PROCESS FOR LOWERING ADHESION LAYER THICKNESS AND IMPROVING DAMAGE RESISTANCE FOR THIN ULTRA LOW-K DIELECTRIC FILM

[0001] 1. Field of the Invention

[0002] Embodiments of the present invention generally relate to the fabrication of integrated circuits. More particularly, the embodiments relate to process for depositing low dielectric constant films for integrated circuits.

[0003] 2. Description of the Related Art

[0004] Semiconductor device geometries have dramatically decreased in size since such devices were first introduced several decades ago. Since then, integrated circuits have generally followed the two year/half-size rule (often called Moore's Law), which means that the number of devices that will fit on a chip doubles every two years. Today's fabrication plants are routinely producing devices having 65 nm and even 45 nm feature sizes, and tomorrow's plants soon will be producing devices having even smaller geometries.

[0005] The continued reduction in device geometries has generated a demand for films having lower dielectric constant (k) values because the capacitive coupling between adjacent metal lines must be minimized to further reduce the size of devices on integrated circuits in Cu dual damascene interconnect process technology. One of the approaches that has been used to obtain an ultra low dielectric constant (k<2.5) is to fabricate hybrid films of a silicon matrix and an organic porogen by depositing the hybrid films from a gas mixture comprising an organosilicon compound and a compound comprising thermally labile species or volatile groups (porogen) and then post-treat the deposited films with UV curing or thermal treatment to remove the thermally labile species or volatile groups of porogen from the deposited films, resulting in nanometer-sized voids in the films which lowers the dielectric constant of the films.

[0006] The nanoporous films are known to have less adhesion to underlying barrier/liner layers than silicon oxides. Improvement of adhesion may be obtained by depositing an adhesion layer of oxide, which can enhance adhesion at the interface. To further improve adhesion, it has been suggested to use a gradient layer with increasing carbon content gradually between adhesion and main low-K film deposition step. However, uncontrolled transition of both silicon and porogen flow in this gradient layer can cause undesirable gas phase reaction (due to variable changes of RF power, pressure, and flow rate etc.), causing particle clusters on the film or/and carbon bumps to form in the films or at the interfaces.

[0007] In addition, it has been reported that ultra low dielectric constant films developed as described above exhibit less than desirable mechanical properties, such as poor mechanical strength (modulus≈4 GPa), which renders the films susceptible to damage during subsequent semiconductor processing steps. Moreover, since an oxide adhesion layer that is currently used for better adhesion with ultra low dielectric constant films to underlying barrier/liner layers constitutes major portion for the dielectric film stack and typically has higher dielectric constant (k≈3.5) and very high thickness non-uniformity, the overall dielectric constant and thickness non-uniformity of the resulting dielectric film stack have not been reduced as expected.

[0008] Therefore, there is a need for a process of making ultra low dielectric constant materials with improved mechanical strength, lowered thickness non-uniformity, and minimize k (dielectric constant) increase from initial stages of depositing the ultra low dielectric constant materials, without compromising the controllability for lower application thickness.

[0009] Embodiments of the present invention generally provide a method for depositing an ultra low dielectric constant film with novel process parameters. In one embodiment, the method includes flowing into the processing chamber a gas mixture comprising a flow rate of one or more organosilicon compounds and a flow rate of one or more porogen compounds to deposit an initiation layer (oxide layer) on the substrate by applying a radio frequency (RF) power to the processing chamber, ramping-up the flow rate of the one or more organosilicon compounds until reaching a final flow rate of the one or more organosilicon compounds to deposit a first transition layer on the initiation layer, and while flowing the final flow rate of the one or more organosilicon compounds, ramping-up the flow rate of the one or more porogen compounds until reaching a final flow rate of the one or more porogen compounds to deposit a second transition layer on the first transition layer, wherein the depositions are performed at a low RF power between about 350 W and about 500 W, and a ratio of the RF power to a total flow rate is between about 0.1 W/sccm and about 0.3 W/sccm. Various processing parameters and precursors are further discussed in the detailed description.

[0010] In another embodiment, the method includes providing a substrate bearing a liner/barrier layer, depositing a carbon-containing oxide adhesion layer over the liner/barrier layer at a deposition rate between about 1000 Å/min and about 3500 Å/min, comprising flowing into the processing chamber a gas mixture comprising a flow rate of one or more organosilicon compounds and a flow rate of one or more porogen compounds to deposit an initiation layer on the substrate by applying a radio frequency (RF) power level of about 300 W to about 600 W at 13.56 MHz to the processing chamber, ramping-up the flow rate of the one or more organosilicon compounds until reaching a final flow rate of the one or more organosilicon compounds to deposit a first transition layer on the initiation layer, and while flowing the final flow rate of the one or more organosilicon compounds, ramping-up the flow rate of the one or more porogen compounds until reaching a final flow rate of the one or more porogen compounds to deposit a second transition layer on the first transition layer, depositing a low K film over the adhesion layer, and curing the deposited low K film to form nanopores therein. Various processing parameters and precursors are further discussed in the detailed description.

