SiO2 ACTIVATION ENERGY: AN ELEMENTARY STUDY IN THE MATRIX ISOTROPIC ETCHER. 1 A.J. BALLONI Fundação Centro Tecnológico para Informática Instituto de Microeletrônica Laboratório de Manufatura de Circuitos Integrados C.P. 6162 - Campinas/S.P. FAX: 0192/402029 ABSTRACT. It is presented an elementary study of temperature, power and etching time effects on SiO2 isotropic etching. It is showed that the activation energy of SiO2 appears to be slightly dependent on the rf power input but, the ARRHENIUS Plot in the high limit etches rate, indicates that the activation energy is really not dependent on the rf applied. I - EXPERIMENTAL SET UP. The work was realized in a Matrix 303 Downstream Isotropic Etcher. This system works at pressures between 300mTorr and 3000mTorr and generates isotropic etching through an afterglow of NF3/He/O2 plasma. The configuration is such that no ions can strike the wafer surface and, hence the chemical dry etching reactions occur within separated discharge chamber [01,02]. The power can be changed from 0W up to 500W and the electrode temperature between 5C and C. II - EXPERIMENTAL RESULTS.
2 II.1 - Process Parameter. The following process parameters were utilized, to get the SiO2 etch rate : NF3 flow [sccm/(%)]:25/(25), rf Power [W] :150,250 and 330 He flow [sccm/(%)]:47/(31), Etching time [sec] :5,10 and 15, Pressure [mtorr] :1500, Electrode temperature [C]:20,50 and 80. By keeping constant the NF3 flow, He flow and pressure and, for each value of electrode temperature, power and etching time presented above the SiO2 etch rate was obtained. II.2 - The SiO2 etch rate measurements. The SiO2 etch rate measurements were performed for each set of process parameters presented in section II.1 ( Process Parameter ), resulting in 9 (nine) different experimental conditions. Table II.1 display the values measured and, the figures 1 reports for each value of rf power and etching time, the SiO2 etch rate versus electrode temperature. TABLE II.1 - THERMAL OXIDE ETCH RATE AND PROCESS PARAMETER VALUES. P = 150 [W] P = 250 [W] P = 330 [W] t [sec] T [C] THERMAL OXIDE ETCH RATE[A/min] 20 240 336 312 5 50 360 450 500 80 468 588 588 20 278 381 410 10 50 443 591 617 80 592 784 781 20 250 404 456 15 50 476 660 692 80 720 892 952
SiO2 ETCH RATE [A/min] SiO2 ETCH RATE [A/min] Apresentado e publicado nos ANAIS do 10TH CONGRESS OF THE BRAZILIAN MICROELETRONICS 800 600 400 200 0 20 40 60 80 t=5sec, P=150W t=5sec, P=250W t=5sec, P=330W t=10sec, P=150W t=10sec, P=250W t=10sec, P=330W t=15sec, P=150W t=15sec, P=250W t=15sec, P=330W T(C) Figure 1.: SiO2 etch rate vs. electrode temperature. t=etching time. P=rf power. 3 We note in each case an increase of the etches rate as the temperature is increased: this means a priori that the etching is (at least) partially chemical, i.e., isotropic. Since a purely chemical etching, verify the ARRHENIUS law: K (T) = K (T)exp( -E a K T ) 0 B Where: K(T) =etch rate, depending on temperature, K 0 (T) =high temperature limit etch rate, E a =activation energy of the chemical reaction, T=Temperature K B =Boltzman constant. and, The ARRHENIUS plots for curves of figure 1 are presented in figures 2, 3 and 4, respectively. The slop of each curve in these figures, give straightforward the activation energies for the chemical reactions. The table II.2 shows the values of Ea(K) and Ko for each figures 2,3 and 4. t=5sec, P=150W t=5sec, P=250W t=5sec, P=330W 2.8 2.9 3.0 3.1 3.2 3.3 3.4 /T(K) Figure 2.: Arrhenius Plot. t=etching time.p=rf power. 3.5
SiO2 ETCH RATE [A/min] SiO2 ETCH RATE [A/min] Apresentado e publicado nos ANAIS do 10TH CONGRESS OF THE BRAZILIAN MICROELETRONICS 4 t=10sec, P=150W t=10sec, P=250W t=10sec, P=330W 2.8 2.9 3.0 3.1 3.2 3.3 3.4 /T(K) Figure 3.: Arrhenius Plot. t= etching time. P=rf power. 3.5 t=15sec, P=150W t=15sec, P=250W t=15sec, P=330W 2.8 2.9 3.0 3.1 /T(K) Figure 4.: Arrhenius Plot. t=etching time. P=rf power. 3.2 3.3 3.4 3.5 From each curve in figures 2, 3 and 4 we get the activation energies Ea(K) and the high temperature limit etch rate (Ko), cf. Table II.2. TABLE II.2 - ACTIVATION ENERGY Ea(K) FOR THERMAL OXIDE, AND HIGH TEMPERATURE LIMIT ETCH RATE Ko(A/min). P1 150 W P2 = 250W P3 = 330 W = Ea(K) Ea(K) Ea(K) Ko(A/min) Ko(A/min) Ko(A/min) t1 = 5sec 392.5 1578.3 431.0 1802.0 479.0 1960.0 t2 = 10sec 541.0 2121.0 694.0 2745.0 640.0 2450.0 t3 = 15sec 790.0 2995.0 841.0 3268.0 850.0 3350.0
