FIELD OF THE INVENTIONThe present invention relates to an apparatus and method for the production of films by chemical vapour deposition.
BACKGROUND OF THE INVENTIONMetal or metalloid containing films, such as gallium nitride (GaN) films, have applications in a range of devices from light emitting diodes (LEDs) to ultraviolet detectors to transistor devices.
These films have commonly been produced by techniques including molecular beam epitaxy (MBE), metal organic chemical vapour deposition (MOCVD) and remote plasma enhanced chemical vapour deposition (RPECVD or RPCVD). RPECVD has been employed to produce films of high quality at considerably lower temperatures than those used in MOCVD which thereby reduces process costs and allows the use of temperature sensitive preferred substrates for film deposition.
One problem which must be addressed during film production using any chemical vapour deposition (CVD) technique is to obtain an even and controlled distribution of reagents across the substrate of the surface onto which the film is to be grown to thereby achieve uniform thin film growth. At least part of the solution to this problem may be addressed by the design of the distribution systems. For example, in RPECVD a shower head or lattice design may be employed to obtain an even distribution of metal organic reagent across the substrate while a baffle may be used to enhance the even distribution of the plasma stream of active nitrogen species. One such baffle design is disclosed in WO/2010/091470, the disclosure of which is hereby incorporated in its entirety, wherein an ‘inverse pagoda’ style of baffle is used to diffuse and filter the plasma stream.
Many of these approaches focus on growth of a single film and so the reagent, e.g. metal organic, distribution lattice and plasma channel with baffle are generally centred over the substrate location to thereby provide a generally homogeneous distribution of both materials over the entire substrate surface. This form of chamber design is not so effective when it is desired to grow a plurality of films within the same growth chamber to improve productivity.
The use of multiple substrates is particularly desirable in those film deposition techniques where the growth rate is extremely slow. For example, atomic layer deposition (ALD) is a useful growth technique based upon the sequential pulsing of chemical precursor vapours to thereby achieve one atomic layer per pulse. Due to the sequential pulsing arrangement of ALD each reagent pulse reacts with the deposition surface until the reaction is completed. A purge gas is used to carry away excess reagent and reaction side products after each pulse in an attempt to minimise impurities being deposited in the film.
ALD is of interest due to the ability to produce thin uniform films with a high degree of control over film thickness and composition. One of the drawbacks of ALD is the amount of time required to grow a useful film since only a monolayer may be deposited in each complete deposition cycle. The time required for each cycle is limited by the switching speed of the reagent release valves as well as the time taken to purge after each half cycle and rotate the substrate into place. This results in each full cycle taking from 0.5 to a few seconds, further contributing to the slow production.
Further, the purging cycle is not entirely effective which often means that an amount of the metal organic reagent, such as trimethylgallium (TMG), remains in the growth chamber during pulsing of the plasma containing the active second reagent, such as nitrogen. This may result in carbon impurities being incorporated into the growing film, thereby reducing its quality.
Minimising the extent of incorporation of both carbon and oxygen, as impurities, in the growing film is a major challenge in CVD film production. As well as altering the desired chemical composition of the film these impurities disrupt the lattice matching of forming layers thereby causing defects within the film and negatively impacting on the overall quality of the product.
MOCVD approaches have been relatively more successful than certain other CVD techniques at lowering oxygen incorporation into the growing thin films but levels of carbon incorporation are not ideal. More particularly, MOCVD often involves growth temperatures of about 1000° C. to 1200° C., which thereby results in high equipment costs and rules out the use of temperature sensitive preferred substrates for film deposition.
It would thus be desirable to provide for a CVD apparatus and method which could provide for the advantages in control of film growth afforded by ALD while minimising the drawbacks of that technique. Particularly, it would be useful to provide for a CVD apparatus and method allowing for a reduction in the levels of incorporation of carbon and oxygen as impurities in the film product and preferably a CVD apparatus and method which can be run at lower temperatures than those employed in a standard MOCVD approach.
SUMMARY OF THE INVENTIONIn a first aspect, although it need not be the only or indeed the broadest form, the invention resides in an RPCVD apparatus for forming a film, the apparatus including a growth chamber comprising:
- (a) a Group VA′ plasma inlet located in a first deposition zone of the growth chamber to introduce a Group VA plasma thereto;
- (b) a Group IIIA reagent inlet located in a second deposition zone of the growth chamber to introduce a Group IIIA reagent thereto;
- (c) an additional reagent inlet adjacent the Group IIIA reagent inlet to introduce an additional reagent selected from the group consisting of ammonia, hydrazine, di-methyl hydrazine and a hydrogen plasma into the second deposition zone such that the additional reagent and Group IIIA reagent mix prior to deposition; and
- (d) a substrate holder adapted to support one or more substrates and rotate each substrate between the first and second deposition zones.
Preferably, the additional reagent inlet is an ammonia inlet.
Preferably, the Group VA plasma inlet, the Group IIIA reagent inlet and the additional reagent inlet open into the growth chamber at a distance between about 1 cm to about 30 cm from a growth surface of the one or more substrates. More preferably, between about 1 to about 20 cm or 1 to about 10 cm.
Preferably, a ceiling of the growth chamber is located less than about 30 cm vertically above the location of the substrates, more preferably less than about 25 cm, even more preferably less than about 20 cm, still more preferably less than about 10 cm. Values of 5 cm and 7.5 cm may be useful with 3 cm to 4 cm as the lower end values.
In certain embodiments at least one of the Group VA plasma inlet, the Group IIIA reagent inlet and the additional reagent inlet end flush with the ceiling of the growth chamber which is located between about 1 to about 30 cm, 1 to 20 cm, 1 to 10 cm vertically above a growth surface of the substrates, preferably, between 4 to 15 cm, 4 to 10 cm, 4 to 8 cm.
Suitably, an opening of the additional reagent inlet opens into the growth chamber in close proximity to the one or more substrates.
The additional reagent, inlet may extend downwardly from the ceiling of the growth chamber to end in close proximity to the growth surface of the one or more substrates.
In one embodiment, the additional reagent inlet opens into the growth chamber through a side wall thereof at a height suitable to enable a flow of additional reagent entering therethrough to have a flow path passing over and substantially adjacent to the growth surface of the one or more substrates.
In a preferred embodiment there is a direct flow path between the Group VA plasma inlet and the one or more substrates.
Suitably, the direct flow path between the Group VA plasma inlet and the one or more substrates extends to an unimpeded path between a plasma generator for generating the Group VA plasma and the one or more substrates.
In an embodiment, the Group VA plasma inlet and the Group IIIA reagent inlet end flush with a ceiling and/or side wall of the growth chamber through which they extend.
Preferably, the first deposition zone is substantially isolated from the second deposition zone.
Preferably, rotation of the substrate holder causes the one or more substrates to pass sequentially from the first deposition zone to the second deposition zone.
Preferably, the substrate holder is of a turntable design whereby it rotates around a central pivot and is provided with a plurality of recesses, each adapted to hold a substrate, around its periphery.
The Group VA plasma inlet and the Group IIIA reagent inlet may be located centrally within the growth chamber.
When the Group VA plasma inlet and the Group IIIA reagent inlet are located centrally within the growth chamber one or both thereof may be provided with a flow control device to direct the corresponding plasma or reagent into the appropriate first or second deposition zone.
The flow control device may be a flow barrier blocking one or more reagent flow paths within the Group VA plasma inlet or the Group IIIA reagent inlet or a directing portion, such as a shroud, continuous with the first or second reagent inlet.
In one embodiment, the apparatus may further comprise a baffle associated with the Group VA plasma inlet such that the plasma substantially passes therethrough.
The baffle may comprise the flow control device which may be a flow barrier blocking one or more outlets of the baffle.
Preferably, the additional reagent inlet opens into the growth chamber substantially adjacent to the opening of the Group IIIA reagent inlet to promote mixing of said reagents prior to their contacting the one or more substrates.
Suitably, the Group VA plasma inlet is in fluid communication with a plasma generator producing a Group VA plasma comprising an active species.
Preferably, the Group, VA plasma is a nitrogen plasma comprising active nitrogen species.
Suitably, the Group IIIA reagent is a Group IIIA metal organic reagent.
In one particularly preferred embodiment, the Group VA plasma inlet and the Group IIIA reagent inlet are located peripherally within the growth chamber.
Suitably, the Group VA plasma inlet and the Group IIIA reagent inlet are located substantially at opposite ends of the growth chamber.
The growth chamber may comprise one or more structures associated with the additional reagent inlet and/or the Group IIIA reagent inlet to promote mixing of said reagents immediately prior to their contacting the one or more substrates.
The apparatus may further comprise one or more heating devices to heat the additional reagent inlet and/or the Group IIIA reagent inlet prior to entering the growth chamber.
In a second aspect the invention resides in a method of forming a thin film on a substrate by RPCVD including the steps of:
- (a) introducing a Group VA plasma through a Group VA plasma inlet into a first deposition zone of a growth chamber;
- (b) introducing a Group IIIA reagent through a Group IIIA reagent inlet into a second deposition zone of the growth chamber, the second deposition zone being substantially isolated from the first deposition zone;
- (c) introducing an additional reagent selected from the group consisting of ammonia, hydrazine, di-methyl hydrazine and hydrogen plasma through an additional reagent inlet into the second deposition zone such that the additional reagent and the Group IIIA reagent mix prior to deposition;
- (d) moving the substrate between the first and second deposition zones,
to thereby form a thin film on the substrate.
