2 Electrical and Thermal Conduction in Solids ction involves the motion of charges in a material under the influence Jectric field, A material can generally be classified as a conductor if it umber of “free” or mobile charge carriers. In metals, due to the na onding, the valence electrons from the atoms form a sea of electrons Trove within the metal and are therefore called conduction electrons. In vill treat the conduction electrons in metal as “free charges” that can by an applied electric field. In the presence of an electric Field, the Fore y attain an average velocity, called the drift velocity, that depends on ying Newton’s second Iaw to electron motion and using such con time between electron collisions with lattice vibrations, crystal de- vt we will derive the fundamental equations that govern electrical ids, A key concept will be the drift mobility, which is a measure of the charge carriers in the solid drift under the influence of an external seal conductors, such as metals, are also known to be good thermal conduction of thermal energy from higher to lower temperature re- Jmvolves the conduction electrons carrying the energy. Consequently, tolationship between the electrical and thermal conductivities, which eory and experiments.n4 Current density | definition Definition of drift velocity Current density and drife velocity CHAPTER 2 + ELECTRICAL AND THERMAL CONDUCTION IN SOLIDS 2.1 CLASSICAL THEORY: THE DRUDE MODEL 2.1.1 METALS AND CONDUCTION BY ELECTRONS The electric current density J is defined as the net amount of charge flowing act unit area per unit time, that is Aq var where Aq is the net quantity of charge flowing through an area 4 in time Ar. Figurg shows the net flow of electrons in a conductor section of cross-sectional area i presence of an applied field %,. Notice that the direction of electron motion is opp to that of the electric field £, and of conventional current, because the electrons ex ence a Coulombic force e£, in the x direction, due to their negative charge. We know that the conduction electrons are actually moving around random! the metal, but we will assume that as a result of the application of the electric field they all acquire a net velocity in the x direction. Otherwise, there would be no net fl of charge through area A The average velocity of the electrons in the x direction at time ¢ is denoted « This is called the drift velocity, which is the instantaneous velocity vy in the x di tion averaged over many electrons (perhaps, ~10°* m~*); that is var = slo + at Us to + Uewl where v,; is the x direction velocity of the ith electron, and N is the number conduction electrons in the metal. Suppose that n is the number of electrons per volume in the conductor (n= N/V). In time Ar, electrons move a dist Ax = vue At, so the total charge Ag crossing the area A is enA Ax. This is v because all the electrons within distance A.x pass through A; thus, n(A A.x) is thet number of electrons crossing A in time Ar ‘The current density in the x direction is Aq _ enAv AAI A Ar This general equation relates J, to the average velocity vax of the electrons. It mus appreciated that the average velocity at one time may not be the same as at ancl £,(0). We the time, because the applied field, for example, may be chan fore allow for a time-dependent current by writin; I(t) = envas(t) ti To relate the current density J, to the electric field £,, we must examine the ef ic field on the motion of the electrons in the conductor. To do so, we i of the elec consider the copper crystal ed Allheconducon elecFigure 2.1 Drift of electrons in o ‘conductor in the presence of an applied electric field Electrons drift with an average velocity vax in the x direction. pperatom has a single valence electron in its 4s subshell, and this electron Shound. The solid metal consists of positive ion cores, Cu”, at regular sites, eentered cubic (FCC) crystal structure. The valence electrons detach them- their parents and wander around freely in the solid, forming a kind of elec Or gas. These mobile electrons are free to respond to an applied field, creat Bn density J,. The valence electrons in the electron gas are therefore electrons. active forces between the negative electron cloud and the Cu* ions are re- Or metallic bonding and the existence of the solid metal. (This simplistic fal was depicted in Figure 1.7 for copper.) The electrostatic attraction be- Snduction electrons and the positive metal ions, like the electrostatic attrac the electron and the proton in the hydrogen atom, results in the conduction fing both potential energy PE and kinetic energy KE. The conduction el pout the crystal lattice in the same way that gas atoms move randomly in a tof electrostatic interactions. KE of the conduction electrons in a metal is primarily determined Qstatic interaction of these electrons with the positive metal ions and hh other. For most practical purposes, we will therefore neglect the Wependence of the mean KE compared with other factors that control of the conduction electrons in the metal crystal. We can speculate 1.1, that the magnitude of mean KE must be comparable to the the mean PE of electrostatic interaction? or, stated differently, to the ergy which is several electron volts per atom. If w is the mean speed ion electrons, then, from electrostatic interactions alone, we expect Several electron volts which means that w is typically ~10° m s~'. This is relatively temperature insensitive and much greater than that kinetic molecular theory. The true reasons are quantum mechanical ed in Chapter 4. (They arise from what is called the Pauli exclusion in clssical mechanics called the vitial theorem, which totes that fora collection of poricles, BF He mognitde of te mean PE tthe only forces acing on the particles are such thal they Boor low dependence on the particle-partcle separation (as in Covlombic and gravitationalCHAPTER 2 © ELECTRICAL AND THERMAL CONDUCTION IN SOLIDS ‘Vibrating Cu* ions (e) A conduction electron in the electron gas moves gbout randomly in amet (wih mean speed o ing frequently and randomly scattere Fn dela af th oor pe cbnce of ‘on applied field there is no net drift in any direction. {b) In the presence of an applied field, there is a net drift along the x direction. This net drift ‘long the force of the fel is superimposed on the random motion of he electron. Aer many scattering events the electron has been displaced by a net distance, 4.x, from its inital position toward the pesitve terminal. Figure 2.2. Motion