Trends in Lifetime Measurements

The interpretation of lifetime measurements and new characterization techniques are addressed in this paper. Parameters such as surface/interface recombination, sample thickness, and injection level are discussed. Novel characterization techniques using frequency-dependent capacitance, conductance, or impedance measurements are described and are shown to be useful for thin layer, e.g., epitaxial or silicon-on-insulator layers, characterization. Introduction Lifetime was one of the earliest parameters to be measured during the evolution of semiconductors. For example, one the parameters determined in the classic HaynesShockley experiment during the early 1950s, is the minority carrier lifetime. After the initial flurry of developing numerous lifetime characterization techniques, the topic lost some of its appeal and went into a decline for many years. Of course, people made lifetime measurements, but the equipment was frequently home made, samples had to carefully prepared, and the interpretation was not always unambiguous. It was only when commercial equipment became available, that significant progress was made. Measurements made by Si wafer producers and integrated circuit (IC) companies generated a wealth of data, usually displayed as color maps for any size wafer. These measurements were generally made contactless with all necessary data processing built into the equipment. The silicon crystal growers measured lifetimes to characterize the Czochralski crystal pullers and the growth process and the IC companies used these measurements to follow the cleanliness of their processes. A major impact on lifetime measurements, was the discovery of iron-boron pair formation and dissociation in 1981. Initially, this was merely an interesting phenomenon, but was not exploited. It was only when Zoth and Bergholz pointed out that by measuring the minority carrier lifetime or the recombination lifetime during the Fe-B state and then during the interstitial iron, Fei, state in boron-doped wafers, was it possible to determine the iron density in a wafer. Since iron is also one of the chief metallic impurities introduced during wafer processing, iron being one the main constituents of the ubiquitous stainless steel, this was an ideal combination of simple, contactless diffusion length/lifetime measurements with iron detection in commercial equipment. It spawned several new characterization tools. Now that lifetime measurements are routinely made, attention has shifted from the basic measurements to interpretation of the measured data in today's Si wafer/process technology and development of novel lifetime characterization techniques. Such parameters as injection level, temperature, frequency dependence and others are being actively looked at for lifetime data interpretation. I will focus on several aspects of lifetime characterization and interpretation and will assume the reader to be familiar with the concept of lifetimes and the basic measurement techniques. Epitaxial layer characterization through conventional recombination lifetime measurements is difficult, because the epi layers are typically much thinner than the minority carrier diffusion length, making the data analysis difficult. This difficulty also applies to today's bulk Si wafers. They are so pure that they have exceedingly high lifetimes. This makes surface recombination much more important than in the past. The bulk recombination lifetime τB in its simplest form is given by T th n B N v σ τ 1 = (1) where σn is the minority carrier capture cross section, vth the thermal velocity, and NT the defect density. For σn = 10 cm and vth = 10 cm/s, τB becomes