Published on

Published in: Business, Technology
1 Like
  • Be the first to comment

No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide


  1. 1. Thin Solid Films 326 (1998) 92–98 The preparation of photoluminescent Si nanocrystal–SiOX films by reactive evaporation S. Zhang*, W. Zhang, J. Yuan State Key Laboratory of Si Material Science, Zhejiang University, Hangzhuo, PR China 310027 Accepted 13 February 1998Abstract Si nanocrystals–SiOx films on Si substrates have been prepared by evaporating Si or SiO in the residual gas of the system or in an oxygenatmosphere by introducing oxygen into the deposition chamber. Films with different oxygen concentration and different Si crystallite sizewere obtained by changing the substrate temperature and the oxygen partial pressure. It is observed that only the films prepared byevaporating SiO in an oxygen atmosphere are photoluminescent at room temperature. The possible mechanism for the luminescence isdiscussed, according to the results of X-ray photoelectronic spectroscopy (XPS), infrared (IR) spectra and X-ray diffraction (XRD)measurements. © 1998 Elsevier Science S.A. All rights reservedKeywords: Evaporation; Luminescence; Nanostructures; Silicon oxide1. Introduction luminescence layer and metal contacts used as electrical leads may be finished in the same system without breaking Since the observation of photoluminescence from Si/SiO2 vacuum; (3) because reactive evaporation is an alternativemultilayers [1], Si nanocrystals in SiOX matrix are exten- technique in preparing SiO2, which is extensively used assively studied due to their unique optical properties and the the masks in the Si devices and integrated circuit fabricationsimplicity of producing them, as well as the technological and as the insulating layer in metal-oxide-semiconductorcompatibility with present Si integrated circuits. They may devices, it is possible to make masks or insulating layersprovide a simple way to realize stable Si-based luminescent or luminescent films by the same technique, i.e. it is moredevices, making integrated optoelectronic devices possible compatible to the semiconductor device processing. It is theon Si chips. Several techniques have been employed to purpose of this paper to use a simple and convenient proces-make the structure of Si nanoparticles in SiO2, which sing technique to realize rather difficult results, though theinclude single energy Si ion implantation into SiO2 [2–4], results presented are initial and the processing conditionslow temperature multienergy Si+ ion implantation into SiO2 have not been optimized yet. In term of scientific interests,[5], plasma enhanced chemical vapor deposition [6], two films with a wide range of compositions and crystallinestep thermal annealing with auxiliary use of high pressure structures can be obtained, which might be difficult to rea-[7] and sputtering [8]. However, there is no report on the lize by other techniques. This should be helpful in under-preparation and luminescence from Si nanocrystals – SiOx standing the mechanism of luminescence from Si–SiO2films deposited by simple evaporation (including reactive films.evaporation), which are studied in the present experiment. Compared to other techniques, reactive evaporation hasthe following advantages: (1) the technology is simpler and 2. Experimentalthe system to prepare them is more available, because eva-poration is one of the techniques most often used in semi- Two-inch (111) oriented Si wafers were used as the sub-conductor processing and other film depositions; (2) the strates. They were cleaned using RCA1 solution (NH4OH:- H2O2:H2O = 1:1:5), followed by RCA2 solution (HCl: * Corresponding author. Fax: +86 571 7951360. H2O2:H20 = 1:1:5), each for 15 min at 358 K. Then they0040-6090/98/$19.00 © 1998 Elsevier Science S.A. All rights reservedPII S0040-6090 (98 )0 0532-X
  2. 2. S. Zhang et al. / Thin Solid Films 326 (1998) 92–98 93were dipped into 10% hydrofluoric acid (HF) solution for 10 persed by a double-grating 0.64 m monochrometer (HRD1)s to remove native SiO2 on the Si surface. The substrate was and the slits were set at a width of 300 mm. A GaAs photo-immediately transferred into the chamber from the solution multiplier in the photon counting mode was used as the lightand the chamber was pumped down. The substrate was detector. Data were recorded at intervals of 1 nm in 0.5 s.heated from the backside by thermal radiation of coiled The sample was kept in air and remained at room tempera-tungsten in closely arranged parallel quartz tubes. For uni- ture.form substrate temperature, the radiator was 2 cm away Infrared (IR) spectra were measured at room temperaturefrom the substrate and the area of the radiator was three in a single beam mode. The absorption by Si substrate wastimes larger than that of the substrate. Three groups of sam- removed by subtracting the spectrum of the bare substrateples were prepared. Groups 1 and 2 were made by