Initial conditions and long time dynamics for a complex system: Hierarchical properties of the spin glass decay

This dissertation explores the influence of the initial state on spin glass dynamics. The initial state is defined as the state prepared before aging. To probe the effects of different initial states, we develop six cooling protocols. The initial state is quantified by an effective cooling time, teffc . We vary teffc from 19s to 406s. The initial state is found to influence Thermoremanent Magnetization (TRM) decays with waiting times less than 1000s.; Aging in spin glasses, the dependence of the magnetic relaxation on the time spent at a particular field and temperature configuration before a magnetic field change, is defined through a "waiting time" tw. When magnetic relaxation curves are scaled as t/ tw, they do not collapse onto each other. This had been interpreted by comparison with aging in strain creep effects in amorphous glasses. Subaging, tmw , mu < 1, was found to be characteristic of TRM decays. We will show that subaging is actually a byproduct of an improperly prepared initial state, to which we refer as contamination. Simplifying the initial state with appropriate cooling protocols allows us to minimize teffc , and improve t/tw scaling. Using a cooling protocol with teffc = 19s we find almost perfect t/tw scaling and therefore Full Aging.; A second experimental protocol used to simplify the initial state is the Isothermal Remnant magnetization (IRM). The sample is first cooled in zero magnetic field. After cool down, a delay of the first waiting time, tw1, takes place before the field is applied. The magnetic field is applied for a second waiting time, tw 2. Any contamination of the system during the cooling process should thereby contribute only zero magnetization. The magnetization curves are scaled with tmw2 . We observe super aging, defined by mu > 1. The scaling characteristics of the IRM are dependent on tw1.; Using special cooling protocols, we are able to obtain very short teffw decay curves that are independent of tw at long times. The tw independent decay appears at the end of aging effects, when curves of different tws overlap. We fit a log function to the tw independent part of the decay. The fitted decay is extrapolated to zero magnetization. The maximum spin glass correlation length is determined from the time the extrapolated curve goes to zero magnetization. A comparison of the calculated spin glass correlation length with Transmission Electron Microscope images of our sample find the lengths to be comparable. This suggests that a maximum spin glass correlation length is set by the size of the largest crystallites in the sample.