Scribd
Upload
347 views46 pages

Yariv Yeh Chapter 3

optical modes in circular waceguide

Uploaded by

Pavitra pathak
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
347 views46 pages

Yariv Yeh Chapter 3

optical modes in circular waceguide

Uploaded by

Pavitra pathak
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
CHAPTER GUIDED WAVES IN DIELECTRIC SLABS AND FIBERS 3.0 INTRODUCTION So far. we have discussed wave propagation in free space. including the propagation of plane waves and beams. As a result of diffraction, a beam of light with a finite cross section will spread as it propagates in free space. Lenses or gradient index media may be employed at appropriate locations to focus the beams. Generally, dielectric media of high refractive index can be employed to confine the propagation of the beam. In this chapter, we will show that both dielectric slabs and circular fibers can support confined electromagnetic propagation. These modes of propagation are the so-called guided waves (or guided modes), and the structures that support guided waves are called waveguides. In this chapter, we first discuss the propagation of guided waves in dielectric slabs. As we know, any light beam with a finite transverse dimen- sion will diverge as it propagates in a homogeneous medium. This divergence disappears in guiding dielectric structures under the appropriate conditions. In dielectric waveguides, the transverse dimension of these modes of propagation is determined by the dielectric waveguide. We shall derive first the properties of guided modes in a dielectric slab structure. Optical modes are presented as the solution of the eigenvalue equation, which is derived from Maxwell’s equations subject to the boundary conditions imposed by waveguide geometry. Both transverse electric (TE) and transverse magnetic (TM) modes of propagation are derived. The physics of confined propagation is explained in terms of the total internal reflection of plane waves from the dielectric interfaces. After discussion of the slab wave- guides, we will cover the important subject of guided waves in circular fibers. We will then introduce the simple theory of effective index, which is useful in understanding waveguiding in two-dimensional structures. In the last part of the chapter, we will take up the subject of signal corruption due to chromatic dispersion in optical fibers, and the signal attenuation due to scattering and absorption. The subject of coupling between waveguides and the method of dispersion compensation in fiber transmission will be discussed later in this book 3.1. TE AND TM CONFINED MODES IN SYMMETRIC SLAB WAVEGUIDES Dielectric slabs are the simplest optical waveguides. Figure 3.1 shows a typical example of a slab waveguide, It consists of a thin dielectric layer (called the guiding layer, or simply the core) 110 3.1. TE and TM Confined Modes in Symmetric Slab Wavegt Figure 3.1. Schem metric slab waveguide. The waveguide consists of a guiding layer of thickness d with a retractive index 1p, surrounded by media of refractive index n, ~ ky= o/c sandwiched between two semi-infinite bounding media (clad). Generally, the index of refrac- tion of the guiding layer must be greater than those of the surrounding media, In addition, the thickness of the guiding layer is typically on the order of a wavelength. In symmetric slab ‘waveguides, the two bounding media are identical. The simplest example will be a thin glass film (or layer) immersed in air or another fluid (or solid) of a lower index of refraction. ‘The following equation describes the index profile of a symmetric dielectric waveguide: . fe I 44. In other words, the propa- gation constant B of a confined mode must be such that p> me G.1-5) where we recall that n, is the index of refraction of the bounding media. On the other hand, the continuity of the field requires that the magnitude of the field E,,(x) attain a maximum, value. The existence of a maximum requires that the Laplacian of the field be negative. In other words, the propagation constant of a confined mode must be such that 20 Be G16 Thus we will find confined modes whose propagation constant satisfies these conditions, Equations (3.1-5) and (3.1-6). The modes can also be classified as either TE or TM modes. ‘The TE modes have their electric field perpendicular to the xz plane (plane of incidence, or plane of propagation) and thus have only the field components £,, H,, and H,. The TM ‘modes have the field components H,, E,, and E.. Guided TE Modes The electric field amplitude of the guided TE modes can be written in the form Ey, Eq(x) expli(ot ~ B2)] G7) Ina manner very similar to the wavefunction of a particle in a square-well potential, the mode function E,,(x) is taken as ‘Asin hx + B cos hx, [xl< 4d C expl-qr), x>dd (G18) Dexplqr), x<-td E,(0) where A, B, C, and D are constants, and the parameters h and q are related to the propagation (ei “Peay The parameter h may be considered as the transverse component of the wavevector in the guiding layer. To be acceptable solutions, the tangential component of the electric and magnetic G19) 3.1 TE and TM Confined Modes in Symmetric Slab Waveguides 113 fields, E, and H,, must be continuous at the interfaces. Since H, = (i/op)(QE,/2x), we must match the magnitude as well as the slope of the TE mode functions E,,(x) at the interfaces. This leads to A sin($hd) + B cos(4hd) = C exp(-$ad) A cos(-Fhd) ~ hB sin( Ehd) = “A sin( hid) + B cos(-}hd) = D expl—+4qd) A cos(4hd) + hB sin(+ hd) = gD exp(-had) from which we obtain 2A sin(4hd) = (C - D) exp(-4.qd) G.1-10) INA cos( $d) = -q(C ~ D) expl—$4d) Gut 2B cost Lhd) = (C + D) exp(-4.qd) 1-12) 2hB sin( thd) = q(C + D) exp(—5qd) G.1-13) By examining the above equations, we find there are two sets of solutions. (a) Symmetric modes (A = 0 and C = D): Equations (3.1-12) and (3.1-13) yield Atan($hd)=q (for symmetric TE modes) (3.1-14) (b) Antisymmetric modes (B = 0 and C = —D): Equations (3.1-10) and (3.1-11) give ‘hcot($hd)=—q (for antisymmetric TE modes) G.1-15) Note that both Equations (3.1-14) and (3.1-15) cannot be satisfied simultaneously since the elimination of q would lead to a pure imaginary h and a negative q. However, these two equations can be combined into a single equation (see Problem 3.14): (3.1-16) —¢ The solutions of TE modes may thus be divided into two classes. For the first class, | C=D, tan hd) =q Gu-17) and for the second class, C=-D, heot($hd) =-9 G.1-18) Note that the solutions in the first class have symmetric wavefun¢ 1s, whereas those of the second class have antisymmetric wavefunctions. The propagation constants of the TE modes are found from a numerical or graphical solution of Equations (3.1-17) and (3.1-18), with the definition of h and q given by Equa- tion (3.1-9). A very simple and well-known graphic solution is described here, since it clearly shows the way in which the number of TE modes depends on both the thickness d and the difference of indices of refraction. By putting u= $/d and v = 44d, Equation (3.1-17) becomes tan w= v, with G.1-19) 114 Chapter 3. Guided Waves in Dielectric Slabs and Fibers Figure 3.2 Graphic solution of Equations (3.1-17) and (3.1-18) for three values of V. Solid curves are v = wtanu, and the dotted curves are v = -1 cot u. ice u and v are restricted to positive values, the propagation constants may be found in this case from the intersection of both the curve v = w tan w and a circle of known radius, V = (n} - n})!?(nd/A) in the first quadrant of the uv plane. A similar graphic construction for the solution of Equation (3.1-18) can be obtained by plotting v = -w cot w and the circle on the wv plane. Figure 3.2 shows such a graphic method for three values of V. For V = 1.2, there is only one solution—the TE, mode. There are two solutions (TE, and TE,) when V = 2.5 and four solutions when V = 5. Note that the number of solutions depends on the value of V. From Figure 3.2, it is clear that the number of confined TE modes depends on the magnitude of the parameter V. For V between zero and 47, there is just one TE mode of the first class. The first mode of the second class appears when the parameter V is greater than +n. As this parameter V increases, confined modes appear successively, first of one class and then of the other. Figure 3.3 plots the wavefunctions of a symmetrical slab waveguide with my = 1.6, n, = 1.5, d= 5pm, and 4 = 1.55 um. According to Equation (3.1-19), the parameter V = 5.64. This waveguide supports four TE modes. It is not difficult to see from Figure 3.3 that, when ordered according to the propagation constant f, the mth wavefunction has m— 1 nodes. We also notice that the wavefunctions are either symmetric or antisymmetric with respect to the origin x = 0. It follows from the discussion earlier that the wavefunctions are divided into two classes (see Equations (3.1-17) and (3.1-18)). This division is a direct, consequence of the fact that the index profile n(x) is symmetric about x Knowledge that the solution possesses