STRIPPING, PASSIVATION AND CORROSION INHIBITION OF SEMICONDUCTOR SUBSTRATES
by
Jian Chen, James S. Papanu, Steve Mak, Caπnel Ish-Shalom,
Peter Hsieh, Wesley Lau, Charles S. Rhoades, Brian Shieh,
Ian Latchford, Karen Williams, and Victoria Yu-Wang
This application claims priority from U.S. Patent Application Serial No.
08/191,828, filed on February 3, 1994, by Rhoades, et al.; U.S. Patent Application Serial No. 08/268,377, filed on June 29, 1994, by Chen, et al.; and U.S. Patent Application
Serial No. , filed on January 9, 1995, by Chen, et al. — all of which are incorporated herein by reference.
In the manufacture of integrated circuits, electrically conductive features are formed on semiconductor substrates by depositing a layer comprising a metal on the substrate, forming a resist composed of polymer or oxide on the layer, and etching the exposed portions of the layer. Especially when halogen-containing etchants, (e.g., Cl2, BC13, CC14, SiC-4, CF4, NF3, SF6 and mixtures thereof, as described for example in Silicon Processing for the VLSI Era. Vol. 1, Chapter 16, by Wolf and Tauber, Lattice Press, 1986, the disclosure of which is incorporated herein by reference) are used to etch the substrate, e.g., in reactive ion etching, the etched substrate is contaminated by etchant residues and byproducts. These contaminants, particularly in conjunction with ambient moisture, can corrode the conductive features, especially when the features are composed of alloys which can galvanic couple, such as Al-Cu or Ti-W. The contaminant byproducts are, for example, produced by reaction between residual halogen from the etchant, metal in the conductive features and/or polymeric resist material. They may be in the form of sidewall deposits formed by condensation of reaction byproducts on sides of the conductive features. Also, after etching, remnant resist which is not etched by the etchant gases can remain on the substrate.
It is known to treat the etched substrate to reduce the adverse effects of such contaminants and to remove the remnant resist on the substrate. The substrate treatments can include (i) removing the remnant resist (usually referred to as stripping), (ii) removing or changing the contaminant (usually referred to as passivation, for example, through exposure to a CF4 plasma), and/or (iii) forming a protective layer over part or all of the conductive feature (usually referred to as inhibition, for example through exposure to a CHF3 plasma). However, known stripping, passivation and inhibition treatments require excessively long treatment times and/or require the use of materials or equipment which are expensive, difficult or dangerous to use.
Also, known passivation and stripping methods can become ineffective and allow corrosion of the substrate too quickly, e g., in 1 to 5 hours. It is generally important to prevent corrosion of the etched features on the substrate for at least until the next step in the processing of the substrate (typically a stripping treatment which removes the resist and/or at least some of the contaminants). If the treatment which prevents corrosion is effective for only a short time, this places serious constraints on the timing of the production process and can result in the loss of an entire batch of wafers if there is an unexpected delay in production caused for example by an equipment failure.
SUMMARY
We have discovered, in accordance with a first aspect of the present invention, an improved process for treating an etched substrate to reduce the adverse effect of etchant byproducts. The process comprising multiple passivating cycles. In the first cycle, passivating gas is introduced into the chamber 52 and plasma is generated from the passivating gas. Thereafter, the flow of passivating gas is stopped, and the plasma in the chamber is extinguished. In the second passivating cycle, passivating gas is again introduced into the chamber, and plasma is again generated in the chamber. In each cycle, the plasma activated passivating gas reacts with the etchant byproducts 24 on the substrate 20 to form gaseous byproducts which are exhausted from the vacuum chamber. The multiple cycle process provides a substrate having improved resistance to corrosion in air.
We have also discovered, in accordance with a second aspect of the invention, a multicycle passivation and stripping process, which allows treating an etched substrate to reduce or remove etchant byproducts and remnant resist thereon. In the passivating step, passivating gas is introduced into the chamber 52 and plasma is generated from the passivating gas to passivate the substrate 20. In the stripping step, stripping gas is introduced into the chamber, and a plasma is generated from the stripping gas to strip the remnant resist on the substrate 20. The stripping and passivating steps are repeated at least once, preferably in the same order that steps were originally done, to yield the multicycle process. In the process, the passivating step can be performed before the stripping step, or vice versa.
DESCRIPTION
The description of the drawings below include a recitation of prefeπed and exemplary features. It is to be understood that each of the features can be used in the invention in general, not merely in the context of the particular drawings, and the invention includes any combination or two or more of these features.
Figure la is a schematic view in vertical cross-section of a substrate with etched metal-containing features, showing etchant byproducts, remnant resist, and sidewall deposits thereon;
Figure lb is a schematic view in vertical cross-section of the substrate of Figure la after passivating and stripping showing removal of the etchant byproducts and remnant resist;
Figure lc is a schematic view in vertical cross-section of the substrate of Figure lb after removal of the sidewall deposits;
Figure Id is a schematic view in vertical cross-section of the substrate of Figure lc, after inhibition of the substrate with amines, showing amines adsorbed on the substrate; and
Figure 2 is a schematic view in vertical cross-section of a vacuum chamber suitable for practicing the process of the present invention.
