Вы находитесь на странице: 1из 22

Chapter 9 Coherence

In this chapter, we discuss coherence ( ), the correlation between phases of monchromatic radiation. Beams with random phase relationships are incoherent beams, whereas beasms with a constant phase relationship are coherent.

9-1 Fourier Analysis


For any function f(t) , of period T, we have
g a0  am cos m[ t  bm sin m[ t , f (t ) ! 2 m !1 2 t T (9.2) a0 ! 0 f (t )dt T t0 2 t T (9.3) am ! 0 f (t ) cos m[ t dt T t0 2 t T (9.4) bm ! 0 f (t ) sin m[ t dt T t0

[ ! 2T f ! 2T / T

(9.1)

For complex form, we use Eq.(9.5) & (9.6). For Fourier integral, we uses Eq.(9.7)-(9.10).
f (t )

g ([ )

1 2T

g ([ )e i[ t d[ f (t )ei[ t dt

(9.7) (9.8)

9-2 wave train with a lift time 0 and a frequency [0, say Consider aFourier Analysis ofXa Finite Harmonic
f (t ) ! e i[0t ,  X 0 / 2 t X 0 / 2 ! 0, elsewhere

Wave Train
(9.11)

The Fourier spectrum is calculated from Eq.(9.8)


g ([ ) ! 21 T

g f (t )e

 i [ 0t

dt ! 21 T

X 0 / 2 e

X0 / 2

i ([ [0 )t

dt ! ...

Use the L Hospitle rule, at


X lim g ([ ) ! 0 2T [ [0

X sin[(X 0 / 2)([  [0 )] ! 0 2T (X 0 / 2)([  [0 )


[ p [0

(9.12)

, we have (9.13) (9.14)


2 nT X0

g ([ ) ! 0 when [ ! [0 s

The resulting Fourier spectrum & power spectrum are shown in Fig.9.6. It is seen that, most energy of the wave train is carried by the waves of frequencies present in the central maximum, of width 4T / X . The smaller the lift time, the wider the central 0 maximum of Fig.9.6. Accordingly, we define the frequency band ([ ! 2T / X 0 or (f ! 1 / X 0 , centered around [0to specify the harmonic wave train.

9-2 Fig.9.5 Fourier Finite harmonic wave train f (t )

Analysis of a Finite Harmonic Wave Train

Fig.9.6 Resulting Fourier spectrum


g ([ )

& power spectrum


g ([ )
2

Clearly, there 9-3 Temporal Coherence and Lind Widththe form shown are no perfectly monochromatic sources. A real source is of in Fig.9.7.

In the figure, we see many phase discontinuities between finite wave trains. These phases changes reflect the erratic ( ) process by which excited atoms in a light source undergo transitions between energy levels, producing brief and random radiation wave trains. These temporal wave may or may not induce the temporal coherence, also named as longitudinal coherence. A given source can be characterized by an average wave train lifetime X 0, called its coherence time. Thus Eq.(9.15) means that the frequency width (f of a spectral line is inversely proportional to the coherence time of the source. The greater the coherence time, the more monochromatic the source. The coherence length l , of a wave train is the t length of its coherence pulse, or (9.16) lt ! cX 0 Combining Eqs.(9.15) & (9.16), the coherence length is
lt ! c / (f

9-3 Temporal Coherence and Lind Width The frequency band (f can be related to the line width (P by taking the differential of the relation f ! c / P , say (f !  (c / P2 )(P . Then lt } P2 / (P (9.17) (9.18) (P } P2 / lt Eq.(9.18) is called the uncertainty principle, lt (P }P2 , say a wave pulse can be used to represent the location of an electron such that one cannot precisely indentify both (P & lt simultaneously. Since the line width of spectral sources can be measured, average coherence times and coherent lengths may be surmised ( ). White light has a line width (P of 300nm, extending roughly from 400 to 700 nm. Taking the average wavelength at 550 nm, Eq.(9.17) gives I t ! [(550) 2 z 300]nm } 1000nm } 2P , a very small coherence length indeed, of around a millionth of a centimeter or two wavelengths of white light. Understandably, interference fringes by white light are difficult to obtain since the difference in the path lengths of the interfering beams should not be greater than the coherence length for the light. Sodium or mercury gas-discharge lamp sources are far more monochromatic and coherent. For example, green line of mercury at 546 nm may have a line width of around 0.025nm, giving a coherence length of 1.2 cm. One of the most monochromatic gas-discharge sources is a gas of the krypton 86 (Kr 86) isotope ( ), whose orange emission line at 606 nm has a line width of only 0.00046nm. The coherence length of this radiation is 78cm! (it is used to define a length unit because of this special coherence length!)