of a conduction electron in a metal In general, the copper crystal will not be perfect and the atoms will not be s tionary. There will be crystal defects, vacancies, dislocations, impurities, etc., whi will scatter the conduction electrons. More importantly, due to their thermal ener the atoms will vibrate about their lattice sites (equilibrium positions), as depicted Figure 2.2a. An electron will not be able to avoid collisions with vibrating ato ‘scattered” from one atom to another. In the absence of consequently, it will be applied field, the path of an electron may be visualized as illustrated in Figure 2. from lattice vibrations causes the electron to move randomly in lattice. On those occasions when the electron reaches a crystal surface, it becom deflected” (or “bounced”) back into the crystal. Therefore, in the absence of where scatterin: jeld, after some duration of time, the electron crosses its initial x plane posit again. Over a long time, the electrons therefore show no net displacement in any a irection. ‘When the conductor is connected to a battery and an electric field is applied tot! crystal, as shown in Figure 2.2b, the electron experiences an acceleration in the direction in addition to its random motion, so after some time, it will drift a finite di tance in the x direction. The electron accelerates along the x direction under the acti of the force e£,, and then it suddenly collides with a vibrating atom and loses t gained velocity. Therefore, there is an average velocity in the.x direction, which, if culated, determines the current via Equation 2.2. Note that since the electron expe ences an acceleration in the x direction, its trajectory between collisions is a parabul like the trajectory of a golf ball experiencing acceleration due to gravity To calculate the drift velocity vi, of the electrons due to applied field 4 consider the velocity v,; of the ith electron in the x direction at time 1. Suppose its collision was at time 1); therefore, for time (1 ~ 1;), it accelerated free of collisions,2.1 CLAssicaL Turory: THe Drupe Mopet 7 ined alo Present time Va M3 SLast collision ¥ ec 1. lecton 3 1, Freetime 5 1 a 1 Velocity gained in the x direction at time #from the electric field (£,) for three electrons, INelectrons to consider in the meta Gigure 2.3. Let ua; be the velocity of electron i in the x direction just after We will call this the initial velocity. Since e£,/m, is the acceleration of Mihe velocity v,; in the x direction at time f will be Ug = Wai t= 8) Ihis is only for the ith electron. We need the average velocity vg, for all along x. We average the expression for = | to N electrons, as in Equa- sume that immediately after a collision with a vibrating ion, the electron any random direction; that is, it can just as likely move along the nega x, so that w,; averaged over many electrons is zero. Thus, Drift velocity [og + ta $s + Den is the average free time for NV electrons between collisions. ml r is the mean free time, or the mean time between colli ean scattering time). For some electrons, (1 ~ f;) will be greater than it will be shorter, as shown in Figure 2.3. Averaging (1 — ¢;) for N be the sume as r. Thus, we can substitute r for (¢ — 1) in the previous ons (also [23] 3 shows that the drift velocity increases linearly with the applied field proportionality ez/m, has been given a special name and symbol. Itis mobility j.,, which is defined as Definition of Var = a3 (24) rift mobility 7 Drift mobility bes 12.5] and mean free < time 5 relates the drift mobility of the electrons to their mean scattering le, r, which is also called the relaxation time, is directly related tone Ohm's law Unipolar conductivity Drift velocity CHAPTER 2 + ELECTRICAL AND THERMAL CONDUCTION IN SOLIDS the microscopic processes that cause the scattering of the electrons in the metal; tat lattice vibrations, crystal imperfections, and impurities, to name a few From the expression for the drift velocity vj., the current density J, followsi mediately by substituting Equation 2.4 into 2.2, that is, Jy = enw. Therefore, the current density is proportional to the electric field and the cond tivity o is the term multiplying £,, that is, o=enu It is gratifying that by treating the electron as a particle and applying chanics (F = ma), we are able to derive Ohm’s law. We should note, however, that assumed r to be independent of the field Drift mobility is important because it is a widely used electronic parameter semiconductor device physics. The drift mobility gauges how fast electrons will d when driven by an applied field. If the electron is not highly scattered, then the me free time between collisions will be long, r will be large, and by Equation 2 drift mobility will also be large; the electrons will therefore be highly mobile and able to “respond” to the field. However, a large drift mobility does not necessadl imply high conductivity, because o also depends on the concentration of conduct electrons n The mean time between collisions r has further significance. Its reciprocal | represents the mean frequency of collisions or scattering events; that is, |/r is mean probability per unit time that the electron will be scattered (see Example Therefore, during a small time interval 5r, the probability of scattering will bed The probability of scattering per unit time 1/z is time independent and depends o on the nature of the electron scattering mechanism. There is one important assumption in the derivation of the drift velocity Equation 2.3. We obtained vj. by averaging the velocities v,; of N electrons al at one instant, as defined in Equation 2.1. The drift velocity therefore represents average velocity of all the electrons along x at one instant; that is, vg, is a number erage at one instant. Figure 2.2b shows that after many collisions, after a time inter At > r, an electron would have been displaced by a net distance Ax along x term Ax/Ar represents the effective velocity with which the electron drifts along x is an average velocity for one electron over many collisions, that is, over a long tf (hence, Ar > r), so Ax/At is a time average. Provided that Ar contains many cil sions, it is reasonable to expect that the drift velocity Ax/Az from the time avera one electron is the same as the drift velocity v,, per electron from averaging for electrons at one instant, as in Equation 2.1, or The two velocities are the same only under steady-state conditions (Ar >> r). proof may be found in more advanced texts.1 Dupe MopeL 19 2.1 