evapor- from the spectra of samples.ating ultra-pure Si (99.999%), respectively in the residual We have observed the film surfaces using scanning elec-gas of the deposition chamber with a base pressure of tron microscopy. Smooth surfaces were often obtained,5 × 10−3 Pa and in an oxygen atmosphere by introducing except for the films which contained large size of Si crystal-pure oxygen (99.99%) into the chamber. Group 3 was lites. Some voids were observed in those samples. Highmade by evaporating pure SiO (99.99%) in the oxygen resolution transmit electronic microscopy (HRTEM) wasatmosphere. Both Si and SiO were evaporated using an e- not performed for the samples.gun. The beam current of the gun in preparing groups 1 and2 was maintained at 0.3 A. In evaporating SiO, the beamcurrent was reduced to 0.1 A. The deposition started by 3. Resultsopening the shutter after the substrate temperature, the Sievaporation rate and the reactive gas pressure were stable. 3.1. Films in group 1The film thickness determined with a deflected stylus variedfrom 1 to 13 mm for different samples, due to various eva- Fig. 1 shows XRD spectra for three samples preparedporating materials and different deposition time, as well as without intentionally introducing oxygen into the systemthe effect of substrate temperature (Ts) and the partial pres- at substrate temperatures (Ts) of 673, 1013 and 1113 K,sure of reactive gases on the deposition rate. The effect of respectively. Fig. 1a shows one peak only at 2v = 28.34°.thickness on the film properties have not been studied yet. The peak position and FWHM are the same as the substrateNo other elements were found in the resulting films by X- without any film deposited (bare substrate). However, theray photoelectronic spectroscopy (XPS) except for Si, O and intensity of the peak from this sample is about 1/8 of theC which is unavoidable in XPS measurement. bare substrate because the film reduces the diffraction inten- X-Ray diffraction (XRD) was used to determine film sity. The result implies that the peak is due to the diffractionstructures. Ka lines from Cu were used as the X-ray source. from the substrate, and the deposited film is amorphousData were taken at intervals of 0.02 s. The instrument was which is not thick enough to show its amorphous structureset so that Si (111) peaks resulting from the diffraction of in the XRD spectrum.Ka1 and Ka2 lines could be distinguished obviously. For the sample prepared at Ts = 1013 K, six peaks occur XPS, performed in ESCALAB MKII, was used to deter- as shown in Fig. 1b. They are at 2v = 28.34, 47.12, 55.8, 69,mine the film composition and possible bounding configura- 76 and 88° and are related to the diffraction from Si (111),tions. A monochromatic X-ray source from Al Ka1,2 (220), (311), (400), (331) and (422) planes, respectively.guaranteed an instrument resolution of 0.8 eV full width Except for the Si (111) peak, the other five peaks areat half maximum (FWHM). The angle between the sample obviously from the deposited film. Considering the factsand the analyzer input was arranged so that an electron that the bottom of Si (111) peak is much broader than thatescape depth was about 2–5 nm, depending on the sample. in Fig. 1a or than that from the bare substrate, and that theIt was observed that serious oxidation occurred on the sur- intensity part (or narrow part) disappear in a 12 mm thickfaces of the samples with low oxygen concentration in sample, it is believed that the broad part is diffracted fromgroups 1 and 2 after the samples were exposed to air. the film, while the intensity part is still from the substrate.Hence, a 12 nm thick surface layer was removed by Ar The relative intensities for the last five peaks in this curveion sputtering in the preparation chamber of ESCALAB are close to those in powder Si. Hence, it is concluded thatMKII before the measurement was carried out. This may the film contains Si crystallites which grow at arbitraryintroduce error in determining the composition due to pre- orientation, instead of epitaxially. The reason why Si doesferential sputtering effect. However, it should not signifi- not grow epitaxially will be discussed later. From FWHM ofcantly affect the following qualitative analysis on the the peaks in Fig. 1b, we can estimate the diameter d of thecomposition. crystallite using the formula [9]: Photoluminescent measurements were carried out at 09l d= :room temperature and in air. Samples were excited by a B cosv (1)413.10 nm line from krypton ion laser with an incidentpower density of about 120 W/cm2. The spectrum was dis- where l is the wavelength of the X-ray used, B is the value