a definite symmetry sometimes simplifies the determination of the propagation constant, since we need only find the solution for positive x. Even solutions have zero slope and odd solutions have zero value at the origin x = 0. Thus the wavefunction of the even solutions can be written as cos(/ix), whereas those of the odd solutions can be written as sin(/x). Both types of solutions decay exponentially in the region |x| > 4d. The solutions are then obtained by matching the value and the slope at |x| = $d. 3.1. TE and TM Confined Modes in Symmetric Slab Waveguides 115 TE, ° 10 ° 10 Figure 3.3 Wavefunctions of a symmetrical slab waveguide with n= 1.6 and n, = 1.5. The thickness of the core is equal to d= 5 Hm, and 4 = 1.55 jum. The normalized propagation constants are 1.5946, 1.5785, 1.5521, and 1.5175. The fundamental mode has the largest propagation constant. The confine- ‘ment factors (fraction of energy inside the core) for the modes are T= 0.9914, 0.9631, 0.9033 and 0.7511. The fundamental mode has the highest confinement, For the purpose of describing and comparing the confined modes, it is convenient to define the normalized propagation constant as po (1-20) oye Such a normalized propagation is often called the effective index of refraction of the mode, ‘ny and is related to the phase velocity of the mode: B12 where 1, is the phase velocity of the mode, v, = «/B. Thus, for confined modes, the normal- ized propagation constant B or the effective index nis between n, and 7). Guided TM Modes We now consider the TM modes whose magnetic field vector is perpendicular to the plane of propagation (xz plane). The derivation of the confined TM modes is similar in principle to that, of the TE modes. The field amplitudes are written Hy (x, 2) = Hf) expli(or~ Be) minor on a: EQ) 1-22) Ex, 2, 116 Chapter 3. Guided Waves in Dielectric Slabs and Fibers The wavefunction H,,(1) is (A sin hx + Boos hx, [xl< 4d Hy(2) =4C expl-qn), xodd 3.123) D expla). x<-td where A, B, C, and D are constants, and the parameters h and q are given by Equation (3.1-9). The continuity of H, and E, at the two interfaces x = +4d leads, in a manner similar to Equations (3.1-14) and (3.1-15), to the following eigenvalue equation: 2 hian(thd)=“2g — for even solutions nj > G24 Weta) =" 4. foradd soars These two equations can also be combined into a single equation, tan(ind) ~ ee 1.25, where ne G.1-26) Equation (3.1-24) can also be solved by using the graphic method described earl Figure 3.4 shows the dispersion relation (effective index neg versus normalized frequency V) of a typical symmetric waveguide. The frequency at which q = 0 is a cutoff frequency. For a mode with q = 0, the field is no longer exponentially decaying in the cladding region and the propagation is no longer confined. Referring to Figure 3.4, we note that TE, and TM, modes have no cutoff frequency. In other words, these two modes are always confined in a symmetric waveguide. V = n/2 is 19 18 a 7 16 | 18 Z 0 1 2 3 4 5 6 Figure 3.4 The effe with n= 1.5 and n, = 2.0. index ng Versus normalized frequency V of a typical symmetric waveguide 3.1 TE and TM Confined Modes in Symmetric Slab Waveguides 117 the cutoff frequency for TE, and TM, modes. For frequency in the range V < m/2, the wave- ‘guide can only support TE, and TMg modes. At a higher frequency in the range m/2 1,00/c, it follows directly from Equation (3.1-4) that (1/E)(0E/2x°) > 0 everywhere, 3.2. TE and TM Confined Modes in Asymmetric Slab Waveguides 119 Figure 3.6 Schematic drawing of an asymmetric slab waveguide with a core mdex ot my. and E(2) is exponential at all three regions of the waveguides. If we take E(x) = exp(-ga), which decays to zero at x = +e2, because of the need to match both E(x) and its derivatives at the two interfaces, the resulting field distribution is infinite at x=—se, as shown in Figure 3.7a. Such a solution would correspond to a field of infinite energy. It is not physically realizable and thus does not correspond to a real wave. For n,(c/c) < B < n,(w/e), as in Figures 3.7b and 3.7c, it follows from Equation (3.1-4) that the solution is sinusoidal in the core (-r 0. Two such solutions are shown in Figures 3.7b and 3.7c. The energy carried by these modes, as represented by the Poynting vector, is confined to the vicinity of the guiding layer, and, consequently, we will refer to them as confined or guided modes. In fact, only a small fraction of the energy is flowing out- side the guiding layer. From the preceding discussion, it follows that a necessary condition, for their existence is that (o/c), n,(«/c) < B < ns(«o/c), $0 that confined modes are possible only when ny > m,, ny; that is, the inner layer (core) possesses the highest index of refraction. In some basic sense, the confined modes in this regime are reminiscent of quantized states of an electron in a potential well, in which the electron is trapped by the potential well. Mode solutions for n,(«o/c) < B < n,(co/c) (regime (d) in Figure 3.7) correspond, according to Equation (3.1-4), to exponential behavior in the region x > 0 and to sinusoidal behavior in the regions x < 0, as illustrated in Figure 3.74, In this regime, almost all the energy is flowing in the substrate, We will refer to these modes as substrate radiation modes. For 0< B <1 (0/c), as in Figure 3.7e, the solution for E(x) becomes sinusoidal in all three regions. These are the so-called radiation modes of the waveguides. A solution of Equation (3.1-4), subject to the boundary conditions at the interfaces given in what follows, shows that while in regimes (d) and (e) of Figure 3.7, B is a continuous variable, the values of allowed B in the propagation regime n,(«o/c) < B 0, and the mode is poorly confined. As the values of 1/2 increase, so does the value of p, and the mode becomes increasingly confined to layer 2. This is reflected in the normalized propagation constant or the effective index, BA/2n, which at cutoff is equal to n3 and for large 1/A approaches n,. In a symmetric waveguide (n, ~ n,), the lowest order modes TE, and TM, have no cutoff and are confined for all values of #/2. The confinement is, however, poor when 17h becomes small. The total number of confined modes that can be supported by a waveguide depends on the value of 1/A. To study the number of confined modes, we defined the parameter vem Jni—nt 215) Let us now consider what happens about the TE modes in a given waveguide (ic., fixed nm, ing, ng, and f) as the wavelength of the light decreases gradually, assuming that the medium remains transparent and the indices of refraction n,, n,, and n, do not vary significantly. Since Gle = 2n/h, the effect of decreasing the wavelength is to increase the value of w/c. At long wavelengths (low frequencies), such that 3.216) 0< vba ( 2 the value of 1/2. is below the cutoff value, and no confined mode exists in the waveguide. As the wavelength decreases such that 210 one solution exists to the mode condition (3.2-5). The mode is designated as TEy and has a transverse h parameter falling within the range Ochen so that it has no zero crossings in the interior of the guiding layer (~t p,) and is therefore more tightly confined to the guiding slab. It follows from Equation (3.2-4) that By > B, so that the phase velocity 19 = W/By of the TE, mode is smaller than that of the TE, mode. We can now generalize and state that the mth mode (TE or TM) satisfies (m= Ie a where A and B are constants, and J, and K, are Bessel functions. For confined modes (rpky < B< mo), the electric wavefunction is oscillatory inside the core and evanescent in the cladding region. The Bessel functions K;,(qr) decay exponentially in the cladding region along, the radial direction, We assume that E, < E,. The magnetic field components are then given, according to Appendix A, by 33-6) The longitudinal component of the electric field vector E is related to H, according to the Maxwell equation V x H = e dE/ar: (83-7) where we used Equation (3.3-6) in arriving at the last equation, We note that the field com- ponents £, and H, are zero in this solution, The other four field components can be expressed in terms of E,, In order to calculate H, and E,, we need to carry out the differentiation with respect to.x and y, respectively, according to Equations (3.3-6) and (3.3-7). Since E, is of the form (3,3-5), we need the relations a_aa ama oe x Oxo ae) and aaa aa By dy ar * dy 80 Coe) By using the definition of r and ras yl? (3.3-10) o= www) 3-11) x we obtain 33-12) 3.3. Step-Index Circular Dielectric Waveguides (Linearly Polarized Modes in Optical Fibers) 129 (33-13) (33-14) and eer os (33-15) ay For ‘We now substitute Equation (3.3-5) for E, in Equations (3.3-6) and (3.3-7) and carry out the differentiation, using Equations (3.3-8)-(3.3-15). After some laborious algebra and using the following functional relations of the Bessel function, Ui) — Jind] (3.3-16) 71K 12) + Ki] Hp) + Ja] 3-17) 1 £i(3) = HK, 09) — Ky] we obtain the following expressions for the field components: Core (r < a): E,=0 