Patent Application
20070048456
Plasma enhanced chemical vapor deposition (“PECVD”) systems may be used, for example, in semiconductor manufacturing processes to deposit thin films of silicon onto a substrate. Conventional PECVD systems include a deposition chamber with two or three electrodes that, when excited by a voltage, ionize the reactant gas between the electrodes to create a plasma. In many instances, the reactant gas is supplied directly into the high intensity plasma region through one of the electrodes, which is commonly referred to as a “showerhead” electrode.
Although PECVD has proven to be a useful process, the present inventors have determined that conventional PECVD processes are susceptible to improvement. More specifically, the present inventors have determined that the deposition rates of conventional PECVD processes must be kept relatively low in order to produce acceptable film quality and, given the fact that the cost of conventional PECVD systems is comparable to the cost of deposition systems such as sputtering systems that have higher deposition rates, the per unit area cost of films produced by conventional PECVD processes is relatively high. The present inventors have also determined that conventional PECVD processes consume reactant materials (e.g. silane) inefficiently because the concentration of the silane in the reactant gas (e.g. silane and hydrogen) is only marginally higher than the concentration of the reactant material in the exhaust. As such, the vast majority of the silane flows through the system and is not utilized by the deposition process, thereby being wasted. The present inventors have also determined that silicon particles can form in the plasma unless conventional PECVD processes are operated at low gas pressures, low reactant material concentrations, and low excitation power, all of which result in low deposition rates. The formation of silicon particles is problematic, thereby necessitating the low deposition rates, because the particles can damage the vacuum pumps that draw exhaust gasses from the deposition chamber and can also damage the devices being formed. The vacuum pumps must also be relatively large so that the slightly used gas within the deposition chamber can be rapidly withdrawn before the silane concentration becomes too low or the silane distribution becomes non-uniform. The present inventors have also determined that conventional PECVD processes require the reactant gas to flow through the length (or width) of the entire chamber before the reactant gas is exhausted. This results in a long dwell time for silane molecules within the chamber that exacerbates the formation of silicon particles and also increases the formation of higher order silanes (e.g. Si2H6). A significant concentration of higher order silanes results in very poor device quality. Therefore, the flow rates are kept high to exhaust the higher order silanes quickly and avoid their accumulation. As a result, most of the silane flows through the system and is exhausted rather than used efficiently in the reaction to deposit silicon. The present invention fulfills these needs and provides other related advantages.
In accordance with one embodiment of a present invention, a substrate processing system is disclosed, comprising a deposition chamber; and a plurality of tubular electrodes positioned within the deposition chamber, defining plasma regions adjacent thereto, and having internal lumens and apertures that connect the internal lumens to the deposition chamber.
In accordance with another embodiment of the present invention, a substrate processing system is disclosed, comprising: a deposition chamber; first and second substrate carriers located within the deposition chamber; a plurality of spaced elongate electrodes positioned between the first and second substrate carriers; and a power supply operably connected to each of the electrodes and adapted to drive adjacent electrodes out of phase from one another.
In accordance with another embodiment of the present invention, a method of forming a film is disclosed, comprising the steps of: generating a plasma region having a relatively high intensity and a plasma region having a relatively low intensity; and introducing a reactant including film layer material into the relatively low intensity plasma region.
In accordance with another embodiment of the present invention, a substrate processing system is disclosed, comprising: means for generating a plasma region having a relatively high intensity and a plasma region having a relatively low intensity; and means for introducing a gas including film layer material into the relatively low intensity plasma region.
In accordance with another embodiment of the present invention, a substrate processing system is disclosed, comprising: a deposition chamber; at least one substrate carrier located within the deposition chamber and adapted to guide a substrate in a substrate travel direction; and a plurality of elongate rod electrodes spaced from one another in the substrate travel direction and defining respective longitudinal axes that extend in a direction that is at least transverse to the substrate travel direction.
In accordance with another embodiment of the present invention, a substrate processing system is disclosed, comprising: a deposition chamber defining an interior having a length and a height; first and second substrate carriers located within the deposition chamber adapted to position first and second substrates apart from one another by a distance that is no more than one-tenth of the height and no more than one-fifteenth of the length measured in a direction that is perpendicular to the length and the height; and an electrode assembly located between the first and second substrate carriers and adapted to create plasma between the first and second substrate carriers.
In accordance with another embodiment of the present invention, a method of forming a film on a substrate is disclosed, comprising the steps of: generating a plasma within a deposition chamber; introducing a reactant including film layer material into the plasma at a reactant input rate; depositing the film layer material onto the substrate; evacuating exhaust from the from the deposition chamber; measuring the amount of film layer material in the exhaust; and adjusting the reactant input rate in response to the measured amount of film layer material in the exhaust.