Preferably, the additional reagent is ammonia.
Suitably, the additional reagent is introduced into the second deposition zone substantially adjacent the opening of the Group IIIA inlet.
In one embodiment, the additional reagent is introduced into the growth chamber through a side wall thereof.
In one embodiment, the additional reagent is introduced into the growth chamber to form a substantially horizontal flow path passing over and substantially adjacent to the growth surface of the substrate.
The additional reagent and the Group IIIA reagent are preferably being introduced into the growth chamber simultaneously.
Suitably, the Group IIIA reagent is a Group IIIA metal organic reagent.
Preferably, the Group IIIA metal organic reagent is a Group IIIA metal alkyl reagent.
Preferably, the Group IIIA metal alkyl reagent is selected from the group consisting of trimethyigallium, triethylgallium, trimethylindium and trimethylaluminium.
The method may further include the step of heating one or more of the reagents prior to their entering the growth chamber.
The method may further include the step of promoting the mixing of the metal organic reagent and the additional reagent adjacent the one or more substrates.
Suitably, the Group VA plasma inlet is in fluid communication with a plasma generator.
Preferably, the Group VA plasma is a nitrogen plasma comprising active nitrogen species.
The isolation of the deposition zones substantially prevents the mixing of the Group VA plasma and Group IIIA reagent.
The method may further include the step of controlling the flow of one or more of the Group VA plasma or Group IIIA reagent upon exiting the associated inlet to direct that flow to a desired deposition zone.
The method may further include the step of controlling the temperature to be between about 400 to about 1200° C., preferably between about 500 to about 1000° C. (inclusive of a temperature of about 500° C., 600° C., 700° C., 800° C., 900° C. or 1000° C.), more preferably between about 500 to about 850° C.
In combination with the presence of an additional reagent gas, preferably ammonia, it has been found that the power of the plasma generator has an effect on carbon incorporation into the thin film and so the method may also include the step of controlling the power of the plasma generator to be between about 500 W to about 5000 W from a single source. This may be combined with a growth pressure of 2-5 torr and a nitrogen plasma flow of 2000-3000 sccm with an ammonia flow of about 15 to about 1500 sccm, preferably about 20 to about 200 sccm, preferably about 20 to about 100 sccm, more preferably about 20 to about 50 sccm.
The growth pressure may be between 2-5 torr, 2-4 torr or about 3 torr.
Preferably, the power of the plasma generator is between about 100 watts to about 3000 watts with a nitrogen flow rate of 1000-3000 sccm extending to 100-20000 sccm in a commercial unit. A preferred metal organic reagent flow rate is 1200-2000 sccm which may extend to 100-10000 sccm in a commercial unit. A value for the plasma generator power of about 500 to 5000 W, 500 to 4000, 500 to 3000, 500 to 2000, 500 to 1000, 500 to 900 W, 500 to 800 W, 600 to 1000 W, 600 to 900 W, 600 to 800 W, 700 to 1000 W, 700 to 900 W and preferably about 800 W is preferred and each value or range of which may be independently coupled with an ammonia flow rate of any one of between 15 to 1500 sccm. For relatively small growth chambers it is found that ammonia flows of 10 to 75, 10 to 60, 10 to 50 10 to 40, 10 to 30, 15 to 75, 15 to 60 15 to 50, 15 to 40, 15 to 35, 15 to 30, 20 to 75, 20 to 60, 20 to 50, 20 to 40, 20 to 30, including values of about 15, 20, 25, 30, 35, 40, 45 and 50 sccm are particularly useful in lowering carbon incorporation, however, in moving to commercial scales higher powers and multiple plasma sites are envisaged as being useful.
In a third aspect the invention resides in a method of forming a thin film having a carbon impurity content of less than about 5E+17 atom/cm3, on a substrate by RPCVD including the steps of:
- (a) introducing a Group VA plasma through a Group VA plasma inlet into a first deposition zone of a growth chamber wherein a direct flow path is provided between the Group VA plasma inlet and a substrate located in the first deposition zone;
- (b) introducing a Group IIIA reagent through a Group IIIA reagent inlet into a second deposition zone of the growth chamber, the second deposition zone being substantially isolated from the first deposition zone;
- (c) introducing an additional reagent selected from the group consisting of ammonia, hydrazine, di-methyl hydrazine and hydrogen plasma through an additional reagent inlet into the second deposition zone such that the additional reagent and the Group IIIA reagent mix prior to deposition;
- (d) moving the substrate between the first and second deposition zones,
to thereby form a thin film on the substrate having a carbon impurity content of less than about 5E+17 atom/cm3.
Preferably, the carbon impurity content is less than about 3E+17 atom/cm3, even more preferably less than about 2E+17 atom/cm3, yet more preferably less than or about 1E+17 atom/cm3. A lower limit may be considered to be about the SIMS detection limit for carbon impurities in such films.
In one embodiment, the thin film also has an oxygen impurity content of less than about 8E+17 atom/cm3, even more preferably less than about 6E+17 atom/cm3, yet more preferably less than about 4E+17 atom/cm3, still more preferably less than about 2E+17 atom/cm3, or even less than or about 1E+17 atom/cm3. A lower limit may be considered to be about the SIMS detection limit for oxygen impurities in such films.
The statements made above in relation to the second aspect apply equally well to the third aspect.
In a fourth aspect the invention resides in a film made by the method of the second or third aspects.
In a fifth aspect the invention resides in use of a film of the fourth aspect in a semiconductor device.
Further features of the present invention will become apparent from the following detailed description.
Throughout this specification, unless the context requires otherwise, the words “comprise”, “comprises” and “comprising” will be understood to imply the inclusion of a stated integer or group of integers but not the exclusion of any other integer or group of integers.
BRIEF DESCRIPTION OF THE FIGURESIn order that the invention may be readily understood and put into practical effect, preferred embodiments will now be described by way of example with reference to the accompanying figures wherein:
FIG. 1 shows a schematic representation of a typical RPCVD apparatus for depositing a metal nitride film on a substrate;
FIG. 2 shows a perspective sectional view of one embodiment of an apparatus for depositing a metal nitride film on a substrate when employing an inverse pagoda baffle and multiple substrates;
FIG. 3 shows a schematic sectional representation of one embodiment of an apparatus for forming a film according to the present invention;
FIG. 4 shows a schematic sectional representation of one preferred embodiment of an apparatus for forming a film according to the present invention;
FIG. 5 shows a partial perspective sectional view of the apparatus for forming a film, as represented inFIG. 4;
FIG. 6 shows a schematic sectional representation of a highly preferred embodiment of an apparatus for forming a film according to the present invention;
FIG. 7 shows a partial perspective sectional view of the apparatus for forming a film, as represented inFIG. 6;
FIG. 8 shows a partial perspective sectional view of an alternative embodiment of the apparatus for forming a film to that shown inFIG. 7;
FIG. 9 shows a partial perspective sectional view of an alternative embodiment of the apparatus for forming a film to that shown inFIG. 7;
FIG. 10 shows a schematic representation of an RPCVD apparatus for depositing a film on a substrate, according to another embodiment of the invention;
FIG. 11 shows a schematic representation of an alternative RPCVD apparatus for depositing a film on a substrate to that shown inFIG. 10;
FIG. 12 shows a schematic representation of a further alternative RPCVD apparatus for depositing a film on a substrate to that shown inFIG. 10;
FIG. 13 shows a schematic representation of yet a further alternative RPCVD apparatus for depositing a film on a substrate to that shown inFIG. 10;
FIG. 14 shows a partial perspective sectional view of an apparatus for forming a film, according to a further embodiment of the present invention;
FIG. 15 is a graphical representation of the carbon levels incorporated into films under varying conditions;
FIG. 16 is a SIMS graphical analysis of the typical impurities found in a film produced by a method and apparatus of the invention and an underlying GaN template;
FIG. 17 is a SIMS graphical analysis of the level of carbon, as an impurity, found in a film produced by a method and apparatus of the invention on an underlying GaN template with varying ammonia flow rates; and
FIG. 18 is a SIMS graphical analysis of the level of oxygen, as an impurity, found in a film produced by a method and apparatus of the invention on an underlying GaN template with varying ammonia flow rates.
DETAILED DESCRIPTION OF THE INVENTIONThe present inventors have identified a particular RPCVD apparatus and process conditions for the production of high quality films which results in improvements to the film growth rate and growth control, by comparison to standard ALD techniques and other CVD processes, and, importantly, which provides for a surprising level of reduction in oxygen and carbon-based film impurities due to reagent side reactions.