The process of the present invention is performed on an etched substrate 20 typically comprising a semiconducting material, such as silicon or gallium arsenide. The conductive features 22 on the substrate 20 preferably comprise a metal layer, for example, aluminum, an aluminum alloy such as an Al-Cu alloy, copper, and optionally silicon, and may also include a diffusion barrier layer comprising for example Ti, W, a Ti-W alloy, or
TiN, and/or an antireflective layer comprising for example Si, TiN, or a Ti-W alloy. The substrate 20 has etched conductive features 22 with (i) etchant byproducts 24, (ii) remnant resist 26, and (iii) sidewall deposits 27 on the sidewalls of the features 22, as schematically shown in Figure la. The etchant byproducts 24 on the features 22 typically comprise residual halogen containing radicals and compounds that are formed during etching of the substrate 20. The remnant resist 26 are those portions of the resist that remain on the substrate after etching. The sidewall deposits 27 on the features 22 typically comprise organic compounds containing (i) carbon and hydrogen, (ii) metal from the metal-containing layers, such as aluminum, and (iii) etchant gas species such as boron and nitrogen.
An apparatus 50 suitable for passivating, stripping and inhibiting corrosion of the substrate 20 is schematically shown in Figure 2. The apparatus 50 comprises an etch chamber (not shown) connected by a load-lock transfer area maintained in vacuum (also not shown) to a vacuum chamber 52 having a plasma generation zone 54 and a vacuum zone 56. Process gases enter the vacuum chamber 52 through a gas inlet 60, and are uniformly distributed in the vacuum zone 56 by a "showerhead" type diffuser 62. A substrate support 64 which can comprise a "basket" hoop support (as shown), or a pedestal (not shown), is provided for holding the substrate 20 in the vacuum chamber, and a focus ring 70 maintains the process gas flow around the substrate 20. A heat source, such as infrared lamps 72, can be used to heat the substrate 20. Gaseous byproducts and spent process gas are exhausted from the vacuum chamber 52 through an exhaust 74 via an exhaust system (not shown) capable of maintaining a pressure of at least about 1 mTorr in the vacuum chamber 52.
A microwave plasma generator assembly 80 connected to the plasma generation zone 54 of the apparatus 50 can be used to generate a plasma from the process
gas. A suitable microwave generator assembly 80 is an "ASTEX" Microwave Plasma Generator commercially available from the Applied Science & Technology, Inc. , Woburn, Massachusetts. Typically, the microwave generator assembly 80 comprises a microwave applicator 82, a microwave tuning assembly 84, and a magnetron microwave generator 86. Alternative plasma generating sources, such as RF-generated plasmas and inductive coupled plasmas are also effective.
To perform the process of the present invention, an etched substrate 20 is transferred to the vacuum chamber 52 which is maintained at suitable temperature and pressure. A multicycle process comprising either (i) multiple passivating steps, or
(ii) multiple passivating and resist stripping steps, is then used to treat the etched substrate to reduce the adverse effect of etchant residues and byproducts on the substrate.
The multicycle passivation process comprises at least two passivating cycles, and is desirable when there are only etchant byproducts on the substrate, or when it is desirable to strip the substrate in a separate stripping step. In the first cycle, passivating gas as described below is introduced into the chamber 52 and plasma is generated from the passivating gas. The plasma activated passivating gas reacts with the etchant byproducts 24 on the substrate 20 to form gaseous byproducts which are exhausted from the vacuum chamber. Thereafter, the flow of passivating gas is stopped, and the plasma in the chamber is extinguished. In the second passivating cycle, passivating gas is again introduced into the chamber, and plasma is again generated in the chamber.
More typically, it is desirable to use a multicycle process comprising both stripping and passivating steps so that remnant resist and etchant byproducts can be simultaneously removed from the substrate. In the passivating step, passivating gas is introduced into the chamber 52 and plasma is generated from the passivating gas to passivate the substrate 20. In the stripping step, stripping gas as described below, is introduced into the chamber, and a plasma is generated from the stripping gas to strip the polymeric resist or oxide hard mask on the substrate 20. The stripping and passivating steps are repeated at least once, preferably in the same order that steps were originally done
to yield the multicycle process. In the process, the passivating step can be performed before the stripping step, or vice versa. Typically, a stabilization step is performed between each process step, during which the process conditions for the successive step are stabilized.
Generally, in either of the multicycle processes, a larger number of cycles provides higher corrosion resistance of the processed substrate 20, even if the total time duration of the entire process is not increased. However, because a smaller number of cycles provides faster process throughput, particularly when the duration of the stabilization period between each process step is long, the number of cycles is preferably from about 1 to about 10 cycles, and more preferably from about 2 to about 5 cycles.
Preferably, each passivating step or optional stripping step of the multicycle process has a duration ranging from about 1 to about 60 seconds, more preferably from 2 to 30 seconds, and most preferably from 2 to 20 seconds. In the multicycle passivating and stripping processes, the passivating step within each cycle has typically the same duration as the stripping step within the same cycle, however, these steps can also have different durations.