[0017] The present invention provides a method of depositing a low dielectric constant film. The low dielectric constant film comprises silicon, oxygen, hydrogen and carbon. Embodiments of the invention have been proved to be able to significantly lower the K impact (dielectric constant) of an adhesion layer to an ultra low dielectric constant film stack by reducing the thickness of the adhesion layer. By lowering the adhesion layer thickness to about or less than 200 Å, the thickness non-uniformity for an ultra low dielectric film stack (<2 kÅ) is also reduced to less than 2%. As will be discussed below, the improved oxide adhesion layer is deposited at lower deposition rate and lower plasma density in combination with higher total flow rate, resulting in better packing/ordering of the co-deposited species during film deposition which causes higher mechanical strength and lower porosity. The improved adhesion layer provides high adhesion energy for better adhesion with ultra low dielectric constant films to underlying barrier/liner layers. The resulting low dielectric film has nanometer-sized pores and tighter pore-size distribution. The low dielectric constant film has a dielectric constant of about 3.0 or less, preferably about 2.5 or less. The low dielectric constant film may have an elastic modulus of at least about 6.5 GPa or above.

[0018] FIG. 1A schematically illustrates a cross-sectional view of a dielectric film stack 100 formed according to embodiments of the present invention. While not shown here, it is contemplated that the dielectric film stack 100 of the present invention can be used as an inter-metal dielectric layer in a dual damascene structure, which may generally include one or more nanoporous inter-metal dielectric layers (not shown) and one or more etch stop layers (not shown) of silicon oxide, silicon nitride, silicon oxynitride, or amorphous hydrogenated silicon carbide that are deposited in an alternating or desired order. An anti-reflective coating (not shown) and a trench photomask (not shown) comprising a photoresist layer are then respectfully deposited over the deposited film layers and patterned by conventional photolithography techniques in a manner to develop a metallization structure to be filled with a desired metal such as copper. The dual damascene formation process may be repeated to deposit a desired number of interconnection levels. An exemplary dual damascene structure that may be benefited from the present invention is further described in the commonly assigned U.S. Pat. No. 7,547,643 issued on Jun. 16, 2009 to Francimar Schmitt et al., which is incorporated by reference in its entirety.

[0019] Generally, the dielectric film stack 100 as shown in FIG. 1A comprises a substrate 102 bearing a liner/barrier layer 104, which acts as an isolation layer between a subsequent adhesion layer 106 and the underlying substrate surface 103 and metal lines 108 formed on the substrate surface 103. A low K layer 110 is deposited over the adhesion layer 106, which is capped by a capping layer 112. Methods of depositing such a dielectric film stack 100 according to various embodiments of the invention will be described briefly with respect to FIG. 2 in conjunction with FIG. 1B.

[0020] FIG. 2 is a process flow diagram 200 illustrating a method of depositing a dielectric film stack 100 according to one embodiment of the invention. Generally, a typical porous dielectric film requires simultaneous deposition of one or more organosilicon compounds, which becomes Si backbone, and one or more unsaturated non-silicon compounds having thermally labile groups, which acts as a sacrificial porogen. In step 202, a substrate 102 bearing a liner/barrier layer 104 is positioned on a substrate support in a processing chamber capable of performing Plasma-Enhanced Chemical Vapor Deposition (PECVD) process. The liner/barrier layer 104 may be deposited by a PECVD process from a plasma comprising a organosilane compound, ammonia, oxygen and inerts. The deposition process can include a capacitively coupled plasma or both an inductively and a capacitively coupled plasma in the processing chamber according to methods known in the art. The plasma can be generated using inert gases, such as He, Ar, and N₂. An inert gas such as helium is commonly used in the PECVD deposition to assist in plasma generation.

[0021] In step 204, a gas mixture having a composition including one or more organosilicon compounds, one or more porogen compounds, and one or more oxidizing gases is introduced into the processing chamber through a gas distribution plate, such as a showerhead. An initial gas composition of oxygen and/or helium may be introduced into the processing chamber before initiation of the RF power to stabilize the conditions for the subsequent depositions.

