over corresponding nodal points. Also, first and second focal lengths are equal in ‘magnitude, and the equations, ail —_ wtp mem are valid, with symbols properly reinterpreted. Except for one negative sign, re~ quired by the sign convention, these are identical with the thin-lens equations. Example Determine the focal lengths and the pris ints for a 4-cm thick, bicon- vex lens with refractive index of 1.52 and radii of curvature of 25 cm, when the lens caps the end of a long cylinder filled with water (x = 1.33), Solution Use the equations for the thick lens in the order given: 11s = 1.33 152-1 (52 = ds2= 1.334 fi 28) 125) 1.52) 255) or fi = ~35.74 cm to the left of the first principal plane. Then (4-6) fa - (32 )c3s 74) = 47.53 cm to the right of the second principal plane, and 52 = 1.33, (52-25) 21 _— Asn 25) 475 = ~2.60 em ‘Thus the principal point M; is situated 0.715 em to the right of the left vertex of the lens, and Hy is situated 2.60 em to the left of the right vertex. (-35.74)(4) = 0.715 em 4-2 THE MATRIX METHOD ‘When the optical system consists of several clements—for example, the four or five lenses that constitute a photographic lens—we need a systematic approach that facili tates analysis. As long as we restrict our analysis to paraxial rays, this systematic ap- proach is well handled by the matrix method. We now present a treatment of image formation that employs matrices to describe changes in the height and angle of a ray as it makes its way by successive reflections and refractions through an optical sys- tem. We show that, in the paraxial approximation, changes in height and direction of a ray can be expressed by linear equations that make this matrix approach possi- ble. By combining matrices that represent individual refractions and reflections, a given optical system may be represented by a single matrix, from which the essential properties of the composite optical system may be deduced. The method lends itself to computer techniques for tracing a ray through an optical system of arbitrary com- plexity. Figure 4-4 shows the progress of a single ray through an arbitrary optical sys- tem, The ray is described at distance 2» from the first refracting surface in terms of its height yo and slope angle c relative to the optical axis. Changes in angle occur at ‘each refraction, such as at points 1 through 5, and at each reflection, such as point 6. ‘The height of the ray changes during translations between these points. We look for Sec. 42 The Matrix Method 6Figure 44 steps in taxing a ay though an opi tem, Progress of ny cn ‘be srbed by changes ins elevation and iectien a procedure that will allow us to calculate the height and slope angle of the ray at any point in the optical system, for example, at point 7, a distance xy from the mir- ror. In other words, given the input data yo, ao at point 0, we wish to predict values of yx, a at point 7 as output data 4-3 THE TRANSLATION MATRIX ‘Consider a simple translation ofthe ray in a homogeneous medium, as in Figure 4-5. Let the axial progress of the ray be Z., as shown, such that at point 1, the elevation and direction of the ray are given by “coordinates” y, and aj, respectively. Evi- dently, = a0 and yy = yo + L tan ay ‘These equations may be put into an ordered form, where the paraxial approximation tan ag = av has been used: 31 = Dye + Daw 4 = Oyo + (Daw 7) In matrix notation, the two equations are written [2 7 [3 ‘Ae] (48) Evidently, the 2 X 2 ray-transfer matrix represents the effect of the translation on the ray. The input data (yo, a) is modified by the ray-transfer matrix to yield the correct output data (yi, a1). Figure 48. Simple translation of « ny. 4-4 THE REFRACTION MATRIX Consider next the refraction of a ray at a spherical surface separating media of refrac- {ive indices n and n’, as shown in Figure 4-6. We need 0 relate the ray coordinates 66 Chap. 4 Matrix Methods in Paraxial Opticsigure 46 Refraction of a ray al a spherical surlice. (0, ««’) after refraction to those before refraction, (y, a). Since refraction occurs at a point, there is no change in elevation, and y = y’. The angle a’, on the other hand, is by inspection of Figure 4-6 Incorporating the paraxial form of Snell's law, 10 = n'0" wwe have ‘The appropriate linear equations are then y= My + Oa on le ie pb i (Ze (49) yy We use the same sign convention as designed earlier. If the surface is instead concave, R is negative. Furthermore, allowing R —» % yields the appropriate refrac- tion matrix for a plane interface. or, in matrix form, 4-5 THE REFLECTION MATRIX Finally, consider reflection at a spherical surface, illustrated in Figure 4-7. In the ‘case considered, a concave mirror, R is negative. We need to add a sign convention for the angles that describe the ray directions. Angles are considered positive for all Sec. 45 The Reflection Matrix 7igure 47 Reflection of ray at «spherical surface. The inset strates the sign convention for ray angles. rays pointing upward, either before or after a reflection; angles for rays pointing, downward are considered negative. The sign convention is summarized in the inset of Figure 