Diffraction by a Slit As an application of Huygens’s wavelets we will derive the intensity distribution for diffraction by a slit. Consider a beam of light incident on a slit in a screen of width a as shown. The beam is assumed to be much wider than the slit and can be modeled as a monochromatic plane wave. The incident radiation will be assumed to be polarized in a direction parallel to the long axis of the slit, i.e. along the z-direction. At any time the wave amplitude at the location of the slit is E(slit, ) = E0 sin (k· 0 - ) =-E0 sin ( ) = -E0 z sin( ) where z is a unit vector in the z direction. For conceptual purposes, imagine the slit divided into N thin regions as shown, where N is some large integer. Each region has a width y = a/N. We will number the regions with an index n = 0, 1, 2, ... N-1. 7.1 The center of region n has coordinate yn = -a/2 + (n + ½ ) y where the origin of the y coordinate is placed at the center of the slit. We will suppose that N is so large that the wave originating in each region can be considered to be a Huygens wavelet. Ultimately this requires that we take the limit y 0. The slit width a is assumed to be very small compared to the screen-to-screen separation D: D>>a. This is called the Fraunhofer limit and the resulting diffraction pattern is called a Fraunhofer diffraction pattern. Let P be an arbitrary point on the observing screen characterized by its distance s from the midplane of the slit. In general, we think of D as measured in meters, s in centimeters (or less) and a in millimeters (or less). The intensity at P at any instant t will be proportional to E1 +E2 + E3 + ...2 where En is the wave amplitude at P at time t due to the Huygens wavelet originating at region n. All of these electric fields will point in the z-direction so we can write the instantaneous intensity as proportional to E1 +E2 + E3 + ...2 where En = En z. The wavelet from region n takes a time rn / c to get to P, where rn is the distance from the center of region n to P, so that a spherical wave arriving at P from this region has wave amplitude (7.1) En(P,t) sin [ (t - rn /c)] / rn sin (k rn - t) / rn The proportionality constant in (7.1) is determined by the rate at which energy is radiated from region n. Wider regions radiate more energy than thinner ones and a reasonable estimate for a thin region is that En y. We can then write (7.1) as 7.2 (7.2) En(P,t) sin (k rn - t) y / rn The proportionality constant in (7.2) depends only on the incident field E0 and will not be needed. The distance rn from the center of region n to P is given by rn = [D2 + (s-yn)2 ] = [D2 + s2-2syn + yn2 ] Let r [D2 + s2 ] = distance from center of slit to P. Then rn [r2-2syn + yn2 ] = r [1 - 2syn /r2 + yn2/r2 ]1/2 This can be simplified using the approximation that <<1< (1+ )p 1 + p . Here we take = - 2syn /r2 + yn2/r2 and p = ½. Then rn r [1 - syn /r2 + yn2/2r2 ] = r - syn / r + yn2/2r . Note that s /r s /D = sin( ), so that (7.3) rn r - yn sin + yn2/2r The total electric field at P at time t is given by N-1 (7.4) E(P,t) sin (k rn - t) y / rn sin (k rn - t) y / r n=0 where we have replaced rn in the denominator with r. This is permissible because 1/rn is very close to 1/r with our assumptions about the geometry. This is not true for the numerator where we expect sin (krn - t) to go through one full cycle of oscillation when rn changes by one wavelength. Since r is constant we can take it out of the summation and incorporate it into the overall proportionality constant and write, using 7.3, N-1 (7.5) E(P,t) sin (k rn - t) y n=0 = sin [k r -kyn sin( ) + kyn2/2r - t] y This is still a rather complicated summation to perform. The quadratic term in the argument of the sine function above has maximum value ka 2/2r ka2/2D = a2/ D. This will be very small when D is large. We define the Fraunhofer condition as (7.6) a2 << D. When this condition is met we can write 7.3 N-1 (7.7) E(P,t) sin [(kr - t) - kyn sin ] y n=0 Now take the limit N . In this limit yn becomes a continuous variable y and the sum becomes an integral a/2 (7.8) E(P,t) , sin[(kr - t) - kysin ] dy -a/2 y = a/2 = (k sin )-1 cos[kr - t - kysin ] y = -a/2 = (k sin )-1 { cos[kr - t - ½ kasin ] - cos[kr - t + ½ kasin ] } Define a new quantity (7.9) u ½ k a sin = a sin / so that (7.10) E(P,t) (k