E,=Adj(hrje® expli(wr ~ Bz)] 4 ahr + J, (hryel"“) expliteor — Br] (3.3-18) Famine expli(or —B2)} ina A inet? — 7, (ire) explitor — op 2 Vette Jea(hryel*) expli(ot — Br)] Cladding (r > a) E,=0 Ey = BK (qrie"® expli(oor ~ Bz)] B z [Kyalgrne™® — Ky (qe!) expli(ar — Bz)] 33-19) Peakiane expli(@r ~ Bz)] B =~ Bia, ane + Ky Cane] expitar— Bo 130 Chapter 3. Guided Waves in Dielectric Slabs and Fibers In arriving at Equations (3.3-18) and (3.3-19), we have also used B = nko ~ mko, since Ingky < B-< mky and m, ~ my <1. Note that E, and H, are the dominant field components because, in the limit (3.3-1b), 4, q < B. In other words, the field is essentially transverse. The constant B is given by p- Atiti, 83-20) Kiga) to ensure the continuity of E, (Ey % £,) at the core boundary r = a. The constant A is then determined by the normalization condition. The field solution (3.3-18) and (3.3-19) isa y-polarized wave (E,=0). For a complete field description, we also need the mode with the orthogonal polarization (i.e., an -polarized wave). The field components E, and E, of this orthogonal mode are taken as p= | Mitre! explior - Bey], rca E,= 33-21 BK,(qrje"® expli(wr - Bz)], r>a eet 33-22) (33-23) ; 0 and #, ~ 0 in this solution. By ing Equation (3.3-21) for E, in Equation (3.3-23) and carrying out the differen- tiation, using Equations (3.3-8)~(3.3-15), we obtain, again after some laborious algebra and using the relations (3.3-16) and (3.3-17), the following expressions for the field amplitudes: Core (r< = Al (hrye'® expli(cr ~ B2)] E,=0 Stra (dne™* J, (he) explo ~ 0 3-24) B io = A (hr)e"® expli(or — Be) op Mite explivor Be) H.= op Meath + Sather) '| expli(wr — Bz)] 3.3. Step-Index Circular Dielectric Waveguides (Linearly Polarized Modes in Optical Fibers) 131 Cladding (r > a): BK (gre expli(ot ~ Bz)] [Kiatqne* + (gre!) expli(or ~ Bz)] sis whe (33-25) B il 4, nel ~B2 ap BRiarel explo ~ Bey) B et) /itt-1 a7 Klar! ® — Ky s(qrel-] expli(wr ~ B2)) In arriving at Equations (3.3-24) and (3.3-25), we again made the assumption that = mky = myky because of Equation (3.3-1b). We note that E, and H, are again the dominant field components in this solution, Therefore the mode is again nearly transverse and linearly polarized along the x direction. The constant B is again given by Equation (3.3-20) to ensure the continuity of E, (E, s E,) at the core boundary r= a. We have obtained the field expressions for two types of guided modes whose transverse fields are polarized orthogonally to each other. Based on the circular symmetry of the fiber, we can be sure that these two guided modes must have the same propagation constant and the same intensity and power distribution, The field expressions for the linearly polarized modes are solutions of Maxwell’s equations, provided the tangential components of the field vectors are continuous at the dielectric interface r = a. The continuity of Ey at r= a leads to Equa- tion (3.3-20). The H,, components are proportional to the E, components, according to the field expressions (3.3-18), (3.3-19), (3.3-24), and (3.3-25) in this approximation. Therefore the continuity of £,, results in the continuity of H1,. We now consider the continuity of E, at =a. Since the continuity condition must hold for all azimuth angles 6, we must equate the coefficients of expfi(/ + 1)$] and expli(! — 1)9] separately. Using the field expressions (3.3-18) and (3.3-19) and (3.3-20), we obtain the following mode conditions: (ha) _ | Kia(ga) 33-26) Jha) Ky(ga) a) and Jia(ha) —— Kuas(aa) nq tO = gq Kiva 3.3.27) (ha) Kya) a ‘The same equations result from the continuity of H,. In addition, if we use the field expressions (3.3-24) and (3.3-25) for the x-polarized mode, we will arrive at the same mode conditions (3.3-26) and (3.3-27), This means that these two transversely orthogonal modes are degener- ate in the propagation constant B. The mode condition (3.3-27) is mathematically equivalent to condition (3.3-26) if we use the recurrence relation of the Bessel functions (3.3-17).. The mode condition (3.3-26) obtained in this approximation is much simpler than the exact expression for HE and EH modes in Appendix B. The exact mode condition (B-11) has twice as many solutions as the simple one (3.3-26) because condition (B-11) is quadratic Ji(ha)/J,(ha). This indicates that each solution of Equation (3.3-26) is really twofold degener- ate, In fact, the propagation constants of the exact HEj,,, and EH,_;,, modes are nearly 132 Chapter 3 Guided Waves in Dielectric Slabs and Fibers degenerate (6). They become exactly the same in the limit n, -> m,. This can also be seen from the expressions of the field components E, and H, in Equations (3.3-18), (3.3-19), (3.3-24), and (3.3-25). Comparison of the linearly polarized mode expressions with the exact modes (B-6)-(B-9) shows that the linearly polarized modes are actually a superposition of HE}. » and EH, ;», modes [6]. Two independent linear superpositions lead to the x-polarized and y-polarized modes. The total number of modes is the same in both theories. The eigenvalues obtained from Equation (3.3-26) are labeled B,, with /= 0, 1, 2,3,...and m=1,2,3,..., where the subscript m indicates the mth root of the transcendental Equation (3.3-26). The modes are designated LP;,,. The lowest order mode is LPp, with a propagation constant labeled Bp). This mode corresponds to the HE,, mode of the exact solutions. It is important to note that LP,o, and LP,o, are degenerate and have the same propagation constant By. This is consistent with the circular symmetry of the waveguide. In practical fibers, the cores are not exactly circular due to production imperfection. Such an asymmetry breaks the degener- acy and leads to Bo, # B,o1- In other words, the x-polarized LP mode and the y-polarized LP mode are propagating at a slightly different propagation constant and thus a slightly different group velocity. This leads to a phenomenon known as polarization mode dispersion (PMD), which will be described in Chapter 7. The mode conditions for those linearly polarized waves (3.3-26) or (3.3-27) can also be solved numerically. Here we examine the case of ! = 0. For the purpose of discussion we define an important parameter aoe fc i V = kya(n? = n3)!? = Bot = n3)!2 = (hay? + (gay? (3.3-28) where a is the core radius. As we will see later, this parameter determines how many confined modes can be supported by the step-index circular fiber. If we replace ga in Equation (3.3-26) with ./V? - (ia)?, we can then plot both sides of Equation (3.3-26), for a given V, as functions of ha. The intersections are the solutions. Figure 3.11 shows such a graphical Figure 3.11 Graphical method of determining ha of LP modes with = 0. The left side of Equa- tion (3.3-26) is plotted as the dotted lines. The right side is plotted for V = 2, 5, and 8. For V=2, only ‘one solution exists (LPp, mode). For V = 8, there are three solutions: LPy). LP. and LP; modes. 3.3. Step-Index Circular Dielectric Waveguides (Linearly Polarized Modes in Optical Fibers) 133 14628 LP) LP, we LPs, LP LP, LP 1.4600, fy T 3 4 3 6 Figure 3.12 Normalized propagation constant ng as function of normalized frequency V for some of the guided modes of the optical fiber, nj, = Bik. The fiber parameters are n, = 1.4628, n= 1.4600, and 7m. method of determining the solution for ! = 0 for the cases of V = 2, 5, and 8. For the left side, the denominator vanishes at roots of Jg(ha). Thus discontinuity at +e occurs at ha = 2.405, 5.520, 8.654, .... The numerator vanishes at roots of J;(ha), which occur at ha ), 3.832, 7.016, .... So, near ha = 0, the left side is a positive increasing function of ha starting from zero and reaching +0 at ha = 2.405. For the right side, we note that both Ky(qa) and K,(qa) are positive functions. Near ha = 0, the right side starts from a positive value of VK,(V)/Ko(V) and decreases as a function of ha, reaching zero at ha = V. As a result, there is at least one intersection even if V is very small. As V increases, the number of intersections increases. Figure 3.12 shows the normalized propagation constant (or effective index) of several modes as a function of the normalized frequency V. We note that for V < 2.405, there is only one mode (LPp,). For V = 6, there are six modes: LP), LP), LP}, LP}, LP2y, and LP3,. We note that the LP, mode always exists regardless of the fiber parameter V. As V increases beyond V = 2.405, the LP,, mode starts to appear. Thus we call V = 2.405 the cutoff value for the LP,, mode. As V increases beyond V = 3.832, LP. and LP) modes start to appear. Thus wwe call V = 3.832 the cutoff value for LPs, and LP9, modes. In general, the mode cutott cor- responds to the condition q = 0, which, according to Equation (3.3-27), leads to the condition, JV)=0 (33-29) where V = koa(n} — 13: 28). It follows that the lowest order mode, est root of the equation 2na(n? — n3)'/2. is the fiber parameter defined in Equation (3.3- ‘haracterized by = 0, has a cutoff given by the low- IV) = IV) =0 (33-30) Hence V = 0. In other words, the lowest order mode does not have a cutoff. This is the HE,, mode and is now labeled LPo,. The next mode of the type /=0 cuts off when J\(V) next equals zero, that is, when V ~ 3.832. This mode is labeled LP. The cutoff values of V for some low-order LP,,, modes are given in Table 3.1. 