The reagents which may be employed with the present apparatus and method, and hence the nature of the films which can be formed, are not particularly limited. Although the embodiments discussed herein generally employ a nitrogen plasma and a metal organic (typically a gallium containing metal organic such as trimethylgallium) as the reagents, the utility of the present invention is not so limited. The Group IIIA (otherwise known as Group 13 under the current IUPAC system) reagent may comprise an element which is selected from the group consisting of boron (B), aluminium (Al), gallium (Ga), indium (In) or thallium (TI). The Group VA (otherwise known asGroup 15 under the current IUPAC system) plasma may be generated from any suitable reagent containing a Group VA element selected from the group consisting of nitrogen (N), phosphorus (P), arsenic (As), antimony (Sb) and bismuth (Bi).
The term “deposition zone” as used herein is used to refer to a distinct region, section or segment of the growth chamber into which one or more reagents are introduced. Individual depositions zones, such as first and second deposition zones, are isolated one from the other such that a substrate or growing film will only be substantially exposed to a particular reagent introduced only into one deposition zone when the substrate or growing film actually enters that deposition zone. The separation or isolation of deposition zones may be spatial only or may be effected by partial or complete physical barriers.
In the embodiments described the reagents employed will be trimethylgallium, a nitrogen plasma and ammonia but the person skilled in the art will appreciate the principles disclosed herein could be applied mutatis mutandis to other reagent combinations.
Without wishing to be bound by any particular theory, the present inventors have postulated that an experimentally observed dramatic reduction in the levels of oxygen and carbon incorporated into gallium nitride films produced in an RPCVD apparatus are due to a choice of conditions, including, primarily, the supply of an additional gaseous reagent, preferably ammonia gas, to mix with the Group IIIA reagent and preferably in close proximity to the substrate, which favours formation of a trimethylgallium:ammonia Lewis acid:base adduct. This adduct breaks down to form gallium nitride with the release of methane gas which is not incorporated into the growing film to the same extent as methyl radicals would be.
In typical prior art CVD approaches, and particularly MOCVD due to the high temperatures employed, a molecule of trimethylgallium injected into the growth chamber will decompose thermally to finally produce a gallium atom and three methyl radicals. The gallium will react with the nitrogen source, which may be ammonia or a nitrogen plasma, to form the GaN film. The reactive methyl radicals are often incorporated into the growing film as an impurity thereby increasing strain and lowering the overall quality of the film product. Additional hydrogen-containing reagents, such as ammonia gas, are typically only introduced with the plasma stream and the benefits produced herein are not seen with such an approach.
When the growth temperature is instead kept below the thermal decomposition point of trimethylgallium and an additional reagent, preferably ammonia, is introduced into the growth chamber then the two components form a Lewis acid-base adduct of proposed formula (CH3)3Ga:NH3. It is further proposed that this adduct reacts to form an intermediate (CH3)2Ga:NH2+CH4. A further step in this pathway leads to the formation of an adduct of formula [(CH3)2Ga:NH2]3with eventual formation, from that structure, of three molecules of GaN and six molecules of CH4gas. The methane gas is less reactive than a methyl radical and is easily removed via the exhaust of the growth chamber.
The inventors have further postulated that such adduct formation and preference of the formation of methane over methyl radicals can be encouraged by minimising the extent of the reactions which occur in the gas phase i.e. in the upper and central regions of the growth chamber and instead maximising the mixing of the reagents only in the immediate vicinity of the substrates. This can be achieved by introducing the ammonia or other additional reagent into the growth chamber with the Group IIA reagent in a manner which causes it to be present or available only adjacent the substrates and hence the growth surface of the growing film.
FIG. 1 shows a schematic representation of atypical RPCVD apparatus100 for depositing a Group IIIA nitride film on a substrate. Theapparatus100 comprises agrowth chamber105 inside which film growth will occur. Located within thegrowth chamber105 is asubstrate110 which is supported by asubstrate holder115 which may include or be connected to a heater to allow thesubstrate110 to be adjusted to growth temperatures. Aplasma inlet120, located at a distance from thesubstrate110, allows for entry ofplasma130 formed in thehigh frequency generator125 into thegrowth chamber105. Thehigh frequency generator125 acts on a region of theapparatus100 receiving nitrogen from anitrogen source135. A Group IIIA reagent source which is usually a Group IIIA metalorganic reagent source140, which is usually also at a distance from thesubstrate110 i.e. not adjacent thereto, introduces the metal organic into aflow path145 which delivers the reagent to a metalorganic injector150 for dispersion into thegrowth chamber105.
It can be seen that the plasma enters an area of thegrowth chamber105 directly above the metalorganic injector150 and so, in operation, the plasma containing active neutral nitrogen species and the metal organic reagent mix and react to form the particular metal nitride, such as gallium nitride, which is deposited on the substrate to form the film. Excess reagents, carrier gases, contaminants etc are removed via awaste outlet155.
Carbon and oxygen are inevitably incorporated as impurities into the film but, these aside, this approach is generally satisfactory for the formation of a film on a single substrate. However, it is often desirable to have the capacity to generate a number of such films at the same time. Hence, an apparatus such as that shown inFIG. 2 may be useful.
FIG. 2 shows a perspective sectional view of one embodiment of anapparatus200 for depositing a metal nitride film on a substrate which essentially corresponds to the simple representation shown inFIG. 1 but with the use of a baffle and multiple substrates. Theapparatus200 comprises agrowth chamber205 partially formed fromouter housing210.
Aplasma generator215 receives nitrogen through anitrogen inlet220 and the active nitrogen plasma formed passes throughplasma inlet225, which once again is remote from the substrates, and into thegrowth chamber205 via abaffle230, which in the embodiment shown takes the form of an inverse pagoda style baffle as described in WO/2010/091470. The plasma passes through thebaffle230 and is evenly distributed by its concentric ring-like structure. The distributed plasma flow then passes over a metalorganic injector235 where the metal organic reagent is introduced and mixes with the plasma. The metal nitride formed will then deposit on one or more ofsubstrates240 located on asubstrate holder245. Thesubstrate holder245 may be of a turn table design and so may be rotating at high speed throughout the deposition process. Waste is removed viaoutlet250.
It will be appreciated that the central placement of theplasma inlet225 will likely result in the bulk of the plasma flow being focused on the centre of thesubstrate holder245, even with the use of a distribution system such as theinverse pagoda baffle230. It is critical to quality film growth that the reagents be distributed evenly across the surface of theappropriate substrate240 and the deficiency in thisapparatus200 will not be solved by rotation of thesubstrate holder245. This type ofapparatus200 also does not typically provide the advantages of control over film growth and thickness that is provided by atomic layer deposition (ALD).
FIG. 3 shows a schematic representation of one embodiment of anapparatus300 for forming a film, according to the present invention. The actual components of theapparatus300 are much the same as those displayed inFIG. 2 but with two notable exceptions being that one region of the baffle is blocked to plasma flow and a number of ports of the Group IIA reagent inlet (referred to herein as the metal organic injector) are either removed or closed to reagent flow.
As forFIG. 2, theapparatus300 shown inFIG. 3 comprises agrowth chamber305 having aplasma inlet310 to receive plasma flow comprising active neutral nitrogen species from aplasma generator315. AlthoughFIG. 3 is merely a schematic representation, theplasma inlet310 in this embodiment will be located physically closer to the level of the substrates than in the prior art apparatus. The plasma will flow into abaffle320 which may be of any suitable design but in the embodiment shown has an inverse pagoda shape, as shown inFIG. 2. This time the baffle is provided with aflow barrier325 formed around one side of thebaffle320 so as to prevent plasma from exiting along that side. This will result in the plasma flow being directed towards the opposite side of thegrowth chamber305 from the side of thebaffle320 bearing theflow barrier325.
The active nitrogen species then pass by a Group IIIA reagent injector in the form of metal organic reagent (e.g. trimethylgallium)injector330. InFIG. 3, those circles which are black inside represent ports or valves of the metalorganic reagent injector330 which are open to reagent flow i.e. they areopen ports335 while those circles which are white inside (not filled) represent ports or valves of the metalorganic reagent injector330 which are closed to reagent flow i.e. they areclosed ports340. In reality, the parts of the metalorganic reagent injector330 which are represented as closed may simply not be present in theapparatus300 and so only those regions of thegrowth chamber305 having theopen ports335 will actually be provided with a metalorganic reagent injector330 structure.
Located beneath the metalorganic reagent injector330 are a number ofsubstrates345 which are supported by a substrate holder350. The substrate holder350 may hold any desired number of wafers, for example, from 2 to 20 individual substrates, preferably 3 to 10, more preferably 5, 6 or 7. The substrates may have a crystal structure that is suitable for growth of the particular film desired. In particular examples, thesubstrates345 may comprise sapphire, SiC, silica, soda lime glass, borosilicate glass, Pyrex®, silicon, glass, synthetic sapphire, quartz, zinc oxide, nitride coated substrates and other materials as are well known in the art including free standing bulk semiconductor substrates and nitride templates. As is indicated by the arrow inFIG. 3, the substrate holder is adapted to rotate relative to theplasma inlet310 and metalorganic reagent injector330 thereby controlling growth and deposition uniformity. Waste materials may be removed throughwaste outlet355.