[0022] In one embodiment, the one or more organosilicon compounds are introduced into the chamber at a flow rate between about 200 milligrams/minute to about 5000 milligrams/minute, for example, between about 350 milligrams/minute and about 2500 milligrams/minute; the one or more oxidizing gases are introduced into the chamber at a flow rate between about 100 sccm and about 1000 sccm, for example, between about 125 sccm and about 550 sccm; and the one or more porogen compounds are introduced into the chamber at a flow rate between about 50 milligrams/minute to about 5000 milligrams/minute, for example, between about 150 grams/minute and about 1500 grams/minute. A radio-frequency (RF) power is applied to an electrode, such as the showerhead, in order to provide plasma processing conditions in the chamber. Suitable RF power may be a power in a range of about 10 W to about 2000 W, such as about 300 W to about 600 W at a frequency of about 13.56 MHz. The gas mixture is reacted in the chamber in the presence of RF power to deposit an initiation layer 106a comprising an oxide layer that adheres strongly to the underlying liner/barrier layer 104.

[0023] The gas mixture may optionally include one or more carrier gases. Typically, one or more carrier gases are introduced with the one or more organosilicon compounds and the one or more porogen compounds into the processing chamber. Examples of carrier gases that may be used include helium, argon, carbon dioxide, and combinations thereof. In one embodiment where helium is used as the carrier gas, the helium gas are introduced into the chamber along with one or more organosilicon compound at a flow rate between about 1500 sccm and about 8000 sccm, for example, between about 3500 sccm and about 5500 sccm. With one or more porogen compounds, helium gas are introduced into the chamber at a flow rate between about 300 sccm and about 1800 sccm, for example, between about 700 sccm and about 1250 sccm.

[0024] The initiation layer 106a generally includes a silicon oxide layer. As will be discussed below, the initiation layer 106a and a first and second transition layers 106b, 106c (FIG. 1B) constitute the adhesion layer 106 that enhances adhesion between the underlying liner/barrier layer 104 and the subsequent low K layer 110. In one embodiment, the initiation layer deposition may have a time range of between about 0.5 seconds and about 10 seconds, as long as the deposition period is long enough to ensure cohesion of the entire film. In one example, the initiation layer deposition may last for about 1 second. The initiation layer 106a may be deposited to a thickness in a range of about 5 Å to about 100 Å, preferably about 10 Å to about 50 Å. It is contemplated that the times for the various periods described in this disclosure may be adjusted depending on the needs of particular embodiments. For example, while a time range of about 0.5 seconds to about 10 seconds is described, in some embodiments, the initiation period may last for 0 seconds. An initiation period of 0 seconds means that changing flow rates of gas streams begins immediately upon introducing them to the chamber. Thus, embodiments with no initiation period are contemplated.

[0025] Prior to deposition of the low K layer 110, a separate transition step is performed to prevent any unwanted particle clusters from forming in the films due to undesirable gas phase reaction of both silicon and porogen flows occurring at the gas distribution plate. It has also been observed that the smooth transition of the liquids into the chamber can significantly reduce the occurrence of carbon bumps. These issues may be addressed by separating the transition of two liquid precursors (i.e., organosilicon compounds and porogen compounds) at a desired ramping rate. In a first period of the separate transition step 206, or simply referring to step 206, the flow rate of the one or more organosilicon compounds is gradually increased at a ramp-up rate between about 100 mgm/sec. and about 5000 mgm/sec., for example, between about 800 mgm/sec. and about 1200 mgm/sec., such as about 1000 mgm/sec., in the presence of the RF power, to deposit a first transition layer 106b (see FIG. 1B, which is a close up of the cross-section of the film stack shown in FIG. 1A) on the initiation layer 106a until reaching a predetermined organosilicon compound gas mixture. In embodiments where a helium carrier gas is used, the flow rate of the one or more organosilicon compounds and the helium gas may be decreased to a range between about 2500 sccm and about 4000 sccm. In one embodiment, the first transition layer deposition may have a time range of between about 0.5 second and about 10 seconds. In one example, the first transition layer deposition time may be about 1 second. The first transition layer 106b may be deposited to a thickness in a range of about 10 Å to about 300 Å, for example, about 50 Å to about 200 Å.

[0026] In a second period of the transition step 208, or simply referring to step 208, while keeping the predetermined organosilicon compound gas mixture constant, the flow rate of the one or more porogen compounds is gradually increased at a ramp-up rate between about 100 mgm/sec. and about 5000 mgm/sec., for example, between about 200 mgm/sec. and about 350 mgm/sec., such as about 300 mgm/sec., to deposit a second transition layer 106c (FIG. 1B) onto the first transition layer 106b until reaching a predetermined final gas mixture. In embodiments where a helium carrier gas is used, the flow rate of the one or more porogen compounds and helium gas may be increased to a range between about 800 sccm and about 2000 sccm. In one embodiment, the second transition layer deposition may have a time range of between about 1 second and about 180 seconds. In one example, the second transition layer deposition time may be about 3 seconds. The second transition layer 106c may be deposited to a thickness in a range of about 10 Å to about 600 Å, preferably, about 100 Å to about 400 Å.