134 Chapter 3. Guided Waves in Dielectric Slabs and Fibers TABLE 3.1 Cutoff Values of V for Some Low-Order LP Modes m=1 m=2 m=3 m=4 0 3.832 7.016 10.173 2.405 5520 8.654 1792 3.832 7.016 10.173 13.323 5.136 8417 11.620 14.796 6379 9.760 13017 16.224 Figure 3.13 The regions of the par- ameter V for modes of order /=0, 1 Alll these values are zeros of the Bessel function. For high-order modes, the cutoff value of V is given approximately according to Equations (3.3-29) and (A-13) 3.331 Figure 3.13 shows the regions in which a given mode is the highest one allowed for a given 1 value group, labeled in LP mode designation. Also shown in the figure are the associated HE, EH, TE, and TM mode notations that are the exact modes. Figure 3.14 shows the field distribution of the LP, , modes [6]. We note that the intensity distribution consists of two lobes inside the core. This is distinctly different from that of the fundamental mode (the LP), mode), which has a radially symmetric field distribution J,(hr) in the core. One of the most important advantages of using the linearly polarized mode is that the modes are almost transversely polarized and are dominated by one transverse electric field ‘component (E, or E,) and one transverse magnetic field component (H, or H,). The E vector can be chosen to be along any arbitrary radial direction with the H vector along a perpen- dicular radial direction. Once this mode is chosen, there exists another independent mode with E and H orthogonal to the first pair. Power Flow and Power Density We now derive expressions for the Poynting vector and the power flow in the core and cladding. The time-averaged Poynting vector along the waveguide is, according to Equa- tion (1.3-17) $RelE,H?—E,H*] 3.32) Substituting the field components from Equations (3.3-18) and (3.3-19) or (3.3-24) and (3.3-25) into Equation (3.3-32), we obtain 3.3. Step-Index Circular Dielectric Waveguides (Linearly Polarized Modes in Optical Fibers) 135 Figure 3.14 Sketch of the four possible field distributions of LP, modes inside the core of the fiber. ‘The arrows indicate the polarization of the electric field vector. Bary Parsjim, r 1, the confinement factor at cutoff is obtained by taking the limit ga — 0. This leads to the following limiting value of confinement factor at cutoff: T(qa > 0) (for 1 > 1 at cutoff) (33-43) 1 For example, the confinement factor of the LP,, mode at cutoff is 3. Further discussion of the confinement factor for high-order LP modes can be found in Problem 3.11 3.4 Effective Index Theory 137 08 06 04 02 Figure 3.15 Confinement factor I versus V. Note that, at cutoff, the confinement factor is zero for LP modes with /= 0, 1; whereas the confinement factor is + for = 2; 3 for 1=3; 4 for (=4; and so on. 3.4. EFFECTIVE INDEX THEORY Although simple layered structures such as dielectric slabs can be used for waveguiding purposes, the coufiiement of energy is only limited Ww une dimension. In practice, unure complicated waveguide structures are used. For example, the waveguides used in integrated optics or guided-wave optics (also known as planar light circuits (PLCs)) are usually two- dimensional waveguides (c.g., channel waveguides, ridge waveguides). Figure 3.16 shows examples of such two-dimensional waveguides. Exact analytical treatment of these waveguide structures is not possible, except for some special cases. Although numerical solutions can be obtained by various methods, there are several approximate analytical approaches. Here, we will introduce one of the simplest approaches, the effective index theory. Referring to Figure 3.16c, we consider the guiding of electromagnetic radiation in a ridge waveguide. The rectangular waveguide chown in Figure 3.16b can be considered as a special case of this ridge waveguide by taking d= 0, where d is the thickness of the guiding layer on both sides of the ridge. The thickness of the guiding layer at the ridge is ¢, which is chosen to be greater than d because we are interested in the confinement of electromagnetic radiation at yenan y=al2 » © Figure 3.16 Schematic drawing of several two-dimensional waveguides: (a) a rectangular strip of dielectric medium embedded in another dielectric medium of lower index of refraction; (b) ridge waveguide structure; and (c) another ridge structure. 138 Chapter 3 Guided Waves in Dielectric Slabs and Fibers the ridge. The width of the ridge is a. For confined propagation, the index of refraction n, is, greater than either m, or 73. We now divide the structure into three regions for the purpose of introducing the concept of effective index theory. These regions are y <~—4a, -4a < y < 4a, and 4a ;. This is always true when the thickness f is greater than d because, according to Figure 3.7, the effective index of refraction of any confined mode is an increasing function of thickness. To illustrate further the lateral guiding, we will consider the following example. GaAs Ridge Waveguide We consider a ridge waveguide made of GaAs layers and AlGaAs substrate. Let the indices of refraction be m= 1, ny = 3.5, and ny = 3.2. The thicknesses are 1 = 0.40 and d = 0.252. According to a numerical solution of Equation (3.2-5), the effective indices are ny=3.301 and ny =3.388 ‘We note that a step height of 0.152. at the ridge gives rise to an increase in the effective index of 0.087. Such an index difference is sufficient to provide lateral waveguiding. In fact, if we take a = 0.52 as the width of the ridge and solve for the confined TE modes of such a symmetric waveguide, we obtain a single TE mode with a normalized propagation constant of 3.348. The wavefunction of such a mode is similar to the fundamental mode shown in Figure 3.3. This wavefunction shows the lateral confinement. Itis important to note that the effective index theory is a good approximation, provided the index difference hetween the core and cladding is emall so that the cealar wave approximation is valid. In the scalar approximation, we ignore the vector nature of the electromagnetic waves. Ridge Waveguides or Two-Dimensional Waveguides The ridge waveguide structures described in Figure 3.16 can be obtained by several different approaches, including etching and diffusion. In the following, we describe some examples of channel waveguides that are important in optical communications. ‘Thermal Indiffusion A channel waveguide in a LiNbO, crystal can be obtained by using a conventional thermal indiffusion process. Prior to the diffusion process, a thin strip of metallic Ti layer about 100 nm 3.4 Effective Index Theory 139 LINO, LiNbO, (a) Masking (b) Thermal diffusion Figure 3.17 Fabrication steps in Ti-diffused LiNbO, channel waveguide. (a) First, a thin strip of Ti ‘metal is deposited on the optically polished surface of the LiNbO, crystal wafer. Such a strip or a more ‘complicated planar optical circuit pattern can be obtained by a photolithographic method that defines the channel waveguide circuit. (b) The water is then heat treated at an elevated temperature. As a result of thermal diffusion, ‘Ti atoms migrate into the crystal wafer. This leads to an optical circuits with a higher refractive index at the core of the waveguide circuit. in thickness is deposited on the surface of a LiNbO, wafer by a vacuum evaporation technique. A planar optical circuit can be patterned by a conventional photolithography technique. The wafer is then heat treated at around 1000 °C for a period of time (e.¢., 20 hours). As a result of the heating-assisted diffusion, the Ti atoms migrate into the crystal. ‘The presence of Ti atoms in the LiNbO; crystal increases both the ordinary and extraordinary index of refraction. This creates a channel waveguide in the region of Ti-atom concentration, Figure 3.17 illustrates the steps in a Ti-diffused LiNbO, channel waveguide. ‘A channel waveguide in silica (SiO,) wafer can be obtained by a similar thermal indiffu- sion process using Pb or other metals. The presence of Pb atoms in silica increases the index of refraction and thus creates a channel waveguide. Proton Exchange in Lithium Niobate ‘A channel waveguide in a lithium niobate wafer can also be obtained by using the proton exchange process (see Figure 3.18). Fst, a patterned material 1s applied to the optically polished surface of the LiNbO, crystal. The pattern can be obtained by a photolithographic method that defines the channel waveguide optical circuit. The wafer is then immersed in a bath of liquid acid that provides a source of protons that exchanges with lithium ions, creating a thin layer of protons at the surface. The protons at the LiNbO, surface can penetrate further into the crystal via diffusion during an anneal process at high temperature, Once cooled, the channel waveguide is extremely stable, As a result, the extraordinary index of refraction is increased in the proton-exchanged regions and the ordinary index is reduced. This leads to a channel ‘waveguide that supports confined propagation with the E field polarized along the c axis of the crystal. There are other techniques that can be employed to increase the index of refraction and to produce optical waveguides. These include ion implantation and the sol-gel process. a -" on, (a) Masking (b) Proton exchange (©) Anneal Figure 3.18 Fabrication steps in proton-exchange process in a LiNbO, channel waveguide. 