Classica THEORY: 'SCATTERING PER UNIT TIME AND THE MEAN FREE TIME If 1/r is defined Etre bility pr unit time that an electron is scattered, show that the mean time be- itesimally small time interval ds at time r. Let NV be the number of unscattered i. The probability of scattering during dr is (1/r) dr, and the number of scat- ing dt is N(1/t) de. The change dN in N is thus wwe-n(!)a sign indicates a reduction in NV because, as electrons become scattered, N' de- 8 this equation, we can find N’ at any time , given that at time ¢ = 0, No is fof unscattered electrons. Therefore, Unscattered v=, ee(~!) aaron r concentration represents the number of unscattered electrons at time f. It reflects an expo- for the number of unscattered electrons. The mean free time 7 can be calcu: thematical definition of 7 ~ IN dt Mean free i= f; time i N = Noexp(~1/r). Clearly, 1/¢ is the mean probability of scattering per MOBILITY IN METALS Calculate the drift mobility and the mean scatter Etna electrons in copper at room temperature, given that the conductivity of copper fem”! The density of copper is 8.96 g cm” * and its atomic mass is 63.5 g mol Nat Iu, from o = enj because we already know the conductivity #. The number In per unit volume can be taken as equal to the number of Cu atoms per unit me that each Cu atom donates one electron to the conduction electron gas in le of copper has N, (6.02 x 108) atoms and a mass of 63.5 g. Therefore, the ‘atoms per unit volume is 8.96 gcm~>, and My = atomic mass = 63.5 (g mol”). Substituting for 8.5 x 10 electrons em? drift mobility is therefore « 5.9 x 10° 2*! em” en ((1.6 x 10-™ CVB.S x 107 em] =B4em? v's!hl RyiaL. CONDUCTION IN SOLIDS 120 CHAPTER 2 © ELFCTRICAL AND From the drift mobility we can calculate the mean free un Equation 25. (9.1 x10) PC (43.4 < 10m? V - (16 ion electrons is about 1.5 x 10° ms. Note that the mean speed w of the conduct path is about 37 nm. } their mean free Wt DRIFT VELOCITY AND MEAN SPEED What is the applied electric field that will impose sapeity equal to 0-1 percent of the mean speed w(~10" m § ") of conduction electro a Cu wire of diam Jing current density and current thro copper? What is the correspond SOLUTION where zy is the drift mobility. 0.001u = 10° ms~', wehave jon electronsis vy. = Hu The drift velocity of the conducti 2.2). With v ‘opper is 43.4em? V"' s" (sce Example fore 230 KV = Bax lot mves! jeld in a metal, Given the 10° ms Bare This is an unattainably large electric fi conductivity «of coppet equivalent current density is Jy = OF, = (5.9 x 107 Qt m}2.3 x 10° Vm") = 14x10" Am or 14x 107 Amm 1 mm diameter wire! It is clear from A through a ighest working currents and W ven under the hij he mean speed of the electrons. C practical purposes, th fans a current of 1.1 x 10) \ctical purposes, © hh smaller than ied to a conductor, for all This me example that for all prai ges, the drift velocity is muc! when an electric field is appli an speel | ———— ae Wt DRIFT VELOCITY IN A FIELD: A CLOSER LOOK There 's another way to explain the obse dependence of the drift velocity on the field, and Equayidn 2.3. Consider the path of acon tion electron in 2.4, Suppose that at time ¢ = 0 the tron has just been scattered fi al velocity in ection just after this initial mediately after ‘as shown in Figure attice vibration. Let x1 be the initia ‘a collision number of 2er0 a random an applied field rom a li 1 collision (to which we assig ‘collision, the velocity of the Since e er) will be din will assume that imn tion. Suppose that the tance s; covered in the x electron is in first collision occurs at time f ‘m, isthe acceleration, th Distance direction during the free tim traversed ‘Ys | along x before sy =u t5( ha lattice vibration (its first collision), and th .en repeated during the next inte jong.x, and So on. collision At time 1), the electron collides with The whole process is th on traverses a distance s> al tering events, We ain to become and the electro is randomized which lasts for a free time f: distance traversed by the elect the overall ron after p such scatl sum2.1 CLASSICAL ORY: THE DRUDE M DEL 121 1 —>s Finish Collison e " Figure 2.4. The motion ofa single elecron inthe presence of on elecic Feld ©. During time intra he > electron traverses a distance s; along x. Ax After p collisions, it has drifted a distance Distance difted in total time Ar sax for p fee time i 1(e&\r. 5 3 = lath tate ot Mant + (SYP tee te) 2) Collision the “initia!” velocity u, is always random, the first term has u, val ly negative and positive, so for many collisions (large p) the first term on the Equation 2.8 is nearly zero and can certainly be neglected compared with the afier many collisions, the net distance s = Ax traversed in the x direction is 2.8, which is the electric field induced displacement term. Distance drifted after p yon (SE) scattering 2\'m, events Mean square is he mean free time between collisions, where r + tpy/p. free time forward elementary statistics it can be shown that = 2(7 27. Soin definition free time r between collisions, the overall distance s = A.x drifted in the ip collisions is pr’) the total time Ar taken for these p scattering events is simply pr, the drift len by Ax/Ar ors/(pr), that is, = Drift velocity s [29] and mean free time me expression as Equation 2.3, except that r is defined her as the average le electron over a lon ne, that is, over many collisions, whereas previously fe time averaged over many electrons. Further, in Equation 2.9 vy, is an electron over a long time, over many collisions. In E uation 2.1 vy, is theDrift mobility and conducti vity and mean free path CHAPTER 2 + ELECTRICAL AND THERMAL CONDUCTION IN SOLIDS average velocity averaged over all electrons at one instant. For all practical purposes, the are equivalent. (The equivalence breaks down when we are interested in events over ai scale that is comparable to one scattering, ~10-" second.) The drift mobility eg from Equation 2.9 is identical to that of Equation 2.5, eg = er Suppose that the mean speed of the electrons (not the drift velocity) is u. Then an ele moves a distance ¢ = ut in mean free time r, which is called the mean free path. Ted ‘mobility and conductivity become, and = erty = ta Equations 2.3 and 2.10 both assume that after each collision the velocity is randomia process, latice scatter ing. In general not all electron scattering processes can randomize the velocity in one scat process. If it takes more than one collision to randomize the velocity, then te electrons carry