9-3 far surpassed Laser radiation has Temporal (Coherencethe coherence Width ) even and Lind of this gas-discharge source. For example, the short-term stability of commercially available CO2 laseres is such that line widths of around 1 X 10-5 nm are attainable at the infrared emission wavelength of 10.6 Q m. These numbers give a coherence length of 11km!. Under carefully control conditions, He-Ne lasers can improve this figure by another order of magnitude. Note that, the common He-Ne laser may not have coherence lengths much greater than its cavity length, due to random temperature fluctuation and mirror vibrations (that change the cavity length randomly multiple mode oscillations a bad effect upon the coherence length).

9-4 Partial Coherence As pointed out previously, when the phase difference between two waves is constant, they are mutually coherent waves. In practice, this condition is only approximately met, and we speak of partial coherence. Fig.9.8 shows a general situation where the T interference is produced at P between two beams. Consider E & T being superposed 1P T E2 ( )at point P in terms of the field E at the source S. Moreover, wePconsider the case in S T T T which both , so E& E2maintain the same polarization as E S that we can represents 1P P the fields by scalar function. 1 E S (t ) ! E (t )  E * (t ) ! Re( E (t )) For convenience, we consider (9.19) 2 where E (t ) ! E0e i[t eiJ (t ) (9.20) Here, J (t )models the departure form monochromaticity of the source field. Similarly, (9.21) E1P (t ) ! E1 (t )  E1 * (t ) / 2 ! e( E1 (t )) (9.22) E2 P (t ) ! E2 (t )  E2 * (t ) / 2 ! e( E2 (t )) The relation between source E (t ) and field E1 (t )& E2 (t )are shown below. (9.23) E1 (t ) ! F1E (t  T1 ) ! F1 E0 e  i[ (t T1 ) eiJ (t T1 )
E2 (t ) ! F 2 E (t  T2 ) ! F1E0 e i[ (t T2 ) eiJ (t T2 ) Here F & F are constant factors due to splitting and reflection & transmission before 1 2

they reach pt. P.

Fig.9.8 Interference at P due to waves from S traveling different paths. The waves are redirected at S1 & S2, including reflection, refraction, and diffraction ( ).

T1 & T2 are the time of flight for the light fields propagating along two paths, respectively.

9-4 Partial Coherence

With Eq.(9.23), we have I P ! I 0c E1P  E2 P 2 ! I 0c E1P 2  E2 P 2  2 E1P E2 P


I c * * * * ! I1 p  I 2 p  0 E1E2  E1 E2  E1E2  E1 E2 2 2 . I 2 P ! I 0 c E2 P

(9.24)

2 where I1P ! I 0c E1P The last term of Eq.(9.24) is the interference terms since its value determines whether the irradiance at P is more than, less than or equal to the sum of the irradiances of the field being superposed. Not that, due to the time average of sine & cosine factors that oscillate * * E1E 2 ! 0 & E1 E2 ! 0 at 2[ t * * * Then Eq.(9.24) becomes I P ! I1 p  I 2 p  I 0c E1E2  E1 E2 ! I1 p  I 2 p  I 0c E1E2

And I P ! I1 p  I 2 p  I 0c 1 2 E (t  T1 ) E2 (t  T2 ) with the assumption of real F1 & F 2 Performing the time shifting so that T1 is considered as the origin and let X ! T1  T2. *
I P ! I1 p  I 2 p  I 0 cF1F 2 Re E (t ) E (t  (T1  T2 )) ! I1 p  I 2 p  I 0 cF1F 2 Re E (t ) E * (t  X )

The remaining time average has the form of a correlation function, say (9.25) +(X ) ! E (t ) E * (t  X ) This correlation function, which determines the size of the interference, depends on the amount of correlation that exist in the value of the source field at two different times.

9-4 Partial Coherence The normalized correlation function, or correlation coefficient (or degree of coherence) is
I cF F K (X ) ! 0 1 2 2 I cF F ! 0 1 2 2 I1P I 2 P +(X ) e[K (X )] E (t ) E * (t  X ) I1P I 2 P

I P ! I1P  I 2 P  2 I1P I 2 P

(9.26 & 9.27)

The coefficientK (X )is the heart of the interference term and is a function of X and therefore of the location of point P. Now we see that the time difference between two paths, relative to the average coherence time X 0 of the source, is crucial to the degree of coherence achieved. If, X " X 0 , some coherence between the two beams will be lost! Assume that X 0 is a constant coherence time rather than an average. Such a wave train is as follows (Fig.9-9a) that has regular discontinuity in phase, separated by the time interval X 0. Now use the relations of Eq.(7.10) & (7.11)
I c I1P ! 0 F1E0 2 2 I c I 2 P ! 0 F 2 E0 2 2