  3. 3. 94 S. Zhang et al. / Thin Solid Films 326 (1998) 92–98of FWHM and v the Bragg diffraction angle of the peak.The Si crystallite size for this sample is about 6.5 nm onaverage, estimated from (220), (311) and (331) three dif-fraction peaks. The accuracy of FWHM is about 15%. Theexperiment shows that films containing smaller crystallitescan be prepared by reducing Ts. When Ts was 813°K, thecrystallite in the resulting films was about 5 nm calculatedfrom pretty weak (220) and (311) diffraction peaks (theXRD curve is not shown). Compared to Fig. 1b,c shows one more peak at2v = 58.76°, which is associated with the second order dif-fraction from (111) planes of the Si substrate. FWHMs forall other peaks in Fig. 1c are narrower than those of corre-sponding peaks in Fig. 1b, i.e. the crystallite in this sampleis larger than that of the sample prepared at Ts = 1013°K. IfEq. (1) is still used, then the diameter of the crystalliteshould be Ͼ12 nm. These results show that the crystallite size significantlydepends on Ts, i.e. increasing with increasing Ts. Fig. 2a is an XPS spectrum for the sample prepared at1013 K and measured at a depth of 12 nm. Fig. 2b is theenlarged XPS spectrum of the Si2P peak in Fig. 2a. No largedifferences were observed in the depth profile of the oxygenconcentration, except on the surface. The Si2P peak can bedeconvoluted into two components as shown by Fig. 2b,which are respectively located at the binding energy of99.99 and 102.5 eV. The peak at 99.99 eV is attributed toelement Si and the peak at 102.5 eV to the Si which hasbound to the oxygen. Due to the surface charge effect, both Fig. 2. (a) The XPS spectrum for the sample represented by Fig. 1b. (b) Enlarged Si2p peak in (a) and its two deconvoluted components. C1s and Si2P peaks in this sample shift about 0.7 eV. The difference of the binding energies between element Si and bound Si with O is about 2.5 eV, which is 1.7 eV smaller than that between element Si and Si in SiO2. This can be explained by the low oxygen concentration in the film. From the area of Si2p and O1s peak the ratio of Si to O is calculated and equal to 11:2 for his sample. From the deconvoluted two Si2P peaks, it is roughly estimated that 27% of the total Si is in the form of SiOx. The above calculation suggests that x ≈ 0.7 in SiOx, even in the regions where Si and O have combined, i.e. the oxygen concentration in the film is very low. The dependence of the oxygen concentration on Ts for the measured samples is illustrated in Fig. 3. From the figure it can be seen that the oxygen concentration increases with increased substrate temperature. The oxygen concentration is about 28% for the sample prepared at the substrate tem- perature of 1113 K, compared to 11% prepared at 553 K. 3.2. Films in group 2Fig. 1. XRD spectra for three typical samples in group 1, made at Ts = 673 To introduce more oxygen into the film, oxygen wasK (a), 1013 K (b) and 1113 K (c), respectively. filled into the deposited chamber in evaporating Si. Fig. 4
  4. 4. S. Zhang et al. / Thin Solid Films 326 (1998) 92–98 95 Fig. 5. XRD spectra for two samples in group 3. The sample for (a) was made at an oxygen flow rate of 100 sccm and Ts = 513 K, while the sampleFig. 3. The dependence of oxygen concentration on Ts in the resulting films for (b) was prepared at an oxygen flow rate of 150 sccm and Ts = 673 K.of group 1. the diameter of the crystallite is estimated to be less than 5shows an XRD spectra for three samples prepared at a sub- nm. In Fig. 4a, (220) and (331) peaks overlap and turn into astrate temperature of 1113 K and oxygen flow rate of 150 broad one, which together with the broad bottom of (111)sccm (Fig. 4c), and at a substrate temperature of 1013 K, peak suggests that the film contains very small crystallitesoxygen flow rate of 100 sccm (Fig. 4b) and 150 sccm (Fig. with diameter Ͻ3 nm, estimated after the broad peak is4a), respectively. As discussed above, the peaks at 28.34° in deconvoluted. From Fig. 4b,c we observed again thatFig. 4 (also in Fig. 5) are the contributions of deposited films FWHM is reduced, or the crystallite size is increased withas well as the substrate, while the peak at 58.7° is due to the increasing Ts for the same oxygen partial pressure or oxygensecond order diffraction from (111) planes of the Si sub- flow rate. The oxygen concentration for the sample repre-strate. All other peaks diffracted from the films will be used sented by Fig. 4c is about 56%, the highest value in group determine the crystallite size in the following. If Figs. 1cand 4c are compared, it is noted that the corresponding 3.3. Films in group 3peaks in Fig. 4c are broader. In Fig. 4b, (220) and (331)peaks can be clearly distinguished and from these two peaks Fig. 5 shows XRD spectra for two typical samples in this group. The sample for Fig. 5a was made at an oxygen flow rate of 100 sccm and Ts = 513 K, while the sample forFig. 4. XRD spectra for three typical samples in group 2. They wereprepared at Ts = 1113 K and oxygen flow rate of 150 sccm (c), and at Tsof 1013 K, oxygen flow rate of 100 sccm (b) and 150 sccm (a), respec- Fig. 6. IR absorption spectra. (a–c) relate to the samples represented by thetively. curves in Fig. 4, while (d,e) relate to the samples in Fig. 5.