Suitable passivating gases for use in the multicycle process include any gas capable of reacting with the etchant byproducts 24 on the substrate 20 to form gaseous byproducts which can be exhausted from the vacuum chamber 52. For example, when the etchant byproducts 24 comprise chlorine, the passivating gas can comprise a hydrogen- containing gas which reacts with the chlorine to form etchant byproducts such as hydrochloric acid vapor, which is exhausted from the chamber 52. Suitable passivating gases, include (i) ammonia and oxygen, or (ii) water vapor, with optional oxygen and nitrogen, can be used in this step. When the passivating gas comprises ammonia and oxygen, the volumetric flow ratio of ammonia to oxygen is preferably from about 1:1 to about 1:20, more preferably from about 1:5 to about 1:15, and most preferably about 1:10. For a 5-liter capacity chamber 52, a preferred gas flow comprises 300 seem NH3 and 3000 seem O2. When an ammonia and oxygen passivating gas was used, a two-cycle process
comprising 10 second passivation and 10 second strip steps provided an optimal combination of high corrosion resistance and high process throughput efficiency.
Alternatively, a passivating gas comprising only water vapor can be used to passivate the etchant byproducts 24. When the vacuum chamber 52 has a five-liter capacity, the water vapor flow rate is preferably from about 100 to 1000 seem, and more preferably about 500 seem. For passivating gas comprising water vapor, optimal corrosion results were achieved using a three-cycle multicycle process comprising passivation and stripping steps having a duration of 20 seconds each. The water vapor is formed in a boiler or bubbler 100 containing water which is connected to the vacuum chamber 52 by the feedline 102. The boiler or bubbler is maintained at a sufficiently high temperature and at a sufficiently low pressure to vaporize the water. When a boiler is used, the water in the boiler is heated to a temperature close to the boiling point of water. Typically, the pressure in the boiler ranges from about 50 Torr to about 200 Torr, and more typically ranges from 100 Torr to 150 Torr. When a bubbler is used, an inert carrier gas such as argon or helium can be passed through the bubbler to transport water vapor to the vacuum chamber 52.
Preferred gases which can be used in the multicycle processes to either passivate, or passivate and strip, the substrate comprise water vapor, oxygen, and nitrogen. Depending on the ratio by volume of (i) water vapor (Vmo) to (ii) oxygen and nitrogen together (V02 + VN2), the process gas can have primarily a passivating function, or primarily a stripping function. When the ratio by volume of Vm0:(yo2 + Nm) is fr m about 1:2 to about 2:1, preferably 0.8:1 to 1:0.8, and especially about 1:1, the process gas functions primarily as a passivating gas; and a separate resist stripping step is used to remove the resist on the substrate. When the ratio by volume of VH20:(Vo2 + V ) is from about 1:4 to about 1:40, preferably 1:6 to 1:20, and especially about 1:10, the process gas functions primarily as a stripping gases and a separate passivation step can be used to passivate the substrate. To serve primarily as a stripping gas, the water vapor content should be less than about 20% by volume of the combined oxygen and nitrogen gas content to provide adequate stripping rates. It is believed that the water vapor addition results in passivation of some of the etchant byproducts, thus allowing simultaneous stripping and
passivating of the substrate. In either process gas mixture, the ratio by volume of oxygen to nitrogen is preferably from about 1:1 to about 50:1, and more preferably from 1:1 to 20: 1, and especially about 10:1.
A suitable stripping gas to strip polymeric resist in the multicycle stripping and passivating process comprises (i) oxygen, and (ii) an oxygen activating gas or vapor, such as water vapor, nitrogen gas, or fluorocarbon gas, the fluorocarbon gases including CF4, C2F6, CHF3, C3H2F6, and C2H4F2 (as for example disclosed in U.S. Patent No. 5,221,424, to Rhoades, and U.S. Patent No. 5,174,856 to Hwang, et al., both of which are incorporated herein by reference). For example, a preferred stripping gas comprises oxygen and nitrogen in a volumetric flow ratio of about 6:1 to about 200:1, and more preferably from 10: 1 to 12:1, for example, a suitable gas flow rate for a 5-liter vacuum chamber 52 comprises 3000 to 3500 seem of O2 and 300 seem of N2.
Stripping gases suitable for stripping oxide hard mask include halogen containing gases, for example CF4, C2F6, CHF3, C3H2F6, C2H4F2 and HF. Other halogen gases, such as BC13, CC14, or SiCl4, can also be used to facilitate removal of sidewall deposits, however, when chlorine containing gases are used, the oxide stripping process should be performed in the etching chamber (not shown) of the apparatus 50, to prevent contamination of the passivation chamber with chlorine gas. Typically, only a portion of the oxide hard mask is stripped during the stripping step, and a subsequent process step is used to deposit a dielectric or insulative layer on the substrate.
In the multicycle process, the pressure and temperature of the vacuum chamber 52 can be varied between successive passivating or stripping steps, or varied between successive cycles, or maintained substantially constant. Preferably, the vacuum chamber 52 is maintained at a pressure ranging from about 1 to about 100 Torr, more preferably from 1 to 10 Torr, and most preferably at 2 Torr. Optionally, in a second step or cycle the pressure in the chamber 52 is reduced to a second lower pressure of less than about 1 Torr, more typically less than 500 mTorr, and most typically less than about 100 mTorr. Typically, the substrate 20 is heated using the lamp heaters 72 to a temperature of
from about 150°C to about 400°C, and more preferably from 200°C to 380°C. Preferably, the change in pressure and temperature between successive sieps is minimized to increase process throughput efficiency.