9-4 Use Eq.(9.20)&(9.26) to gives,

I cF F K (X ) ! 0 1 2 2

Partial Coherencet ) E0ei[ (t X )e iJ (t X ) E0e i[ t eiJ1(


2 (I 0c / 2) 2 ( F1F 2 E0 )

where 1 T (9.28) ei[N (t )N (t X )] ! ei[N (t )N (t X )]dt 0 T Where T is a sufficiently long time. The function e i[N (t )N (t X )] is pictured in Fig.9.9b and is seen to be a series of regularly spaced rectangular pulses with random magnitude falling between , where the  2T . Consider the first coherence time interval 2T 0, 0 t X 0 X impulse may be expressed by
N (t )  N (t  X ) ! H1 , X 0  X t X0

i[X i[N (t )N (t X ) e After simplification, we have K (X ) ! e

In the successive intervals, the expression is similar, except for the value of H1. The normalize coherence function for a large number of N of intervals is
K (X ) ! e i[X 1 X 0 X i (0) X X e dt  0 eiH1 dt  similar terms for rest N  1intervals X 0 X NX 0 0 (X 0  X  XeiH1 )  (X 0  X  XeiH 2 )  ...

ei[X ! NX 0 ei[X ! NX 0

A
(9.29)

N N (X 0  X )  X eiH j ! (1  X )ei[X X0 j !1

9-4 Partial Coherence

Re[K (X )] ! (1 

Take on a maximum value of 1 when X ! 0 (equal path length), a value of 0 when X ! X 0 (path difference = coherence length), and values between 0 and 1 for 0 X X 0 . X | K (X ) |! (1  ) Now (9.31) X0 K (X ) !| K (X ) | e i[X Or (9.32), (9.33)
Re(K (X )) !| K (X ) | cos [X

X ) cos [X X0

9-4 Partial Coherence (9.30)

The value of | K (X ) | is plotted in Fig.9.10 Use the empirical expression of visibility,


V ! ( I max  I min ) /( I max  I min )

Fig.9.10

(9.34)

Finally, we have the following special cases: 1. complete incoherence: X X 0 K ! 0, I p ! I1  I 2 ,

2. Complete coherence: X ! 0 & K ! 1

! 2 I 0 , or equal beams

! (2 I 0  2 I 0 ) /(4 I 0 ) ! 0 I p ! I1  I 2  2 I1 I 2 cos [X I max ! 4 I 0 , for equal beams I min ! 0, for equal beams V ! ( I max  I min ) /(4 I 0 ) ! 1

X 0 Partial 3. partial coherence: 0 X 9-4 1 " K " 0

I p ! I1  I 2  2 I1 I 2 e(K )

Coherence

I P ! 2 I 0 [1  e(K )] , or equal beams I max ! 2 I 0 (1 K ), I min ! 2 I 0 (1 K ), or equal beams ! 4 I 0 K /(4 I 0 ) ! K

9-5 Spatial Coherence Now we consider the spatial, or lateral coherence: the correlation in phase between spatially distinct points of the radiation field. This type of coherence is important when using a wavefront-splitting device, such as a double slit. The quality of the interference pattern in the double-slit experiment depends on the degree of coherence between distinct regions of the wave field at the two slits. Consider that depicted in Fig.9-11 where light from a source S passes through a double-slit and is also sampled by a Michelson interferometer located nearby.

Spatial coherence between 9-5 Spatial Coherence is insured as long as the wavefront points A & B at the slits source S is a true point source. In that case, all rays emanating form S are associated with a single set of spherical waves that have the same phase on any given wavefront. A clear distinguishable fringes then formed on a screen near P1 if the optical paths SAP1 & SBP1 is both temporally & spatially coherence. Temporally coherence requires that the optical path difference ( ! SAP  SBP2 is shorter 1 than the coherence length of the radiation lt . This is equivalent to a comparison of coherence along any fixed radial direction of light propagation form the source at two wavefronts separated by the same path difference (where the temporal coherence can be measured by the Michelson interferometer). If ( lt , clear interference fringes are formed at P1; If ( u lt , interference is poor or even absent. In practice, S is always an extended source, so that rays reach A & B from many points of source. For a non-laser light sources, lights emitted from different sources are not correlated so the coherence may be poor. Thus, the spatial coherence of light at slits A & B depends on how closely the source S resembles a point source of light. For the spatially coherence, consider Fig.9.11, we need we need a spatial coherence dimension l s , that distance s between source S1 & S2 should satisfy s ! ls P / U (9.35)

