JP4744141B2 - Manufacturing method of CMOS device by implantation of N and P type cluster ions and anions - Google Patents

Description

RELATED APPLICATIONS This application claims priority to and benefits from US Provisional Application Nos. 60 / 392,271 and 60 / 391,847, both filed June 26, 2002. This patent application is also owned by the present applicant, filed Sep. 16, 2002, and currently filed in U.S. Patent Application No. 10 / 244,617 and U.S. Patent Application No. filed on Sep. 20, 2002. Claim priority to 10 / 251,491.

  The present invention relates to an ion implantation system and a semiconductor manufacturing method for implanting an ion beam formed from a cluster of N-type dopant cluster ions and a negatively charged cluster ion beam.

  In the manufacture of a semiconductor device, impurities are partially introduced into a semiconductor substrate in order to form a doped region. The impurity element is selected to appropriately combine with the semiconductor material to produce an electrical carrier and change the conductivity of the semiconductor material. The electrical carrier can be an electron (generated by an N-type dopant) or a hole (generated by a P-type dopant). The conductivity of the resulting region is determined by the concentration of the introduced dopant impurity. Many such N-type and P-type impurity regions should be created to form transistor structures, isolation structures, and other such electronic structures that collectively function as semiconductor elements.

A conventional method for introducing a dopant into a semiconductor substrate is by ion implantation. In ion implantation, a feed material containing a desired element is introduced into an ion source, energy is introduced to ionize the feed material, and a dopant element (eg, elements ⁷⁵ As, ¹¹ B, ¹¹⁵ In, ³¹ P, Alternatively, ions containing ¹²¹ Sb) are created. In general, an accelerating electric field is provided to extract and accelerate positively charged ions, thus creating an ion beam. The ion beam is then directed at the semiconductor substrate using mass spectrometry to select the species to be extracted, as is known in the art. The accelerating electric field provides the kinetic energy of the ions so that they can penetrate into the target. The depth of penetration into the target is determined by the energy and mass of the ions, and the higher the energy of the ions and / or the smaller the mass, the greater the velocity, and the deeper the penetration into the target. Ion implantation systems are important in implantation processes such as ion beam energy, ion beam mass, ion beam flow (charge / unit time), and ion dose at the target (total number of ions per unit area penetrating into the target). Configured to carefully control variables. In addition, the beam angle spread (variation in the angle at which ions strike the substrate) and the spatial uniformity and extent of the beam should also be controlled to maintain semiconductor device yield.

  It has recently been recognized that anion implantation has advantages over cation implantation (eg, DC Jacobson, Konstantin Bourdel, HJ Gossman, M. Sosnowski, MA Albano, V). Babaram, J. M. Poate, Aditya Agarwal, Alex Perel, and Tom Horsky, “An Alternative Method of Ultralow Energy Ion Implantation of Decaborane”, 13th International Conference on Ion Implantation Technology, Proceedings of the Alps, Austria Bakk, 2000, N. Kishimoto et al., “High Current Anion Implanter and Its Application for Nanocrystal Fabrication in Insulators”, 12th International Conference on Ion Implantation Technology, Proceedings of Kyoto, Japan, 6 Moon 2–26, 1998, (1999) 342-345, N. Tsubouchi et al., “Beam Characterization of Mass-Separated Low Energy Positive and Negative Ion Deposition Devices”, 12th International Conference on Ion Implantation Technology, IEEE Proceedings, Kyoto, Japan, June 22-26, 1998, (1999) 350-353, and Junzo Ishikawa et al., “Anion Implantation Technology”, Nuclear Instrumentation and Methods in Physics Research, B96 (1995) Year) see pages 7-12). One very important advantage of negative ion implantation is that it reduces the surface charge induced by ion implantation of VLSI devices in CMOS fabrication. In general, high current injections of cations (on the order of 1 mA or more) result in positive potentials that can easily exceed the gate oxide damage threshold on the gate oxide and other components of the semiconductor device. Generate. When the cations collide with the surface of the semiconductor device, not only a net positive charge is deposited, but also secondary electrons are released at the same time, and the charging effect is doubled. Therefore, equipment suppliers of ion implantation systems have already developed state-of-the-art charge control devices, so-called electron flood guns, for introducing low energy electrons into the positively charged ion beam and onto the device wafer surface during the implantation process. Yes. Such an electronic flood system introduces additional variables into the manufacturing process and cannot completely eliminate yield losses due to surface charge. As semiconductor devices become smaller and smaller, transistor operating voltage and gate oxide thickness also decrease, resulting in a lower damage threshold in semiconductor device manufacturing, further reducing yield. Thus, anion implantation provides a significant yield improvement over conventional cation implantation for potentially many advanced processes. Unfortunately, this technique is not yet commercially available and in fact, to the best of the inventors' knowledge, anion implantation has not been used in integrated circuit fabrication, even in research and development.

