Solutions Manual Chapter 10 Problems 10.1, Why might pure chemical etching, such as in wet etching, be adequate for patterning the silicon nitride layer used to define the field oxidation area in the CMOS process flow given in Chapter 2? Answer: For a purely chemical process, the etching is usually assumed to be isotropic with the degree of anisotropy Ay equal to 0. By Eqn 10.7, this would give an etch bias equal to the film thickness (overetching my add S- 10% to this.) But the nitride layer for the field oxidation process is only about 80 nm, of 0.08 jm. Thus the undercutting is only about 0.08 pm. And the field oxidation delineation is not that critical. So a 0.08 ym undercutting may be adequate for this step. 10.2. If the etch anisotropy is 0, as in wet etching, what is the undercut or etch bias when etching a 0.5 1m thick film? What is the undercut when the anisotropy is 0.752 Assume no overetch in each case. Answer: ‘The anisotropy in this case is given by Eqn. 10.7. Rearranging gives: b=x;(1-A;) For Ar equal to 0, the undercutting or etch bias, b, equals the thickness of the film being etched, of 0.5 jm. For Ar equal to 0.75: b= 0.5(1-0.75)= 0.125 wm 10.3, Ina certain process, it is desired that the pitch of metal lines be equal to or less than 1.0 um (the pitch equals one metal line width plus one spacing between metal lines, measured at top of features). Assume that the metal line width and spacing are equal (that is, 0.5 pm each). The height of such structures is also 0.5 jm, and the minimum lithographic dimension is 0.25 um. a. What minimum degree of anisotropy is needed in an etch process in order to produce such a structure? SILICON VLSI TECHNOLOGY 136 © 2000 by Prentice Hall Fundamentals, Practice and Modeling Upper Saddle River, NJ. By Plummer, Deal and GriffinSolutions Manual b. What minimum pitch could be obtained for such a structure with wet etching? (Again with minimum lithograph dimension of 0.25 um, thickness of 0.5 um, and equal metal width and spacing.) Answer: b= 28; 4 b= Ste 4 iy T Bsa a. Metal pitch = S-+ Sp= 1 um, or $= 0.5 pm. x= 0.25 um = 0.5 pm From Eqn. 10.8, Sp = x+2t,(l-Ay), 0.5=0.25+2(0.5) (I-A) -all in um which leads to As, the degree of anisotropy, equal to 0.75. (this could not be achieved by wet etching.) b, Ifwet etching is used, we assume the degree of anisotropy is equal to 0. Again, Sp =x+2t-(1-Ar) so that S¢=0.25 +2 (0.5) (1-0) -all in wm. = 1.25 pm The pitch equals 2 * S¢~2.5 pm 10.4. What are the advantages and disadvantages of reactive ion etching versus sputter etching? Cite a hypothetical example of when you might want to use sputter etching rather than RIE? Answer: SILICON VLSI TECHNOLOGY 137 © 2000 by Prentice Hall Fundamentals, Practice and Modeling Upper Saddle River, NJ. By Plummer, Deal and GriffinSolutions Manual RIE gives more selectivity; sputtering etches almost everything, at about the same rate. You might want to use sputter etching when etching multilayer stack of different materials on the substrate, when etching substrate a little doesn’t matter. (You might also use sputter etching in a dep/etch/dep process to remove overhang, exploiting angle-dependent sputtering behavior. Selectivity is not important here so there is no need to develop an RIE process for this.) 10.5. Explain how loading effects can affect endpoint detection. Answer: 10.6. The signal to be detected from either the reactants or products will depend on how much material is being etched. Therefore the critical point, or threshold, for the endpoint could depend on the etching area, or circuit layout. Also, as the etching nears completion, the etch rate can increase dramatically. It is found that a certain plasma etch chemistry in a certain RIE etch system produces vertical sidewalls with zero etch bias when etching a particular film. Adding chemical A to the etch chemistry results in non-vertical sidewalls, and an etch bias. Adding chemical B to the original etch chemistry results in non- vertical sidewalls, but with zero etch bias. Explain what may be going on. Answer: 10.7. Adding A results in a higher active species to inhibitor ratio (either by increasing free radical production or decreasing inhibitor/polymer formation), reducing the sidewall inhibitor formation relative to the etch rate, and leading to more isotropic etching. Adding B results in a lower active species to inhibitor ratio, increasing the the