  5. 5. 96 S. Zhang et al. / Thin Solid Films 326 (1998) 92–98Fig. 1b was prepared at an oxygen flow rate of 150 sccm and high frequency with increased x. When x = 2, the peakTs = 673 K. The measurement was done after the samples occurs at 1070 cm − 1. The peak at 1200 cm − 1 in Fig. 6b iswere annealed at 1073 K for 1 h in a 99.99% nitrogen atmo- from the crystobalite. The absorption peak of the crystoba-sphere. Both curves show a broad diffraction peak near 22°, lite in single crystal Si locates near 1220 cm − 1 [11]. Thewhich is absent in Figs. 1 and 4 and is assigned to the reason why the peak occurs at a smaller wavenumber herediffraction from (101) plane of crystobalite. The assignment may be due to the non-single crystalline structure of SiO2, asfor the peak is consistent with the result of infrared (IR) indicated by the XRD peaks in Fig. 5. Peaks near 486 andabsorption measurement given in Fig. 6. In addition, Fig. 800 cm − 1 are other vibration modes for SiOx (see Ref. [14]).5a shows a broad peak between 40 and 60°, similar to Fig.4a. Hence, the crystallite should be smaller than 3 nm. The 3.4. PhotoluminescenceSi (111) peak in Fig. 4b is obviously diffracted from thefilm, because this film is thick, which absorbs most of the X- Fig. 7 shows room temperature photoluminescence (PL)rays incident or reflected from the substrate. No other Si spectra for typical samples in three groups. The two bottomdiffraction peaks can be observed in curve b, from which curves (Fig. 7a,b) are related to the sample representedand the broad (111) peak it is suggested that the film has respectively by Figs. 1b and 4c. Fig. 7c,d are related toamorphous structure, or at least the size of the crystallites (if the samples respectively represented by curves in Fig. 5.any) in the film should be smaller than that in the sample All measurements were carried out after the samples wererepresented by Fig. 4a. annealed as indicated above. Except that the oxygen con- XPS measurement shows that the films in this group can centration in the annealed films was slight changed, nohave more stoichiometric SiO2 structure. The highest oxy- nitrogen was detected either by XPS or by IR. This is rea-gen concentration can be up to 63% (the sample for Fig. 5b), sonable when considering the inactivity of N2. So far, no PLwhich is much larger than the corresponding value in group at room temperature has been observed in the films of1 (28%) and in group 2 (56%). groups 1 and 2. In group 3, samples with Si crystallite The differences between three groups are also compared size Ͻ3 nm show a PL peak near 725 nm like Fig. 7c,using IR measurements. Fig. 6 shows IR absorbance spectra while other samples with Si crystallite size ≥4 nm orfor five samples. Fig. 6a–c is related to the samples in group amorphous structure do not show obvious PL peaks, like2, represented by the curves in Fig. 4, while Fig. 6d,e is Fig. 7d.related to the samples in group 3 in Fig. 5. All data weretaken after an annealing in pure nitrogen for 1 h at a tem-perature of 1073 K, or at 1173 K for the sample prepared at 4. DiscussionTs Ͼ 1073 K. The curves show absorption bands centeredbetween 970 and 1060 cm − 1. In addition, both samples in The results presented above shows that the substrate tem-group 3 (Fig. 6d,e) have a peak or apparent shoulder at 1200 perature and the gas in the deposition system are verycm − 1 as indicated by a dashed line in the figure, which was important factors in controlling film structures and proper-absent or very weak if any in groups 2 and 1 (the IR spectra ties. The oxygen and water vapor in the deposition chamberfor group 1 are similar to group 2 and are not shown in the play the role of oxidizing the Si substrate and the evaporatedfigure). The peaks between 970 and 1070 cm − 1 are the Si on the substrate, which respectively prevent an epitaxialstretching mode for SiOx [10], where