The combined effect of the directing action offlow barrier325 on the plasma pathway and the release of metal organic reagent from only thoseopen ports335, as can be seen fromFIG. 3, means that, mixing between the metal organic reagent and plasma species is minimized. Reduction of dead zones within the chamber could be accomplished by arranging the reagent inlets to be flush with the upper surface of the chamber minimising premature mixing of reagent gasses. The design of theapparatus300 has thus effected a physical separation of the regions into which plasma and metal organic reagent are released into a first deposition zone and a second deposition zone, respectively, which are substantially isolated from one another. It will be appreciated that rotation of the substrate holder350 causes thesubstrates345 to pass sequentially from the first deposition zone to the second deposition zone in a repetitive, continuous manner to thereby be exposed to the plasma and metal organic reagents, one after the other.
The sequential exposure of eachsubstrate345 to the metal organic reagent and the active nitrogen species will result in formation of subsequent layers of a film, much in the manner of ALD. However, the formation of separate deposition zones means that the delays experienced in ALD in both waiting for valves to be switched and the removal of one reagent by a purge gas before introduction of the second reagent, are avoided. Instead, the growing surface of the film is exposed to each reagent with a minimum of downtime in between due to the ability of the substrate holder350 to rotate at very high speeds. This greatly accelerates the growth of the films while maintaining control over sample growth.
The substrate holder350 may be adapted to rotate continuously. Preferably, the substrate holder is capable of rotating at speeds of between 10-2000 rpm. A preferred rotation speed may be between 25 to 100 rpm, more preferably about 50 rpm. The skilled addressee will understand that film growth will be controlled by a combination of the speed of rotation of the substrate holder350 and the reagent flow rates in the deposition chamber. Higher rotation speeds of the substrate holder350 will require a higher flow rate of reagents to ensure an overall increase in the growth rate of the film is produced.
It will be appreciated that although only oneplasma inlet310, and associatedbaffle320, and one region ofopen ports335 have been shown inFIG. 3 theapparatus300 may in fact comprise multiples of each component. For example, when looking down on thegrowth chamber305 from above the circular substrate holder350 could be imagined to be split into quadrants with a plasma inlet and associated baffle, if required, sitting above two adjacent or diagonally opposite quadrants and the same relationship for two distinct regions ofreagent injector330 which are open to release of metal organic or other reagent.
It will also be understood thatbaffle320 is not an essential feature but may be preferred, under certain process conditions, to prevent or reduce etching due to active nitrogen species which may have relatively high kinetic and/or potential energies. If thebaffle320 was not present inFIG. 3 then some form of structure, such as a shroud, could be used in its place to direct and contain plasma flow to one isolated deposition zone. When thebaffle320 is employed then it may take a variety of forms other than the inverse pagoda style shown which are well known in the art such as a plate with tortured multiple pathways therethrough shower head design etc. Whatever alternate style of baffle is ultimately used it may have either closed pathways or some form of flow barrier or flow directing means to ensure the plasma passes only into a discrete deposition zone and substantially avoids mixing with the other reagent.
Theapparatus300 may further comprise one or more heaters to heat the growth chamber and/or one of the reagent inlets. This may be useful to promote increased reaction rates, assist with quality of the growing film or to break or otherwise activate one or more of the reagents before exposure to the substrates.
As was mentioned above, the present apparatus and method are not particularly limited in the type of reagents suitable for use therein. Any reagents which are suitable for use in ALD may be appropriate. A wide range of reagent classes including, nitrogen plasma, nitrogen/hydrogen plasma, ammonia plasma and metal organics may be suitable. When a metal organic reagent is used then preferred examples include alkyl Group IIIA reagents such as but not limited to one or more of trimethylgallium, trimethylindium, trimethylaluminium as well as employing various well known Mg, Si and Zn precursors as dopant sources.
FIG. 4 shows a schematic representation of one embodiment of anapparatus400 for forming a film, according to the present invention. The majority of the components are as described forFIG. 3 and so will only be referred to briefly. Agrowth chamber405 is provided with aplasma inlet410, as forFIG. 3 being located relatively close to the substrates, which is continuous with aplasma generator415. Plasma introduced into thegrowth chamber405 will pass through abaffle420, which again in the embodiment shown is an inversepagoda style baffle420, before passing by metalorganic reagent injector425 which has closed ports430 (circles with white inside) and open ports435 (circles with black inside). Again, the regions of the metalorganic reagent injector425 having closedports430 may simply not be present. A number ofsubstrates440 are placed upon a substrate holder445 which rotates relative to thegrowth chamber405 and unwanted reactants and reaction products are vented viawaste outlet450.
The key difference inFIG. 4, compared to the embodiment shown inFIG. 3, is the physical location of theplasma inlet410 and associatedbaffle420 relative to theopen ports435. WhereasFIG. 3 showed a modified design whereby a typical central placement of theplasma inlet310 was manipulated with the additional feature of aflow barrier325 formed around a portion of the border of thebaffle320,FIG. 4 represents a radical shift in thegrowth chamber405 design by comparison to standard ALD set ups.
When looking at a section of thegrowth chamber405, as represented inFIG. 4, theplasma inlet410 and baffle420 have been shifted to the left hand side of thechamber405 to form a discrete first deposition zone which is substantially separate from the second deposition zone formed under and adjacent theopen ports435 of thereagent injector425.
The schematic representation shown inFIG. 4 is reproduced, in part, in three dimensions inFIG. 5 which shows a partial perspective sectional view of theapparatus400. For the sake of clarity many of the components of theapparatus500, such as the housing and high frequency generator, have been removed to focus on the key relationship betweenplasma inlet505,baffle510 and metalorganic reagent injector515.
In the embodiment shown inFIG. 5, as forFIG. 4, theplasma inlet505 and associatedbaffle510 are located peripherally to sit more or less directly above, i.e. adjacent, afirst substrate520 which can therefore be described as being within a first deposition zone receiving active nitrogen species. The baffle illustrated may be replaced with a showerhead or shroud or similar distribution system common to the field. Thereagent injector515 is only disposed on the opposite side of the chamber to theplasma inlet505, generally above asecond substrate525 and located within a second deposition zone receiving only metal organic reagent, for example trimethylgallium and/or trimethylindium. Thus, assubstrate holder530 rotates thefirst substrate520 will have been contacted with a first reagent (in this case active nitrogen species from the plasma) before proceeding out of the first deposition zone and entering the second deposition zone to then be contacted by the second reagent (in this case the metal organic). Thesecond substrate525, and all substrates located upon thesubstrate holder530, will undergo a similar cycle of sequential exposure to one reagent and then the other. This allows epitaxial crystal layers to be deposited sequentially to build up a film with a high degree of control. Alternating the exposure of the substrates to reagents by controlling the speed of rotation of thesubstrate holder530 provides for finer control than the rotating, pulsing and purging arrangement employed in ALD.
Although inFIGS. 3 to 5 the plasma and metal organic reagent inlets are shown as being vertically above the substrates it will be appreciated by the person skilled in the art that this is not necessarily the case. For example, the plasma inlet may inject plasma into the growth chamber from the side of the housing i.e. the plasma is injected parallel to the substrates and then proceeds to deposit down upon them. References herein to a plasma inlet or metal organic reagent inlet or additional gas inlet are meant to address the point at which the plasma or plasma activated reactant or metal organic reagent or additional gas enters the reaction chamber proper.
It will be appreciated that in the embodiments described inFIGS. 3 to 5 the two streams of reagents do not, to any notable degree, come into contact with one another. The physical separation of the reagent inlets assists with minimising reagent mixing such that the amount of oxygen and carbon-based impurities, which may be formed by such mixing and incorporated into the growing films, is reduced compared with a standard ALD or other CVD approach.
However, even employing the approach discussed above it has been found that sufficient amounts of oxygen and carbon impurities are still incorporated in the thin film formed to thereby reduce its quality. The use of RPCVD approaches, while more convenient in many ways than a standard MOCVD approach, are generally accepted in the art as inevitably resulting in moderate levels of oxygen and carbon impurities in the films produced as compared with high end MOCVD produced films. Very low impurity limits can be considered as being at least relatively close to the SIMS detection limits as set out in table 1.
| TABLE 1 |
|
| SIMS detection limits of selected elements in GaN under normal depth |
| profiling conditions |
| O2/SIMS | Cs/SIMS | Cs/SIMS |
| Positive Secondary Ion | Negative Secondary Ion | Positive Secondary Ion |
| Detection | Detection | Detection (CsM+) |
| Detection | | Detection | | Detection |
| Limit | | Limit | | Limit |
| Element | (atoms/cm3) | Element | (atoms/cm3) | Element | (atoms/cm3) |
|
| Be | 1E+14 | H* | 8E+16-2E+17 | Mg | 5E+15 |
| Li | 1E+14 | C* | 5E+15-2E+16 | Zn | 1E+16 |
| B | 1E+15 | O* | 1E+16-3E+16 |
| Na | 5E+14 | Si | 3E+15 |
| Mg | 5E+14 | As | 5E+15 |
|
| *Varies with vacuum conditions |
However, the present inventors have found that the levels of these impurities can be lowered significantly by the use of an apparatus as shown inFIGS. 6 to 9.
FIG. 6 shows a schematic sectional representation of one preferred embodiment of anapparatus600 for forming a film according to the present invention whileFIG. 7 is a partial perspective sectional view of thesame apparatus600. TheRPCVD apparatus600 comprises agrowth chamber605 which is provided with aplasma inlet610. It is clear fromFIGS. 6 and 7 that theplasma inlet610 is physically adjacent to the level of the substrates. Theplasma inlet610 is continuous with a plasma generator615 (detail not shown).