Prior art anion sources have relied on so-called negative affinity sputter targets. A heavy inert gas such as xenon is supplied to a plasma source that generates Xe ⁺ ions. As Xe ⁺ ions are generated, they are attracted to a negatively biased sputter target coated with cesium vapor or other suitable alkaline material. Strong Xe ⁺ ions sputter neutral target atoms, some of which pick up electrons while leaving the target surface due to the negative electron affinity of the cesium coating. When the target ions are negatively charged, they are repelled from the target and can be collected and focused from the ion source by the electrostatic ion optical element to become an anion beam. Although it is possible to produce semiconductor dopant ions such as boron by this method, the ion current tends to be low, the beam radiation tends to be large, and the alkali metal is very The presence of cesium vapor presents an almost unacceptable risk to wafer yield because it is considered a serious contaminant. Therefore, there is a need for an anion source technology that is more feasible at the commercial level.

  Of particular interest in the semiconductor manufacturing process is the formation of a PN junction in a semiconductor substrate. This requires the formation of adjacent regions of N-type and P-type doping. One common example of junction formation is the implantation of N-type dopants into a semiconductor region that already contains a uniform distribution of P-type dopants. In such cases, an important parameter is the junction depth, which is defined as the depth from the semiconductor surface where N-type and P-type dopants have equal concentrations. This junction depth depends mainly on the mass, energy and dose of the implanted dopant.

  An important aspect of modern semiconductor technology is the continuous evolution towards smaller and faster devices. This process is called scaling. Scaling is driven by continuous improvement advances to lithographic processing, and allows for the definition of further miniaturization of features in semiconductor substrates including integrated circuits. A generally accepted scaling theory has been developed to assist chip manufacturers in the appropriate sizing of all aspects of semiconductor device design simultaneously, ie at each technology or scaling node. The greatest effect of scaling on the ion implantation process is the scaling of the junction depth, which requires increasingly shallow coupling as device dimensions are reduced. This requirement for increasingly shallow junctions as a measure of integrated circuit technology translates to the requirement that ion implantation energy should be reduced at each scaling step. In recent years, the implantation energy required for many critical implants is that traditional ion implantation systems, originally developed to produce a much higher energy beam, are not effective in providing the necessary implants. It has decreased to. These extremely shallow junctions are referred to as “ultra shallow junctions” or USJs.

The limitations of conventional ion implantation systems with low beam energy are most evident in the extraction of ions from the ion source and subsequent transport through the implanter beam line. Ion extraction is governed by the Child-Langmuir relationship, where the extracted beam current density is proportional to the 3/2 power of the extraction voltage (ie, the beam energy at the time of extraction). FIG. 1 is a graph showing the extracted voltage with respect to the maximum extracted arsenic beam current. For simplicity, it is assumed that only ⁷⁵ As ⁺ ions are present in the extraction beam. FIG. 1 shows that the extraction current drops rapidly as the energy decreases. In conventional ion implanters, this “extraction limited” operating regime is seen with less than about 10 keV energy. Similar constraints occur in the transfer of low energy beams. The smaller the energy, the lower the ion beam moves, and thus the ions approach each other for a given value of beam current, i.e. the ion density increases. This can be seen in the relationship J = nev, where J is the ion beam current density in mA / cm ² units, n is the ion density in cm ⁻³ units, and e is the charge (= 6.02 × 10 6 ^-19 coulombs) and v are average ion velocities in cm / s. Since the electrostatic force between ions is inversely proportional to the square of the distance between them, this repulsive force is much stronger at low energies, thus dispersing the ion beam. This phenomenon is called “beam blow-up”. Low energy electrons present in the implanter beam line are trapped by the positively charged ion beam and tend to help compensate for the space-charge blow-up during transport, but the blow-up still occurs and is tied up This is most noticeable when there is an electrostatic condenser lens that has a tendency to peel off the compensation electrons that are loose and have high mobility. Low energy beam transfer can be difficult with large atoms such as arsenic (75 amu), because for a given ion energy, the ion velocity is slower than with light atoms. In the case of boron, which is a P-type dopant, there are severe difficulties in extraction and transfer. Boron transfer is an extremely low implantation energy (eg, less than 1 keV) required by some state-of-the-art processes, and most of the ions extracted and transferred from a typical BF ₃ source plasma are desirable ions. ^This is made difficult by the fact that it is not ¹¹ B ⁺ but rather ion fragments such as ¹⁹ F ⁺ and ⁴⁹ BF ₂ ⁺ which serve to increase the charge density and average mass of the extracted ion beam. Considering the future of VLSI semiconductor manufacturing, these problems in transferring significant currents of low energy As and B combine to make USJ formation very difficult.