with it some velocity gained from a previous collision and hence possesses a mobility. In such eases one needs to consider the effective mean free path a carrer has tom to eventually randomize the velocity gained; this isa point considered in Chapter 4 when calculate the resistivity at low temperatures 2.2 TEMPERATURE DEPENDENCE OF RESISTIVITY: IDEAL PURE METALS When the conduction electrons are only scattered by thermal vibrations of t metal ions, then r in the mobility expression 1 = et/m, refers to the mean ti between scattering events by this process. The resulting conductivity and resistivi are denoted by o7 and pr, where the subscript T represents “thermal vibration sc tering. To find the temperature dependence of o, we first consider the tempera dependence of the mean free time r, since this determines the drift mobility. Ancl tron moving with a mean speed u is scattered when its path crosses the cred sectional area S of a scattering center, as depicted in Figure 2.5. The scattering cent Figure 2.5 Scattering of an electron from the thermal vibrations ofthe atoms The electron travels o mean distance ¢ = ur between collisions. Since the scattering cross: sectional area is 5, in the volume S¢ there least one scaterer, N, [Sue] = 1 \ Electron2.2 TEMPERATURE DEPENDENCE OF RESISTIVITY: IDEAL PURE METALS ling atom, impurity, vacancy, or some other crystal defect. Since r is taken for one scattering process, the mean free path € of the electron ing processes is ur. If N, is the concentration of scattering centers, ime Sé, there is one scatteririg center, that is, (Sur)N, = 1. Thus, the is given by Mean free [2111 time between collisions L SuN. speed u of conduction electrons in a metal can be shown to be only mature dependent In fact, electrons wander randomly around in the ith an almost constant mean speed that depends largely on their cor hence on the crystal material. Taking the number of scattering centers tobe the atomic concentration, the temperature dependence of r then ly from that of the cross-sectional area S. Consider what a free electron caches a vibrating crystal atom as in Figure 2.5. Because the atomic xdom, the atom covers a cross-sectional area ra”, where a is the am- fbrations. If the electron’s path crosses 7a”, it gets scattered. Therefore. between scattering events z is inversely proportional to the area za’ electron, that is, ro 1/za ial vibrations of the atom can be considered to be simple harmonic fhe same way as that of a mass M attached to a spring. The average Of the oscillations is } Maw, where w is the oscillation frequency: fic theory of matter, this average kinetic energy must be on the order re Mao? = LAT tively, this is correct because raising the temperature increases the am. mnic vibrations. Thus, perature-independent constant. Substituting for r in ug = et/me, We ec Me nT of a metal is, «@ metal is only weckly temperature dependent can be proved from no elecrons in @ metal (see Chapter 4), This result contrast olecdor theory of goses (Chapter 1], which predicts that the mean speed of molecules is the tine being, we simply use 0 constant mean speed u forthe conduetion electrons in @128 CHAPTER 2 + ELFCTRICAL AND THERMAL CONDUCTION IN SOLIDS Pure metal that is, resistivity due pr =AT t 10 thermal where A is a temperature-independent constant, This shows that the resistivity of a purea wire increases linearly with the temperature, and that the resistivity is due simply to the scl thermal vibrations of the atoms. We term this conduct ity. vibrations of the crystal ing of conduction electrons by the lattice-scattering-limited conduct Ete ‘TEMPERATURE DEPENDENCE OF RESISTIVITY What is the percentage chang of a pure metal wire from Saskatchewan's summer to winter, neglecting the changes in the mensions of the win SOLUTION (°C for the summer and perhaps —30°C for the winter, from R cx p = AT, web Assuming 2 (20 +273) =0.171 or 17% Notice that we have used the absolute temperature for 7. How will the outdoor cable po losses be affected? DRIFT MOBILITY AND RESISTIVITY DUE TO LATTICE VIBRATIONS Given that the mean s of conduction electrons in copper is 1.5 x 10° m s~! and the frequency of vibration of the * estimate the drift mobility of electrons per atoms at room temperature is about 4 x 10" s the conductivity of copper. The density d of copper is 8.96 g cm™* and th 63.56 g mol! atomic mass SOLUTION ng the drift mobility and hence the conductivity is based on evalu The method for calculati Equation 2.11, that is, 7 = 1/SuN,. Since t is due to scattering fi the mean free time v atomic vibrations, N, is the atomic concentration, dNq _ (8.96 x 10° kg m™*)(6.02 x 10" mol”! M, 63.56 x 10-* kg mol! =85 x 10% m> a? depends on the amplitude a of the thermal vibrations y KE, associated with a vibrating mas. where « is the angular frequency of ion to the vibrating atom and equal The cross-sections shown in Figure 2.5. The average kinetic ener atached to a spring is given by KEy = + vibration (w = 224 x 10 rad s~'). Applying this equ the average kinetic energy KEy, to 4kT. by virtue of equiparttion of energy theorem, we a? = 2kT/Mo* and thus 2 (1.38 x 10°? J K~')(300 K) S=na x 4 x 10" rad s“')2.3 MATTHIESSEN’S AND NoRDHEIM’S RULES © G9 10-= m(L.S x 10% ms-)(8.5 x 10 ms) 20x 10" s (1.6 x 10° €)(2.0 x 10-5) (O.1 x 10" key = 35x 103m? V's"! = 35em?V"!5 ty is then enjtg = (1.6 x 10- C)(8.5 x 10 em=3)(35 em? V-! 5!) = 4.8 x 10° 9°! m= ally measured value for the conductivity is 5.9 x 10° 2-! em™', so our sed on Equation 2.11 is actually only 18 percent lower, which is not bad for might have surmised, the agreement is brought about by using reasonable speed w and the atomic vibrational frequency «. These values were taken janical calculations, so our evaluation for t was not truly based on classical ITESSEN’S AND NORDHEIM’S RULES ESSEN’S RULE AND THE TEMPERATURE duction that considers scattering from lattice vibrations only works tals; unfortunately, it fails for metallic alloys. Their resistivities rature dependent, We must therefore search for a different type of ism, tal alloy that has randomly distributed impurity atoms. An electron red by the impurity atoms because they are not identical to the host ed in Figure 2.6. The impurity atom need not be larger than the host ler. As long as the impurity atom results in a local distortion of the Will be effective in scattering. One way of looking at the scattering impurity is to consider the scattering cross section. What actually n is a local, unexpected change in the