Other features and advantages of the present invention will become apparent from the following more detailed description, taken in conjunction with the accompanying drawings which illustrate, by way of example, the principles of the invention.
Detailed description of embodiments of the inventions will be made with reference to the accompanying drawings.
The following is a detailed description of the best presently known modes of carrying out the inventions. This description is not to be taken in a limiting sense, but is made merely for the purpose of illustrating the general principles of the inventions. It should also be noted that detailed discussions of the various aspects of PECVD systems that are not pertinent to the present inventions have been omitted for the sake of simplicity. Additionally, although the inventions are described in the context of the formation of thin films of silicon (Si) from silane (SiH4), they are not limited to any particular types of films or input reactant material. By way of example, but not limitation, the inventions also have application in the deposition of silicon carbide (SiC), amorphous silicon Si(H), micro-crystalline silicon Si(H), silicon germanium (SiGe) and other semiconductor materials, all with hydrogen (H) incorporated. Doped semiconductor materials can also be fabricated. The dopant is most easily input to the system as a gas, but could also be introduced by including a solid piece of doped silicon in the plasma region. Gas sources for doping materials include, for example, tri-methyl borane (B(CH3)3) and phosphine (PH3).
As illustrated for example in
Turning to
The interior of the deposition chamber 102 in the exemplary embodiment is relatively narrow. More specifically, the distance between the substrates 120a and 120b is substantially less than the length of the chamber (measured in the direction of arrows A) and the height of the chamber (measured in the direction perpendicular to arrows A). For example, the distance between substrates 120a and 120b may be one-tenth or less of the length and height dimensions. The substrates 120a and 120b will also preferably extend from end to end in the length dimension of the deposition chamber 102 and from top to bottom in the height dimension. As a result, the substrates 120a and 120b will be between the electrode assembly 104 (and the plasma created thereby) and the large interior surfaces of the chamber and will cover the vast majority of the interior surface of the deposition chamber 102.
The deposition chamber 102 is not limited to any particular size. Nevertheless, in one exemplary implementation of the deposition chamber 102 that is suitable for commercial applications and is oriented in the manner illustrated in
There are a number of advantages associated with deposition chambers that are configured in this manner. For example, the relatively small spacing between the substrates 120a and 120b, as compared to the relatively large dimension in the direction of substrate travel and the dimension perpendicular to substrate travel increases the percentage of the plasma generated silicon that is deposited onto the substrates and decreases the amount that is deposited onto the chamber walls, as compared to conventional deposition chambers. As a result, the reactant materials are consumed more efficiently. The downtime and expense associated with deposition chamber cleaning and maintenance is also reduced. The close spacing between the electrode assembly 104 and the substrates 120a and 120b also facilitates rapid diffusion in the smallest dimension to be the dominant process for transporting atomic hydrogen created at the center of the deposition chamber 102 to the substrates, where the atomic hydrogen can react with silane to create the pre-cursors that result in the deposition of good quality semiconductor material onto the substrates. The configuration of the deposition chamber 102 also allows rapid diffusion to equalize the concentrations of all species throughout the plasma, including the rapid diffusion of the input reactant gas, to obtain a uniform concentration.
The exemplary electrode assembly 104 illustrated in
With respect to plasma formation, the electrode assembly 104 may be used to create high intensity plasma between the substrate carriers 106a and 106b (as well as substrates 120a and 120b). The high intensity plasma is created when the rod electrodes 126 are energized by power such as, for example, RF or DC power from the power supply 108. The energy is supplied in alternating phases from one rod electrode 126 to the next adjacent rod electrode, as is represented by the alternating series of “+” and “−” signs in
It should be noted that the rod electrodes 126 may, alternatively, be driven in phase with each other. Here, the substrates 120a and 120b are held at ground potential or at ground with a small DC bias. This will create a relatively uniform electric field and plasma in each of the two areas between the central plane CP and the substrates 120a and 120b.