4-7. From the geometry of Figure 4-7, with both « and a’ positive, y a= 04 h=04% and a =0'- 6 e o Using these relations together with the law of reflection, @ = 0", ete: rea? ae Hoe ee ee and the two desired linear equations are y= Dy + Oa (2) +e @an In matrix form, CHET] wn 4-6 THICK-LENS AND THIN-LENS MATRICES ‘We construct now a matrix that represents the action of a thick Jens on a ray of light, For generality, we assume different media on opposite sides of the lens, having re~ fractive indices n and n',, as shown in Figure 4-8. In traversing the lens, the ray un- dergoes two refractions and one translation, steps for which we have already derived matrices. Referring to Figure 4-8, where we have chosen for simplicity a lens with of curvature, we may write, symbolically, [2] =m [2] forthe first refraction [Z]=a[2"] ort wantin 68 Chap. 4 Matrix Methods in Paraxial Optics,[2] = » {>| for the second refraction ‘Telescoping these matrix equations results in [2] =m] Evidently the entire thick Jens can be represented by a matrix M = MyMyM,. Re- calling that the multiplication of matrices is associative but not commutative, the de- scending order must be maintained. The individual matrices operate on the light ray in the same order in which the corresponding optical actions infiuence the light ray as it traverses the system. Generalizing, the matrix equation representing any num- ber WN of translations, reflections, and refractions is given by [2 = MyMy 1 wm] ®] @13) e and the ray-transfer matrix representing the entire optical system is M = MyMy-s ++ MM, 14) ‘We apply this result first to the thick lens of Figure 4-8, whose index is m. and whose thickness for paraxial rays is 1. The correct approximation for @ thin lens is then made by allowing ¢ —> 0. Letting ® represent a refraction matrix and J represent a translation matrix, the matrix for the thick lens is, by Eq. (4-14), the composite matrix M = FIR, “be Es] = We simplify immediately for the case where 1 is negligible (¢ lens is surrounded by the same medium on either side (n =n’). Then eed) igure 4.8. Progress of ray through thik lens Sec. 4-6 Thick-Lens and Thin-Lens Matrices cc)‘TABLE 4-1 SUMMARY OF SOME SIMPLE RAY-TRANSFER MATRICES “Torstation matrix: [i | Refraction max, spherical interface: Refraction matrix, plane interface: “Thin-lens matrix Spherical miro 70 Chap.4 Matrix Methods Paraxial Optics‘When multiplied together, (417) ‘The matrix element in the first column, second row, may be expressed in terms of the focal length of the lens, by the lensmaker’s formula, Paes 10 Me [:: ‘ (4-18) f ‘As usual, f is taken as positive for a convex lens and negative for a concave lens. ‘This matrix together with those previously derived are summarized for quick refer- cence in Table 4-1. 4-7 SYSTEM RAY-TRANSFER MATRIX By combining appropriate individual matrices in the proper order, according to Eq. (4-14), itis possible to express any optical system by a single 2 2 matrix, which we call the system matrix. Example Find the system matrix for the thick Jens of Figure 4-8, whose matrix before ication isexprese by Ea. 4-18), and pei the hick ens exatly by 45.cm, Ro = 30 em, ¢ ='S em, m. = 1.60, and 2% 2 opi spo BS + ello ft 21° "" [a 30 10 16 Folate The elements ofthis composite raptransfer matrix, usually referred to in the symbolic form _[ae [05 describe the relevant properties of the optical system, as we will see. Be aware that the particular values of the matrix elements of a system depend on the location of the ray at input and output. In the case of the thick lens just calculated, the input plane was chosen at the left surface of the lens, and the output plane was chosen at its right surface. If éach of these planes is moved some distance from the lens, the system matrix will also include an initial and a final translation matrix incorporating Sec. 4-7 System Ray-Transfer Matrix nthese distances. The matrix elements change, and the system matrix now represents this enlarged “system.” In any case, the determinant of the system matrix has a very useful property: Det M = Ab ~ Be =" (4-19) where ny and ry are the refractive indices of the initial and final media of the optical system. The proof of this assertion follows upon noticing first that the determinant ofall the individual ray-transfer matrices in Table 4-1 have values of either n/n’ ot unity and then making use of the theorem? that the determinant of a product of ma- trices is equal to the product of the determinants. Symbolically, if M = MiMaMs - ++ My, then Det (M) = (Det M,(Det M,)(Det Ms) - + - (Det Mu) (4-20) [In forming this product, using determinants of ray-transfer matrices, all intermediate refractive indices cancel, and we are left with the ratio no/ny, as stated in Eq. 19). Most often, as in the case of the thick-lens example, no and nboth refer to air, and Det (M) is unity. The condition expressed by Eq. (4-19) is useful in checking the correctness of the calculations that produce a system mat 4-8 SIGNIFICANCE OF SYSTEM MATRIX ELEMENTS We examine now the implications that follow when each of the matrix elements is zero. In symbolic form, we have, from Eq. (4-13), [2l-[¢ sla] «an ‘hich is equivalent to the algebraic relations y= Ayo + Bao a = Oyo + Daw (4-2) 1. D = 0. Inthis case, ay = Cyo, independent of ao. Since yo is fixed, this means that all rays leaving 2 point in the input plane will have the same angle ay atthe ‘output plane, independent of their angles at input. As shown in Figure 4-9, the input plane thus coincides with the first focal plane of the optical system. 