potential energy PE of the oaches the impurity, because the force experienced by the electron an impurity atom of a different size compared to the host atom is sta lattice, the impurity atom distorts the region around it, either by126 Overall frequency of scattering CHAPTER 2 + ELECTRICAL AND THERMAL CONDUCTION IN SOLIDS Strained region by impurity exerts a force F'=—d(PE) /dx Fae 2 rosters TO C-OQ|O O-O scattering processes involving scomalag hen reperieg ae id Oo O-OL0-O- OQ | from thermal vibrations alone. pushing the host atoms farther away, or by pulling them in, as depicted in Figure J The cross section that scatters the electron is the lattice region that has been elastic distorted by the impurity (the impurity atom itself and its neighboring host atoms) that in this zone, the electron suddenly experiences a force F = —d(PE)/dx duet sudden change in the PE. This region has a large scattering cross section, since the tortion induced by the impurity may extend a number of atomic distances. These im rity atoms will therefore hinder the motion of the electrons, thereby increasing resistance We now effectively have two types of mean free times between collisions: one, for scattering from thermal vibrations only, and the other, r,, for scattering from i purities only. We define ty as the mean time between scattering events arising fr thermal vibrations alone and r, as the mean time between scattering events ais from collisions with impurities alone. Both are illustrated in Figure 2.6. In general, an electron may be scattered by both processes, so the effective m free time r between any two scattering events will be less than the individual scat ing times rr and r). The electron will therefore be scattered when it collides with eit an atomic vibration or an impurity atom. Since in unit time, I/r is the net probabil of scattering, 1/rr is the probability of scattering from lattice vibrations alone, I/t1 is the probability of scattering from impurities alone, then within the realm elementary probability theory for independent events, we have ip 4 =—+ In writing Equation 2.13 for the various probabilities, we make the reasonable sumption that, to a greater extent, the two scattering mechanisms are essentially i pendent. Here, the effective mean scattering time r is clearly smaller than both r 1). We can also interpret Equation 2.13 as follows: In unit time, the overall nun2.3 MATTHIESSEN’S AND NoRDHEIM’S RULES } is the sum of the number of collisions with thermal vibrations alone number of collisions with impurities alone (1/1) Mobility jz, depends on the effective scattering time t via ug 1B can also be written in terms of the drift mobilities determined by the g mechanisms. In other words, et/ mes t=“ —+— [2.14] Ma Me lattice-seattering-limited drift mobility, and jz; is the impuri lited drift mobility. By definition, 4, = etr/me and py = ety/me 1 1 nag en, enw p=prt+pr (2.15) jy is defined as the resistivity due to scattering from thermal vibrations, the resistivity due to scattering from impurities, or 1 1 pr=—— and pp=— enw enys Presult in Equation 2.15 simply states that the effective resistivity p is the Shributions. First, pr = 1/enj1, is the resistivity due to scattering by ther- fof the host atoms. For those near-perfect pure metal crystals, this is the Ohiribution. As soon as we add impurities, however, there is an additional © 1 /enj.;, which arises from the scattering of the electrons from the Hirst term is temperature dependent because ty x T~' (see Section term is not ifime 1, between scattering events involving electron collisions with im- depends on the separation between the impurity atoms and therefore on on of those atoms (see Figure 2.6). If ¢; is the mean separation between then the mean free time between collisions with impurities alone will be Hemperature independent because €, is determined by the impurity con- ¢. {) = N,"”), and the mean speed of the electrons u is nearly con- Inthe absence of impurities, r, is infinitely long, and thus p; = 0. The of resistivities from different scattering mechanisms, as shown by js called Matthiessen’s rule. talso be electrons scattering from dislocations and other crystal defects, ‘grain boundaries. All of these scattering processes add to the resistiv- ist as the scattering process from impurities. We can therefore write the ity of a metal as p= pr+Pr [2.16] 127 Effective drift mobility Matthiessen's rule Resistivities due to lattice and impurity scattering Matthiessen’s ruleDefinition of temperature coefficient of resistivity Temperature dependence of resistivity CHAPTER 2 + ELECTRICAL AND THERMAL CONDUCTION IN SOLIDS where p¢ is called the residual resistivity and is due to the scattering of electrons impurities, dislocations, interstitial atoms, vacancies, grain boundaries, etc. (Whi means that px also includes p,). The residual resistivity shows very little temperat dependence, whereas py = AT, so the effective resistivity p is given by prAT+B where A and B are temperature-independent constants. Equation 2.17 indicates that the resistivity of a metal varies almost linearly wi the temperature, with A and B depending on the material. Instead of listing A and B resistivity tables, we prefer to use a temperature coefficient that refers to small, m malized changes around a reference temperature. The temperature coefficient y (TCR) ao is defined as the fractional change in the resistivity per unite perature increase at the reference temperature Ty, that is, I [2] % PoLdT Sy, where po is the resistivity at the reference temperature To, usually 273 K (0 °C) 293 K (20°C), and 5p = p — pg is the change in the resistivity due to a small inc in temperature, 67 = T — Tp When the resistivity follows the behavior p © AT + B in Equation 2.17, th according to Equation 2.18, ao is constant over a temperature range Ty to T, and Eq tion 2.18 leads to the well-known equation, P= poll + ao(T — T)] taf Equation 2.19 is actually only valid when ao is constant over the tempera range of interest, which requires Equation 2.17 to hold. Over a limited temperatu range, this will usually be the case. Although it is not obvious from Equation 2] we should note that a» depends on the reference temperature To, by virtue of depending on To The equation p = AT, which we used for pure-metal crystals to find the chan in the resistance with temperature, is only approximate; nonetheless, for pure meial it is useful to recall in the absence of tabulated data. To determine how good formula p = AT is, put it in