A plasma is formed from the stripping or passivating gas using the microwave plasma generator 80 of the apparatus 50. When the plasma causes the substrate 20 to heat up, the power level of the microwave generator 80 is regulated so that the temperature of the substrate remains substantially constant. Typically, the power output of the microwave generator 80 ranges from 500 to 2500 Watts, and more preferably from 800 to 1500 Watts.
The multicycle passivation process can provide faster process throughput and superior corrosion resistance than prior art processes. The ability to achieve corrosion resistance equivalent or superior to that obtained using single cycle processes, in a shorter duration process, is an unexpected commercial advantage of the multicycle process. It is believed the multicycle passivating process can provide fast throughput because it allows faster etchant byproduct removal because of faster diffusion mechanisms. In the first passivation step, etchant byproducts 24 are removed from the surface of the substrate 20. Thereafter, passivating species must diffuse into the remnant resist 24, and sidewall deposition 27 to react with the etchant byproducts 24 therein, and the reaction products must diffuse out of the resist or sidewall, and thereafter desorb. When the flow of passivating gas is stopped in multicycle passivation process more rapid desorption of passivating byproducts should occur. The sequential depletion and diffusion process allows faster and more effective passivation of the etchant byproducts 24.
The multicycle passivating and stripping processes have demonstrated faster throughput, and superior corrosion resistance than prior art processes. It is believed the stripping steps remove portions of the remnant resist 26 and sidewall deposits 27 thereby making the remnant resist 26 and the sidewall deposits 27 thinner and more porous, and enhancing both diffusion of passivating species into the remnant resist 26, and diffusion of reaction products out of the resist 26. The multicycle process provides a substrate that is
resistant to corrosion by ambient moisture for at least 24 hours, and more typically at least 48 hours, after passivation of the substrate. This is a substantial improvement over prior art processes, which typically provide corrosion resistance for only about 1 to 2 hours. The improved corrosion resistance allows more efficient processing schedules and reduces manufacturing losses.
The corrosion resistance obtained from the multicycle process can be further improved by exposing the substrate, under vacuum, to an alkylamine of the formula:
R, - N - R2 !
R3 wherein R3 is an alkyl group, preferably an alkyl group containing 1 to 5 carbon atoms, e.g., methyl, ethyl, or propyl; and each of R2 and R3, which may be the same or different, is a hydrogen atom or an alkyl group, preferably an alkyl group containing 1 to 5 carbon atoms, e.g., methyl, ethyl, or propyl. The vapor pressure of the amine in the vacuum chamber is preferably such that part of the amine is present in gaseous form and part of the amine is adsorbed on the substrate. It is believed that in the amine inhibition step, the amine forms a passivating layer adsorbed on the surface of the features that serves to inhibit coπosion of the features. The amine inhibition layer is useful for features 22 containing metal alloys that exhibit high galvanic coupling activity, to prevent coπosion of these metals.
To perform the amine inhibition step, amine vapor is introduced into the vacuum zone 56 of the vacuum chamber 52 for a sufficient time to adsorb sufficient amine on the substrate 20 to inhibit coπosion of the substrate 20 for at least about 24 hours when the substrate 20 is exposed to the atmosphere. Generally, during this 24-hour period, the substrate 20 undergoes additional processing steps which eliminate the necessity for inhibiting coπosion of the substrate 20. The longer the substrate 20 is exposed to the amine the more effective is the coπosion inhibition. However, for process throughput efficiency, the amine is exposed to the substrate 20 for less than about 120 seconds, more preferably for less than about 90 seconds, and most preferably for less than about 60 seconds. During
the amine exposure step, the vacuum zone 56 is preferably maintained at a pressure ranging from about 1 Ton to about 100 Ton, and more preferably ranging from 1 Ton to 10 Ton.
The vapor pressure of the amine should be sufficiently high that at least a portion of the amine is gaseous in the vacuum chamber 52 and sufficiently low that at least a portion of the amine is adsorbed onto the substrate 20 in the vacuum chamber 52. The alkyl moieties of the amine preferably comprise alkyls, such as methyl, ethyl and propyl, and each alkyl preferably comprising from 1 to 5 carbon atoms. Suitable amines include mono-alkyl, di-alkyl and tri-alkyl substituted methylamines, ethylamines, propylamines, such as monomethylamine, dimethylamine and trimethylamine, because these amines have low boiling points and are readily commercially available. Particular amines which can be used include trimethylamine (BP 2.9°C), which is prefeπed, diethylamine (BP 7.4°C), and monomethylamine (BP -6.3°C). Preferably, the amine comprises at least two alkyl moieties, and more preferably three alkyl moieties. Of these amines, trimethylamine is prefeπed because it is believed that the tertiary (trialkyl) amines are more effective inhibitors than the secondary (dialkyl) amines, which in turn, are more effective inhibitors than the primary (monoalkyl) amines. However, the toxicity and commercial availability of the amines may also control selection of an appropriate amine.
The amine vapor is formed from an amine source 104, comprising an amine gas or amine liquid. A liquid amine source 104 comprises a boiler or bubbler fluidly connected to the vacuum chamber 52 by the feed line 102, and maintained at a sufficiently high temperature and sufficiently low pressure to vaporize at least a portion of the liquid amine. Preferably, the boiler or bubbler is maintained at room temperature. The pressure in the boiler is preferably from about 50 Ton to about 200 Ton, and more preferably from 100 Ton to 150 Ton. When a boiler is used, the boiler is maintained at a temperature substantially equal to the boiling point of the liquid amine to form amine vapor. When a bubbler is used, a carrier gas, such as argon or helium, bubbled through the bubbler transports the amine vapor to the vacuum chamber 52.
