62415@IEC:2010 -7 5.2 Via chains This is a chain of vias between metal layers connected in series. The via chain test structure shall contain at least 10 vias (see Figure 2). As an option, test structures may be used where the contacts between metal layers are formed by a number of vias in parallel.The number of vias per contact may be determined by the following requirement: Jvia test - Jvia-use. (1) Jine_test Jline_use Via size shall be theminimum design dimensions.Metal line lengthbetween vias shall exceed the Blech length,to avoid stress induced atomic back diffusion counteracting electromigration. EAAEDYBOOK SPLYUAU For line lengths <200 μm the Blech effect shall be verified. Via current density is defined as the current divided by the via area (ignoring current crowding). Voltage terminal Current 88 [88 88 X terminal O-F Voltage terminal IEC1122/10 Figure2-TEGofelectromigrationevaluationforvias 5.3 Contact chains This is a chain of contacts to n+ in substrate or p-well, or p+ in n-well. The number of contacts shall be kept low as the voltage required to force the stress current is limited by the junction breakdownvoltage. Contact size shall be the minimum design dimension. Metal length between contacts shall exceed the Blech length. For line lengths <200 μm the Blech effect shall be verified. Contact current density is defined as the current divided by the contact area (ignoring current crowding). 6 Testconditions Current density values are determined by the constraints that too low currents cause long test times, and too high currents may cause non-uniform heating and irrelevant failures.Practical values are in the order of 105A/cm2-1o6A/cm2 for both Al and Cu lines.For contacts and vias, 10 times the design limit is typically used. It shall be verified if Joule heating is significant. This verification is done by determining the temperaturecoefficient ofresistance ofthemetalline,and comparing theresistance atthe test condition with the resistance at low current density. When Joule heating is significant the line temperature shall be corrected for Joule heating [4] and data shall be available to demonstrate that the failure mechanism has not changed. - 8 - 62415@IEC:2010 The typical test conditions shown above guarantee usually sufficient degradation in a reasonable time (days orweeks). 7 Failure criteria Open failure: typically 10 % - 30 % resistance change. Shortfailure:contactdetection in extrusion monitors. Contact spiking:a substrateleakagecurrent increase oftwo decades. 8 Data analysis The time to failure is estimated by fitting a lognormal distribution through the data points (see Figure 3). For plotting the use of the failed fraction according to the mean rank method is NSERT - 0,3)/(N + 0,4)), is allowed but shall be reported. Fitting to be done with the least squares or maximum likelihood methods. Calculate the each failure time t(F%). The confidence interval is determined using the t-distribution. The confidence level used shall be reported Graph fitted lognormal distribution /J1 / J2 JJ3 Cumulative % failure F(A1%) I: t1 t2 : t3 Log (failure time) IEC1123/10 Key J1,/2.J3(A/cm2) stress current density to line or via J, > J2 > J3 (A/cm2) t,t2,tg(h): failure time when the cumulative failure reaches A1 percent. Figure 3-Graphfitted lognormal distribution 62415@IEC:2010 Extrapolationto otherconditionsis doneusingBlack'sequation withno linewidthterm: t(x%)= A· j-n .exp(Ea/(k T) (2) where A is a process-dependent factor, j is the current density, n is the current exponent, Ea is the activation energy, k is the Boltzmann constant, and T is the absolute temperature. RENSEERT It is assumed that this formula holds for all fail percentages, in other words that the spread of the distribution is not affected by the acceleration. D1 For the determination of the activation energy Ea, three temperatures, for the and determination of the current density exponent n, three current densities should be used. The power exponent"n