Equation 2.19, which leads to ay = Ty '. If we take t reference temperature Ty as 273 K (°C), then a is simply 1/273 K; stated differen Equation 2.19 is then equivalent to p = AT Table 2.1 shows that p x T is not a bad approximation for some of the fami pure metals used as conductors (Cu, Al, Au, etc.), but it fails badly for others, such indium, antimony, and, in particular, the magnetic metals, iron and nickel The temperature dependence of the resistivity of various metals is shown in Fj ure 2.7, where it is apparent that except for the magnetic materials, such as iron nickel, the linear relationship p ox 7 seems to be approximately obeyed almost all way to the melting temperature for many pure metals. It should also be noted that the alloys, such as nichrome (Ni-Cr), the resistivity is essentially dominated by residual resistivity, so the resistivity is relatively temperature insensitive, with av small TCR.2.3 MATTHIESSEN’S AND NORDHEIM’S RULES 129 iy, thermal coefficient of resistivity a» at 273 K (°C) for various metas, The ily index nin p oT” for some of the metels is also shown, 1.80 Magnetic metal; 273 < T < 1043 K 1.72 Magnetic metal; 273 < T < 627K the resistivity versus temperature behavior of pure metals can be sented by a power law of the form. ry" Resistivity of resistivity at the reference temperature J, and n is a characteristic fits the data. Table 2.1 lists some typical values for various pure met It is apparent that for the nonmagnetic metals, n is close to unity erto 2 than | for the magnetic metals Fe and Ni. In iron, for example. Sleciron is not scattered simply by atomic vibrations, as in copper, but netic interaction with the Fe ions in the lattice. This leads to a Bperature dependence. ir oversimplified theoretical analysis predicts a linear p = AT + B feresistivity down to the lowest temperatures, this is not true in reality190 EMAPTER 2+ ELECTRICAL AND TutRMAL CONDUCTION 18 SOLIDS 2009) Incone-8 ee Wire 1000 iron — Tungsten Monel-400 pet Tin 100 4 Platinum \ Copper Nickel — Silver 10 100 1000 10000 ‘Temperature (K) Figure 2.7 The resistivity of various metals as a function of temperature above 0 °C Tin melts at 505 K, whereas nickel and iron go through a magneticto- ‘nonmagnetic (Curie) transformation at about 627 K and 1043 K, respectively. The theoretical behavior (1 ~ Tis shown for reference. SOURCE: Data selectively exracied from various sources, including sections in Meols Handbook, 10th ed., 2and 3. Metals Park, Ohio: ASM, 1991 as depicted for copper in Figure 2.8. As the temperature decreases, typically bek ~100 K for many metals, our simple and gross assumption that all the atoms vibrating with a constant frequency fails. Indeed, the number of atoms that are vl ing with sufficient energy to scatter the conduction electrons starts to d with decreasing temperature, so the resistivity due to scattering from the tions becomes more strongly temperature dependent. The mean free time r = 1/Su becomes longer and strongly temperature dependent, leading to a smaller resist than the p o T behavior. A full theoretical analysis, which is beyond the scope of chapter, shows that p oc T°. Thus, at the lowest temperature, from Matthiessen’s ni the resistivity becomes p = DT‘ + pg, where D is a constant. Since the slope of p sus T is dp/dT = SDT*, which tends to zero as T becomes st2.3) Marriiessey’s ap Norbit 120 40. 60 80 100 TIK) 100 1000 10000 Temperature (K) The resistivity of copper from lowest to highest temperatures (ner pertue, 1358 K) on a loglog plot 100 K, pc, whereos ot low lemperatures, » x TS, and at the lowest appccoches the residual resistivity pg. The inset shows the p vs. T Bbsow 100K on linear plot. (nit 100 small on this scale.) eases toward 0K. This is borne out by experiments, as shown in er. Therefore, at the lowest temperatures of interest, the resistivity ring from impurities and crystal defects.* s of nickel are added to copper, the resistivity increases by virtue of r+ be pr, where pr isthe resistivity due to scattering from ther } trace impurities, etc. sad @, it resistivity arisi 135 K for ce Feodkist upercenduciviy in thich he rests vonthes [p= O}, even ine presence of ity and is quantum mechanical origin wll be explained in Chapter 8,CHAPTER 2 + ELECTRICAL AND THERMAL CONDUCTION IN SOLIDS 605 (Cu-3.32%Ni Cu-2.16%Ni (Cu-1.12%Ni (Defa Cu-L.12%Ni Resistivity ( 100%Cu (Deformed 100%Cu (Annealed Figure 2.9 Typical temperature dependence of the resistivity of ennealed and cold worked (deformed) copper containing various amounts of Ni in atomic percentoge. ia an | SOURCE: Data adapted from 0. ‘Ann Physik, 5, 219 (Germany, 1932 Temperature (K) alone (scattering from Ni impurity regions). Since p, is temperature independent, for amounts of Ni addition, p; will simply shift up the p versus T' curve for copper, by an amouni portional to the Ni content, , o Ni, Where Ny is the Ni impurity concentration, This i ap in Figure 2.9, where the resistivity of Cu-2,16% Ni is almost twice that of Cu-1.12% Ni working (CW) or deforming a metal results in a higher concentration of dislocations and ther increases the residual resistivity px by cw. Thus, cold-worked samples have a resistivity d that is shifted up by an additional amount pew that depends on the extent of cold working ‘TEMPERATURE COEFFICIENT OF RESISTIVITY «x AND RESISTIVITY INDEX nm If ci is the perature coefficient of resistivity (TCR) at temperature Ty and the resistivity obey’ t show that rr] What is your conclusion? Experiments indicate that n = 1.2 for W. What is its «ry at 20 °C? Given that, expe tally, a = 0.00393 for Cu at 20°C, what is 1’? SOLUTION Since the resistivity obeys p = y(T/T,)" , we substitute this equation into the definition of wliel= ale] Itis clear that, in general, ay depends on the temperature T, as well as on the refer temperature Ty. The TCR is only independent of T when n = 12.3 MATTIIESSEN’S AND NorDHEIM’s RULES: or n=ah B12, soatT = 7, = 293 K, we have aosx 0.0045, frequently found in data books. B= 0.00393. so that x 0.0041, which agrees reasonably 1.15, which agrees with the experimental value of n REFERENCE TEMPERATURES If a is the temperature coefficient of re BM exanrte 29 | perature T, and ay is the TCR at 7 show that T+ a7 vty at temperature 7 in terms of a and er pla —T)] and p= pill +a(T -7)) ms are expected to hold at any temperature T, so the first and second equa- respectively, give E roll + aT — 7) pill +a (To ~ T)1 ions can be readily solved to el THE FILAMENT OF A LIGHT BULB IW, 120 V incandescent