Since the rod electrodes 160 present a load having a capacitive reactance (due to the length of the rod electrode being less than one-quarter wavelength of the excitation frequency), the RF energy is coupled to the rod electrode in parallel with an inductive reactance so as to create a predominantly resonant circuit. In the embodiment of
With respect to materials, the rod electrodes 126 illustrated in
Turning to size and shape, the rod electrodes 126 in one implementation that is suitable for commercial applications are cylindrical in shape, are about 1.2 cm in diameter and about 60 cm in length. The rod electrodes 126 are positioned parallel to one another about every 2 cm (i.e. 2 cm between the longitudinal axes of adjacent rod electrodes) in the direction of substrate travel and in the central plane CP of the deposition chamber interior. Thus, in the illustrated embodiment, the central plane CP is also the electrode plane. So configured and arranged, there will be forty-six (46) of the rod electrodes 126 in a 100 cm long deposition chamber that has small electrode-free areas near the inlets and outlets. In another exemplary implementation, smaller rod electrodes that are about 0.6 cm in diameter and about 60 cm in length are positioned parallel to one another about every 1 cm (i.e. 1 cm between the longitudinal axes of adjacent rod electrodes) in the direction of substrate travel and in the central plane CP of the deposition chamber interior. So configured and arranged, there will be ninety-two (92) of the smaller rod electrodes 126 in a 100 cm long deposition chamber that has small electrode-free areas near the inlets and outlets. It should also be noted that, for both rod electrode 126 sizes, the spacing between adjacent rod electrodes is less than one twenty-fifth ( 1/25) of the length and the height of the interior of the deposition chamber 102 and relatively short as compared to the distance over which silicon particles and higher order silanes form.
The rod electrodes 126 are not, however, limited to these configurations and arrangements. For example, the rod electrodes may be other than circular in cross-sectional shape, as are the exemplary cylindrical rod electrodes 126. There may also be instances where the spacing between the rod electrodes 126 will vary, where some or all of the rod electrodes are slightly offset from the central plane CP and/or where some of the rod electrodes are not parallel to others. The cross-sectional size of the rod electrodes (e.g. the diameter where the rod electrodes are cylindrical) may also be varied from electrode to electrode to suit particular applications.
There are a number of advantages associated with the present electrode assembly 104. For example, the arrangement of the plurality of closely spaced rod electrodes 126 allows higher RF frequencies to be used to excite the plasma in the present PECVD system 100, as compared to the frequencies that can be used in conventional PECVD systems, when the systems are of commercial production size (i.e. where the substrates are relatively long and at least 0.5 m wide). The series of parallel rod electrodes 126, with alternating phases of applied RF power, forms a series of well characterized electronic transmission lines capable of supporting high frequency RF excitation in the range of 27-81 MHz. It has been shown in laboratory experiments that RF power in the 27-81 MHz excitation frequency range can provide higher deposition rates (i.e. about 0.5 nm/sec.) and better material quality than the conventional excitation frequency of 13.5 MHz. Conventional electrode designs are not conducive to these higher frequencies in commercial production sized systems because they create poorly controlled standing waves, which results in non-uniform plasma intensity and non-uniform deposition rates. Conversely, the present electrode assembly 104 produces well controlled standing waves and only minor variations in plasma intensity when excited to a frequency of 80 MHz over relatively long substrates that are at least 0.5 m wide.
Other advantages are associated with the creation of high intensity plasma regions 128 along the central plane CP (
It should also be noted that a series of rod electrodes that are arranged in the manner described above does not create a uniform electric field and plasma in the substrate travel direction indicated by arrows A and, instead, will create an electric field and plasma that varies periodically in the travel direction from the area closet to a rod electrode to the midpoint between two rod electrodes. The deposition rate and semiconducting properties of the deposited material could, therefore, vary periodically in the travel direction. The illustrated embodiment eliminates this periodic variation in electric field and plasma intensity in a variety of ways. Periodic variations are reduced to a large extent by insuring that the distance between adjacent rod electrodes 126, as well as the distance between the rod electrodes and the substrates 120a and 120b, is within a diffusion length. For example, in the exemplary embodiments, the spacing between adjacent rod electrodes 126, is less than half of the distance from the central plane CP to the substrates. In fact, the spacing between adjacent rod electrodes 126 and from the rod electrodes to the substrates 120a and 120b should be minimized so that rapid diffusion can further reduce variations in the deposition rate. Finally, if necessary, the substrates 120a and 120b can be moved relatively rapidly in the non-uniform direction (i.e. the direction indicated by arrows A) to average out any small, remaining variations in the deposition rate.