[0027] The deposition periods of the initiation layer 106a and the first and second transition layers 106b, 106c preferably result in deposition of a thin portion 106 of the film stack (106a, 106b, 106c) as shown. This thin portion 106 of the film stack serves as an adhesion layer for better adhesion with ultra low dielectric constant films to underlying barrier/liner layers. In most embodiments, the thickness of this portion is reduced by almost half, e.g., less than about 200 Angstroms. Deposition of the thin portion 106 of the film stack (106a, 106b, 106c) may be achieved through relatively short duration and/or low deposition rate. In one embodiment, the deposition rate for the thin portion of the film stack is between about 1000 Angstroms/minute to about 3500 Angstroms/minute, such as about 2500 Angstroms/minute. As the thin portion 106 of the film stack (106a, 106b, 106c) constitutes significant portion of an ultra low K nanoporous film stack (106a, 106b, 106c, 110) having a thickness less than 2000 Angstroms, the thickness non-uniformity for the dielectric film stack 100 can be reduced to less than 2% by lowering the thickness of the thin portion of the film stack. Most importantly, the reduced thickness of the thin portion 106 of the film stack (106a, 106b, 106c) minimizes the K impact to the overall nanoporous film stack.

[0028] In step 210, upon reaching the final gas mixture composition, a plasma of the final gas mixture comprising a flow rate of one or more organosilicon compounds and a flow rate of the one or more porogen compounds is formed to deposit a porogen-containing organosilicate dielectric layer, i.e., the low K layer 110. In one embodiment, the low K layer deposition may have a time range of between about 15 second and about 180 seconds. In one example, the final layer deposition time may be about 130 seconds. The low K layer 110 may be deposited to a thickness in a range of about 200 Å to about 10,000 Å until the RF power is terminated. Not wishing to be bound by theory, it is believed that by separating the ramp-up rates of the organosilicon compounds and the porogen compounds, a more stable and manufacturable process can be obtained, yielding organosilicate dielectric layers with significantly less defect issues, such as carbon bumps.

[0029] Alternatively, step 208, depositing the second transition layer 106c, may be combined with step 210, depositing the final porogen silicon oxide layer. In such an embodiment, the porogen compound flow rate is continuously ramped-up while flowing the predetermined organosilicon compound gas mixture during the porogen silicon oxide layer deposition. The combination of step 208 with step 210 may have a time range of between about 1 second and about 180 seconds. In this manner, the final porogen silicon oxide layer may have a gradient concentration of porogen where the concentration of porogen in the silicon oxide layer increases as the porogen silicon oxide layer is deposited. This gradient layer may be deposited to a thickness in a range of about 50 Å to about 10,000 Å, preferably, about 100 Å to about 5000 Å, until the RF power is terminated.

[0030] During the processes described above, the substrate is typically maintained at a temperature between about 100° C. and about 400° C., for example between about 200° C. and about 350° C. The chamber pressure may be between about 1 Torr and about 20 Torr, for example between about 7 Torr and about 9 Torr, and the spacing between a substrate support and the chamber showerhead may be between about 200 mils and about 1500 mils, for example, between about 280 mils and about 450 mils. A RF power level of between about 100 W and about 600 W for a 300 mm substrate may be used. The RF power is provided at a frequency between about 0.01 MHz and 300 MHz, such as about 13.56 MHz. The RF power may be provided at a mixed frequency, such as at a high frequency of about 13.56 MHz and a low frequency of about 350 kHz. The RF power may be cycled or pulsed to reduce heating of the substrate and promote greater porosity in the deposited film. The RF power may also be continuous or discontinuous, depending upon application.

[0031] In certain embodiments, lower plasma density in combination with higher total flow rate is used. To obtain a lower plasma density, a RF power level of between about 300 W and about 600 W, for example, between about 350 W and about 500 W, may be used. In cases where the RF power level of between about 350 W and about 500 W is used, a RF power/total flow rate of about 0.1 W/sccm to about 0.3 W/sccm is preferred. Alternatively, a RF power/total volume flow of about 0.2 W/cm³ to about 0.5 W/cm³ is preferred. The term “total flow rate” or “total volume flow” as used herein is intended to refer to the flows/volumes of the gas mixture and optional carrier gases introduced into the processing chamber during the deposition, as discussed previously. It has been observed by the present inventors that the use of lower plasma density in combination with higher total flow rate may allow for a denser packing of the co-deposited species during film deposition, resulting in higher mechanical strength, smaller pore size (<10 Å), and tighter pore-size distribution. This leads to significant improvement in the mechanical integrity of the film by increasing the damage resistance of the film to subsequent device manufacturing processes.