sidewall inhibitor formation relative to the etch rate. This increases the inhibitor deposition to etch ratio, enough so that excess inhibitor forms. This leads to sloped sidewalls, as illustrated in Figure 10-14 or 10-24c, with no undercutting, or etch bias. a, In a particular etch process, if selectivity is the biggest concern, which type(s) of etch equipment should be used? b. If the biggest concern is ion bombardment damage, which type(s) of etch equipment should be used? c. If the biggest concern is obt: equipment should be used? d. If the biggest concerns are selectivity and vertical sidewalls, which type(s) of etch equipment should be used? e. What about selectivity and vertical sidewalls and damage, while maintaining reasonable etch rate? ing vertical sidewalls, which type(s) of etch SILICON VLSI TECHNOLOGY 138 © 2000 by Prentice Hall Fundamentals, Practice and Modeling Upper Saddle River, NJ. By Plummer, Deal and GriffinSolutions Manual Answer: . Ifbiggest concem is selectivity? barrel or remote plasma or wet etching b. If biggest concem is damage? barrel or remote or wet c. If biggest concem is vertical sidewalls? sputtering or RIE/HDP 4. What about selectivity AND vertical sidewalls? RIE/HDP e. What about selectivity AND vertical sidewalls AND damage, while maintaining reasonable etch rate? HDP 10.8. It is observed that the sidewall slope in an etch process becomes more sloped as the temperature is reduced. Why? Answer: Plasma etch processes are rather independent of temperature, but inhibitor deposition usually goes up when the temperature decreases (less desorption as temperature decreases). The etch/deposition ratio thus decreases, and the slope is increased. 10.9. If the anisotropy of an etch process is 0.45, sketch the etch profile. What percentage of the etch rate in the vertical direction is due to the chemical component and what percentage is ionic/physical, assuming a linear etch mechanism? State all assumptions. Answer: By Eq. 10.5, A= 1-na/ten 0.45 = 1+ taltver — Ta/tven = 0.55 (= amount of bias for each um deep). The profile would look like Figure 10-3b. For a linear etch mechanism we assume that the etch rate just equals the sum of the physical and chemical etching at each point following Eqa. 10.15, with no coupling between the physical and chemical components. ‘The physical etching occurs only in the vertical direction and the (spontaneous) chemical etching occurs equally in both lateral and vertical directions (assuming that the physical etching is completely anisotropic, the chemical etching is perfectly isotropic, and the etch rate of each component is constant with time). tyeltyer = 0.55 = chem (lat) /[chem(vert)+phys(vert)]. But chem(lat) = chem(vert), so SILICON VLSI TECHNOLOGY 139 © 2000 by Prentice Hall Fundamentals, Practice and Modeling Upper Saddle River, NJ. By Plummer, Deal and GriffinSolutions Manual 0.55 = chem (vert) ((chem(vert)+phys(vert)) 0.55 = chem (vert) total(vert) So the chemical etch component in the vertical direction = 0.55 and the physical etch component in the vertical direction = 0.45 10.10. For a particular plasma etch process in which the linear etch model is applicable, a degree of anisotropy of 0.8 or better is desired. If the unobstructed ionic flux on a flat surface is 3 x 10" atoms cm” sec”! (with Ki ‘equal to 1), what unobstructed chemical flux would result in an anisotropy of 0.8. For this process Sc is 0.01 and Kris 0.1. Answer: Tine a. The degree of anisotropy is given as: A For an Ar value of 0.8, ae =0.2. Following the example given in the text, for the linear etch model the etch rate in the lateral direction is given by Eqn 10.15 with only the chemical flux term. The etch rate in the vertical direction is given by Eqn 10.15 with both the chemical flux and ionic flux terms. Assuming no shadowing by the mask of the ion species in the center of the etch window (usually a reasonable assumption for the very directed ion species), we can let F; equal the unobstructed flux of the ion species for the vertical etch rate. We also let F, equal the unobstructed flux of the chemical species for both the vertical and lateral etch rates. This assumes completely isotropic behavior of the chemical species giving equal flux everywhere on the surface, even for narrow etch features, due to a low S.. Thus: (S.K:.) ta NY SK gy he (SERRA) 8,K/ + KR ND Rearranging gives: SKE 02 = 8.KeR = 0.2(6,KeF, 1K) SEKVF, + KF, => 0.88. K;F,)=0.2(KjR) > S.K¢ 5p, —220KiK) * S.Kp 0.25(K\F,) Plugging in the numbers gives: SILICON VLSI TECHNOLOGY 140 © 2000 by Prentice Hall Fundamentals, Practice and Modeling Upper Saddle River, NJ. By Plummer, Deal and GriffinSolutions Manual 0.25(KiF,) S.Kp 0.25((+ 1x10" 0.01 0.1 =2.5x 10! cm? sec"! ‘Any higher chemical flux would give a lower anisotropy. 