After the stripping, passivating, and optional amine inhibition process steps, the passivated and stripped substrate 20 is removed from the chamber and etched in an etchant solution to remove the sidewall deposit 27 that forms during the etch process. Conventional wet chemical etchant solutions are suitable. In this process, the wafers were chemically etched either in an "ACT" 900 series amine-based liquid etchant, commercially available from Advanced Chemical Technologies, Allentown, New Jersey, or in a HF- containing wet chemical etchant. After wet etching, the wafer was rinsed in deionized water to remove residual wet etchant.
EXAMPLES
The following examples illustrate the process of the present invention. All the examples were performed in an "AMAT PRECISION 5000 METAL ETCHER" system, commercially available from Applied Materials, Santa Clara, California. The "PRECISION 5000" apparatus had an etching chamber (not shown) connected to the passivating and stripping chamber (as shown), so that the substrate can be transfened from the etching chamber to the passivating and stripping chamber without exposure to the atmosphere. The etch chamber of the apparatus (not shown) also comprised an inductive coil encircling the chamber for optionally generating a magnetic field to enhance the intensity of the plasma in the chamber. The experiments were performed on silicon wafers having a diameter of about 200 mm (8 inches) and a thickness of 0.73 mm.
After passivating and stripping, the coπosion resistance of the substrates was tested by exposing the processed substrates to atmosphere or exposing the substrate to elevated moisture levels ranging up to about 40% relative humidity for set intervals of time, and thereafter examining the substrates in a microscope under dark field conditions, or in a scanning electron microscope. Conosion of the substrate was visible as specks of scattered light caused by the conoded hydroscopic species formed on the metal features. The coπosion was measured either directly after the passivating and stripping steps, at intervals of 2 hours to 7 days; or alternatively, after a wet etching step was performed and at intervals of 1, 2, and 3 weeks.
Examples 1-11
In these examples, either single or multicycle passivating and strip processes were performed on substrates having features comprising (i) TiW barrier layer; (ii) an aluminum-silicon-copper conductive metal layer approximately 550 nm thick, the aluminum alloy containing 1.5% silicon and 0.5% copper; and (iii) an antireflective layer of titanium. The substrates previously were etched in an reactive ion etching process using a BC13, Cl2, and N2 gas mixture.
The process conditions and the results of the conosion tests for Examples
1-11 are described in Table I. The passivating step was effected using water vapor flowed at a volumetric flow rate of 500 seem. The stripping step was performed using a stripping gas comprised of oxygen, nitrogen, and optionally water vapor in the described flow rates. In all the processes, the vacuum chamber was maintained at a pressure of 2 Ton. Generally, the multicycle passivating and stripping processes provided superior coπosion results compared to the single cycle processes.
The multicycle passivating and stripping process used for Example 7 provided the best conosion resistance. In this process, each passivating and stripping step was 20 seconds in duration, and the passivating and stripping steps were repeated three times.
Coπosion resistance of greater than 72 to 96 hours was obtained.
Comparison of Examples 6 and 7 suggests that for the same total multicycle process time, an increased number of cycles provide better coπosion resistance.
In comparing Examples 7 and 9, it is observed that a final stripping step substantially increases the coπosion resistance of the substrate. In Example 9, where the final stripping step was not performed, a conosion resistance of 3 to 7 hours was observed; whereas in Example 7, a coπosion resistance exceeding 72 hours was observed. It is believed that a final stripping step increases corrosion resistance because the oxygen in the
stripping gas oxidizes the aluminum in the features, forming a thin protective layer of aluminum oxide on the metal features.
Example 11 demonstrates that the fastest total processing time can be achieved by increasing the number passivating and strip cycles.
TABLE I
Total Hours of
8,0 Passivating Step Stripping Step1 Cycles Time Corrosion Resistance
RF Gas
Example Power Temp Duration Composition Temp Duration (No) (Sec) (Hours) (Watts) <°C) (Sec) (seem) (°) (Sec)
1 O,:3000
None None None N,:200 300 40 40 ≤2 8,0:300
2 800 250 40 None None None 40 ≤2
3 800 250 90 None None None 90 ≥2 ≤24
4 800 250 60 O,:3000 250 60 120 ≤2 N2:200
5 1400 250 150 None None None 150 ≤2
6 1400 220 30 O,:3000 250 30 2 120 ≤7 N,:200
7 O,:3000 ≥72 to
1400 200 20 N2:200 220 20 3 120 962 8,O:300
8 O,:3000
1400 200 10 N,:200 220 10 3 60 ≤2 8,0:300
9 02:3000 ≥3
1400 200 20 N2:200 220 20 3/23 100 ≤7 8,0:300
10 O,:3000 ≥3
1400 200 20/10 N,:200 220 20/10 2 80 ≤24 8,0:300
11 O,:3000
1400 220 10 N,:200 220 10 5 100 ≥72 8,0:300
(1 ) For all stripping steps, the RF power applied to the microwave generator was maintained at 1400 Watts.
(2) Repeatability was demonstrated for this example.
(3) Three passivating cycles + two stripping cycles.