One way to benefit from the above Child-Langmuir equation is to increase the mass of the ions by ionizing molecules that contain the associated dopant, rather than the dopant atom, for example, as shown in FIG. 1a. . In this way, although the kinetic energy of the molecule is greater during transport, the molecule decomposes into constituent atoms as it enters the substrate and shares the molecular energy among the individual atoms according to the mass distribution, resulting in dopants. The atom implantation energy is much smaller than its original transport kinetic energy. Consider the dopant atom “X” linked to the radical “Y” (for the sake of discussion, ignore the question of whether “Y” affects the device formation process). If ions XY ⁺ are implanted instead of X ⁺ , XY ⁺ should be extracted and transported with greater energy increased by a factor equal to {(XY mass) / (X mass)}. This keeps the speed of X the same. The space-charge effect described in the above Child-Langmuir equation is superlinear with respect to ion energy, and therefore the maximum transportable ion current increases. Traditionally, the use of polyatomic molecules to address the problem of low energy implantation is known in the art. A common example is to use BF ₂ ⁺ molecular ions for low energy boron implantation instead of B ⁺ . By this process, the FB ₃ supply gas is decomposed into BF ₂ ⁺ for injection. In this way, the ion mass is increased to 49 AMU, increasing the extraction and transfer energy by almost a factor of five (ie 49/11) over the use of a single boron atom. However, when boron energy is implanted, it decreases by the same factor of (49/11). This technique shows that the current density in the beam does not decrease because there is only one boron atom per unit charge in the beam. In addition, this process also implants fluorine atoms into the semiconductor substrate along with boron, which is known to adversely affect semiconductor devices.

Also, Jacobson et al. “Alternative method of ultra-low energy ion implantation called decaborane”, IEEE Proceedings of the 13th International Conference on Ion Implantation Technology, Alpsbach, Austria, 300-303 (2000), and Yamada “ Use of decaborane as a polyatomic molecule for ion implantation as reported in "Applications of Gas Cluster Ion Beams for Material Processing", Materials Chemistry and Engineering, A217 / 218, pp. 82-88 (1996) Molecular ion research has been conducted. In this case, the implanted particles are ions of the decaborane molecule B ₁₀ H ₁₄ containing 10 boron atoms and are therefore “clusters” of boron atoms. This technique not only increases the mass of the ions, but the decaborane ion B ₁₀ H _x ⁺ has 10 boron atoms per unit charge, so that the injected dose rate is substantially reduced for a given ion current. Increase. This is a very promising technique for the formation of USJ P-type metal oxide semiconductor (PMOS) transistors in silicon, and generally the implantation of ultra-low energy boron. Reducing the current carried in the ion beam significantly (1/10 for decaborane ions) not only reduces the space charge effect of the beam, but also reduces the wafer charging effect. It is known that charging of wafers, especially gate oxide, by cation beam bombardment reduces device yield by damaging sensitive gate isolation, so the reduction of such current by using cluster ion beams is not possible. It is very attractive for USJ device manufacturing, which is increasingly required to adapt to very low gate threshold voltages. It should be noted that in these two examples of P-type molecular implantation, ions are generated by simple ionization of the feed material, not by agglomeration into the feed material cluster. It should also be noted that no similar technology has been developed to date to produce N-type molecular dopant ions. The future success of complementary metal oxide semiconductor (CMOS) processing will depend on the commercialization of viable N-type and P-type polyatomic implantation technologies. That is, there is a need to solve two distinct problems facing the semiconductor manufacturing industry today: wafer charging and low productivity in low energy ion implantation.

  Ion implanters are conventionally divided into three basic types: high current, medium current, and high energy implanters. Cluster beams are useful for high current and medium current injection processes. More specifically, today's high current injection devices are primarily used for the formation of low energy, high dose regions of transistors such as drain structures and doping of polysilicon gates. They are generally batch implanters, i.e., process many wafers mounted on a rotating disk while the ion beam remains fixed. High current beamlines tend to be simple, incorporate a large acceptance of the ion beam, and at low energy and high current, the beam at the substrate tends to be large and have a large angular divergence. Medium current implanters typically incorporate a continuous (one wafer at a time) processing chamber that provides a high tilt capability (eg, up to 60 ° from the substrate normal). The ion beam is typically electromagnetically scanned across the wafer in an orthogonal direction to ensure dose uniformity. In order to meet commercial implant dose uniformity and reproducibility requirements, which are typically only a few percent of dispersion, the ion beam has excellent angular and spatial uniformity (eg, the angle of the beam on the wafer below 2 °). Uniformity). Because of these requirements, medium current beamlines are designed to provide superior beam control at the cost of limited acceptance. That is, the efficiency of ion transmission through the implanter is limited by the amount of ion beam radiation. Currently, the generation of high current (about 1 mA) ion beams at low energy (<10 keV) is the wafer throughput for several low energy implants (eg, in the generation of source and drain structures in state-of-the-art CMOS processing). There are many problems with continuous infusion devices so that is unacceptably low. For batch implanters (which process many wafers mounted on a rotating disk), similar transfer problems exist with low beam energy of less than 5 keV per ion.

