New Research


Computational materials design for semiconductor spintronics

Design of ‘co-doping’ for (Ga, Mn)As as a Route to higher TC

Prof. Hiroshi Katayama-Yoshida, Guest Associate Prof. Kazunori Sato

  In the present semiconductor electronics, information processing is realized by controlling the electric current. In principle, we can achieve high-speed computation by downsizing the processors and shortening the travel distance of electrons. However, it is easily recognized that this direction of development should see a dead-end when the scale of the device element reaches atomic scale. Moreover, increase of the leakage current due to the thinner insulating layer directly means explosion of Joule heating and uncontrollable thermal management.

  We avoid this difficulty by using another freedom of electrons, which is ‘spin’. Spin degree of freedom of electron normally does not play an important role in semiconductors, since the semiconductors are non-magnetic. In 1989, Munekata and Ohno showed that we can make a semiconductor ferromagnetic by fabricating (Ga, Mn)As. This discovery of the dilute magnetic semiconductors (DMS) initiated the semiconductor spintronics. Nowadays, domain wall motion, magnetization reversal and electric field effects on these materials are studied intensively.

  However, Curie temperature, TC, (below which the system shows ferromagnetic property) of this system is not high enough from application point of view and a fabrication of DMS with high-TC has been desired. Low-TC in DMS is mainly due to the difficulty of high concentration doping of Mn atoms into semiconductor host materials. Low concentration of magnetic impurity means long average distance between magnetic impurities, therefore we cannot expect strong magnetic coupling between magnetic impurities. According to the first-principles studies, it is shown that high-TC can be achieved only when high concentration doping (~30%) of magnetic impurities is realized. However, when we consult a phase diagram of DMS, we immediately recognize that it is prohibitively difficult task.

  In our group, based on first-principles calculations and Monte Carlo simulations we propose clever doping and processing method to realize high-TC DMS. We have shown that by using a ‘co-doping’ of donor impurities in addition to the magnetic impurities (Mn in case of (Ga, Mn)As), we can control solubility limit of Mn in (Ga, Mn)As. However, due to the carrier induced nature of the ferromagnetism in this compound, the compensation by the co-dopants always lowers the TC, therefore the co-dopants which are used to increase the Mn concentration should be removed after the crystal growth. The advantage to use light interstitials, such as Li interstitials, as a co-dopant is that after the crystal growth we can remove co-dopants easily by using low temperature annealing. This annealing process is also designed by the combination of first-principles calculations and Monte Carlo simulations.
  Realizing high- TC DMS is so important that our materials and process design give great impact in the community of semiconductor spintronics. At the same time the present research indicates powerful potential of materials design based on the computational methods.

Yoshida Lab. web page
◎About this site
Go back to page top