Thegrowth chamber605 is also provided with a Group IIIA reagent inlet and, more specifically, in the embodiment discussed herein, a metalorganic reagent inlet620 and an additional reagent inlet which may be a hydrazine inlet, a di-methyl hydrazine inlet or a hydrogen plasma inlet but is preferably anammonia inlet625. As was described forFIG. 4 theplasma inlet610 and metalorganic reagent inlet620 are physically distant forming a first and second deposition zone, respectively, withsubstrates630 arranged within each zone by asubstrate holder635 which rotates relative to thegrowth chamber605. Unwanted reactants and reaction products are vented viawaste outlet640 to which access is provided bygap645 provided between the circumference of thesubstrate holder635 and the inner walls of thegrowth chamber605. Theammonia inlet625, however, is immediately adjacent the metalorganic reagent inlet620 and so the ammonia will be introduced into the second deposition zone along with the metal organic reagent.
Plasma introduced into thegrowth chamber605 will directly contact thesubstrate630 placed in the first deposition zone as no baffle, shroud or like blocking or distributing device is in place in the embodiment shown. The present inventors have found that, when using such an apparatus under conditions of relatively low power of the plasma generator (around 500 W to 2500 W) and temperature (about 700° C. to 800° C.) no significant degree of etching was observed. The use of this arrangement with an injection of between about 15 to about 50 sccm ammonia resulted in a substantial reduction of the levels of oxygen and carbon incorporated into the film product.
Further process runs were carried out varying the power output from the plasma generator. At a power output of about 800 W the level of carbon incorporated into the film was reduced to levels approaching the actual detection limit of secondary ion mass spectrometry (SIMS). Oxygen levels have been reduced to those observed using MOCVD wherein oxygen is effectively removed as an impurity of concern. Such a reduction in the levels of oxygen and carbon in RPCVD produced films has hitherto not been shown.
It is envisaged that the method may also include the step of controlling the power of the plasma generator to be between about 500 W to about 5000 W from a single source. This range would be suitable with a growth pressure of 2-3 torr, a nitrogen plasma flow of 2000-3000 sccm and an ammonia flow of between about 15-1500 sccm.
Preferably, the power of the plasma generator is between about 100 watts to about 5000 watts, preferably about 500 to about 3000 W with a nitrogen plasma flow rate of 1000-3000 sccm extending to 100-20000 sccm in a commercial unit. A preferred metal organic reagent flow rate is 1200-2000 sccm which may extend to 100-10000 sccm in a commercial unit. A value for the plasma generator power of about 500 to 1000 W, 500 to 900 W, 500 to 800 W, 600 to 1000 W, 600 to 900 W, 600 to 800 W, 700 to 1000 W, 700 to 900 W and preferably about 800 W is preferred.
Such power levels may be independently coupled with an ammonia injection (in sccm) of between 15 to 1500 sccm. Ranges of between about 15 to 200, preferably 15 to 150, 15 to 100, 15 to 75, 15 to 60, 15 to 50, 15 to 40, 15 to 30, 20 to 150, 20 to 100, 20 to 75, 20 to 60, 20 to 50, 20 to 40, 20 to 35, 20 to 30, 25 to 150, 25 to 100, 25 to 75, 25 to 60, 25 to 50, 25 to 40, 25 to 30, including values of about 15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 85, 90, 95, 100, 120, 140, 160, 180 and 200 sccm have been found to be particularly useful in lowering carbon incorporation for relatively small growth chamber sizes, for example, of dimensions in the order of 7×2″. However, in moving to commercial scales higher powers and multiple plasma sites are envisaged as being useful. For growth chambers of larger dimensions of, for example, 56×2″ then ammonia injection flows of between about 200 to about 1500 sccm, inclusive of 200 to 1300, 200 to 1100, 200 to 1000, 200 to 900, 200 to 800, 200 to 700, 200 to 600, 200 to 500, 200 to 400, 300 to 1500, 300 to 1300, 300 to 1100, 300 to 1000, 300 to 900, 300 to 0.800, 300 to 700, 300 to 600, 300 to 500, 300 to 400, 400 to 1500, 400 to 1300, 400 to 1100, 400 to 1000, 400 to 900, 400 to 800, 400 to 700, 400 to 600, 400 to 500, 500 to 1500, 500 to 1300, 500 to 1100, 500 to 1000, 500 to 900, 500 to 800, 500 to 700, 500 to 600, 600 to 1500, 600 to 1300, 600 to 1100, 600 to 1000, 600 to 900, 600 to 800, 600 to 700, 700 to 1500, 700 to 1300, 700 to 1100, 700 to 1000, 700 to 900, 700 to 800, 800 to 1500, 800 to 1300, 800 to 1100, 800 to 1000, 800 to 900, 900 to 1500, 900 to 1300, 900 to 1100, 900 to 1000, 1000 to 1500, 1000 to 1300, 1000 to 1100, are appropriate.
It will be appreciated that higher power output from the plasma generator can be tolerated when a baffle or like device is employed whereas lower values will be preferred when there is an unimpeded flow path between the plasma inlet and the substrates to minimise etching.
FIGS. 8 and 9 show partial perspective sectional views of alternative embodiments of the apparatus for forming a film to that shown inFIG. 7. Like parts have been given like numbers betweenFIGS. 7 to 9 and it will be appreciated that the most important difference is in the placement and/or design of theadditional reagent inlet625, which is preferably an ammonia gas inlet.FIG. 8 also demonstrates an embodiment of theapparatus600 wherein thewaste outlet640 is actually provided within the hollow centre of the axis about which thesubstrate holder635 rotates.
FIG. 8 shows theammonia reagent inlet625 placed behind or collinear with the metalorganic reagent inlet620. This is a preferred embodiment as the location of theammonia reagent inlet625 is such that the introduced ammonia will be directed more or less to the centre of the substrates as they rotate on thesubstrate holder635. This ensures good delivery of the ammonia to the surface of the substrate whereas in the embodiment shown inFIG. 7 the side by side placement of theammonia reagent inlet625 and the metalorganic reagent inlet620 means that theammonia reagent inlet625 is somewhat offset from being directly vertically above the centre of the rotating substrates.
FIG. 9 represents a slightly different operational set up in that theammonia reagent inlet625 enters thegrowth chamber605 from the side and so is at more or less a right angle to the substantially vertical metalorganic reagent inlet620. The metalorganic reagent inlet620 has been cut away inFIG. 9 to better show the design of theammonia reagent inlet625. On approaching the metalorganic reagent inlet620 theammonia reagent inlet625 then has a bend such that its terminal portion finishes vertically and in a similar position to that shown inFIG. 8. The horizontal placement of theammonia reagent inlet625 may have operational advantages in use.
Although not shown inFIGS. 7 to 9 in one embodiment it may be preferred that theplasma inlet610, the metalorganic reagent inlet620 and theammonia reagent inlet625 all finish flush with the ceiling of thegrowth chamber605. To maintain the inlets close to the substrates the ceiling will therefore be located at a lower level than in a typical RPCVD apparatus. For example, in one embodiment the ceiling may be located approximately less than 30 cm, preferably less than 25 cm, more preferably less than 20 cm, still more preferably less than about 10 cm vertically above the location of the substrates. Values of about 5 cm and 7.5 cm may be useful with 3 cm to 4 cm as the lower end values.
As discussed earlier, the present inventors postulate that it is important for the minimisation of carbon and oxygen impurities in the final film to minimise reactions which occur in the gas phase above the substrates. Instead, it is preferable to encourage the key film forming reactions to occur on or as close to the actual substrate surface as possible. Directing the reactions to occur on the substrate surface may improve the scavenging of the oxygen and carbon impurities.
Thus the lowering of thegrowth chamber605 ceiling, with reagent inlets formed therein and having their openings flush with said ceiling, results in delivery of the reagents more quickly and effectively to the substrate surface.
Minimising deadspots and, particularly, optimising the flow of the plasma and reagents in relation to the substrates are considered inFIGS. 10 to 14 which are further embodiments on the inventive apparatus and method already discussed.
FIG. 10 shows a schematic representation of anRPCVD apparatus1000 for depositing a film on a substrate, according to one embodiment of the invention. Theapparatus1000 comprises agrowth chamber1005 inside which film growth will occur. Anexhaust1010 is provided at a lower extent of thegrowth chamber1005 for the removal of excess reagents and waste products.
Aplasma generator1015 is located externally to thegrowth chamber1005 which may be a high frequency generator acting upon nitrogen received from a nitrogen source (not shown). The nitrogen plasma thereby generated enters thegrowth chamber1005 atplasma inlet1020 which ends flush with the ceiling of thegrowth chamber1005 i.e. theplasma inlet1020 does not, to any significant extent, extend into the interior of thegrowth chamber1005. Theplasma inlet1020 may, if required, open into a baffle, shroud, impeller or the like to modify the flow path and energy of the plasma.