EXAMPLES 12-40 In these examples, a multicycle process using (i) a passivating gas comprising 3000 seem and 300 seem of ammonia, and (ii) a stripping gas comprising 3000 seem of oxygen and 300 seem of nitrogen was used.
The features on the substrates used in these examples comprised sequentially (i) a 3,500 A thick layer of aluminum alloy containing 0.5% copper, (ii) a 450 A layer of TiN, (iii) a 1,000 A layer of Ti, and (iv) a 365 A layer of TiN. The wafers were previously etched in a reactive ion etching process that used an etchant gas comprising BC13, Cl2 and N2, and optionally CF4.
For all the examples, an initial stabilization step of about 10 seconds in duration was performed. In this step, the substrate was transfened to the vacuum chamber. The temperature of the chamber was ramped up from a temperature of about 60 to 100°C to a temperature of about 325 °C, and the pressure of the chamber ramped from a pressure of about 10 to 50 mTon to a pressure of about 2 Ton. Passivating gas comprising 3000 seem O2 and 300 seem NH3 was flowed in the chamber. After the stabilization step, process gas conditions suitable for passivating and stripping the substrates were maintained as described below.
Examples 12-21
Examples 12-21 were processed using an L9 (four process variables with three levels for each variable), orthogonal factorial design experiment. Table II describes the four process variables and the three levels used for each variable. Table III describes the actual process conditions used to process each of the ten wafers. Example 21 was run at the same process conditions as Example 12 to verify the repeatability of the experiment. In these examples, the first passivating and stripping cycle was performed at a temperature of about 325 °C, and the temperature of the second passivating cycle was varied as shown in Table III. In Table III, the variable PS/PT represents the duration in time for a single passivating and strip cycle divided by the total multicycle process time. The variable
P/(P+S) represents the total time of a single passivating step divided by the total time for a passivating and strip cycle.
The wafers are inspected under 100X magnification in an optical microscope to examine the photoresist remaining on the wafer immediately after stripping, and to identify conosion of the wafers after the wafers were exposed to the atmosphere for 6 and 24 hours. No photoresist was seen on any of the wafers, and no signs of coπosion were observed on any of the wafers after the 6 and 24 hour test intervals. These examples demonstrate the low variability in coπosion performance obtained using multicycle passivating and stripping processes.
TABLE II
Level Total Process Time Temperature of Second PS/PT P/(P + S) (Sec) Cycle (°C) (%) (%)
1 60 380 63 75
2 50 353 50 63
3 40 325 38 50
TABLE III
Example Total Process Time Temperature of Second PS/PT PΛP + S) (Sec) Cycle (°C) (%) (%)
12 60 380 63 75
13 60 353 50 63
14 60 325 38 50
15 50 380 50 50
16 50 353 38 75
17 50 325 63 63
18 40 380 38 63
19 40 353 63 50
20 40 325 50 75
21 60 380 63 75
Examples 22-31
Examples 22-31 were also processed also using an L9 orthogonal factorial design experiment. The factorial design variables and levels used are listed in Table IV. Table V shows the process conditions used to process each of the Examples 22-31.
Example 31 was run at the same process conditions as Example 22 to verify repeatability of the experiment.
After processing, each of the wafers was examined at 100X magnification in an optical microscope. No photoresist was visible on any of the wafers, and none of the examples exhibited any conosion after 6 and 24 hours, with the exception of Example 30, which exhibited severe coπosion.
These experimental results demonstrate that the conosion resistance of the substrates is improved with longer passivating and strip process times and higher stripping process temperatures.
TABLE IV
Total Process Time Temperature of Second PS/PT
Level P/(P + S) (Sec) Cycle (°C) (%) (%)
1 80 380 63 75
2 70 353 50 63
3 60 325 38 50
TABLE V
Example Total Process Time Temperature of Second PS/PT P/(P + S) (Sec) Cycle (°C) (%) (%)
22 80 380 63 75
23 80 353 50 63
24 80 325 38 50
25 70 380 50 50
26 70 353 38 75
27 70 325 63 63
28 60 380 38 63
29 60 353 63 50
30 60 325 50 75
31 80 380 63 75
Examples 32-40
In Examples 32-40, two sets of wafers were processed using multicycle passivating and stripping processes. The first set of wafers were processed to examine the conosion resistance of the wafers immediately after the stripping and passivating processes. The second set of wafers was processed to examine the coπosion resistance after the stripped and passivated wafers were wet chemical etched.
The first set of wafers used for Examples 32-36 were processed using a two cycle passivating and strip process. The passivating and stripping steps in each cycle were run for about ten seconds, providing a total multiple processing time of 40 seconds. The total processing time for the multiple process, including three second intervals between each passivating and strip step, was 49 seconds. The passivating and stripping steps of the first cycle were both performed at a temperature of 325 °C, and the passivating and stripping steps of the second cycle were both performed at a temperature of 380° C.
After processing, the wafers were stored at room temperature in a 40% relative humidity atmosphere. An optical microscope was used to examine residual resist and coπosion on the wafers after set intervals of time. The wafers were examined after one, two, three, four, and seven days; all five of the wafers being examined after the first day, four after the second day, three after the third day, and so on until only one wafer was examined on the seventh day. No conosion was visible on any of the wafers after these time intervals.