light bulb. The tungsten filament is 0.381 m long and fof 33 jum. Its resistivity at room temperature is 5.51 x 10-* © m. Given that, of the tungsten filament varies at 7", estimate the temperature of the bulb ated atthe rated voltage, that is, when it is lit directly from a power out latcally in Figure 2.10. Note that the bulb dissipates 40 W at 120 V. clectrical power dissipated in the tungsten wire is radiated from the sur- jen. The radiated e tic power at the absolute temperature T can igure 2.10 Power radiated from a light bulb is equal to the electrical power dissipated in the filament134 cHA ER 2 + ELECTRICAL AND THERMAL CONDUCTION IN SOLIDS be described by Stefan’s law, as follows: P, susos = C75 A(T* as; is Stefan’s constant (5.67 x 10-* W m-? K-*), € is the emissivity of thes (0.35 for tungsten), A is the surface ar temperature (293 K). For T* > T, of the tungsten filament, and Tp is the the equation becomes 4 = cosAT Assuming that all the electrical power is radiated as electromagnetic waves from surface, estimate the temperature of the filament and compare it with your answer in pa SOLUTION 4a. When the bulb is operating at 120 V, itis dissipating 40 W, which means that the cure P 40W =i = 0.333 A Vv 10V ‘The resistance of the filament at the operating temperature 7 must be Vv 120 R=—= = 3602 1 0333 Since R = pL/A, the resistivity of tungsten at the operating lemperature T must be D?/4) _ 360: 22(33 x 10-* m) ron( Sea") The surface area is A = L(x D) = (0.381)(133 x 10-*) = 3.95 x 10-9 mv Then [fae] [ 40W cos | (0.35)5.67 x 10 Wm? K9G.95 x 10m) = [5.103 x 10"]""=2673K or 2400°C the two methods is less than 3 percent. 2.3.2 SOLID SOLUTIONS AND NORDHEIM’S RULE Inan isomorphous alloy of two metals, that is, a binary alloy that forms a solid solut we would expect Equation 2.15 to apply, with the temperature-independent imput contribution p; increasing with the concentration of solute atoms. This means that as alloy concentration creases, the resistivity p increases and becomes less tempera2.3 MATTHIESSEN’S AND NORDHEIM’S RULES 135 ble2.2_ The effect of alloying on the resistivity Resistivity at 20°C @ at 20 (nm) alk) o 0.006 129 0,003 1120 0.0003 400. 300. 204 » 100 Solid solution 0 —— D % «8 10 0. 2 40 6 8 100 Ni 100% Ni 100% Cu aL% Ni 100% Ni iogram of the Cu-Ni alloy system. (b The resistivity ofthe Cu-Ni alloy os © nicer incl he ud pose function of Ni content (ot) at room i Sregion, the quid (Lf a temperature. pees coun whereas below the = ely fe solid phase (a solid er (Cu-Ni alloy system ’ tos Handbook, 10th ed, 2 and 3, Metals Park, Ohio: ASM, 1991, ond M. Hansen ond York: MeGraw Hil, 1958. Sbverwhelms pr, leading tow < 1/273. This is the advantage of alloys ts. Table 2.2 shows that when 80% nickel is alloyed with 20% SSistvity of Ni increases almost 16 times. In fact, the alloy is called GS widely used as a heater wire in household appliances and industrial sample of the resistivity of a solid solution, consider the copper— nickel sgram for this alloy system is shown in Figure 2. 1a. Itis clear that the phase solid solution for all compositions. Both Cu and Ni have the structure, and since the Cu atom is only stightly larger than the Ni atom (easily checked on the Periodic Table), the Cu-Ni alloy will there- put with Cu and Ni atoms randomly mixed, resulting in a solid solu- added to copper, the impurity resistivity p, in Equation 2.15 will i concentration. Experimental results for this alloy system are shown Should be apparent that when we reach 100% Ni, we again have a pure fvity must be small. Therefore, p versus Ni concentration must pass . which for the Cu-Ni alloy seems to be at around ~50% Ni.Nordheim’s rule for solid solutions CHAPTER 2 + ELECTRICAL AND THERMAL CONDUCTION IN SOLIDS There are other binary solid solutions that reflect similar behavior to that der in Figure 2.11, such as Cu-Au, Ag-Au, Pt-Pd, Cu-Pd, to name a few. Quite oft use of an alloy for a particular application is necessitated by the mechanical propet rather than the desired electrical resistivity alone. For example, brass, which is Cu-30% Zn in solid solution, has a higher strength compared to pure copper, as 4 itis a suitable metal for the prongs of an electrical plug. An important semiempirical equation that can be used to predict the resistvi an alloy is Nordheim’s rule which relates the impurity resistivity ; to the a! fraction X of solute atoms in a solid solution, as follows: pr = CX(1 = X) where C is the constant termed the Nordheim coefficient, which represents the tiveness of the solute atom in increasing the resistivity. Nordheim’s rule assumes the solid solution has the solute atoms randomly distributed in the lattice, and random distributions of impurities cause the electrons to become scattered as whiz around the crystal. For sufficiently small amounts of impurity, experiments that the increase in the resistivity p, is nearly always simply proportional to thei rity concentration X, that is, p; o X, which explains the initial approximately equil crements of rise in the resistivity of copper with 1.11% Ni and 2.16% Ni addition shown in Figure 2.9. For dilute solutions, Nordheim’s rule predicts the same linea havior, that is, py = CX for X <1 Table 2.3 lists some typical Nordheim coefficients for various additions to cy and gold. The value of the Nordheim coefficient depends on the type of solute and solvent. A solute atom that is drastically different in size to the solvent atom will in a bigger increase in p; and will therefore lead to a larger C. An important assump Table 2.3 Nordheim coefficient C (at 20 °C] for dilute alloys obtained from pr= CXand X< 1 at." Solute in Solvent c (element in matrix) (12 m) Aun Cu matin 3500 Mn in Cu matrix 2900 Niin Cu matrix 1200 Sn in Cu matrix 2900 Zain Cu matrix 300 Cu in Au matrix 450 Man in Au matrix, 2410 Niin Au matrix 790 Sn in Au matrix 3360 Zain Au matrix 950 “INOTE: For many isomorphous olloys C may be diferent ot higher concentrations; tht is, it may depend on the composition of the alloy. SOURCES: D.G. Fink ond D. Chrisionsen, eds., Electronics Engineers’ Hondbook, 2nd ed New York, McGraw, 1982. JK. Sloniey, Elecricol and Magnetic Properties of Metals, M Park, OH, American Society for Metals, 1963. Solubility data from M. Hansen ond K. Ande, Constitution of Binary Alloys, 2nd ed., New York, McGrawil, 1985.’s RULES 197 DHE 2.3 MATTHIESSEN’S AND No! in Equation 2.21 is that the alloying does not significantly vary the jon electrons per atom in the alloy. Although this will be true for al- valency, that is, from the same column in the Periodic Table (e.g it will not be true for alloys of different valency, such as Cu and Zn, here is just one conduction electron per atom, whereas each Zn atom duction electrons. As the Zn content in brass is increased, more con- sbecome available per atom. Consequently, the resistivity predicted by high Zn contents is greater than the actual value because C refers to iget the correct resistivity from Equation 2.21 we have to lower C, ent to using an effective Nordheim coefficient Cu that decreases as mereases. In other cases, for example, in Cu-Ni alloys, we have to in- fi concentrations to account for additional electron scattering mech- pp with Ni addition, Nonetheless, the Nordheim rule is still useful for tivities of dilute alloys, particularly in the low-concentration region. 