[0032] In any of the embodiments described herein, a porogen-containing organosilicate dielectric layer is deposited from a process gas mixture comprising an organosilicon compound and a porogen. The organosilicate layer may be used as a dielectric layer. The dielectric layer may be used at different levels within a dual damascene structure or a suitable device. For example, the dielectric layer may be used as a premetal dielectric layer, an inter-metal dielectric layer, or a gate dielectric layer. The organosilicate layer deposited in accordance with various embodiments of the present invention has been proved to be able to provide a low dielectric constant less than 3.0, for example, about 2.5.

[0033] A wide variety of process gas mixtures may be used to deposit the organosilicate dielectric layer, and non-limiting examples of such gas mixtures are provided below. Generally, the gas mixture includes one or more organosilicon compounds (e.g., a first and a second organosilicon compound), one or more porogen compounds, a carrier gas, and an oxidizing gas. These components are not to be interpreted as limiting, as many other gas mixtures including additional components such as hydrocarbons (e.g., aliphatic hydrocarbons) are contemplated.

[0034] The term “organosilicon compound” as used herein is intended to refer to silicon-containing compounds including carbon atoms in organic groups. The organosilicon compound may include one or more cyclic organosilicon compounds, one or more aliphatic organosilicon compounds, or a combination thereof. Some exemplary organosilicon compounds include methyldiethoxysilane (mDEOS), tetramethylcyclotetrasiloxane (TMCTS), octamethylcyclotetrasiloxane (OMCTS), trimethylsilane (TMS), pentamethylcyclopentasiloxane, hexamethylcyclotrisiloxane, dimethyldisiloxane, tetrasilano-2,6-dioxy-4,8-dimethylene, tetramethyldisiloxane, hexamethyldisiloxane (HMDS), 1,3-bis(silanomethylene)disiloxane, bis(1-methyldisiloxanyl)methane, bis(1-methyldisiloxanyl)propane, hexamethoxydisiloxane (HMDOS), dimethyldimethoxysilane (DMDMOS), and dimethoxymethylvinylsilane (DMMVS), or derivatives thereof. The one or more organosilicon compounds may be introduced into the processing chamber at a flow rate in a range of about 200 milligrams/minute to about 5000 milligrams/minute, for example, between about 350 milligrams/minute and about 2500 milligrams/minute.

[0035] The term “porogen compound” as used herein is intended to refer to compounds that comprise thermally labile groups. The thermally labile groups may be cyclic groups, such as unsaturated cyclic organic groups. The term “cyclic group” as used herein is intended to refer to a ring structure. The ring structure may contain as few as three atoms. The atoms may include carbon, nitrogen, oxygen, fluorine, and combinations thereof, for example. The cyclic group may include one or more single bonds, double bonds, triple bonds, and any combination thereof. For example, a cyclic group may include one or more aromatics, aryls, phenyls, cyclohexanes, cyclohexadienes, cycloheptadienes, and combinations thereof. The cyclic group may also be bi-cyclic or tri-cyclic. In one embodiment, the cyclic group is bonded to a linear or branched functional group. The linear or branched functional group preferably contains an alkyl or vinyl alkyl group and has between one and twenty carbon atoms. The linear or branched functional group may also include oxygen atoms, such as in a ketone, ether, and ester. The porogen may comprise a cyclic hydrocarbon compound. Some exemplary porogens that may be used include norbornadiene (BCHD, bicycle(2.2.1)hepta-2,5-diene), 1-methyl-4-(1-methylethyl)-1,3-cyclohexadiene (ATP or alpha-Terpinene), vinylcyclohexane (VCH), phenylacetate, butadiene, isoprene, cyclohexadiene, bicycloheptadiene, 1-methyl-4-(1-methylethyl)-benzene (Cymene), 3-carene, fenchone, limonene, cyclopentene oxide, vinyl-1,4-dioxinyl ether, vinyl furyl ether, vinyl-1,4-dioxin, vinyl furan, methyl furoate, furyl formate, furyl acetate, furaldehyde, difuryl ketone, difuryl ether, difurfuryl ether, furan, and 1,4-dioxin, and fluorinated carbon derivatives thereof. The one or more porogen compounds may be introduced into the processing chamber at a flow rate in a range of about 50 milligrams/minute to about 5000 milligrams/minute, for example, between about 150 milligrams/minute and about 1500 milligrams/minute.

[0036] As discussed previously, the gas mixture may optionally include one or more carrier gases. Typically, one or more carrier gases are introduced with the one or more organosilicon compounds and the one or more porogen compounds into the processing chamber. Examples of carrier gases that may be used include helium, argon, carbon dioxide, and combinations thereof. The one or more carrier gases may be introduced into the processing chamber at a flow rate less than about 20,000 sccm, depending in part upon the size of the interior of the chamber. Preferably the flow of carrier gas is in a range of about 500 sccm to about 5000 sccm. In some processes, an inert gas such as helium or argon is put into the processing chamber to stabilize the pressure in the chamber before reactive process gases are introduced.