2. A = 0. This case is much like the previous one. Here yy = Bao implies that yy is independent of yo, so that all rays departing the input plane at the same angle, regardless of altitude, arrive at the same altitude yy at the output plane. ‘As shown in Figure 4-9b, the output plane thus functions as the second focal plane. 3. B = 0. Then yy = Ayo, independent of ao. Thus all rays from a point at height Jo in the input plane arrive at the same point of height yy in the output plane. ‘The points are then related as object and image points, as shown in Figure 4-9e, and the input and output planes correspond to conjugate planes for the The theorem can eatily be verified forthe rod of two matrices ad generalized by induction to the prodct of any number of matrices Frmsl proof canbe found in ay sada txtbook on ms tices tnd determinants, fr oanple, ET. Browne, Iroduction to the Theory of Determinants nd ‘Matrices (Chapel Hil: University of Noch Calin, 1958) R Chap. 4 Matrix Methods in Paraxial Optics” ” ‘ie plone pine ® Figure 49 Diagrams ilstatng the significance of the vanishing of system ma- trix elements. (a) When D = 0, the input plane coresponds to te first focal plane ‘ofthe optical system. (by When A 0, the euipt plane corresponds to the second focal plane ofthe optical system. (6) When B = 0, the output plane is the image plane conjugate to the input plane, and A is the linear magnification. (8) When C= 0, a pralel bundle of rays tthe input pane is parallel atthe ouput plane, ‘dD is the angular magnification, optical system. Furthermore, since A = yy/yo, the matrix element A repre- sents the linear magnification. aC 1. Now ay = Dav, independent of yo. This case is analogous to case 3, with directions replacing ray heights. Input rays, all of one direction, now produce parallel output rays in some other direction. Moreover, D = «/ao is the angular magnification, A system for which C = 0 is sometimes called a “telescopic system,” because a telescope admits parallel rays into its objective and outputs parallel rays for viewing from its eyepiece. Example We illustrate case 3 by an example. We place a small object at a distance of 16 cm from the left end of a long, plastic rod with a polished spherical end of radius 4 cm, as indicated in Figure 4-10. The refractive index of the plastic is {or ray-transfer matrix methods. Sec. 48 Significance of System Matrix Elements Figure 4-10 Schematic defining an example B1.50 and the object is in air. Let the unknown image be formed at the output reference plane, a distance x from the spherical cap. We desire to determine the image distance x and the lateral magnification. The system matrix consists of the product of three matrices, corresponding to (1) a translation Jy in air from object to the rod, (2) a refraction at the spherical surface, and (3) a translation 1 in plastic to the image. Remembering to take the matrices in re- verse order, we have M = T2RT1 Lex 1 oVfi i M 0 eal 1 41.50) 1.50. 2 16-5 M= -t 12 12 3 with the unknown quantity x incorporated in the matrix elements. According to this discussion, when B = 0, the output plane isthe image plane, so thatthe image distance is determined by setting 2x 1e-F=0 o x=2em Further, the linear magnification is then given by the value of element A: ‘We conclude that the image occurs 24 em inside the rod, is inverted, and has the same lateral size as the object. This illustrates how the system matrix can be used to perform the customary job of finding image locations and sizes, al- though this may usually be done more quickly by using the Gaussian image formulas derived earlier. 49 LOCATION OF CARDINAL POINTS FOR AN OPTICAL SYSTEM ” Since the properties of an optical system can be deduced from the elements of the system ray-transfer matrix, it follows that relationships must exist between the ma- trix elements, A, B. C, and D and the cardinal points of the system. In Figure 4-11, wwe generalize Figure 4-3 by defining distances locating the six cardinal points rela” tive to the input and output planes that define the limits of an optical system. The fo- cal points are located at distances f, and fs from the principal planes and at distances and q from the reference input and output planes, respectively. Further, measured from the input and output planes, the distances r and s locate the principal points and the distances v and w locate the nodal points. Distances measured to the right of their reference planes are considered positive and to the left, negative. The principal and nodal points often occur outside the optical system, that is, outside the region defined by the input and output planes. Chap. 4 Matrix Methods in Paraxial Optics