Over the past few years, ferroelectric thin films have been widely considered for the development of tunable microwave device [1]. One of the best candidates identified so far for such applications is BaxSr1-xTiO3 (BST) material, which exhibits a significant variation of permittivity as the electrical field increases, with quite low dielectric losses, as shown in figure 1.STMicroelectronics recently developed a solution of integrated tunable capacitor based on BST material, which effectively offers excellent RF performances, low power consumption and high linearity required in adaptive radiofrequency tuning applications. These tunable capacitances are controlled through a bias voltage, typically ranging from 2 to 20 V, which lead to a capacitance tuning ratio up to 4. When embedded in mobile phone circuits, these advanced tunable chips enable to maintain an efficient transfer of energy from the handset amplifier to the antenna under various operating conditions, making the performance almost insensitive to the external environment.During the qualification of BST thin film capacitor process, the evaluation of reliability performances has to be carefully carried on, to ensure that BST material effectively fulfills the reliability expectations required in application mission profiles. This paper focuses on the development of a predictive reliability model, established from highly accelerated tests performed directly at wafer level. This reliability model is representative of Time-Dependent Dielectric Breakdown (TDDB) failure mechanism, which leads to an irreversible loss of the dielectric insulating properties.To develop such a model, the time-to-breakdown (Tbd) evolution of BST capacitors were recorded under various stress conditions, applying voltage stresses from 22V to 30V, at temperature levels ranging from 160°C to 200°C. In such conditions, failures are observed to occur before a few hours, which effectively represent pretty high accelerated test conditions compared to the product lifetime expected to exceed five years. Each test was performed on at least 30 capacitors. The statistical distributions of time-to-breakdown data were fitted by the Weibull law. Indeed, the Weibull approach is known to be the most appropriate one to describe the statistical behavior of breakdown data, not only for silicon oxide dielectrics, but also for high-k dielectrics such as ferroelectric materials, as recently demonstrated [2]. To describe the capacitors time-to-breakdown evolution as a function of stress level, a mathematical model structure can be proposed, associating an Arrhenius law to account for the temperature acceleration, with a power law for the voltage acceleration effect [3]: Tbd = C . exp(Ea/kT) . V-n The lifetime simulation results provided by the model fit perfectly with the experimental data obtained at wafer level, but it also turns out to be in very good agreement with additional reliability results obtained at package level, in moderate acceleration conditions, as shown in figure 2. We believe that such agreement validates the relevance of the reliability approach adopted in this study, and the consistency of the mathematical structure of the model.To conclude, this study presents the development of a predictive model that enables to simulate the reliability performances of a new generation of tunable capacitors, based on BST dielectric material. The applications of such reliability model are numerous; it represents a key asset for any manufacturer. The lifetime predictions provided by the model turn out to be perfectly compliant with RF tunable applications, as demonstrated on three different qualification lots. Besides, by implementing a wafer level monitoring test in-line, this reliability tool offers the opportunity to monitor the lifetime performance of any sample, and to guaranty the reliability of the whole production line.[1] A.K.Tagantsev et al., Journal of Electroceramics, 2003.[2] M-T.Chentir et al., Microelectronics reliability, 2009.[3] S-C.Huang et al., Journal of Applied Physics, 1998.
Read full abstract
Sep 21, 2023
Jiamin Lin + 5
Open Access