot 1
 2       Wire       87.0       1469.0      0.110     Compare with Plot 1
 3       Edge       60.0       1469.0      0.000     Compare with Plot 2
 4       Slit       88.0       1469.0      1.800     Visual comparison
 5       Slit       88.0       1469.0      1.000     Visual comparison
 6       Slit       88.0       1469.0      0.500     Visual comparison
 7       Wire       88.0       1469.0      0.300     Visual comparison

Video      ( Theoretical parameters used in the video: )
Number
 1       Wire       88.0       1469.0   0.10 to 0.90     2:00 min
 2       Wire       88.0       1469.0   0.05 to 0.22     2:00 min
 3       Wire       88.0       1469.0   0.80 to 1.70     2:00 mim
 4       Slit       88.0       1469.0   0.10 to 0.90     2:00 min
 5       Slit       88.0       1469.0   0.80 to 1.70     2:00 min
On every printed figure the undisturbed laser intensity is dotted and a thick black bar indicates the boundaries of the geometrical shadow of the slit, wire or straight edge.

When we compare Fig 1 with Plot 1, the match is almost exact except for a minor intensity kink in the node at position X = 17 mm. On Fig 2 we can clearly see that the pattern of nodes is qualitatively identical to plot 1, notice also how closely the horizontal scale matches Plot 1.

For position X > 10 mm, out of the geometrical shadow of the wire, the pattern resembles a beat wave. Beat waves are normally encountered when two simple harmonic functions with similar frequencies overlap producing a very low frequency amplitude modulation at the difference frequency.

The Babinet principle states that the optical amplitudes produced by the diffraction from two complementary apertures should sum to the orginal undisturbed wavefront. From a visual comparison of of the figures generated by the computer program it becomes obvious that for identical parameters the diffraction patterns of complementary apertures (e.g. slit and wire) are not as similar as we would expect from the Babinet principle. However once it realised that the principle applies only to the vector sum of complex AMPLITUDES it becomes obvious that an important missing parameter in all the previous plots is the PHASE information, which is lost when we take the square of the amplitude to obtain the irradiance. With the correct phase, constructive and destructive interference between complementary patterns should compensate for the mismatch in visual intensities.

CONCLUSION

The theoretical patterns produced by the program quantitatively match the experimentally obtained data. The two dimensinal simulations of the diffraction pattern on video screen bear a remarkable similarity to the actual laser pattern observed visually.

REFERENCES

  1. Klein,M.V.: 1970, 'Optics'
  2. Boyer,R. Fortin,E.: 1972, 'Intensity measurements in a fresnel diffraction Pattern', American Journal of Physics, 40, 75.
  3. Siegman, 'Introduction to Lasers and Masers'
  4. Boersma,J.: 1960, 'Computation of Fresnel Integrals', Mathematics of Computation, 14, 380.
  5. Abramowitz,M., Stegun,A.: 'Handbook of Mathematical Functions', p.321
  6. Press et al,: 1988, 'Numerical recipes in C', p.188
  7. Fowles,G.R.: 1975, 'Introduction to Modern Optics', p.110
Return to Diffraction page.