The Table II.2 shows that the SiO2 activation energy is different by at least a factor of 2 (two) from those values found in the literature [02,03,04,05]. To make this clear, the Table II.3 presents some values of Ea(K) found in the literature, as well the etches rate and temperature values got from its ARRHENIUS Plot. We can see that the etches rates presented in Table II.3 are at least 5 time slower from those values presented in references [02,03], 10 times slower from those presented in references [04,05] and, nearly the same as those presented by reference [06]. May be the high etch rate of thermal oxide should be responsible for the lower value in the activation energy presented in this work. The possible influences of the thermal oxide etch rate in function of activation energy should be checked. 5 TABLE II.3 - REFERENCES VALUES FOR SiO2 ETCH RATE VERSUS TEMPERATURE AND ACTIVATION ENERGY. REFERENCES TEMPERATURE [C] SiO2 ETCH RATE [A/min] SiO2 ACTIVATION ENERGY [K] 20 50 [02] 50 2 80 170 20 50 [03] 50 1890 80 170 20 15 [04] 50 45 1624 80 70 20 18 [05] 50 28 1542 80 42 20 200 [06] 50 400 850 80 680 20 Cf. Table II.1 [THIS WORK] 50 Cf. Table II.1 Cf. Table II.2 80 Cf. Table II.1 The figure 5 shows the Ea(K) versus rf Power. It appears that Ea(K) is slightly dependent on the rf power. This same figure shows that increasing the etching time increases the activation energy. May be, increasing the etching time the activation energy could increase up to a saturation value found in the literature.
Ea[K] Ea(K) Apresentado e publicado nos ANAIS do 10TH CONGRESS OF THE BRAZILIAN MICROELETRONICS 900 700 500 300 200 300 400 P(W) Figure 5.: Activation energy vs. rf power. t=etching time. t=5sec t=10sec t=15sec 6 Figure 6 reports 3 curves of Ea(K) in function of Ko(A/min) (very high temperature limit etches rate)-cf. Table II.2. The trends of these curves indicate that there is not a dependence of Ea(K) on rf power. 900 700 500 300 1500 2000 2500 3000 3500 4000 t=5sec t=10sec t=15sec Ko[A/m] Figure 6.: Activation energy vs. Ko (very high temperature limit etch rate). Assuming Ea(K) does not depend on the rf power, and making the data used to get the figure 6 as only one set of data (i.e., independent of rf power), we find that Ea(K) appears to be not really dependent on rf power, cf. figure 7. Finally, extrapolating the curve in figure 7, we find in the saturation limit that Ea=1500K, cf. figure 8. This work has showed that the activation energy is mainly dependent only on the electrode temperature (figure 2, 3 and 4), as should be. From extrapolation considerations, figure 8 showed Ea(K)=1500K, that is in very well agreement with those values of Ea(K) found in literature: 1624K and 1542K [04,05], cf. table II.3.
7 III - CONCLUSION. The natural conclusion is that the SiO2 Activation energy is not dependent on the rf power. Furthermore in figure 08 (Ea(k) vs. Ko), where each value of Ea was obtained for 3 different value of rf power and etching time, from extrapolation consideration we find that the value of Ea(K)=1500K is in very well agreement with that found in the literature [04,05], cf. Table II.3. Finally, the activation energy for the thermal oxide (cf. table II.2) there is at least a factor of 2 (two) from those values found in the literature, cf. tables II.2 and II.3. As discussed in section II.2, a possible influence of thermal oxide etches rate in function of Ea(K) should be checked. Furthermore, the figure 5 [Ea(K) vs. rf power] showed a very small dependence of Ea(K) on the rf power. This same figure also showed that the etching time may have a very strong influence on the accuracy of the value obtained for the activation energy. Therefore, a possible influence of the etching time on the activation energy should also be checked. ----------- x ----------- This work has been realised at the IMEC vzw - Kapeldreef, 75/B 3001, Leuven/BELGIUM as part of a 9 months training program in metal, polysilicon, silicon nitride and resist strip dry etching steps in view of the development of a
Double Layer Metal process at FCTI/IM - BRAZIL. This work was supported by RHAE/CNPq and FCTI/IM. 8 R E F E R E N C E S. [01] - D.L.Smith, "High Pressure Etching in VLSI electronics" V.8,N.G.Einspruch, Ed., Academic Press, Orlando, FL-, Chap 9, 253(1980). [02] - Takuo Sugano et all 'Applications of Plasma Process to VLSI Technology", John Wiley & Sons, 123(1985). [03] - Dennis M. Manos and Daniel L. Flamm, Plasma Etching - An Introduction Academic Press, INC. (1989). [04] - Lowenstein & al, JAP 65, (1989). [05] - J.F. Daviet, An elementary study of temperature effect on plasma etching Internal Report Doc No. 031-31027 - IMEC/ASP Micropatterning Group (1992). [06] - A.J. Balloni, Alignment Mark on the Matrix 303 Isotropic Etcher - (IMEC/ASP/MP Internal Report - 1992). Submitted to be presented in the Brazilian Microeletronic Congress- SBu/95 (1995).