the peak shifts to growth along the (111) orientation of the Si substrate, and prevent Si crystallites growing large, and finally leads to Si with non-single crystal structure. Ten percent oxygen on the Si surface was detected when the substrate temperature was set at 1113 K for 30 min in the residual gas with the base pressure of 5 × 10−3 Pa, and then was transferred into the XPS chamber immediately for the measurement [12]. The oxidation effect on the crystallite size can be observed by comparing FWHMs of the corresponding peaks of Fig. 1b and Fig. 4a,b, as well as between two curves in Figs. 1c and 4. A larger oxygen flow rate gives a higher oxygen partial pressure in the system and leads to more oxygen reacting with Si. Then smaller crystallites are produced. The substrate temperature can affect the mobility of eva- porating atoms on the substrate, as well as the reaction rate between Si and reactive gases. When the substrate tempera-Fig. 7. Photoluminescence spectra measured with 413.10 nm laser excita-tion at room temperature for four samples. The bottom two curves are ture is increased for a fixed reactive gas pressure, the mobi-respectively related to the samples in groups 1 and 2, while the top two lity of evaporated atoms on the growing surface iscurves to group 3. enhanced, while the oxidation rate, then the oxygen concen-
  6. 6. S. Zhang et al. / Thin Solid Films 326 (1998) 92–98 97tration in the resulting films (as observed in group 1) will tallite size by XRD is not known, because it has not beenalso be increased. The first effect favors the growth of large verified using another independent technique such assize crystallite. On the other hand, the second effect will HRTEM. Hence, further measurements and investigationprevent the increase in the Si crystallite size. The result are needed to tell the discrepancy and verify the lumines-that the crystallite size is increased with increased Ts sug- cence mechanism suggested here.gests that the mobility of Si on the substrate surface playmore important role in determining the crystallite size thanthe oxidation effect. 5. Conclusions In evaporating SiO, SiO is sublimated before it is decom-posed, because the evaporation temperature is much lower Si–SiOx films on Si substrates have been prepared bythan its melting point. The evaporated SiO is reactive and reactive evaporation of Si or SiO. The reactive gases areeasy to form SiO2 through binding one more oxygen atom. provided by either the residual gas or oxygen intentionallyFor the same Ts and oxygen partial pressure in the system, introduced into the system. It is observed that the oxygenthe probability to form SiO2 is higher by evaporating SiO concentration and the Si crystallite size in the resulting filmsthan by evaporating Si. However, it is still possible that can be controlled by changing the substrate temperature andsmall number of SiO molecules are decomposed into Si the oxygen partial pressure, and both can affect the photo-and O in the evaporation, and the Si atoms are coalesced luminescence significantly. Higher Ts gives larger Si crys-to form crystallite in the deposition or in the annealing (as tallite for other fixed deposition conditions. On the otherobserved). hand, higher oxygen partial pressure leads to smaller Si Though we cannot definitely tell the mechanism for the crystallite and higher oxygen concentration in the film. Noobserved PL, some literature has ascribed PL at long wave- luminescence has been detected at room temperature for thelength (Ͼ700 nm) to the quantum size effect [4,6,13,14]. On samples prepared by evaporating Si either in the residual gasthe one hand, in the experiment we have observed that the or in the oxygen atmosphere. A quite narrow luminescencefilms with large Si crystallites or amorphous structure in any peak near 725 nm at room temperature is observed from thegroup are not