140 Chapter 3. Guided Waves in Dielectric Slabs and Fibers 3.5 WAVEGUIDE DISPERSION IN OPTICAL FIBERS So far in this chapter, we have discussed various optical waveguides, including dielectric slab waveguides and step-index optical fibers. These waveguides can be described in terms of the index of refraction as a function of space. Confined propagation in these waveguides is governed by the wave equation, which provides the wavefunction and the propagation constant once the index distribution of the waveguide is given. The wave equation is often written fv: + where V7 is the transverse Laplacian, E,,(x, y) is the wavefunction of the mth mode, and B,, is the propagation constant, Equation (3.5-1) may be viewed as an eigenvalue problem with 3, as the eigenvalue. Consider a general mode of propagation roy} at 9) = Bi Em YD B51 E, FG. 20 9) expli(@r ~ B,.2)] 35-2) In multimode waveguides, a general wave of propagation can be written as a linear combi- nation of all the modes. For the convenience of discussion, we define an effective index of a mode as, 5-3) It is important to note that both the wavefunction and the propagation constant B,, depend ‘on the waveguide geometry »7(x, y) as well as the angular frequency @. For confined modes, the effective index is always between the core index and the clad index. In other words, Nasag < Mest ore 5-4) For multimode waveguides, each confined mode has a distinct effective index. The funda- mental mode has the largest effective index. Generally, each mode has its own phase velocity and group velocity. As a result, modal dispersion occurs in multimode waveguides. This was discussed in Section 2.9. Ina multimode waveguide, if an optical wave (c.g., a sequence of pulses) is represented by a superposition of a number of modes, then the pulses will spread as they propagate in the waveguide. The spread can be due to (a) modal dispersion, (b) waveguide dispersion, and (©) material dispersion. In single-mode waveguides, the pulse spreading is due to (a) wave- guide dispersion and (b) material dispersion. In this section, we discuss waveguide dispersion and the pulse spreading in single-mode optical fibers. The effective index of a step-index fiber (or slab waveguide) depends on the frequency of light «, even if the indices of refraction of the fiber are independent of the frequency of light ©. This is a result of the different wavefunctions and ray paths inside the waveguide at different frequencies. The dispersion figures in Sections 3.1-3.3 (e.g., Figure 3.12) show the dependence of ni on frequency ©. Thus we may write Nege = Megrly(@), nef), ©] (35-5) where the effective index depends on the frequency directly due to waveguide dispersion and indirectly due to material dispersion through n,(@) and n,(@).. 3.5 Waveguide Dispersion in Optical Fibers 141 We now examine these two contributions of the dispersion. Using the method of vari- ation in perturbation theory, we apply a small variation of the frequency 54 to the wave equation (3.5-1). Let 8B3,, 8n}, and 6n3 be the corresponding variations in the propagation constant of the mth mode, and the refractive indices, respectively. We obtain from the wave equation (3.5-1) nt + Ty 80 92 47,2088 92 056 oa, = 1a? + 2 ba} +12 2 where the first two terms on the right side are due to the changes of the index of refraction nj and 8n3, while the last two terms are due to a change of the frequency 8a; TP, and I are confinement factors given by {I BEE, de dy = E,, dx dy where the integral in the numerator is over the region where the index of refraction is ng. Equation (3.5-6) can be rewritten as (a=1,2) 5-7) Bip Sa M8 ni stn} 658) do 'C ny OO Ney OO Clery 2 where we have dropped the subscript m for i ity. Formula (3.5-8) can be employed for each mode. Using Equation (3.5-3) for B, the above equation can further be written fgg a AM or, do Ty OO? Mey OO ON gy (Tin? + Fyn} = 2) B.5-9) where the first two terms on the right side are due to material dispersion, and the last term is, due to waveguide dispersion, which represents the variation of the effective index of the guided mode with respect to the frequency by keeping both the core and cladding index constant. In other words, the dispersion of a guided wave consists of the sum of an intrinsic material dispersion and the waveguide dispersion. Thus we can write d an, “) () ing, =| Se any 3.5-10 oe FE a Fe os where the material dispersion is given by an, :) my Om | my Om arth grt Oe as. Fe) ee OS ee) and the waveguide dispersion is given by mut) 1 n ot =n} + yn} - ny) 5-12) For single-mode silica fibers used in modern optical communications, the clad and core are both made of silica with n, ~ , so Equations (3.5-10)-(3.5-12) can be written approximately at) on . (3) = 142 Chapter 3. Guided Waves in Dielectric Slabs and Fibers an, | Lat En? —n2—n2 Su =——(n} + Tyln} = n}]- ng) 5-14) ( Bo Jeavegite Oe where n is the index of refraction of the fiber material (silica), and we have used 1, = nr in and T; = 1 ~Ty. We note that the term on the right side of Equation (3.5-13) is purely material dispersion, while Equation (3.5-14) addresses the waveguide dispersion, Using the confinement factor expression in Section 3.3, the waveguide dispersion can be written (see Problem 3.12) (ut (n2y 9 t) (35-15) Ya © Org Trae) Jy, (Fa) Since Jp_,(ha)Jp.,\(ha) is always negative for the confined-modes that satisfy the mode conditions (3.3-26) and (3.3-27), we note that the right side of Equation (3.5-15) is always positive. In other words, the effective index is an increasing function of frequency. This is consistent with the dispersion curves shown in Figure 3.12. In optical networks, the dispersion of a mode of propagation is often written an ‘ang, ; @. * (7) 5-16) and 2, 2, 2, Pre (3 +] + (See nef (5-17) il Consider the propagation of an optical pulse in a single-mode waveguide of length L. Let ny be the effective index of the mode of propagation. The phase shift of the beam at output due to propagation is @ = ner(O/c)L. It can be shown that the group delay (flight time) is nite tg) oifte 28) os The group velocity dispersion (in units of picaseconds per nanometer of bandwidth per Kilometer of fiber) for optical fibers is defined as a(x s(t : 35-19) (2) (gains os Using Equations (3.5-18) and (3.5-19), we obtain 1 (42 d?hey ) 2 Fra | 3.5.20) +" ae) Ca If the waveguide material is dispersive, the group velocity dispersion parameter D can be written, according to Equation (3.5-17), Vf 2m) Ay Png a(® OF J ag YE 521 ti averse Note that the terms inside the parentheses are dimensionless. The parameter D has a dimen- sion of s/m-m. For practical application, D is often expressed in units of ps/nm-km. 3.5 Waveguide Dispersion in Optical Fibers 143, EXAMPLE: PULSE SPREADING. Consider the transmission of 10 Gb/s signals at A= 1500 nm in a single-mode fiber of 100 km, with a group velocity dispersion of D = 17 ps/nm-km. The pulse spreading after a transmission of distance L can be written At=DL AA 5-22) where Ad is the spectral bandwidth of the signals. For 10 Gbys, the pulse width is ty = 100 ps. The spectral bandwidth is approximately Av = 1/t) = 10 GHz. In terms of nanometers, the bandwidth is ar=Zay=0.075 nm ‘The pulse spreading is thus given by, using Equation (3.5-22), At = 128 ps Group velocity dispersion Matra dispenion << Total eispersion I ‘Waveguide dispersion D(psinm-km) ‘Core index ny = 1.4628, clad index n,= 1.4600, core diameter 2a=9.4 jm 2131S Wavelength (jm) Figure 3.19 The material dispersion of silica, the waveguide dispersion of a single-mode GeO,-doped silica fiber, and the resultant total dispersion in the spectral region between I and 1.8 jim. Figure 3.19 shows the material dispersion of silica, the waveguide dispersion of a single-mode GeO,-doped silica fiber with a core diameter of 9.4 jum, and the resultant total dispersion. Not that the total dispersion is zero at 2.