This is not an essential component and the need for such a device will depend upon the power of the radiofrequency generator. A suitable baffle may be as described in the applicant's prior PCT publication WO 2010/091470 which is hereby incorporated by way of reference in its entirety.
A metalorganic reagent source1025 supplies the metal organic reagent which, in a preferred embodiment, is trimethylgallium (TMG) or triethylgallium (TEG). The TMG or TEG enters thegrowth chamber1005 via metalorganic reagent inlet1030 which, in the embodiment shown, is located in a side wall of thegrowth chamber1005 and ends flush therewith i.e. the metalorganic reagent inlet1030 does not, to any significant extent, extend into the interior of thegrowth chamber1005.
A hydrogen-containing,additional reagent source1035 supplies the additional reagent which, in a preferred embodiment, is ammonia. The ammonia enters thegrowth chamber1005 viaadditional reagent inlet1040 which, in the embodiment shown, is located in a side wall of thegrowth chamber1005, beneath the location of the metalorganic reagent inlet1030, and ends flush therewith i.e. theadditional reagent inlet1040 does not, to any significant extent, extend into the interior of thegrowth chamber1005.
Under some conditions it may be preferable that theadditional reagent inlet1040 enters thegrowth chamber1005 through a side wall thereof at a height suitable to enable a flow of additional reagent entering therethrough to have a flow path passing over and substantially adjacent to a growth surface of thesubstrates1050. It is also beneficial that the positioning of theexhaust1010 is at an opposite end of thegrowth chamber1005 to theadditional reagent inlet1040 which further encourages a flow path of the additional reagent which passes over the surface of thesubstrates1050. Thus, the region of injection of the additional reagent and the provision of theexhaust1010 generally opposite that create an environment whereby the reagent is in constant contact with the growth surface of thesubstrates1050 and growing film.
The provision of the three plasma/reagent inlets all with ends flush with either the ceiling (the plasma inlet1020) or the side walls (the metalorganic reagent inlet1030 and the additional reagent inlet1040) or a combination thereof avoids the presence of ‘dead spots’ within thegrowth chamber1005. It is preferred that the Group IIIA reagent inlet is flush with the chamber ceiling. It is further preferred that the additional reagent inlet injects the ammonia, or other gas, through entry points, for example view ports, which physically surround the additional reagent inlet to thereby have these two reagents introduced into the growth chamber together to encourage mixing.
The presence of inlets which extend into thegrowth chamber1005 would result in adjacent regions therein where reagents can collect and be moved around in vortex like movements, due to the spinning ofsubstrate holder1045. Dead spots are considered to be unwanted volumes within the growth deposition chamber where there could be depletion or recirculation of gas which does not help with the growth of the film. This would encourage reaction pathways other than the desirable adduct formation outlined above and would result in TMG or TEG degradation with methyl radical production.
Thesubstrate holder1045 may be adapted to support asingle substrate1050 but it is preferred that it is of a design adapted to supportmultiple substrates1050. Suitably, thesubstrate holder1045 is rotatable.
The design of theapparatus1000 shown inFIG. 10 is one preferred embodiment in that theplasma inlet1020 and metalorganic reagent inlet1030 are substantially separated to thereby avoid any potential degradation of the TMG/TEG or other metal organic by the high energy plasma stream.
As discussed previously, the ceiling height of thegrowth chamber1005 is preferably lowered with respect to a standard RPCVD set up. Suitable heights have been set out previously. This helps minimise undesirable non-adduct forming reactions by minimising the space in which they can occur due to the placing of the plasma and reagent inlets close to thesubstrates1050.
Although not shown inFIG. 10, theapparatus1000 may further comprise one or more heating devices to heat theadditional reagent inlet1040 and/or the metalorganic reagent inlet1030 prior to the reagents entering thegrowth chamber1005. The heating devices may take the form of external heaters surrounding the transport members running between the relevant source and its inlet. Simple heating coils or heating tape placed around the piping may suffice. The heating of, particularly, the ammonia introduced above thesubstrate1050 surface means that it is introduced in an activated state, in anticipation of adduct formation, in the key reaction zone above thesubstrates1050.
Although not shown for the sake of simplicity, thegrowth chamber1005 may comprise one or more structures associated with theadditional reagent inlet1040 and/or the metalorganic reagent inlet1030 to promote mixing of said reagents immediately prior to their contacting the one ormore substrates1050. Particularly, it may be desirable to generate some turbulence in the flow path of the introduced additional reagent, preferably being ammonia. Since this reagent is introduced to generate a flow path just above thesubstrates1050 this ensures rapid and efficient mixing with the TMG or TEG to promote adduct formation prior to contact with the growth surface of the growing film.
The structures themselves may take the form of a baffle-like structure, vanes or any shape which promotes flow turbulence. They may be in direct contact with the corresponding reagent inlets or may be operatively associated with them such that the reagent must flow through the structure before passing close to thesubstrates1050.
FIG. 11 shows an alternative schematic representation of anRPCVD apparatus200 for depositing a film on a substrate to that shown inFIG. 10. Similar numbering to theapparatus1000 ofFIG. 10 has been maintained for like parts and it will be apparent that all of thegrowth chamber2005,exhaust2010,plasma generator2015,plasma inlet2020, metalorganic reagent source2025, metalorganic reagent inlet2030, additional reagent (ammonia)source2035 andadditional reagent inlet2040 are present to supply necessary reagents to thesubstrates2050 supported onsubstrate holder2045 which rotates aroundcentral pivot2055.
The key difference between the embodiments ofFIGS. 10 and 11 is that inFIG. 11 theplasma inlet2020, and associatedplasma generator2015, has been shifted to be closer to the side wall in which the metalorganic reagent inlet2030 andadditional reagent inlet2040 are provided. Although less preferred than the embodiment inFIG. 10 due to the close proximity of TEG or TMG reagent and plasma, this design ofapparatus2000 may still provide a significant improvement in the purity of the growing films over those grown in a typical RPCVD apparatus. There may be advantages in the directing of the plasma directly onto the region of thegrowth chamber205 in which the TEG or TMG and ammonia are encouraged to mix. All other components and reagents in theapparatus2000 may be as described forapparatus1000 inFIG. 10.
FIG. 12 shows a further alternative schematic representation of anRPCVD apparatus3000 for depositing a film on a substrate to that shown inFIG. 10. Once again the components of theapparatus3000 are substantially the same as those discussed in relation toFIGS. 10 and 11 and so will not be repeated here. The key differences between the embodiments ofFIGS. 12 and 10 are that, firstly, theadditional reagent source3035 and associatedadditional reagent inlet3040 are located on the ceiling of thegrowth chamber3005, rather than a side wall as inFIG. 10, and secondly, to ensure the introduction of additional reagent e.g. ammonia, into thegrowth chamber3005 only at a point close to the substrates, theadditional reagent inlet3040 is provided with anextended portion3060. Although not shown in the figures a temperature regulating means may be provided generally adjacentextended portion3060 to control the reagent temperature before contact with the growth surface.
The provision of theextended portion3060 does result in the potential generation of one or two “dead spots” as discussed before but this does not prevent the formation of an improved film product compared with many prior art RPCVD approaches. The design of theapparatus3000 still ensures that the ammonia, or other additional reagent, is only provided in close proximity to the growing film such that adduct formation and the production of methane, as opposed to methyl radicals, is promoted immediately adjacent said film.
FIG. 13 shows yet a further alternative schematic representation of anRPCVD apparatus4000 for depositing a film on a substrate to that shown inFIG. 10. Once again like numbering is employed for like components to those inFIG. 10. In this embodiment theplasma generator4015 and associatedplasma inlet4020 are found in a side, wall opposite that in which theadditional reagent inlet4040 is located. Further, instead of the representation of a metal organic reagent inlet as a single inlet it takes the form of an injector framework. The framework may be operative over the entire area of thegrowth chamber4005 occupied by thesubstrates4050 but, preferably, the injector framework will haveopen ports4065 and closedports4070. Theclosed ports4070 may be those adjacent theplasma inlet4020 to protect the TEG or TMG or other metal organic reagent from exposure to high energy plasma as it exits theplasma inlet4020.
FIG. 14 shows a partial perspective sectional view of anRPCVD apparatus5000 for forming a film, according to one embodiment of the present invention. For the sake of clarity not all components, such as the plasma generator and reagent sources, have been shown but rather only those components required to convey the key aspects of the apparatus.
Thegrowth chamber5005 is defined, in part, by theceiling5010 which, in relative terms, is not very distant from thesubstrates5035 to minimise the chamber mixing space. Aplasma inlet5015 opens into thegrowth chamber5005 through theceiling5010 to deliver a plasma, such as a nitrogen plasma. A metal organic reagent inlet takes the form of an injector framework5020 (details of the port openings not shown) while aadditional reagent inlet5025 opens into thegrowth chamber5005 through a side wall thereof at a point underneath the metalorganic injector framework5020 and a height such that a flow path of a additional reagent, preferably ammonia, is created just above the growth surface of thesubstrates5035 which are rotating with the movement of thesubstrate holder5030.