photoluminescent at room temperature, which samples with the Si crystallite diameter Ͻ3 nm but notsuggests that the crystallite size is important for PL. How- amorphous Si, prepared by evaporating SiO in the oxygenever, it is noted that the Si structure of the films in group 2 atmosphere. More data are required to extract a conclusioncan be the same as that for the sample represented by the top for the PL mechanism, though its possibility has been dis-curve in Fig. 7, but PL efficiency is different, which sug- cussed in Section 4.gests that other factors control the radiative transition effi-ciency. The samples prepared using Si as the evaporatingmaterial have low oxygen concentrations, no or very small AcknowledgementsSiO2 diffraction peaks in XRD (Figs. 1 and 4) and no 1200cm − 1 peaks (or very small if any) in IR absorption (Fig. 6). It This work was supported by the Natural Science Founda-is well known that a stoichiometric SiO2 can give a higher tion of Zhejiang Province, China and the Doctoral Founda-energy barrier for the Si nanocrystal than SiOx with x Ͻ 2. tion of the Chinese Education Committee. TheAccording to the theory of the quantum size effect, the photoluminescence was measured in the Institute of Semi-higher the energy barrier, the more the energy gap increases conductors, Chinese Academy of Sciences.and the more efficiently the excess carriers transit radia-tively in the Si nanoparticle quantum well for the samesize nanoparticles. This may explain why PL at room tem- Referencesperature can be observed in the samples of group 3, whilenot in groups 1 and 2. The result and discussion may favour [1] Z.H. Lu, D.L. Lockwood, J.M. Baribeau, Nature 378 (1995) 258.the mechanism that the photoluminescence is from the [2] H.M. Cheong, W. Paul, S.P. Withrow, J.G. Zhu, J.D. Budai, C.W.quantum size effect. It is interesting to note that Shi et al. White, D.M. Hembree Jr., Appl. Phys. Lett. 68 (1996) 87.[14] recently reported a similar PL peak from Si rich SiO2 [3] S. Guha, M.D. Pace, D.N. Dunn, I.L. Singer, Appl. Phys. Lett. 70films prepared by LPCVD and ascribed it to the quantum (1997) 1207. [4] A. Pifferi, P. Taroni, A. Torricelli, G. Valentini, P. Mutti, G. Ghislott,size effect also. We also find that our observation without L. Zanghieri, Appl. Phys. Lett. 70 (1997) 348.PL from amorphous Si–SiOX films is contradictory to the [5] W. Skorupa, R.A. Yankov, I.E. Tyschenko, H. Frob, T. Bohme, by Lu et al. [1]. They did observe PL from amorphous Leo, Appl. Phys. Lett. 68 (1996) 2410.Si/SiO2 multilayers. In the moment, we do not believe that [6] A.J. Kenyon, P.F. Trwoga, C.W. Pitt, G. Rehm, J. Appl. Phys. 79we can give a clear explanation for the discrepancy, because (1996) 9291. [7] D.P. Karwasz, A. Misiuk, M. Ceschini, L. Pavesi, Appl. Phys. Lett.we are not aware whether the properties of amorphous Si, 69 (1996) 2900.such as defects in different geometric structures (dot and [8] V.G. Baru, A.P. Chermushich, V.A. Luzanov, G.V. Stepanov, L.Y.layer) and prepared by different techniques are compatible Zakharov, K.P. O’Donnell, I.V. Bradley, N.N. Melnik, Appl. Phys.or not. In addition, the accuracy in determining the Si crys- Lett. 69 (1996) 4148.
  7. 7. 98 S. Zhang et al. / Thin Solid Films 326 (1998) 92–98 [9] D. Cullity (ed.), Elements of X-Ray Diffraction, Addison–Wesley, [12] W. Zhang, S. Zhang, J. Yuan, Chinese J. Semicon. 18 (1997) 8. London, 1956, p. 262. [13] A.A. Seraphin, S.T. Ngiam, K.D. Kolenbrader, J. Appl. Phys. 80[10] J.C. Knights, R.A. Street, G. Lucovsky, J. Non-Cryst. Solids 35/36 (1996) 6249. (1980) 279. [14] W.Q. Shi, Y.Z. Liu, Z.J. Chen, S.X. Liu, D.C. Yao, The Proceedings[11] C.J. Varker, J.D. Whitefield, P.L. Fejes, in D.C. Gupta (ed.), Silicon of the 10th National Conference on Semiconductor, Chinese Institute Processing, American Standards Test Manual, Philadelphia, PA, of Electronics, Qindao, 1997, p. 73. 1983, p. 369.