Modified Oxide Properties from Dopant Effects at the Si-SiO 2 Interface

  • PDF / 376,512 Bytes
  • 6 Pages / 420.48 x 639 pts Page_size
  • 86 Downloads / 188 Views

DOWNLOAD

REPORT


Mat. Res, Soc. Symp. Proc. Vol 54.

1986 Materials Research Society

574

accounted for in interpreting ellipsometric data and in explaining the oxide properties, such as defects, stoichiometry and electrical properties. EXPERIMENT Table I is a list of the substrates used in this study. The Si dopants were P, As or B. The substrate orientation was (100) except one (111) sample. The dopants had either uniform (U), diffused (D) or ion-implanted (I)profiles. The diffusions were performed by the Semiconductor Products Division of the Carborundum Company, Niagara Falls, New York, using their solid diffusion sources for As and P. The diffusions all used the same (100), B-doped, 10-20 o-cm substrate. The coding used throughout this paper for the substrates has one or two letters for the dopant, followed by a letter for the dopant profile and then a number to distinguish substrates with the same dopant and profile, but different surface concentrations. Hence, an As-diffused substrate may be labelled AsDl. The substrates were oxidized in one of three processes. Native oxides, referred to by process number - #1, were formed by sitting in air after standard HF etching of the substrate. Process #2 consisted of HF cleaned substrates, ramp heated in air to 840°C using a muffle furnace. Process #3 was a 950 0C, dry 02 oxidation for 20 min using HF cleaned substrates. The implanted samples were annealed prior to undergoing the above oxidation processes. A B-doped sample and an As-diffused sample were also annealed to monitor any process effects. The samples were then coded as substrate - process number. Hence, if the As-diffused substrate mentioned above was ramped to 840 0 C, the sample was coded AsDl -#2. All samples that underwent process #1 or #2 were stored in glass petri dishes, that had been acid etched and peroxide rinsed, to minimize carbon contamination. The samples were characterized via null ellipsometry. First, angular readings were recorded from a Gaertne• L117 Production ellipsometer using an angle of incidence of 700 and 6328A He-Ne light. The data was analyzed using a modified version of the NBS Computer Program for ellipsometry written by Frank McCrackin [8]. The program assumes a single layer, nonabsorbing filw on a Si substrate having an index of refract ion of 3.85 iO.02 at 6328A. The index of refraction of the film (nf) was fixed, film thickness (t) calculated, and an error term generated. The index of refraction is incremented and the error term monitored for a sign change. Once the nf range is narrowed to yield almost zero error, the program calculates both t and nf of the film. The samples were also tested on a Rudolph Auto/EL III, microprocessor controlled ellipsometer with the same angle of incidence and wavelength. The program outputs A and i values and the t and nf of the film, but no error indication is given. Angle resolved ESCA data has been obtained for some of the samples. The work was performed at the Physical Electronics Division of PerkinElmer in Edison, New Jersey using a Phi Model 5100. The in