= 1.3 um [8, 9]. It is important to note that the waveguide dispersion depends on core diameter a as well as the core and clad indices rn, and n3, It is possible to tailor the zero-dispersion wavelength by balancing the positive material dispersion against the negative waveguide dispersion [10]. Thus, by choosing a proper core diameter a between 4 and 5 jim, and a relative refractive index difference of (1, — n/m, > 0.004, the wavelength of zero dispersion can be shifted to the 1.5—1.6 um region, where the propagation loss is minimum [1116] 144 Chapter 3. Guided Waves in Dielectric Slabs and Fibers @ o © @ Figure 3.20 Various index profiles of fibers. (a) Multimode fibers with a core diameter of about 50 um, and a clad diameter of 120 pm. (b) Single-mode fibers with a core diameter of about 10 um. (©) Dispersion-shifted fibers with a core diameter of about 5 jim. (d) Dispersion-flattened fibers with double clad or triple clad and a core diameter of about 6 jim. (e) Bragg fibers with a core of low index of refraction, or a hollow core. We have so far discussed the step-index circular fibers. Closed-form solutions for the modes are available. There are optical fibers with different refractive index profiles. These fibers are designed to provide minimum loss or minimum dispersion in the spectral region of interest. Some of the fibers can provide near-zero dispersion over a broad spectral region of interest. Figure 3.20 shows some of the index profiles. Some of the profiles involve additional cladding layers of different refractive indices [17-21]. These index profiles offer additional degrees of freedom in the design of fibers for specific applications. ‘According to Figure 3.19, we note that at A= 1550 nm the chromatic dispersion is mostly due to material dispersion. If a fiber can be fabricated with an air core, the material dis- persion can be virtually eliminated. To support a confined propagation in a low-index region requires a multilayer cladding with a periodic variation of the clad index in the radial direction. High reflection via Bragg scattering in the radial direction provides the confinement (see Figure 3.20e). The discussion of such fiber is beyond the scope of this book. Interested readers are referred to Reference [22]. 3.6 Attenuation in Silica Fibers 145 3.6 ATTENUATION IN SILICA FIBERS Propagation attenuation exists in virtually all waveguides. As a result of attenuation, si power P(z) will decay exponentially according to the following equation: P(L) = P(O) exp(-aL) 36-1) where P(Q) is the power at the input, L is the distance of propagation, and o: is the linear attenuation coefficient. In optical communications, attenuation is often measured in units of 4B, and the attenuation coefficient is measured in units of dB/km. Thus signal attenuation (AB) per unit length is defined as 0 oo, (PO 7 o Pree A2) 3.6-2) There are many sources of propagation attenuation in optical fibers. In the early days of silica fiber development, propagation attenuation was mainly due to the presence of impu- rities that absorb optical energy. Trace metal impurities, such as iron, nickel, and chromium, are introduced into the fiber during the fabrication. The electronic energy levels of these metal ions are broadened by the random atomic environment in the silica. Some of the electronic transitions fall in the spectral regime of the propagating optical beam. Photons in the optical beam are absorbed by these electronic transitions. The absorption also occurs when a small amount of water is present in the silica glass. Water in silica glass forms a silicon-hydroxyl (Si—OH) bond, which can cause absorption in the spectral regime around A= 1.4 jim. Although the intrinsic absorption due to the vibration of the SiO, is relatively insignificant in the spectral regime of optical communications, propagation attenuation is still present in pure silica fibers. The attenuation is mainly due to manufacturing imperfection of the fiber structure, as well as the fundamental process of Rayleigh scattering. Probably the single most important factor responsible for the emergence of silica glass optical fiber as the premium information transmission medium is the low optical propagation losses in such fibers. Figure 3.21 shows the measured losses as a function of wavelength of a high-quality, GeO,-doped single-mode fiber. The loss peak at around 1.4 jim is due to residual OH contamination of the glass. A low value of loss ~ 0.2 dB/km obtains near 4 = 1.55 jum. In 1986, a low transmission attenuation of 0.154 dB/km was realized [23]. Recently, a pure silica core fiber, with the index profile shown in Figure 3.20d, has been fabricated with a low transmission loss of 0.1484 dB/km at 2. = 1570 nm [20, 21]. Consequently, this region of the spectrum is now favored for long- distance optical communications. Recent experiments have taken advantage of the small pulse spreading near the zero group velocity dispersion wavelength and the low losses to demonstrate high-data-rate transmission (data rate exceeding 400 Mb/s) over a propagation path exceeding 100 km (24, 25] at 4 ~ 1.55 pm. Longer distance and higher data rates are possible provided optical amplifiers are employed to boost the energy attenuation due to propagation as well as chromatic dispersion compensation to eliminate the pulse distortion. For a more detailed discussion of propagation effects in optical fibers, the student can consult Reference 26]. Another source of attenuation in single-mode fibers is the bending loss. This is usually not a problem in the transmission optical networks where the curvature of bending is very small. In optical fiber sensors such as fiberoptic gyros, a long segment of fibers (kilometers) must be coiled in a small box of a few centimeters. The radius of curvature in this case is in the range of centimeters. In Dense Wavelength Division Multiplexing(DWDM) optical networks, there are situations when many passive or active components must be connected by fibers inside a 146 Chapter 3 Guided Waves in Dielectric Slabs and Fibers so Single-mode fiber Ge0,-doped, An = 0.0028, 22= 9.4 um 10 s 3 Infrared 2, sorption Bee Rye scattering = 4] Ultraviolet XT o4 Waveguide jo 00s imperfections oot os 10 12 14 16 18 Wavelength (um) Figure 3.21 Observed loss spectrum of a germanosilicate single-mode fiber. Estimated loss spectra for various intrinsic material effects and waveguide imperfections are also shown. (From Reference [24].) small box. Bending of fibers is unavoidable in these situations. To minimize the loss due to bending, the radius of curvature must be kept as large as possible inside the box. In this section we briefly discuss the loss due to bending of fibers. Referring to Figure 3.22, we consider the propagation of a LP, mode inside a single-mode fiber. In the straight fiber, the field distribution will remain unchanged as it propagates along the fiber. When the fiber is bent with a radius of curvature of R, we may assume that the fiber mode pattern remains the same, provided the radius of curvature of bending is much larger than the fiber core radius a (i.e., a < R). It is known that bent waveguides are intrinsically leaky. The source of leak can be explained as follows. To maintain the same mode pattern in the bent waveguide, the planar wavefront of the mode must be pivoted around the center of curvature O. This leads to a velocity mismatch problem. To keep up with the mode, the wave- front on the outer evanescent tail must travel faster than the wavefront at the center of the ° Suaight fiber Bent fiber Figure 3.22 Schematic drawing of a straight fiber and a bent fiber. To maintain the mode patter in the bent fiber, the wavefront ofthe outer tail of the mode must travel faster than the wavefront of the mode atthe center of the fiber core. R is the radius of curvature of the bending and r is the radial distance ‘measured from the center of curvature O. 3.6 Attenuation in Silica Fibers 147 nr) Straight Figure 3.23 Index profile and mode wavefunction of the straight waveguide and the equivalent index profile of the bent waveguide. The horizontal dashed line indicates the effective index of the fiber mode, Nate core. In fact, the speed of the wavefront at location r’ from the center of the fiber core must be (1 +r'/R) times c/ngy. At some critical distance from the core of the fiber, this speed can be greater than c/n. Since this is impossible, the field beyond this critical distance is coupled to the radiation modes. Thus part of the mode energy will break away from the guide. This leads to propagation attenuation. Based on the above argument and the schematic drawing shown in Figure 3.23, the wave propagation in a bent waveguide can be described by the wave propagation in an equivalent waveguide. Since the outer side of the mode must travel more distance and experience more phase shift for a given propagation constant B, the index profile of the equivalent waveguide can be written Megl?") = nir(1 + Zoos «) where r’ is the radial distance measured from the center of the fiber core, and 9” is the azimuthal angle measured in the cross-sectional plane of the fiber. Note that the radial coordinate r is measured from the center of curvature of the bending O. Figure 3.23 is a sketch of the equivalent index profile. The horizontal dashed line indicates the effective index of the fiber mode, ng. Based on the discussion earlier in the chapter, the effective index of a confined mode must be greater than the index of the clad. Beyond the critical radius resiea)s the equivalent index in the clad is higher than the effective index ng This is the region where the guide mode is coupled to the radiation mode. A guided mode in the fiber can thus tunnel through the triangular barrier between the outer edge of the core and the critical radius. As a result of the equivalent index profile shown in Figure 3.23, all modes of propagation in the bent waveguide are leaky. Solving the wave equation with the equivalent index profile given by Equation (3.6-3) would require extensive numerical techniques. The attenuation coefficient due to the bending of the fibers is usually obtained by using various different approaches [27-41]. Here we describe a result computed with the help of scalar diffrac- tion theory [27]. In this approach, we consider the propagation of light inside the core of a circular ring of fiber. The radius of the ring R is much greater than the core radius of the fiber 148 Chapter 3. Guided Waves in Dielectric Slabs and Fibers @ (a 2, Using the method of steepest descent (the saddle point method) as well as the assumption of 1 < ga n°(ae)?, the CARDO RK wavefunction E(x) can have, at most, one zero 1 ne? crossing. (Hint: Use the continuous nature of E(x) and (I/E)(@°E/2x*) > 0.] (b) Show that focal maxima of |(x)| can only occur in the layers where B < n?