The electrode assembly 104 may, in some implementations of the present inventions, also be used during the deposition process to deliver reactant materials to the deposition chamber 102 and to evacuate exhaust from the deposition chamber. To that end, and referring to
The exemplary lumens 132 in the illustrated embodiment are slightly smaller than the rod electrodes 126. For example, the lumen 132 would be about 1.0 cm in diameter in a cylindrical rod electrode 126 that is itself 1.2 cm in diameter, and about 0.5 cm in diameter in a cylindrical rod electrode that is itself 0.6 cm in diameter. The apertures 134, which are about 350 μm in diameter in the larger rod electrodes 126 and about 200 μm in diameter in the smaller rod electrodes, are positioned about every 0.5 cm along the length of the rod electrodes 126. However, for both the rod electrodes 126 delivering reactant materials and the rod electrodes evacuating exhaust, there is preferably a slight variation in aperture spacing from the longitudinal ends of the rod electrodes 126 to the centers in order to compensate for the pressure drop that occurs between the longitudinal ends, which are connected to the manifold 112a, and the center. More specifically, for 0.6 cm diameter rod electrodes 126 with 200 um apertures 134, there is about 5% less spacing at the center (i.e. about 0.475 cm spacing) and about 5% more spacing at the longitudinal ends (i.e. about 0.525 cm spacing) and the change occurs linearly. This results in a uniform flow rate through the apertures 134 in the rod electrodes 126 from one longitudinal end of the rod electrodes 126 to the other. The apertures 134 may also be aligned with one another from one rod electrode 126 to the next, or staggered, as applications require.
As discussed above with reference to
The reactant gas source 110 may be used to fill the deposition chamber 102 with hydrogen, or a mixture of hydrogen and argon (Ar), at the desired pressure (e.g. 300 mTorr) prior to the excitation of the rod electrodes 126 and the introduction of the silane or other reactant material. The rod electrodes 126 are then excited to initiate the plasma. During the actual deposition process, the reactant gas source 110 supplies pure or highly concentrated silane (rather than the dilute 5-10% silane in hydrogen associated with conventional devices) to the rod electrodes 126 that are supplying reactants by way of the manifold 112a. The apertures 134 direct the pure silane into the low intensity plasma regions 130 and the silane diffuses rapidly (i.e. within a few milliseconds) into the hydrogen already in the deposition chamber 102 to achieve an approximately 7% concentration of silane in hydrogen. The diffusion occurs before the silane reaches the high intensity plasma regions 128 where the silane is consumed by the decomposition into silicon and hydrogen (SiH4→Si+2H2). The rapid diffusion and dilution into the hydrogen atmosphere with the deposition chamber 102 prior to encountering high intensity plasma regions 128, as well as the relatively short rod electrode to adjacent rod electrode distance that the silane travels and correspondingly short residence time within the deposition chamber, also reduces the formation of higher order silanes (Si2H6, Si3H8, etc.) and/or silicon particles within the plasma. The silicon is deposited onto the substrates 120a and 120b, while the hydrogen and a very small amount of unused silane is removed by the apertures 134 in the other rod electrodes 126 and the exhaust device 114.
The input flow rate of the pure silane needs to be only slightly greater than the rate at which the silane is consumed because only a small amount of the silane is wasted. More specifically, when the gas in the deposition chamber reaches the apertures 134 in the rod electrodes 126 that are being used to evacuate exhaust from the deposition chamber 102, the gas is about 6% silane and 94% hydrogen. Additionally, because the deposition reaction is SiH4→Si+2H2, the exhaust gas flow rate should be roughly twice the input gas flow rate in order to maintain a constant pressure in the deposition chamber 102. By calculation, for a given reaction rate (n), the input gas flow rate=1.128 n SiH4 and the exhaust gas flow rate=2.128 n (94% H2+6% SiH4). All of the hydrogen generated in the deposition reaction is removed by the exhaust, as is about 13% of the input silane. Hence, by calculation, 87% of the silane is used in the deposition process. Conventional PECVD systems, on the other hand, convert only about 15-20% of the silane into silicon and hydrogen and the remainder is wasted. Of course, in conventional PECVD systems and the present PECVD system 100, some of the silicon is deposited onto the walls of the deposition chamber. This brings conventional PECVD systems down to about 10-15% utilization efficiency, i.e. about 10-15% of the silicon input as silane gas is actually deposited onto substrates. As noted above, the geometry of the present deposition chamber 102 reduces the percentage of silicon that is deposited onto the walls of the deposition chamber and, accordingly, the overall utilization efficiency of the present PECVD system 100 is about 70%.
Another advantage associated with the supply of pure silane through some of the rod electrodes 126 and the evacuation of exhaust through others is that it facilitates much lower gas flow rates than conventional PECVD systems. For a given reaction rate (n), the input flow rate for conventional PECVD systems (7% silane in hydrogen) is 100 n, the exhaust flow rate (6% silane in hydrogen) is 100 n and, accordingly, the net consumption of silane is (7%-6%)×100 n. In the present system, on the other hand, the input flow rate is 1.128 n (100% silane), the exhaust flow rate is 2.128 n (6% silane in hydrogen) and, accordingly, the net consumption of silane is (1.128 n×100%)−(2.128 n×6%). In other words, the input flow rate of the present PECVD system is almost 100 times less than conventional PECVD systems and the output flow rate is almost 50 times less. The lower flow rates allow for a much lower capacity exhaust device 114 (e.g. vacuum pump) to be used to evacuate the reaction products from the deposition chamber 102 and maintain a constant chamber pressure. The very short travel distance from a rod electrode 126 that is supplying reactant to a rod electrode that is evacuating exhaust (e.g. substantially less than one-twentieth ( 1/20) of the length and/or height of the deposition chamber 102 in the illustrated embodiment) ensures that the dwell time for silane in the reaction chamber 102 is short even though the flow rates are low. The short dwell time minimizes the formation of high order silanes and/or silicon particles.