sin )-1 {cos(kr - t - u) - cos(kr - t + u)} Use identity (A.6) from appendix A cos ( - ) - cos( + ) = 2 sin sin (7.11) E(P,t) (k sin )-1 sin(kr - t) sin(u) sin(kr - t) sin(u) / u The instantaneous intensity is given by (7.12) I(P,t) E(P,t)2 sin2(kr - t) sin2(u) / u2 The average intensity at P is given by averaging the above expression over a full period of oscillation. Since sin2( X - t ) = ½ for any quantity X we have (7.13) I(P) = I0 sin2(u) / u2 where I0 is an overall proportionality constant. Equation (7.13) is the final result of our calculation. A frequently occurring function in optics is sinc(x) sin(x) / x, so a more compact way of writing (7.25) is I(P) = I0 sinc2(u). A plot of sinc2(u) is shown below. 7.4 sinc2(u) This intensity distribution should look familiar. The pattern has a central bright spot surrounded by secondary dimmer bright regions. The intensity zeros occur where u = ±m ( m = 1, 2, 3, ...). The angular extent of the central bright region is - /a < sin < /a. When /a is a small number, this can be written as - /a < < /a. I.e. the angular width of the central bright region is 2 /a. All of the secondary bright regions have an angular width /a. The central bright region always occurs at = 0 while the brightest part of a satellite bright region lies approximately at the location of the maximum of sin(u), i.e. approximately at the angular values sin = ±(m + ½ ) /a where m = 1, 2, 3, ... Problem: Why is the word “approximately” used in the above sentence? Refer to eq. (7.13) Problem: Two lasers of different wavelengths (632.8nm and blue light of an unknown wavelength) produce beams that fall on a beam splitter as shown. A beam splitter is a device such as a lightly silvered piece of glass that reflects a portion of the incident light falling upon it (from either side) and transmits the rest. The combined beam falls on a thin slit and produces a Fraunhofer diffraction pattern on a screen 1.6m away. A. What laser safety consideration exists for this experiment that was not present for the single-slit experiment performed in class? B. If the center of the third satellite red bright region lies at the same position as the intensity zero separating the fourth and fifth blue satellite bright regions, what is the unknown wavelength? 7.5 Diffraction by a Circular Aperture The diffraction pattern depends on the geometry of the diffracting aperture. Consider a circular aperture of diameter a cut in an absorbing screen. The screen occupies the xy plane and a parallel observing screen is located at z = D where D >> a . A point on the aperture can be specified by the circular coordinates , where x = cos , y = sin . Again we divide up the aperture into notional regions, each so small that Huygens’s principle can be applied to find the wave amplitude radiated by each region. It is convenient to divide the aperture into thin annuli of thickness and into thin segments of angular width . The area of a piece of aperture with radius lying between and + and angle between and + is A = . We proceed as in the previous section but for the sake of definiteness we now assume that the particular point P of interest on the observing screen has x-coordinate zero. There is no loss of generality in requiring this, since the entire diffraction pattern has circular symmetry. With the coordinates of P being (x, y, z) = (0, s, D) and the coordinates of the center of the nth region being ( cos , sin , 0), the distance between these two points is (7.14) rn = [D2 + ( cos )2 + (s - sin )2 ]1/2 = [D2 + 2 cos2 + s2 -2s sin + 2 sin2 ]1/2 = [D2 + s2 -2s sin + 2 1/2 ] again, define r = [D2 + s2 ]1/2 = distance from center of aperture to P, so that (7.15) rn = [r2 -2s sin + 2 ]1/2 = r [1 - 2s sin / r2 + 2 / r2] ½ r [1 - s sin / r2 + 2 /2r2 ] = r - s sin /r + 2 /2r = r - sin sin + 2 /2r where we have again written s/r = sin . The wave amplitude at P due to this region is given by an expression identical to that of (7.2) (7.16) En(P,t) sin [ (t - rn /c)] / rn sin (k rn - t) / rn The proportionality constant in (7.16) depends on the intensity of the light leaving the nth region. For small regions, this will be proportional to the area A of that region, so we have (7.17) En(P,t) sin (k rn - t) A / rn 7.6 The net wave amplitude at P is (7.18) E(P,t) sin (k rn - t) A / rn sin (k rn - t) A / r where we have again approximated rn in the denominator by r, which can be removed from the summation and absorbed into the overall proportionality constant. Taking the limit A 0, the summation in (7.18) becomes the double integral = a /2 (7.19) E(P,t) , =0 =2 , = a /2 , = =0 sin (k rn - t) d d =0 =2 , sin (k r - k sin sin + k 2 /2r - t) d d =0 where (7.15) has been used. This integral simplifies in the Fraunhofer limit a2 << D. Assuming this condition we have = a /2 E(P,t) =2 , , =0 =0 sin [(k r - t) - k sin sin ] d d Perform the integration first. Expand the sine function using identity (A.2), sin (A-B) = sin(A) cos(B) -cos(A) sin(B) (7.20) sin [(k r - t) - k sin sin ] = sin (k r - t) cos( k sin sin ) - cos (k r - t) sin(k sin sin ) The second term in (7.34) makes no contribution to E(P,t) since =2 , sin [ C sin ] d = 0 for any constant C (why is this true?) =0 so that we need only evaluate the quantity 2 sin (k r - t) ,cos( k sin sin ) d 0 This integral cannot be expressed in terms of elementary functions. It can be expressed in terms of quantities called Bessel functions. For any number x we have 2 (7.21) ,cos(x sin ) d = 2 J0 (x) 0 where J0 (x) is the Bessel function of order zero and is illustrated below. 7.7 The Bessel Function J0(x) Problem: Explain the general features of the above graph using the integral (7.21) At this point in the derivation we have a/2 (7.22) E(P,t) sin (k r - t) , J0(k sin ) d 0 The Bessel function of order one, J1(x), satisfies , J0(x) x dx = x J1(x) and is illustrated below The Bessel Function J1(x) 7.8 therefore (7.23) E(P,t) sin (k r - t) J1( ½ ka sin ) / ka sin sin (k r - t) J1( u) / u where (7.24) u = ½ ka sin = ( a/ ) sin The light intensity at P is proportional to the time-averaged value of E(P,t)2 so that finally, (7.25) I(P) = I0 [ J1( u) / u ]2 A graph of this intensity distribution is shown below [ J1 (u) / u ]2 This intensity distribution is zero at the values of u satisfying J1(u) = 0. The first few zeros of J1(u) are at u ±3.832, ±7.016, ±10.173. The central maximum of the intensity pattern falls in the angular range -3.832 ( a/ ) sin +3.832. This angular range is usually so small that it is a good approximation to replace sin with . To a good approximation, the angular width of the central diffraction maximum is therefore (7.26) 2.43 ( / a) Problem: Compare (7.26) with the corresponding result for diffraction by a slit. Diffraction is often the factor that limits the resolution of an optical system. Each point on an extended luminous object acts as a source of outgoing spherical waves and when these waves reach a distant imaging system such as a lens or aperture they are effectively plane waves. As such, they form a diffraction pattern like the one discussed in this section. 7.9 Problem: Why are we not usually aware of this fact when using lenses? Rayleigh’s criterion Consider two points P1 and P2 on an object. Suppose that ordinary incandescent light has been sent through a color filter so that the light consists of a narrow band of wavelengths around some central value . If the object is illuminated with this light and then viewed through a circular aperture, the two points will each produce a diffraction pattern and, if the points are sufficiently close together, the corresponding diffraction patterns may overlap so much that the points are indistinguishable. While there is some subjectivity involved here, a common quantitative specification is Rayleigh’s criterion, which states that the two points are just barely distinguishable if the central diffraction maximum of one pattern falls on the first zero of the diffraction pattern of the other. Using (7.26), this criterion implies that two points cannot be distinguished if their angular separations are less than about 1.2 /a. (7.40) Rayleigh’s Resolution Criterion: 1.2 /a Problem: A military spy satellite in orbit at a height of 37,000 km uses a lens to focus an image that is electronically scanned and radioed back to earth. The mission of the vehicle requires it to distinguish tanks from trucks on the ground. What minimum linear resolution is required? What minimum angular resolution is required? If yellow-green light of wavelength 560nm is used, what is the smallest diameter lens that can be used? 7.10