10.11, We want to see how the etch rate in the vertical direction might depend on pressure assuming that the etch follows the saturation/adsorption model. Assume that for a particular etch system that the chemical flux is directly proportional to the pressure, while the ion flux is inversely proportional to the pressure, That is F.=F’*P and F=F//P. (P is normalized to 1 atm and unitless.) Also assume that density = 1 atom/nm’, and that KjFy = SF. = 1 atom/nm’/sec. a, Plot the vertical etch rate versus pressure, P, from P = 0 to 10. b, Repeat with KjF/ = 40 atoms nm” sec” and S,F,’ = 1 atom nm” sec, Answer: In the vertical direction, the etch rate will have contributions from both the chemical and ionic etch components, but acting an a synergistic fashion. For the saturation/adsorption etch model (ion enhanced etching): Etch Rate = density Plugging in F\=F//P and F.-F,’*P, the density = 1 atom/nm’, and that K:Fy’ = SF.’ = 1 atom/nm’/sec gives: Etch rate -1 density ( i 1 | (KRVP SF =P) _ 1 Latom/ nm? /sec - (4) Up “T=P) nm/see SILICON VLSI TECHNOLOGY 141 © 2000 by Prentice Hall Fundamentals, Practice and Modeling Upper Saddle River, NJ. By Plummer, Deal and GriffinSolutions Manual os 08 Boa E S08 £02 ii on o o 2 4 é & 10 Pressure (normalized to 1 atm) b, For KiFi’ = 40 atoms/nm’/sec and S.F.’ = | atom/nm’/sec: Etch rate 1 density(_1 | t+) 1 =) CGRP Se Latom/ nm? /see -laomt TT Tatom/am (1, 1) \ao7P *T=P) 1 m/sec (Pt (Pd) \40° P/ 5 Ba E Ea £2 a ° o % 4 Cn) Pressure (normalized to 1 atm) 10.12. In an etch process, there is a finite amount of purely chemical etching without any ion bombardment (i.e. spontaneous chemical etching). In addition, ion bombardment greatly increases the etch rate by facilitating the breaking up of SILICON VLSI TECHNOLOGY 142 © 2000 by Prentice Hall Fundamentals, Practice and Modeling Upper Saddle River, NJ. By Plummer, Deal and GriffinSolutions Manual the etch precursor. At high ion flux the etch rate saturates. No etching occurs when there is only ion bombardment with no chemical component. a. Write a generalized etch rate equation that can describe this behavior. b. Sketch an etch rate versus ion flux curve for this process for some non-zero chemical flux. ¢. Sketch what the etch profile might look like for this process (ie. etching through a window in a mask as in Figure 10-3). Answer: a, This etch process is a combination of purely chemical etching and ion- enhanced etching (with no purely ionic or physical etching). Purely chemical etching can be given by the first term in the linear etch rate expression, Eq. 10.15, while ion enhanced etching can be written by the series expression given in the text, Eq. 10.20, regardless of the exact physical mechanism. The two parts can be added together to give: Etch Rate = K;S,F, + density UGE S.RJ K, can include any ratio'ing factor between the spontaneous chemical etching and the ion-enhanced chemical component since they will probably have different relative rates. More complex expressions can be derived, such as including any coupling between the spontaneous etching and ion bombardment or surface coverage. b. The etch rate versus ion flux curve would be like that for ion enhanced etching, plus a spontaneous, purely chemical component, which would shift the ion enhanced curve up by that amount of spontaneous chemical etching: 1 08 08 Etch rate 04 0.2 # — Spontancons chemical etch rate (° 0) o os 1 48 2 25 3 35 4 Ton flux * Kj cc, Again, the profile would reflect a linear combination of isotropic purely chemical etching and very directional ion-enhanced etching, giving an anisotropic SILICON VLSI TECHNOLOGY 143 © 2000 by Prentice Hall Fundamentals, Practice and Modeling Upper Saddle River, NJ. By Plummer, Deal and GriffinSolutions Manual etch profile with some lateral etching but with a vertical etch depth greater than the lateral etch distance: PLUS GIVING: SILICON VLSI TECHNOLOGY 144 © 2000 by Prentice Hall Fundamentals, Practice and Modeling Upper Saddle River, NJ. By Plummer, Deal and Griffin