As noted above, in an alternative implementation, the rod electrodes 126 are driven in phase with each other, and the substrates 120a and 120b held at ground potential (or at ground with a small DC bias), to create a relatively uniform electric field and plasma in each of the two areas between the central plane CP and the substrates. Here, the rod electrodes 126 may be rotated ninety (90) degrees from the orientation illustrated in
The reactant gas source 110, which may be used to supply the deposition chamber 102 with hydrogen (or hydrogen and argon) prior the initiation of the deposition process and pure silane during the deposition process, includes a plurality of storage containers G1-GN. Other gasses that may be stored include argon, hydrogen, silane, methane, germane, and silane with dopant gasses such as tri-methyl borane or phosphine. The gasses may be stored under pressure and, to that end, the reactant gas source 110 includes a plurality of valves 136 that control the flow rate of the gasses from the storage containers G1-GN. It should also be noted that the present inventions are not limited to gaseous reactant material. Sources of liquid and/or solid reactants may also be provided if required by particular processes.
The controller 116 may be used to control a variety of aspects of the deposition process. For example, the rate at which pure silane is supplied to the deposition chamber 102 and the rate at which exhaust is evacuated from the deposition chamber may be controlled based upon data from the sensors 118. As noted above, the silane input rate should be slightly greater than the rate at which the silane is consumed (i.e. the deposition rate) because only a small amount of the silane is wasted. Thus, for a particular deposition rate and power level applied to the rod electrodes 126 by the power supply 108 (or “plasma power”), the input flow rate may be adjusted by feedback from the sensors 118 to achieve the desired concentration of silane in the exhaust gas. For an operating point in which the deposition rate is limited by the plasma power, the exhaust gas concentration of silane will typically be about 5-6%. Alternatively, for operating points in which the deposition rate is limited by silane depletion, the input flow rate of the silane is adjusted to be equal to the rate consumed in the deposition and the concentration of silane in the exhaust gas approaches zero. The exhaust rate is also controlled by feedback to maintain the pressure in the deposition chamber 102 at the desired pressure (e.g. about 300 mTorr). The temperature of the substrates 120a and 120b and the frequency and power level of the plasma excitation will also typically be controlled to achieve the desired quality of silicon at the desired deposition rate. Accordingly, the sensors 118 may include a gas concentration sensor associated with the exhaust device 114, a pressure sensor within the deposition chamber 102, and a temperature sensor associated with the substrates 120a and 120b. A sensor that detects the presence of a plasma to verify correct operation may also be provided.
Controlling the PECVD process in the manner described above allows the present PECVD system to perform continuous deposition processes at a stable, steady state with stable temperature, gas flow, gas concentrations, deposition rates, etc. The controller 116 can use feedback from the sensors 118 to adjust the parameters of the stable, steady state and achieve the desired material properties. The combination of steady state operation and parameter adjustment, based on sensors within the system as the deposition process proceeds, together with rapid diffusion to reduce any non-uniformity allows the manufacture of the present system to be much less precise in mechanical tolerances, and less uniform in gas flow. As a result, the present system can be manufactured much less expensively than conventional “batch mode” systems which deposit material with comparable uniformity and semiconducting properties.
The present PECVD system 100 may be used to produce a variety of material layers. By way of example, but not limitation, the PECVD system 100 may be used to form high quality amorphous or nano-crystalline silicon semiconductor layers on very large substrates (e.g. 1 m×0.5 m) that may be utilized in silicon photovoltaic cells and other large area, low cost devices.
Although the present inventions have been described in terms of the embodiments above, numerous modifications and/or additions to the above-described embodiments would be readily apparent to one skilled in the art. It is intended that the scope of the present inventions extend to all such modifications and/or additions.
This application is a continuation of PCT US/2004/030275 filed Sep. 14, 2004.
| Number | Date | Country | |
|---|---|---|---|
| Parent | PCT/US04/30275 | Sep 2004 | US |
| Child | 11553334 | Oct 2006 | US |