The second set of wafers used for Examples 37-40 were also processed using a two-cycle passivating and stripping process. The duration of each passivating and stripping step in Example 37 was 10 seconds; in Example 38 was 5 seconds; in Example 39 was 3 seconds; and in Example 40 was 2 seconds. The passivating and stripping steps of the first cycle were both performed at a temperature of 325 °C, and the passivating and stripping steps of the second cycle were both performed at a temperature of 380°C.
After processing, the wafers were stored at room temperature in a 40% relatively humidity environment. After 24 hours, there was no conosion visible on any of the wafers.
Thereafter, the wafers were wet chemically etched using an acidic solution containing HF. After etching, the wafers were again stored at room temperature in a 40% relative humidity chamber, and examined under an optical microscope at intervals of one, two, and three weeks. No coπosion was observed on any of the wafers even after three weeks.
Examples 37-40 demonstrated that there was no difference in observed conosion results between the wafers processed using two second passivating and strip step durations, and the wafers processed using ten second passivating and strip step durations.
EXAMPLES 41-53 Examples 41-53 were processed using a single or multiple cycle passivating and strip process. In all of these examples, the passivating gas included water vapor, and optionally oxygen and nitrogen, as described below.
The features on the substrates of Examples 41-53 comprised (i) a 100 nm thick barrier layer of Ti, (ii) either a 1000 nm or a 1050 nm thick conductive layer of Al containing 0.5% Cu, and (iii) either a 36 nm or a 45 nm thick antireflective layer of TiN. The wafers were etched in a reactive ion etching process that used a BC13, Cl2, and N2 gas mixture, and before etching, the photoresist had a thickness of about 1.8 to 2 microns.
Although both multicycle and single cycle processes were performed on the substrates, some of the single cycle processes, with total process times of 50 seconds, provided adequate conosion resistance. It is believed that the metal alloys in these features have a sufficiently low galvanic activity that a multicycle process is not needed unless a shorter total process time is desired.
Examples 41-45
In Examples 41-45, a wafer "basket" or hoop was used in the vacuum chamber to hold the substrates. After a substrate was placed in the wafer basket, an initial chamber stabilization step was effected for about 15 seconds. In the stabilization step, process gas was flowed into the vacuum chamber at the flow rates shown in Table VI, and the temperature and pressure of the vacuum zone were maintained at the desired process levels. After stabilization, single cycle passivating and stripping processes, as described in
Table VI, were performed. The passivating step of the process had a duration of 20 seconds, and the stripping step a had duration of 40 seconds.
The results of the coπosion tests for Examples 41 through 45 are listed in Table VI. It was observed that the coπosion resistance of the substrates was not effected
by a reduction in the passivating temperature (compare Examples 41 and 42 and Examples 43 and 44), or by an increase in stripping temperature (compare Examples 44 and 45).
TABLE VI
Passivating STRIP
EXAMPLE RESULTS
RF RF
GAS FLOW TEMP POWER GAS FLOW TEMP POWER (SCCM) CO (WATTS) (SCCM) CO (WATTS)
0,:1000 02:3500 no corrosion > 48
41 8,0:500 245 1400 N2:200 275 400 hr N,:100
0,:1000 0,:3500 no corrosion > 48
42 8,0:500 200 1000 N,:200 275 400 hr N,:100
43 8,0:500 200 1 100 O,:3500 275 400 no corrosion > 48 N,:200 hr
44 H,0:500 245 1400 O2:3500 275 400 no corrosion > 48 N,:200 hr
45 8,0:500 200 1 100 O,:3500 325 500 no corrosion > 48 N2:200 hr
Examples 46-52
In Examples 46-52, the process conditions for the passivating and stripping steps were maintained constant, and the duration of the passivating and stripping steps were varied, as shown in Tables VII and VIII. In Examples 46-50, the passivating step preceded the strip step, and in Examples 51 and 52, the stripping step preceded the passivating step. Example 49 used a two-cycle multicycle passivating and stripping process.
In these examples, a pedestal (not shown) was used to hold the substrate in the vacuum chamber. The pedestal allowed more control over the temperature of the substrate, because the larger mass of the pedestal, as compared to the wafer basket holder, stabilized the temperature of the substrate.
In the passivating step, a process gas comprising 500 seem H2O, 1000 seem of O2, and 100 seem of N2 was used, and the vacuum zone was maintained at a pressure of
about 2 Ton. The power level of the microwave generator 86 was maintained at about 1400 Watts.
In the stripping process, a process gas comprising 3500 seem O2 and 300 seem of N2 was used, and the vacuum chamber was maintained at a pressure of about 2 Ton. The microwave power level was maintained at about 1000 Watts.
These examples demonstrated that a single cycle water based passivating and stripping process using a substrate temperature ranging from 200 to 300 °C can provide effective conosion resistance. For features having the described metal-containing layers, a single cycle process having a total duration of at least about 150 seconds provided adequate conosion resistance, and a multicycle process was not needed. The examples also demonstrated that the order of the passivating and stripping steps did not affect the conosion resistance of the substrate.
TABLE VII
TIME (SECS)
EXAMPLE TEMP CO RESULTS
STABILIZATION Passivating STRIP
46 10 30 50 250 no corrosion > 24 hr
47 10 20 40 300 no corrosion > 24 hr
48 10 30 30 300 very light corrosion after 24 hr
49 10 15' 10' 300 no corrosion > 24 hr 15' 20'
50 5 15 30 300 complete resist removal; no corrosion > 24 hr
Note 1 : These wafers were processed using a two-cycle process. The passivating steps both had ; duration of 15 seconds, and the stripping steps were either 10 or 20 seconds in duration.