(w/c)?. 3.2 Let T; be the fraction of power flowing in medium i (= 1, 2, 3) of a slab waveguide. In particular, T is the fraction of mode power flowing in the guiding layer in, and is often called the mode confinement factor. If the mode is normalized to a power of 1 W, the I’ values are defined as where the integral is over the region occupied by medium {where the index of refraction is 1. (a) Show that, for TE modes, THllq+ lip +p reas Up a ¢ 14g lp +g up we T¥lig +p Pape Ts Note that f) +1, +13 = Equation (3.2-7). (b) Show that, for TM modes, . This proves ug Tllg Hip eg wp Pog elgg sy ee le ¢ Tolg sup Rag Up" ae Trig +p Pe 150 Chapter 3. Guided Waves in Dielectric Slabs and Fibers where +e Fa mp bigs PR nig’ op " (c) Show that TP increases as B increases, and hol as Bo = ‘Thus lower-order modes are more confined than higher-order modes because lower-order modes have a larger propagation constant. [Hint: Use h? + p (a3 n3)(/e)? and I? + g? = (n} — n}(@/e}*.] (a) Compare the mode confinement factor I’, for TE, and TM,, modes. (e) Show that in the lir h-0), it of well confinement (ie., 1+ plq + pt MEL attains GERDRA TIE T+ (tn Dpla) + ORInDpe where p = (o/e)(n} — n3)'® and q = (a/eyn} — 9)". 3.3 The interface between a metal and a dielectric can support the propagation of a surface wave, known as a surface plasmon, provided the metal has a negative dielectric constant. (a) If the metal is viewed as a semi-infinite free electron gas with a dielectric constant given by cad] where 0, is the plasma frequency, show that surface plasmon modes exist only when @ < 4@3, Here, we assume that nj = | for the dielectric, For an electron ‘gas with an electron number density of N, the plasma, frequency is} = Ne*/meg, where m is the electron (b) Show that the propagation constant B is given by oe (tehy® [Hint: Use Equation (3.2-11). Show that B is always greater than nko (i.e., B > nko), provided n} > 0 and ni +n3 <0, Here, ky = a/c.) (©) Show that B? = pq and that the polarization states of the E vector in the two media are mutually orthogonal. (d) Obtain an expression for the z component of the Poynting vector. Show that the Poynting power flow in the positive and negative dielectric media are ‘opposite in direction, Thus a surface plasmon wave propagating along the silver surface along the + direction will have a negative Poynting power flow in the silver and a positive Poynting power flow in the air (©) The complex refractive index of gold at A= 10 jum is n= ix = 7.4 ~ 153.4, Find the propagation constant B® and the attenuation coefficient a of the surface plasmon wave. 3.4 The number of confined modes that can be supported by a circular dielectric waveguide depends on the refractive index profile and the wavelength, (a) Using the cutoff value for the LP,,, mode, show that the mode subscripts (I, m) fora step-index fiber must satisfy the condition m3} where V = kya(n} ~ 13)", Show that each LP}, mode is fourfold degenerate. (b) By counting the allowed mode subscripts (I, m), show that the total number of confined modes that ‘can be supported by a step-index fiber is 4 wets w. °2 (©) Using Equation (2.9-15), show that for a truncated ‘quadratic-index fiber (n(r) =n, for r > a) w Note that the total number of modes in a truncated \dex fiber is one-half that of a step-index (@) Estimate the number of confined modes in a multimode step-index fiber with a = 50 im, and ry = 1.50 at a carrier wavelength of A= I jum. (©) Ina general, truncated graded-index fiber with a core radius @ and a cladding index 1, itis convenient to define an effective V number such that 52, and the number of confined modes is approximately given by Show that this approximation agrees with (b) and (¢) for step-index and quadratic-index fibers, respectively. (f) Show that, according to (e), the number of confined ‘modes in a power-law (power p) graded-index fiber is given by Aid? (42 1 20+ 2p) 242 % ) Show that this expression again agrees with the results obtained in (b) for step-index fibers (p ==) and in (¢) for quadratic-index fibers (p=. 3.5 The numerical aperture (NA) is a measure of the light-gathering capability of a fiber. It is defined as the sine of the maximum extemal angle of the entrance ray (measured with respect to the axis of the fiber) that is trapped in the core by total internal reflection, (a) Show that A= my sin 8, = (n3 ~ 03)! (b) Show tat the solid acceptance angle in sir 2 = n(n} ~n3)"? = nINAY? (c) Show that the solid angle (in air) for a single electromagnetic radiation mode leaving or entering the care apertures 2 pase Ap (4) The total number of modes that the fiber can support, couple to, and radiate into air is therefore = 2 Qos Where the factor of 2 accounts for the two independent polarizations in air. Show that this ‘estimate agrees with Problem 3.4. (©) Find the numerical aperture of a multimode fiber with n, = 1.52 and ny = 1.50. 3.6 A single-mode step-index fiber must have a V number less than 2.405; that is, V= kya(ny ~ 03)! < 2.405 (a) Show that the expression derived in Problem 3.4(b) (N= 4V7/x°) still applies, provided we realize that a single-mode fiber supports two independently polarized HE,, modes (or LPo, modes). (b) With a= 5 jm, n, = 1.50, and A= 1 um, find the ‘maximum core index for a single-mode fiber. [Answer: m= 1.50195.] (©) With ny = 1.501, n= 1.500, and A= 1 um, find the maximum core radius for a single-mode fiber. [Answer: a= 7 um} (a) Show that the confinement factor for a single-mode fiber is Problems 151 note (it) Pv NGhay ‘where ha satisfies the mode condition (3.3-26) ha IAD gg Kaa) Jefha) Kelga) (©) Show that, by using the table of Bessel functions, hha = 1.647 is an approximate solution to the mode condition for V = 2.405. Evaluate the confinement factor T for the LPp, mode of this single-mode fiber [Answer: P = 83%.] Note that this is the maximum confinement factor for a single-mode fiber. Compare this value with the curves in Figure 3.12. 37 (a) Derive the mode condition for ste (Equation (B-11), (b) Derive the expressions for the constants B, C, and D in terms of A (Equation (B-12)). (©) Derive the mode condition for TE and TM modes (Equations (B-17a) and (B-17b)). (@) Show that E for TE modes and H. for TM modes. (©) Show that in the limit m ~ ny < m, TE and ‘TM modes become identical. 38 (a) Derive Equations (3.3-6) and (3.3-7). (b) Derive Equations (3.3 (©) Derive the field components (3.3-23). (@) Derive Equations (3.3-24) and (3.3-25). 3.9 There are several approximations for an analytic form of the effective index of the fundamental mode of a symmetric slab waveguide. Using the definition u = hd/2 and v= gd/2, the mode condition for the fundamental mode can be written tan u = v/u. (a) Show that the mode condition can also be written Vcosu, or sty Vsinu or tanu where V =n} —n} (nd/A) for the slab waveguide. (b) Ifw-< 1, then w= tan“'V is an approximation. Show that the following is also an approximation: (©) The following is also a useful approximation [43]: View? 152 Chapter 3. Guided Waves in Dielectric Slabs and Fibers Compare the approximations in (b) and (c) with the exact mode condition by plotting u versus V. Show that for V 1 (a) Using the asymptotic forms of the Bessel functions, show that, for V < 1, the mode condition reduces to 2 -(4)- 3, Mab = nd rary where 7 is Euler's constant y= 0.5772, and T= exp(y) = 1.782. We note that ny approaches the clad index ny very quickly as V approaches zero (see Figure 3.12). (b)_ Using the asymptotic forms of the Bessel functions, show that, for V > 1, the mode condition reduces to where x, = 2.405 is the first root of Jg(x) = 0. [Hint: See Figure 3.11. ha approaches x, for LPo, mode at large V.] ha = x(1~ WV) (nt =n9(1 3.1L The confinement factor I is defined as I, T,, where Ty = Biet = U (gay? «ger tidd pV Thad aha) (a) Using the asymptotic forms of the Bessel function: show that, at cutoff (ga — 0), the confinement factor is for 151 T\=0 for [=0,1 (b) At cutoff (qa = 0), the solutions of the wave equation in the cladding region are no longer Bessel functions. According to Equation (A-7), the wave equation for the field in the cladding region at cutoff (ga = 0 and B= nko) becomes ‘Show that the solution for the y-polarized LP mode can be written {s Jr explifot ~ Bo), ra where B= a'AJ (lia) = a!AJ(V) due to continuity at the core boundary. (©) Using S, = (B/200p)| B/*r-* for the cladding region, evaluate the following integral: Pa [[[searaw Show that the integral diverges for the case when |= 0 and 1. This is consistent with the zero confinement factor at cutoff. For (> 1, the integral converges to a finite value, leading to a finite confinement factor at cutoff. 