9-5 Spatial Coherence According to these result, we conclude that there exists at any point in the radiation field of a real light source a region of space in which the light is coherent. This region has lateral lt dimensions of and longitudinal dimensions of relative to the source and thus ls 2 occupies a volume of l s lt around the point P. It is from this volume that any interferometer must accept radiation if it is to produce observable interference fringes. 9.6 Spatial Coherence Width With Fig.9.13, consider the spatial coherence at pts. P1 & P2 in the radiation field of a quasi-monochromatic extended source, simply represented by two mutually incoherent emitting pts. A & B at the edges of the source. P1 & P2 may be considered as two slits that propagate light to a screen (including P), where interference fringes may be viewed. Fig.9.13

9-5 Spatial Coherence Each point source, acting alone, produces a set of double-slit interference fringes on the screen. When both sources act together, the fringe systems overlap: 1. If fringes systems overlap with their maximum & minimum falling together enhancement the radiation from the two incoherent sources is considered highly coherent. 2. If fringes system are relatively displaced, with the maximum of one falling on the minimum of the other not visible the radiation from the two incoherent sources is considered incoherent. 3. If the distance between A & B is zero, the results is just the fringes of a single point source. A maximum in the interference pattern occurs at P if P lies on the perpendicular bisector of the two slits. Now BP2 BP1 = AP2 AP1 = 0 If B is moved below A, the fringe systems separate until, at a certain distance s, where
BP2  BP ! ( ! P / 2 1 The maximum in the fringe system at P due to source B is replaced by a minimum, and the composite fringe pattern disappears. If the angle U represents the angular separation of the sources from the plane of the slits, then ( } N where N the distance between the slits, and U } s / r, where r is the distance is U to the source.

It follows ( ! P / 2 ! sNr 9-5!Spatial Coherence (9.36) / or s rP /(2N ) When the distance AB is considered instead to be a continuous array of point sources, the individual fringe systems do not give complete cancellation until the spatial extend AB of the source reaches twice the value of s in Eq.(9.36). If extreme points are separated by an amount s rP / N, then fringe definition is assured. Regarding this result as describing instead the maximum slit separation N given a source , dimension s , we have for the spatial coherence width N, say s N rP / s } P / U (9.37) s As Nis restricted to smaller fractions of this value, the fringe contrast is correspondingly s improved. According to this argument, moving the source B farther should bring the fringe system into coincidence again, so that the degree of coherence | K | between P1 & P2 is a periodic function. In ideal situation, the extended source is represented by a continuous array of elemental emitting areas rather than by two point sources. Results show that outside the coherence width given by Eq.(9.37), the fringe visibility, while oscillatory, is negligible. The significance of Eq.(9.37) is apparent in the case of Young s double-slit experiment, where an extended source is used together with a single slit to render the light striking the double slit reasonably coherent, as in Fig.9.14.

9-6 Spatial Coherence Width In Fig.9.14, we may use Eq.(9.37) to determine how small the single slit ( s of AB, Eq.(9.37)) must be to ensure the coherence and the production of fringes at the screen. The two slits S1 & S2 must fall within the lateral coherence width N due to the primary slit s of width . s
Fig.9.14

9-6 Spatial 1.1 mm in width & that the Suppose the source slit is made exactlyCoherence Width separation abetween two slits S1 & S2 is adjustable. If a N , they fall within a high coherence region and the fringes in the interference s pattern appear sharply defined. As two slits move apart, the degree of coherence | K | decreases and the fringe contrast begins to degrade. When the slit separation a reaches a value of 0.1mm, | K |! 0 & the fringes disappear. Evidently an experimental determination of this slit separation could be used to deduce the size s of the extended source. (This technique was employed by Michelson to measure the angular diameter of stars.) Consider a star as an extended incoherent source with light emanating from a continuous array of points extending across a diameter s of the star (Fig.9.15), then the spatial coherence widthN in Eq.(9.37) becomes s N 1.22P / U s (9.38) Where the factor 1.22 arises from the circular shape of the sources, as it does in the Fraunhofer diffraction (Chapter 11) of a circular aperture. Since the angular diameter U of a star is extremely small, N will be correspondingly large. s The movable split were therefore be designed as in Fig.9.15a, using mirrors that direct widely separated portions of the radiation wavefront into a double-slit-telescope.

9-6 Spatial Coherence Width The spacing of the interference fringes depends on the double-slit separation a , whereas s their visibility depends on the separation N . As Nincreased, the fringes disappear when s equality in Eq.(9.38) is satisfied.

9-6 Spatial Coherence Width

Homeworks: 9.6, 9.8, 9.17

Вам также может понравиться