[0037] The gas mixture also includes one or more oxidizing gases. Suitable oxidizing gases include oxygen (O₂), ozone (O₃), nitrous oxide (N₂O), carbon monoxide (CO), carbon dioxide (CO₂), and combinations thereof. The flow of oxidizing gas may be in a range of about 100 sccm to about 3,000 sccm, depending in part upon the size of the interior of the chamber. Typically, the flow of oxidizing gas is in a range of about 100 sccm to about 1,000 sccm, for example about 450 sccm. Disassociation of oxygen or the oxygen containing compounds may occur in a microwave chamber prior to entering the deposition chamber and/or by RF power as applied to process gas within the chamber.

[0038] After the low dielectric constant film is deposited, the film is post-treated. The film may be post-treated with a thermal annealing, alone or assisted by UV radiation to remove the organic labile and create pore inclusions into the final material. In one embodiment, the low dielectric constant film is post-treated with a UV curing process. The UV post-treatment may be performed in-situ within the same processing chamber or system, for example, transferred from one chamber to another without a break in vacuum. Exemplary UV post-treatment conditions that may be used include a chamber pressure of between about 1 Torr and about 10 Torr and a substrate support temperature of between about 350° C. and about 500° C. The source of ultraviolet radiation may be between about 100 mils and about 1400 mils from the substrate surface. Optionally, a processing gas may be introduced during the ultraviolet curing process. Suitable processing gases include oxygen (O₂), nitrogen (N₂), hydrogen (H₂), helium (He), argon (Ar), water vapor (H₂O), carbon monoxide, carbon dioxide, hydrocarbon gases, fluorocarbon gases, and fluorinated hydrocarbon gases, or combinations thereof.

[0039] The UV radiation may be provided by any UV source, such as mercury microwave arc lamps, pulsed xenon flash lamps, or high-efficiency UV light emitting diode arrays. The ultraviolet radiation may comprise a range of ultraviolet wavelengths, and include one or more simultaneous wavelengths. Suitable ultraviolet wavelengths include between about 1 nm and about 400 nm, and may further include optical wavelengths up to about 600 or 780 nm. Additionally or alternatively, the ultraviolet radiation may be applied at multiple wavelengths, a tunable wavelength emission and tunable power emission, or a modulation between a plurality of wavelengths as desired, and may be emitted from a single UV lamp or applied from an array of ultraviolet lamps. Examples of suitable UV lamps include a Xe filled Zeridex™ UV lamp, a Ushio Excimer UV lamp, a DSS UV lamp, or a Hg Arc Lamp. The deposited low dielectric constant film is exposed to the ultraviolet radiation for between about 10 seconds and about 600 seconds, for example between about 60 seconds and about 600 seconds. The UV radiation may have a wavelength of between about 170 nm and about 400 nm, for example. Further details of UV chambers and treatment conditions that may be used are described in commonly assigned U.S. patent application Ser. No. 11/124,908, filed on May 9, 2005, which is incorporated by reference herein. The NanoCure™ chamber from Applied Materials, Inc. is an example of a commercially available chamber that may be used for UV post-treatments.

[0040] In another embodiment, the low dielectric constant film is post-treated with a thermal or plasma enhanced annealing process. The film may be annealed at a temperature between about 200° C. and about 400° C. for about 2 seconds to about 1 hour, preferably about 30 minutes, in a chamber. A non-reactive gas such as helium, hydrogen, nitrogen, or a mixture thereof may be introduced at a rate of about 100 sccm to about 10,000 sccm. The chamber pressure is maintained between about 1 Torr and about 10 Torr. The RF power during the annealing is about 200 W to about 1,000 W at a frequency of about 13.56 MHz, and the preferable substrate spacing is between about 300 mils and about 800 mils. Annealing the low dielectric constant film at a substrate temperature of about 200° C. to about 400° C. after the low dielectric constant film is deposited volatilizes at least some of the organic groups in the film, forming voids in the film.

[0041] In yet another embodiment, the low dielectric constant film is post-treated with an electron beam treatment. Exemplary electron beam conditions that may be used include a chamber temperature of between about 200° C. and about 600° C., e.g. about 350° C. to about 400° C. The electron beam energy may be from about 0.5 keV to about 30 keV. The exposure dose may be between about 1 μC/cm² and about 400 μC/cm². The chamber pressure may be between about 1 mTorr and about 100 mTorr. The gas ambient in the chamber may be any of the following gases: nitrogen, oxygen, hydrogen, argon, a blend of hydrogen and nitrogen, ammonia, xenon, or any combination of these gases. The electron beam current may be between about 0.15 mA and about 50 mA. The electron beam treatment may be performed for between about 1 minute and about 15 minutes. Although any electron beam device may be used, an exemplary electron beam chamber that may be used is an EBk™ electron beam chamber available from Applied Materials, Inc. of Santa Clara, Calif.