3.12. The mode condition for the LP modes is given by Equation (3.3-26), watha) _ |, Kin(aa) Jha)“ K(ga) (@) Differentiating both sides of the mode condition, show that hal Zessbaatha) _) dha) Jia) wv Kia(aa)K, (ga) _) aCaa) + qa{ Kinasx'ae) _ | ae " ( K}(qay ) wv? [Note: V is proportional to frequency «,] (b) Using (ga)? + (ha)’ = V* and the result in (a), show that 400) ty and ha ttt) = =ny av ayo 4a where I’, is the confinement factor and I fits) (au? ba) Spatha\,,(ha) (c)_ Using the results in (a) and (b) and the definition of the effective index ny = B/(@/c), show that a) 1 tn? + 0nd 3 a aay tlin? + Fad - mo) (5 ac a +8 Iti important to note that the right side is always positive. (@) Show that Ore 2 ety = n3){-—22@ Bo Janegise OME" Ty ithadtha) ‘Show that the right side is always positive for the ‘modes of propagation. 33 (a Solve for the mode of propagation along the z axis by assuming n= 1 (vacuum), The partial differential equation is separable, You may assume a solution of the form Er.) = Aldi explicat + ~ Be) H.r.1) = BU(hr) expli(or + I ~ B2)) and use (b) Obtain expressions for the energy density and Poynting vector. Consider TE, TM, and hybrid modes. REFERENCES References 153 (©) Sketch the energy density and the Poynting vector components (Sy, S,) as functions of r, for /= 0, #1, +#2. Show that for / # 0, the Poynting power flows spirally around the 2 axis, with a handedness that depends on the sign of J. (@) Using the angular momentum defined as ‘where P is the linear momentum defined in Problem 1.5, calculate the time-averaged component of the angular momentum along the direction of propagation (+2). Show that this component of the angular momentum is 1h provided the energy of the electromagnetic wave is normalized to Neo (©) What are the mode orthogonal relationships? xP, ad Using Equations (3.1-14) and (3.1-15), the mode condition can be written Ustan( hd) ~ qth cot(hd) + 4) =0 Show that the above equation reduces to tan(hd) 2hqi(h? — q?). (Hint: tan 20 = 2 tan 6/(1 — tan*6).} BAS (a) Using Equations (3.3-16) and (3.3 17), show that 2 [vianare FU iean-J.andian (b) Similarly, show that : | khan) dra Eiken ~K, fare) where ais an arbitrary constant. [Hint: Differentiate both sides with respect to r.] 1. Yariv, A., and P. Yeh, Optical Waves in Crystals. Wiley, New York, 1984. 2. Yeh, P., Optical Waves in Layered Media. Wiley, New York, 1988. 3. Kao, C. K., and T. W. Davies, Spectroscopic studies of ultra low loss optical glasses. J. Sci Instrum, 1(2):1063 (1968). 4, Kapron, F. P., D. B. Keck, and R. D. Maurer, Radiation losses in glass optical waveguides. Appl. Phys. Lett, 17:423 (1970) 5. Gloge, D., Weakly guiding fibers. Appl. Opt. 10:2252 (1971), 6. See, for example, D. Marcuse, Theory of Dielectric Optical Waveguides. Academic Press, New York, 1974. 7. See, for example, LS. Gradshteyn and I. M. Ryzhik, Table of Integrals, Series, and Products. Academic Press, New York, 1965, p. 634, Equation (5.54-2). 154 Chapter 3. Guided Waves in Dielectric Slabs and Fibers 8. Payne, D. N., and W. A. Gambling, Zero material dispersion in optical fibers. Electron, Let 11:176 (1975). 9. Cohen, L. G., and C. Lin, Pulse delay measurements in the zero material dispersion wave length region for optical fibers. Appl. Opt. 12:3136 (1977). 10, Li T.,Structures, parameters, and transmission properties of optical fibers. Proc. IEEE 68:1175 (1980) 11, Cohen, L. G., C. L, Mammel, and H. M. Presby, Correlation between numerical predications and measurements of single-mode fiber dispersion characteristics. Appl. Opt. 19:2007 (1980). 12, ‘Tsuchiya, H., and N. Imoto, Dispersion-free single mode fibers in 1.5 um wavelength region. Electron. Lett. 15:476 (1979). 13, Cohen, L. G., C. Lin, and W. G. French, Tailoring zero chromatic dispersion into the 1.5-1.6 um low-loss spectral region of single-mode fibers. Electron. Lett. 15:34 (1979) 14, White, K. I, and B. P, Nelson, Zero total dispersion in step-index monomode fibers at 1.30 and 1.55 wm. Electron. Lett. 15:396 (1979) 15. Gambling, W. A., H. Matsumara, and C. M. Ragdale, Zero total dispersion in graded-index. single-mode fibers. Electron, Lett, 15:474 (1979). 16. Bloom, D. M., L. F. Mollenauer, Chinlon Lin, and A. M. Del Gaudio, Demonstrat propagation in km-length fibers. Opt. Lert. 4:297 (1979). 17. Monroe, M., Propagation in doubly clad single mode optical fibres. IEEE J. Quantum Electron 18(4):535 (1992). 18. Li, ¥., C. D. Hussey, and T. A. Birks, Triple-clad single-mode fibers for dispersion shifting IEEE J. Lightwave Technol. 11:1812 (1993). 19, Li, Y. and C. D, Hussey, Triple-clad single-mode fibers for dispersion flattening. Opr. Eng 33:3999 (1994). 20. Nagayama, K., T. Saitoh, M. Kakui, K. Kawasaki, M. Matsui, H. Takamizawa, H. Miyaki, Y. Ooga, I. Tsuchiya, and Y. Chigusa, Ultra low loss pure silica core fiber and its impact on submarine transmission systems. Tech. Digest of OFC2002, PD FA10 (2002). 21, Nagayama, K.,M. Kakui, M. Matsui, T. Saitoh, and Y. Chigusa, Ultra low loss pure silica core fiber and extension of transmission distance. Electron. Lett. 38:1168 (2002). 22. Yeh, P., A. Yariv, and E. Marom, Theory of Bragg fibers. J. Opt. Soc. Am. 68(9):1196 (1978). 23. Kanamori, H., H. Yokota, G. Tanaka, M. Watanabe, Y. Ishiguro, I. Yoshida, T. Kakii, S. tou, Y. Asano, and S. Tanaka, Transmission characteristics and reliability of pure silica core single-mode fibers. J. Lightwave Technol. 4(8):1144 (1986). 24, Miya, T., Y. Terunuma, T. Hosaka, and T. Miyashita, Ultimate low-loss single-mode fiber at 1.55 pm, Electron. Lett, 18:106 (1979). 25. Suematsu, Y., Long wavelength optical fiber communication, Proc. IEEE 71:692 (1983). 26. Miller, S., and I. P. Kaminow (eds.), Optical Fiber Telecommunication II. Academic Press, San Diego, 1988, Chapters 2 and 3. 27. Marcuse, D., Bend loss of slab and fiber modes computed with diffraction theory. IEEE J. Quantum Electron. 29:2957 (1993). 28. Amaud, J. A., Transverse coupling in fiber optics, Part Ill: Bending loss, Bell Syst. Tech. J. 53:1379 (1974), 29. Marcatili, E. A. J., Bends in optical dielectric guides. Bell Syst. Tech. J. 48:2103 (1969). 30, Marcuse, D., Curvature loss formula for optical fibers. J. Opt. Soc. Am, 66:216 (1976) 31. Tsao, C., Optical Fiber Waveguide Analysis. Oxford University Press, Oxford, UK, 1992, Chapter 11. 32. Lewin, L., Radiation from curved dielectric slabs and fibers. /EEE Trans. Microwave Theory Tech. 22:718 (1974). n of pulse Additional Reading 135 33. Gambling, W. A., D. N. Payne, and H. Matsumura, Radiation from curved single-mode fibers. Electron. Lett, 12:567 (1976). 34, Sakai, J., and 7. Kimura, Bending loss of propagation modes in arbitrary-index profile optical fibers. Appl. Opt. 17:1499 (1978). 35. Snyder, A. W., and J. D. Love, Optical Waveguide Theory. Chapman & Hall, New York, 1983. 36. Kawakami, S., M. Miyagi, and S. Nishida, Bending losses of of dielectric slab optical waveguide with double or multiple cladding: theory. Appl. Opt. 14:2588 (1975). 37. Vassallo, C., Scalar-field theory and 2-D ray theory for bent single-mode weakly guiding optical fibers. J. Lightwave Technol. 3:416 (1985), 38. Murakami, Y., and H. Tsuchiya, Bending losses of coated single-mode optical fibers. IEEE J. Quantum Electron. 14:495 (1978). 39. Morgan, R., J. S. Barton, P. G. Harper, and J. D. C. Jones, Wavelength dependence of bending loss in monomode fibers: effect of the buffer coating. Opt. Lett. 15:947 (1990). 40. Valiente, I, and C. Vassallo, New formalism for bending losses in coated single-mode optical fibers. Eleciron. Lett. 25:1544 (1989). 41, Renner, H., Bending losses of coated single-mode fibers: a simple approach. J. Lightwave Technol. 10:544 (1992). 42. See, for example, J. D. Jackson, Classical Electrodynamics, 3rd ed. Wiley, New York, 1999, p. 491, Equation (10-108). 43. Chen, K.-L., and S. Wang, An approximate expression for the effective index in symmetric DH lasers. IEEE J. Quantum Electron. 19:1354 (1983). ADDITIONAL READING Collin, R. E., Field Theory of Guided Waves. McGraw-Hill, New York, 1960.

You might also like