's rule in Equation 2.21, the resistivity of an alloy of composition Combined Matthiessen p and Nordheim By + pr is the resistivity of the matrix due to scattering from thermal rules m other defects, in the absence of alloying elements. To reiterate, the ds on the alloying element and the matrix. For example, C for gold in fferent than C for copper in gold, as shown in Table 2.3 fons, at some concentrations of certain binary alloys, such as 75% (0% Cu-S0% Au, the annealed solid has an orderly structure; that atoms are not randomly mixed, but occupy regular sites. In fact, Ons can be viewed as pure compound—like the solids CusAu and fivities of CusAu and CuAu will therefore be less than the same om alloy that has been quenched from the melt. As a consequence, ‘Nersus composition X curve does not follow the dashed parabolic fgrather, it exhibits sharp falls at these special compositions, as illus- 12. Pratrx + CX(1 — X) (2.22) Figure 2.12 Electrical resistivity vs. composition at room temperature in Cu-AW alloys, The quenched sample (dashed curve) is obtained by quenching the liquid, and the Cu and Au atoms ore randomly mixed. The resistivity ‘obeys the Nordheim rule. When the quenched sample is annealed or the liquid is slowly cooled (solid curve}, certain compositions (CusAu and CuAU) result in on ordered crystaline structure in which the Cu and Au atoms are positioned in an ordered fashion in the crystal and the scattering effect is reduced. 40 50 60 70 80 90 100 positon (at.% Au)Conductivity and mean free path CHAPTER 2 + ELECTRICAL AND THERMAL CONDUCTION IN SOLIDS NORDHEIM’S RULE The alloy 90 wi.% Au-10 wi.% Cu is sometimes used in low-vo electrical contacts, because pure gold is mechanically soft and the addition of copper incre hardness of the metal without sacrificing the corrosion resistance. Predict the resistivity alloy and compare it with the experimental value of 108 n& m. SOLUTION We apply Equation 2.22, p(X) = pax + CX(1 — X) but with 10 wt-% Cu converted atomic fraction for X. If w is the weight fraction of Cu, w = 0.1, and if My, and M atomic masses of Au and Cu, then the atomic fraction X of Cu is given by (see Example Me 0.1/63.55 = = 0.256 W/Mo, + (1 —W)/Ma, (0.1/63.55) + (0.90/197 Given that py, = 22.8 n@ mand C = 450 n@ m, P= pay + CX(I— X) = (22.809 m) + (450 n& m)(0.256)(1 — 0.256) = 108.5 nm This value is only 0.5% different from the experimental value RESISTIVITY DUE TO IMPURITIES The mean speed of conduction electrons in copperis 1.5 x 10° ms". Its room temperature resistivity is 17 n& m, and the atomic concentra in the crystal is 8.5 x 10% cm~*, Suppose that we add I at.% Au to form a solid solution is the resistivity of the alloy, the effective mean free path, and the mean free path duc sions with Au atoms only? SOLUTION According to Table 2.3, the Nordheim coefficient C of Au in Cu is 5500 n& m. With X = (1 at.%), the overall resistivity from Equation 2.22 is «+ CX(1 = X) = 17 n@ m + (5500 HQ m)(O.01)I — 0.01) = 17 nm + $4.45 nO m = 71.45 nQm Suppose that ¢ is the overall or effective mean free path and is the effective mean free tim tween scattering events (includes both scattering from lattice vibrations and impurities + and the effective drift mobility zy = et/me, the expression for the conductivity comes ent _ ent We can now calculate the effective mean free path ¢ in the alloy given that copper has a va of / and the electron concentration m = Ny 1 (1.6 x 10-"CV7(8.5 x 107 TSx 107 Om ~ Ox 10 RIS x 10 ms which gives €=8.8 nm, We can repeat the calculation for pure copper using V/pnasix = 1/(17 x 10°? & m) to find écy = 37 nm. The mean free path is reduced ayy mately by 4 times by adding only 1 at.% Au. The mean free path ¢, due to scattering from purities only can be found from Equation 2.13 multiplied through by 1/u, or by2.4 RESISTIVITY OF MIXTURES AND POROUS MATERIALS Equation 2.14 ss used in low-volté 2 pn of copper incre ict the resistivity B7 nm and ¢ = 8.8 nm, we find é; = 11.5 nm se calculations one step further. If N; is the impurity concentration in the IN, ~ 0.01185 x 10 m3) = 85 x 10® m->. The mean separation d; % Cu converted es canbe estimated roughly from d) ~ 1/N}”, which gives dr ~ 1.0 nm if Ma, and Mo fill Au atoms can be involved in scattering the electrons since ¢, is much jen by (see Example her way to Look ati sto say that it takes more than just one collision with ize the velocity of the electron.) = 0.256 TIVITY OF MIXTURES AND POROUS pont — 0 256) RIALS SENEOUS MIXTURES lonly applies to solid solutions that are single-phase solids. In other for homogeneous mixtures in which the atoms are mixed at the Ectrons in copper is whout the solid, as in the Cu-Ni alloy. The classic problem of mata eflective resistivity of a multiphase solid is closely related to th 1m a solid solution : effective dielectric constant, effective thermal conductivity, effe ean free path due to 4 lus, effective Poisson's ratio, etc., for a variety of mixtures, includ- jte materials as fiberglass. Indeed, many of the mixture rules are Material with two distinct phases « and , which are stacked in layers 500 n&@ m. With X Figure 2.13a. Let us evaluate the effective resistivity for current flow pid — 0.01) Continuous phase Dispersed phase | fective mean free ti ons and impurities), 1 A ae ooo ‘ Jn for the conductivi s eo OC] % + - L e @ o%,e eee in that copper has a vi 5 | i L L - _| aaa) = |r m™)e (a o) a ims) The effective resistivity of ¢ material with a layered structure. pure copper using fe path is reduced ay due to scattering fi by I/u, or by tion perpendicular to the layers. fecon parallel tothe plane of the layers ih o dispersed phase in a continuous matrix