The design inFIG. 14 represents a further variation on the themes discussed in relation to the previous figures and achieves its advantages in much the same way. The embodiment shown inFIG. 14 does have the advantages of theplasma inlet5015 and TEG orTMG injector framework5020 being separated and, other than the relatively small size of theinjector framework5020 itself, minimal structural components being located within thegrowth chamber5005 above thesubstrates5035. By these means degradation of the TEG or TMG and the creation of “dead spots” which might promote the methyl radical production pathway are minimised.
It will be appreciated that, in one embodiment, the additional reagent may only need to supply hydrogen in a reactive form if the plasma is a nitrogen plasma and so can be used as the nitrogen source for adduct formation. This would result in the use of a hydrogen plasma generator and inlet along with a nitrogen plasma generator and inlet. Due to the possibility of arcing between these components it would be preferable to physically separate them as much as possible and so one may be located in the ceiling at one end of thegrowth chamber5005 and the other in a side wall at an opposite end of thegrowth chamber5005.
Although the discussion herein has been of rotation of the substrate holder it will be appreciated that it may be possible for the substrate holder and substrates to remain stationary while the plasma inlet and reagent inlets spin within the growth chamber. This will require a design whereby a rotatable connection mates with each of the plasma inlet, the metal organic reagent inlet and the additional reagent inlet which will both be operated in pulses timed to coincide with the rotation speed to ensure that each is only dispersed where required. Such a design would present greater challenges in operation over those disclosed in the figures which have the substrate holder rotate while the plasma inlet and reagent inlets remain stationary and, hence, is a less preferred approach.
The apparatus may also be adapted to allow additional individual rotations of each substrate relative to the substrate holder for further improvement of thin film growth uniformity.
Thus, from the various embodiments described above, it will be appreciated that the components of the inventive apparatus described herein may be arranged in a number of different ways while still achieving a reduced carbon and/or oxygen level in the growing films compared with that achieved by standard RPCVD approaches. However, all of the embodiments described share at least the feature of the additional reagent being introduced to the growth chamber in close if not immediate proximity to the Group IIIA reagent introduction point and, preferably, the substrates to promote formation of the adduct directly above the growing film surface. This approach has been found to greatly reduce the level of carbon and/or oxygen incorporation into the film. Further common features which assist in further reducing the level of carbon and/or oxygen incorporation include a low chamber ceiling height and therefore a correspondingly reduced chamber volume along with provision of reagent inlets which end flush with the ceiling and/or side walls to minimise dead spots and positioning of the exhaust to encourage a flow path of additional reagent over the substrate surface.
In one highly preferred embodiment of the present invention any one or more and most preferably all of the Group VA plasma inlet, the Group IIIA reagent inlet and the additional reagent inlet all end flush with the ceiling of the growth chamber, as previously discussed. However, if the Group IIIA reagent inlet and the additional reagent inlet are to extend into the chamber then, in one embodiment, it is useful if they extend into the chamber to be between about 1 to about 10 cm vertically above the substrates inclusive of between 2 to 9 cm, 3 to 6, cm and 4 to 5 cm.
The process of film formation as described in relation to any of the aspects herein may also include a doping step, which may be necessary for films to be employed in devices such as LEDs and solar cells. Preferably, the doping step is a p-type doping step. For p-type doping, the dopants could be Mg or Zn or other suitable elements. Suitable reagents that contain these elements, such as diethyl zinc (DEZn), bis(cyclopentadienyl)magnesium (Cp2Mg) can be selected from those known in the art for p-type doping. p-type doping is known in the art to be particularly challenging but it has been found that the present set of process conditions and apparatus features used to reduce carbon impurities in the growing film also allow for better p-type doping. Values obtained for p-type doping (Hall measurements) are: resistivity of 0.9 Ohm-cm, mobility of 2.7 cm̂2/Vs for a carrier concentration of 1.4E18 cm̂-3. For n-type doping, the dopants could be Si or oxygen or other suitable elements. Suitable reagents that contain these elements, such as silane, disilane, di-tert-butylsilane, oxygen can also be used for n-type doping.
It will be appreciated from the foregoing discussion that a number of other factors can be controlled to further contribute to the extent of the reduction of impurities achieved in the film product.
For example, the method may further include the step of controlling the temperature to be between about 400 to about 1200° C., preferably between about 500 to about 1000° C. (inclusive of a temperature of about 500° C., 600° C., 700° C., 800° C., 900° C. or 1000° C.), more preferably between about 500 to about 850° C. This is a relatively low temperature range in comparison to typical MOCVD and even many RPCVD approaches. The lower temperatures favour adduct formation over TMG thermal degradation and so reduce methyl radical reactions at the film surface.
The method may further include the step of promoting the mixing of the metal organic reagent and the additional reagent adjacent the one or more substrates using a flow perturbation device. Once again, the mixing step is to promote immediate formation of the adduct in the vicinity of the film/substrate surface.
It has been found that the power of the plasma generator has an effect on carbon incorporation into the thin film and so the method may also include the step of controlling the power of the plasma generator to be between about 400 W to about 5000 W from a single source. Preferably, the power of the plasma generator is between about 500 to about 3000 W, 500 to 2750 W, 500 to 2500 W, 500 to 1000 W, 500 to 900 W, 500 to 800 W, 600 to 1000 W, 600 to 900 W, 600 to 800 W, 700 to 1000 W, 700 to 900 W and preferably about 800 W is preferred. A value of about 800 W has been found to be particularly useful in lowering carbon incorporation, however, in moving to commercial scales higher powers and multiple plasma sites are envisaged as being useful such as about 2500 W.
Although not wishing to be bound by any particular theory, it is postulated that the surprising results achieved may be as a result of one or more of the following processes. Firstly, it is postulated that the injection of ammonia provides additional available nitrogen to the system and this acts as both a reagent in film formation and also as a scavenger for oxygen and/or carbon. Secondly, it is theorized that the improvement in lowering carbon incorporation into the film when the power output of the plasma generator is increased could be due to the carbon atoms being actively removed in favour of nitrogen. Thirdly, and as alluded to previously, a proposed mechanism for removal of carbon from the system is that the ammonia assists in the formation of an adduct with the trimethylgallium, i.e. initially (CH3)3Ga:NH3, which subsequently releases a methane molecule. The methane is not incorporated into the film as readily as a CH3radical may be. Subsequent decomposition of the adduct releases further methane until all of the carbon of the trimethylgallium has been removed as methane and only GaN is left. It is believed formation of this adduct and subsequent decompositions are occurring at the surface of the substrate. Finally, it is possible that the improved lowering of carbon and oxygen incorporation upon removing the baffle from the system is a consequence of the energy from the UV light emitted from the plasma generator/plasma chamber contacting the growing film ejecting carbon and/or oxygen in favour of nitrogen. In reality, it is possible that all of these mechanisms may play at least some role in providing the results achieved.
In a third aspect the invention resides in a method of forming a thin film having a carbon content of less than about 5E+17 atom/cm3, on a substrate by RPCVD including the steps of:
- (a) introducing a Group VA plasma through a Group VA plasma inlet into a first deposition zone of a growth chamber wherein a direct flow path is provided between the Group VA plasma inlet and a substrate located in the first deposition zone;
- (b) introducing a Group IIIA reagent through a Group IIIA reagent inlet into a second deposition zone of the growth chamber, the second deposition zone being substantially isolated from the first deposition zone;
- (c) introducing an additional reagent selected from the group consisting of ammonia, hydrazine, di-methyl hydrazine and hydrogen plasma through an additional reagent inlet into the second deposition zone such that the additional reagent and the Group IIIA reagent mix prior to deposition;
- (d) moving the substrate between the first and second deposition zones,
to thereby form a thin film on the substrate having a carbon content of less than about 5E+17 atom/cm3.
Preferably, the carbon impurity content is less than about 3E+17 atom/cm3, even more preferably less than about 2E+17 atom/cm3, yet more preferably less than or about 1E+17 atom/cm3. A lower limit may be considered to be about the SIMS detection limit for carbon impurities in such films.
In one embodiment, the thin film also has an oxygen impurity content of less than about 8E+17 atom/cm3, even more preferably less than about 6E+17 atom/cm3, yet more preferably less than about 4E+17 atom/cm3, still more preferably less than about 2E+17 atom/cm3, or even less than or about 1E+17 atom/cm3. A lower limit may be considered to be about the SIMS detection limit for oxygen impurities in such films.
The statements made above in relation to the second aspect apply equally well to the third aspect.
In a fourth aspect the invention resides in a film made by the method of the second or third aspects. Such films will have demonstrably lower levels of oxygen and/or carbon incorporated into their structure in comparison to similar films made by standard RPCVD approaches. In one embodiment, films produced by the method of the present invention may have a carbon content of less than about 10E+16 atoms/cm3. Values of 3E+16 atoms/cm3have been attained and it is believed that values of less than 1E+16 atoms/cm3are attainable with process optimisation.
In a fifth aspect the invention resides in use of a film of the fourth aspect in a semiconductor device.
The examples set out in further detail the process runs using the apparatus of the invention and the results thereby obtained. In the examples nitrogen was used as the Group VA plasma and trimethylgallium as the Group IIIA reagent.