TABLE VIM
TIME (SECS)
EXAMPLE TEMP CO RESULTS
STABILIZATION STRIP Passivating
51 5 30 50 250 remaining resist, and light corrosion > 24 hr
52 5 20 40 300 no corrosion > 24 hr
Example 53
Example 53 illustrates a prefened single cycle stripping and passivating process, the process conditions of which are disclosed in Table IX. This process is advantageous because it provides a reasonably high wafer throughput rate while maintaining effective passivating and stripping quality. The wafer passivated using this process demonstrated coπosion resistance over 24 hours when exposed to the atmosphere.
TABLE IX
ITEM STABILIZATION Passivating STRIP
O2 (seem) 0 0 3500
N2 (seem) 0 0 300
H2O (seem) 500 500 0
Temperature (°C) 250 250 250
Pressure (Ton) 2 2 2
Power (W) Basket 0 1400 500 Pedestal 0 1400 1000
Time (sec)
Basket 15 20 40 Pedestal 5 20 40
Examples 54 and 55
These examples illustrate that a single cycle water vapor passivating process can be used to prevent conosion of a highly conosive, partially etched Ti-W layer on a substrate. In these examples, a complete reactive ion etching, passivating and stripping process sequence is described. The wafers used in these examples had features comprising (i) a barrier layer of Ti-W alloy, (ii) a conductive layer of an aluminum containing alloy, and (iii) a antireflective layer. The features on the wafers were etched through until the
lower Ti-W barrier layer was exposed. The Ti-W layer was not etched through, because the underlying circuit devices can be damaged by the plasma etching process. The partially etched Ti-W barrier layer rapidly conodes when exposed to the atmosphere, because the galvanic coupling of the metals in the alloy promotes coπosion. Thus, immediately after etching, the partially etched barrier layer was passivated using a water vapor containing plasma.
The substrates of Examples 54 and 55 were etched in an etching chamber (not shown) using a two-stage etching process. In the first etching stage, etchant gas comprising BC13 at a flow rate of about 50 seem, Cl2 at a flow rate of 40 seem, and N2 at a flow rate of 20 seem, was introduced into the etching chamber. The pressure in the chamber was maintained at about 200 mToπ. The RF power applied to the cathode in the chamber was maintained at about 400 Watts, and a 40 gauss magnetic field was generated using the inductive coils to enhance the plasma. The first etching stage was effected until the aluminum containing layer on the substrates was etched through, the end point of the etching step measured by optical emission techniques. In the second etching stage, the 1500 A thick Ti-W barrier layer on the substrate was etched until 500 A of the Ti-W layer was etched through, and 1000 A of the Ti-W layer remained on the substrates. In the second etching stage, a process gas comprising 25 seem of BC13, 20 seem of Cl2, and 20 seem of N2, was introduced into the chamber, and the chamber was maintained at a pressure of about 20 mToπ. The RF power applied to the cathode was maintained at a level of 250 Watts, and a 40 gauss magnetic field used to enhance the plasma in the chamber. The second etching stage was affected for about 40 seconds.
After etching, the wafer was transfened from the etching chamber (not shown) to the passivating and stripping chamber. In Example 54, the wafer was passivated and stripped in separate steps. In the passivating step, water vapor was introduced into the chamber at a flow rate of 500 seem. An 800 Watt RF power was applied to the microwave plasma generator and the wafer was heated to 250°C. The passivating process was effected for a total time of 45 seconds. After passivating, the wafer was stripped in a separate stripping step. The stripping step used a stripping gas comprising oxygen at a flow of 300
seem and nitrogen at a flow of 200 to 300 seem. A 1400 Watt RF power level was maintained at the microwave generator, and the temperature of the wafer maintained at 250 °C. In both the passivating and stripping steps, the pressure in the chamber was maintained at 2 Ton. The passivated wafer of Example 54 was substantially coπosion resistant when exposed to the atmosphere.
The wafer of Example 55 was passivated and stripped in a single step process. Process gas comprising oxygen at a flow of 3000 seem, nitrogen at a flow of 200 seem, and water vapor at a flow of 300 seem was introduced into the chamber. A plasma at a power level of 1400 Watts was generated for about 90 seconds to strip and passivate the wafer. The passivated wafer of Example 55, was also observed to be resistant to coπosion under atmospheric conditions. After stripping and passivating, the remaining 1000 A thickness of barrier layer on the passivated wafer, was removed using a wet chemical etching process. In addition to this process, the wafer was chemically etched in "ACT" 900 series liquid etchants, commercially available from Advanced Chemical
Technologies, Allentown, New Jersey, to remove sidewall deposition. After etching, the wafer was rinsed in deionized water to remove residual etchant.
The present invention has been described in considerable detail with reference to certain prefened versions thereof, however, other versions are possible. For example, the multicycle process can be performed using passivating and stripping gas processes other than those disclosed herein. Also, the single cycle water vapor based passivating and stripping processes can be combined with other passivating and stripping processes, to provide greater conosion resistance and process efficiency. Therefore the spirit and scope of the appended claims should not be limited to the description of the prefened versions contained herein.