[0042] The e-beam curing process improves mechanical strength of the deposited film network and also lowers the k-value. The energized e-beam alters the chemical bonding in the molecular network of the deposited film and removes at least a portion of the molecular groups, such as organic components from the ring of the one or more oxygen-free hydrocarbon compounds comprising one ring and one or two carbon-carbon double bonds in the ring, from the film. The removal of the molecular groups creates voids or pores within the film, lowering the K value.

[0043] FIG. 3 shows a cross-sectional, schematic diagram of a chemical vapor deposition (CVD) chamber 300 for depositing a carbon-doped silicon oxide layer. This figure is based upon features of the PRODUCER® chambers currently manufactured by Applied Materials, Inc. The PRODUCER® CVD chamber (200 mm or 300 mm) has two isolated processing regions that may be used to deposit carbon-doped silicon oxides and other materials. A chamber having two isolated processing regions is described in U.S. Pat. No. 5,855,681, which is incorporated by reference herein.

[0044] The deposition chamber 300 has a chamber body 302 that defines separate processing regions 318, 320. Each processing region 318, 320 has a pedestal 328 for supporting a substrate (not seen) within the chamber 300. The pedestal 328 typically includes a heating element (not shown). Preferably, the pedestal 328 is movably disposed in each processing region 318, 320 by a stem 326 which extends through the bottom of the chamber body 302 where it is connected to a drive system 303. Internally movable lift pins (not shown) are preferably provided in the pedestal 328 to engage a lower surface of the substrate. The lift pins are engaged by a lift mechanism (not shown) to receive a substrate before processing, or to lift the substrate after deposition for transfer to the next station.

[0045] Each of the processing regions 318, 320 also preferably includes a gas distribution assembly 308 disposed through a chamber lid 304 to deliver gases into the processing regions 318, 320. The gas distribution assembly 308 of each processing region normally includes a gas inlet passage 340 through manifold 348 which delivers gas from a gas distribution manifold 319 through a blocker plate 346 and then through a showerhead 342. The showerhead 342 includes a plurality of nozzles (not shown) through which gaseous mixtures are injected during processing. An RF (radio frequency) supply 325 provides a bias potential to the showerhead 342 to facilitate generation of a plasma between the showerhead and the pedestal 328. The deposition process performed in the deposition chamber 300 can be either a non-plasma process on a cooled substrate pedestal 328 or a plasma enhanced process. In a plasma process, a controlled plasma is typically formed adjacent to the substrate by RF energy applied to the showerhead 342 from RF power supply 325 (with pedestal 328 grounded). Alternatively, the RF power supply 325 can be provided to the pedestal 328, or to different components at different frequencies.

[0046] The plasma may be generated using high frequency RF (HFRF) power, as well as low frequency RF (LFRF) power (e.g., dual frequency RF), constant RF, pulsed RF, or any other known or yet to be discovered plasma generation technique. The RF power supply 325 can supply a single frequency RF between about 5 MHz and about 300 MHz. In addition, the RF power supply 325 may also supply a low frequency RF between about 300 Hz to about 1,000 kHz to supply a mixed frequency to enhance the decomposition of reactive species of the process gas introduced into the process chamber. The RF power may be cycled or pulsed to reduce heating of the substrate and promote greater porosity in the deposited film. Suitable RF power may be a power in a range of about 10 W to about 5000 W, for example in a range of about 200 W to about 600 W. Suitable LFRF power may be a power in a range of about 0 W to about 5000 W, for example in a range of about 0 W to about 200 W.

[0047] A system controller 334 controls the functions of various components such as the RF power supply 325, the drive system 303, the lift mechanism, the gas distribution manifold 319, and other associated chamber and/or processing functions. The system controller 334 executes system control software stored in a memory 338, which in the preferred embodiment is a hard disk drive, and can include analog and digital input/output boards, interface boards, and stepper motor controller boards. Optical and/or magnetic sensors are generally used to move and determine the position of movable mechanical assemblies.

[0048] The above CVD system description is mainly for illustrative purposes, and other plasma processing chambers may also be employed for practicing embodiments of the invention.