ExamplesProcess Runs with BaffleAn apparatus essentially as set out inFIGS. 6 and 7 was used employing a stainless steel style shower head baffle located below the plasma inlet. The power of the plasma generator was between 500 W to 600 W and a growth temperature of 700° C. was employed. The films were grown onto a GaN template. An initial control run was carried out using a nitrogen plasma and trimethylgallium (TMG) as the organometallic reagent but without the injection of any ammonia. This produced a film as would be expected when made via the apparatus ofFIG. 2 i.e. with standard levels of oxygen and carbon impurities.
A second run was then carried out under essentially similar conditions but with an injection of a 15 sccm ammonia flow into the second deposition zone (the organometallic reagent deposition zone). The ammonia was injected at the same time as injection of the TMG so that the two mixed together prior to deposition. This produced a film with a substantial reduction in the levels of both oxygen and carbon. Specifically, compared with the first run without the injection of ammonia, the level of carbon decreased from about 6E+20 atom per cubic centimetre (atom/cc) to about 3E+20 atom/cc while the level of oxygen decreased from about 3E+20 atom/cc to about 1E+17 atom/cc.
The figure of 1E+17 atom/cc for the oxygen level represents an extremely surprising result in that it ceases, for practical purposes, to be a problematic impurity at that level and the result is comparable to that observed using MOCVD. Although the additional hydrogen provided by the ammonia may be expected to provide some benefit in reducing carbon and oxygen impurities it could not have been predicted, based upon accepted wisdom in the semiconductor field, that such a small injection of ammonia could result in such a large reduction in oxygen and carbon impurities.
This experiment was repeated using different flow rates of ammonia. The results of these process runs, in terms of carbon incorporation into the film, are shown inFIG. 15 wherein the diamond icons (labelled ‘Short Jar A RPCVD’) indicate the levels of carbon in the film.
Process Runs without Baffle
A number of runs were then carried out using the same apparatus and conditions as described above but with the shower head baffle removed. Thus, a direct flow path between the plasma generator, plasma inlet and substrates was established in the first deposition zone.
The distance between the plasma inlet and the substrates was less than 20 cm and no plasma etching was observed. It is postulated this may be due to the relatively low (500-600 W) power output from the plasma generator employed while still providing enough energy to activate the nitrogen.
The results of this run are indicated onFIG. 15 as the smaller square icons (labelled as ‘Short Jar B (no plasma showerhead) RPCVD’). It was observed that, under the same conditions as process runs using the shower head baffle and using the same amounts of ammonia, the levels of carbon incorporated into the film was greatly reduced.
A further process run was carried out under identical conditions (30 sccm ammonia) but with the power of the plasma generator increased to 800 W. Once again, surprisingly, the resulting film was not etched to a significant degree. More surprising, however, was that the level of carbon incorporated into the film was about 1.7E+17 atom/cc. This result is indicated onFIG. 15 as a single point larger square icon. The results for the levels of carbon found in films produced by all of the process runs described above, and as indicated graphically inFIG. 15, are shown in table 2.
| TABLE 2 |
|
| Levels of carbon incorporated into films grown under a variety of |
| conditions |
| NH3 | | | | RPCVD B | |
| flow | Standard | | | higher RF | Standard |
| (sccm) | RPCVD | RPCVD A | RPCVD B | power (800 W) | MOCVD |
|
| 0 | 4.10E+19 | 6.00E+20 | 1.05E+19 | | 2.83E+16 |
| 15 | | 2.00E+20 | | | 2.83E+16 |
| 30 | | 1.40E+20 | 1.67E+18 | 1.70E+17 | 2.83E+16 |
| 50 | | 8.50E+19 | | | 2.83E+16 |
| 70 | | 7.00E+19 | | | 2.83E+16 |
| 100 | | | 9.40E+17 | | 2.83E+16 |
| 150 | | 2.44E+18 | | | 2.83E+16 |
|
It will be appreciated that there will be an upper limit to the plasma generator power which can be employed without etching occurring. If this point is reached and further increases in plasma generator power are desirable then a baffle may once again be placed between the plasma inlet and the substrates.
Further runs designed to optimise the process have resulted in levels of 3E+16 atom/cc of carbon and 3E+16 atom/cc of oxygen in the grown GaN films. It may assist the appreciation of how low these levels are by considering that the SIMS detection limits for carbon are between about 1-2E+16 and between about 1-3E+16 for oxygen. At the above quoted optimal results the present films are approaching the detection limits for carbon and oxygen. SIMS is one of the most sensitive surface analysis techniques available, being able to detect elements present in the parts per billion range, and is the accepted standard for analysis in this field. The low levels of carbon and oxygen impurities achieved by use of the present apparatus and method are comparable with those observed in GaN templates and have never been seen previously using RPCVD.
The data in table 2 also indicates that ammonia actively takes part in GaN formation and carbon removal rather than just scavenging carbon and/or oxygen.
FIG. 16 is a SIMS graphical analysis of the typical impurities found in a film produced by a method and apparatus of the invention as described and grown on a GaN template. Films produced by RPCVD would usually not be comparable with the purity levels achieved in a template, however, in the present instance when formed under the optimal conditions described above it is seen that the films are essentially of equal quality.
The first 0-0.5 μm of the depth profile (indicated on the x axis) represents a film produced by the present apparatus and method while the 0.5-2.7 μm component represents the underlying GaN template produced by an MOCVD process. It is clear that the levels of the various impurities, particularly carbon and oxygen, are similar. The spikes observed in the traces are representative of the interfaces between layers or changes in growth conditions and not an increase in impurity levels.
Triethylgallium ExperimentsA further series of experiments were performed using triethylgallium (TEG) as the Group IIIA reagent and investigating different ammonia injection rates. The process conditions used for these experiments are set out below in table 3. Also noteworthy is that the plasma inlet, TEG inlet and ammonia inlets all ended flush with the growth chamber ceiling to reduce recirculation of gases i.e. deadspot effects. The ceiling was fixed at a height approximately 5.0-7.5 cm above the substrates and the speed of rotation of the substrate holder was 1200 rpm.
| TABLE 3 |
|
| Process conditions for growth runs using TEG |
| | | | N2 | | | | | RF |
| Run | Time | | MO | PL | H2 | NH3 | Growth | Press | power |
| # | (min) | TEG | inj. | (2 + 3) | Shrd | 0.25 L | Temp | (Torr) | (W) |
|
| 1380 | 120 | 120 | 1600 | 2500 | 800 | 30 | 720 | 3.5 | 2500 |
| uGaN |
| 1386 | 120 | 120 | 1600 | 2500 | 800 | 0 | 720 | 3.5 | 2500 |
| uGaN |
| 1388 | 120 | 120 | 1600 | 2500 | 800 | 100 | 720 | 3.5 | 2500 |
| uGaN |
|
| N2 PL (2 + 3) is the nitrogen flow to the plasma inlet. |
| H2 Shrd is the hydrogen flow to the chamber through an outer shroud. |
| MO inj. is the metal organic injector flow of hydrogen which carries the metal organic reagent. |
| NH3 0.25 L is the ammonia flow into the growth chamber in sccm. |
The results of the experiments set out in table 3, in terms of the carbon and oxygen impurity levels in the grown films, are shown graphically in the SIMS data inFIGS. 17, for carbon, and18, for oxygen. The grown films were approximately 1 um thick and were grown on top of GaN MOCVD templates. The thickness of the as grown films means that the SIMS data only needs to be viewed from the left side of the x-axis on the data plots up until the 1 um depth point. The spike at this region is due to the interface between the films and the template.
It can be seen fromFIG. 17, firstly, that the injection of increasing amounts of ammonia causes a very significant decrease in the levels of carbon as an impurity in the grown films when compared with the baseline levels with no ammonia injected (run no. 1386). The injection of 30 sccm of ammonia brings the level of carbon impurities down to under 1017atoms/cm3while, with an injection of 100 sccm of ammonia the level of carbon impurities is seen to actually be reduced to more or less correspond to those in the MOCVD generated GaN templates, a result hitherto unseen when employing RPCVD growth of GaN films.
FIG. 18 shows that the oxygen impurity levels in the grown films were on a par with those observed in the MOCVD grown. GaN template under all conditions.
It will be appreciated from all of the foregoing that the use of separate deposition zones in an RPCVD arrangement can be useful in reducing impurity incorporation into films, such as GaN films, however these impurities will still be found in the produced films in noticeable quantities. The use of a growth chamber arrangement whereby the plasma inlet and/or organometallic reagent inlet are between about 1 to about 30 cm vertically above a growth surface of the substrates. However, the introduction of relatively small quantities of ammonia into the second deposition zone, simultaneously with the organometallic reagent, has been shown to dramatically reduce the level of oxygen and, particularly, carbon in the film. Under similar conditions but with the removal of any impediment to direct flow between the plasma generator and/or plasma inlet and the substrates results in an extremely surprising reduction in the incorporation of, particularly, carbon into the film. As a further level of control it has been shown that increasing the power of the plasma generator output can further lower the level of carbon impurities found in the film.
Throughout the specification the aim has been to describe the preferred embodiments of the invention without limiting the invention to any one embodiment or specific collection of features. It will therefore be appreciated by those of skill in the art that, in light of the instant disclosure, various modifications and changes can be made in the particular embodiments exemplified without departing from the scope of the present invention.