[0049] The embodiments of the present invention demonstrate deposition of ultra low K nanoporous films having dispersed microscopic gas voids. In any of the embodiments described previously, the following process parameters and ranges are generally beneficial to main and/or adhesion layer deposition process:

	Parameters	Range
	Heater temperature (° C.)	200-350
	Deposition time (s)	15-360
	Pressure (Torr)	7-9
	Spacing (mils)	280-450
	HF RF (Watt)	300-600
	mDEOS (mgm)	200-2500
	BCHD (mgm)	200-1600
	O2 (sccm)	125-500
	mDEOS He carrier gas flow rate (sccm)	500-5000
	mBCHD He carrier gas flow rate (sccm)	500-1250

[0050] During the deposition of an adhesion layer, ramp-up rates for one or more organosilicon compounds and one or more porogen compounds in various transitions are generally between 800 mgm/sec and 1200 mgm/sec and between about 200 mgm/sec. and about 350 mgm/sec., respectively. Preferably, only one of the flows of the organosilicon compounds and the porogen compounds is changing during various transition steps to prevent any defects in the films, as discussed previously.

[0051] In one specific embodiment described above with respect to FIGS. 1A and 1B, porogen-containing organosilicate dielectric layers were deposited on a substrate. The film was deposited using a PECVD chamber (i.e., CVD chamber) on a PRODUCER® system, available from Applied Materials, Inc. of Santa Clara, Calif. During deposition the chamber pressure was maintained at a pressure of about 6.5 Torr and the substrate was maintained at a temperature of about 270° C. The substrate was positioned on a substrate support disposed within a process chamber. The substrate was positioned 450 mils from the chamber showerhead.

[0052] The process gas mixture having an initial gas composition of 300 sccm oxygen and 3800 sccm helium was introduced into the chamber and flow rates stabilized before initiation of the RF power. Subsequently, a RF power level of about 600 W at 13.56 MHz was applied to the showerhead to form a plasma of a gas mixture including a methyldiethoxysilane (mDEOS) introduced into the chamber at a flow rate of about 600 mgm to deposit a silicon oxide initiation layer. After initiation of the RF power for about 1 second, the flow rate of mDEOS was increased to 2200 mgm. at a ramp-up rate of about 1000 mgm/sec. for about 1 second. In addition, the flow of helium was decreased to about 3000 sccm.

[0053] Upon reaching and keeping a final mDEOS flow rate of about 2200 mg/min, a flow of BCHD was introduced into the chamber at a ramp-up rate of about 400 mgm/sec. for about 3 seconds to reach a porogen deposition flow rate of about 1300 mgm. The final gas mixture composition also includes 3000 sccm helium and 225 sccm oxygen. Upon reaching the desired thickness of the porogen-containing organosilicate dielectric layer, the RF power is terminated to stop further deposition. After RF power termination, the chamber throttle valve is opened to allow the process gas mixture to be pumped out of the chamber. The separate transition of the liquid precursors into the chamber reduces the defects in the films. A secondary-ion mass spectrometry (SIMS) analysis was performed to analyze the depth profile of element concentrations in a dielectric film stack, as shown in FIG. 4. The depth distribution of carbon shows a smooth phase shift of carbon in the films, suggesting that no carbon bumps occurred in the films using the exemplary process.

[0054] It is contemplated that many variations of the above example may be practiced. For example, other organosilane precursors, porogen precursors, oxidizing gases, and inert gases may be used. In addition, different flow rates and/or ramp rates may be employed. Depending upon application, the flow rates of the various precursors may be adjusted to change the carbon content as portions of the film (e.g., initiation layer 106a and/or transition layers 106b, 106c) are deposited, such that an initial portion of the deposited film has a low carbon content, and is therefore oxide-like, while successive portions have higher carbon content, becoming oxycarbide-like.

[0055] Embodiments of the invention have been proved to be able to significantly lower the K impact of an oxide adhesion layer to an ultra low dielectric film stack by reducing the thickness of the adhesion layer that is deposited with novel process parameters. By lowering the adhesion layer thickness to about or less than 200 Å, the thickness non-uniformity for an ultra low dielectric film stack (<2 kÅ) is also reduced to less than 2%. The improved oxide adhesion layer is deposited at lower deposition rate of about 2400 Å/min. and lower plasma density in combination with higher total flow rate (RF power/total flow between about 0.1 W/sccm and about 0.3 W/sccm), resulting in better packing/ordering of the co-deposited species during film deposition which causes higher mechanical strength of about 6.9 GPa. The improved adhesion layer provides good enough adhesion energy (˜4.5 J/m²) for better adhesion with ultra low dielectric constant films to underlying barrier/liner layers. The resulting ultra low K nanoporous films has smaller pore radius of between about 7 Å and about 10 Å and tighter pore-size distribution with porosity in the range about 15% and about 25%.

[0056] While the foregoing is directed to embodiments of the present invention, other and further embodiments of the invention may be devised without departing from the basic scope thereof, and the scope thereof is determined by the claims that follow.