  Although it is possible to design beam transport optics with almost no aberrations, the ion beam characteristics (spatial range, spatial uniformity, angular divergence, and angular uniformity) are still primarily related to the radiation characteristics of the ion source itself ( That is, it is determined by the beam characteristics at the time of ion extraction that determines the extent to which the beam can be collected and controlled when the implanter optical element is emitted from the ion source. The use of a cluster beam rather than a monomer beam can greatly increase the radiation of the ion beam by increasing the beam transport energy and reducing the current carried by the beam. That is, in addition to increasing the effective dose rate and throughput, the cluster ion and cluster ion source technology of semiconductor manufacturing to provide a more focused, more collimated and more tightly controlled ion beam on the target. There is a need.

US Provisional Patent Application No. 60 / 392,271 US Provisional Patent Application No. 60 / 391,847 US patent application Ser. No. 10 / 244,617 US patent application Ser. No. 10 / 251,491 D. C. Jacobson, Konstantin Bourdelle, HJ. Gossman, M.M. Sosnowski, M.M. A. Albano, V.M. Babaram, J. et al. M.M. Poate, Adiya Agarwal, Alex Perel, and Tom Horsky, “An Alternative Method of Ultra-Low Energy Ion Implantation of Decaborane”, Proceedings of the 13th International Conference on Ion Implantation Technology, Alpsbach, Austria, 2000 N. Kishimoto et al. “High Current Anion Implanter and its Application for Nanocrystal Fabrication in Insulators”, Proceedings of the 12th International Conference on Ion Implantation Technology, Kyoto, Japan, June 22-26, 1998 Year (1999) N. Tsubouchi et al. “Beam Characterization of Mass-Separated Low Energy Positive and Negative Ion Deposition Devices”, Proceedings of the 12th International Conference on Ion Implantation Technology, Kyoto, Japan, June 22-26, 1998, ( (1999) Junzo Ishikawa et al. “Anion Implantation Technology”, Nuclear Instrumentation and Methods in Physics Research, B96 (1995) Yamada “Application of Gas Cluster Ion Beam for Material Processing”, Materials Chemistry and Engineering, A217 / 218, pp. 82-88 (1996) "VLSI Technology" by Sze, McGraw-Hill, pp. 253-254 (1983) J. et al. F. Ziegler, “Ion Implantation Technology Handbook”, North Holland, 455-499 (1992) Y. -K. KIM, HWANG, N.K. M.M. WEINBERGER, M.M. A. ALI and M.M. E. RUDD, “J.CHEM.PHYS.”, 106, 1026 (1997) W. HWANG, Y.M. -K. KIM and M.M. E. RUDD, “J.CHEM.PHYS.”, 104, 2956 (1996). N. DJURIC, D.C. BELIC, M.M. KUREPA, J.A. U. MACK, J.M. ROTHLEITNER and T.W. D. MARK, “Summary”, 12th International Conference on Physics of Atomic and Electrical Collisions, S.C. Edition of DATZ (GATLINBURG, 1981), p. 384 M.M. V. V. S. RAO and S.R. K. SRIVASTAVA, “J. PHYS.”, B25, 2175 (1992)

  An object of the present invention is to provide a method of manufacturing a semiconductor device capable of forming an ultra-shallow impurity doped region of N-type (ie, acceptor) conductivity in a semiconductor substrate, and to perform it with high productivity. That is.

Another object of the invention is an ion implantation in which negatively charged ions of decaborane (B ₁₀ H ₁₄ ) are generated as B ₁₀ H _x ^{− and} are implanted into a semiconductor substrate to form a pn junction. System and method.

Another object of the present invention is the form of As _n H _x ⁺ _where n = 3 or 4 and 0 ≦ x ≦ n + 2 for N-type clusters, and B ₁₀ H _x ⁺ or B ₁₀ H _x for P-type clusters. ^- use the N or P-type clusters of N or P type (i.e., acceptor or donor) to form a super shallow impurity doped region of the is to provide a method of manufacturing a semiconductor device.

Still another object of the present invention is to provide a method for implanting arsenic cluster ions in the form of As ₃ H _x ⁺ and As ₄ H _x ⁺ , which can form an N-conductivity type ultra-shallow implantation region in a semiconductor substrate. It is to be.

Yet another object of the present invention is to ionize the PH ₃ feed gas and then inject this phosphorus cluster into the semiconductor substrate to achieve N-type doping, so that 0 ≦ x at n = 2, 3, or 4. It is to provide a method of making phosphorus cluster ions in the form of P _n H _x ⁺ _where ≦ 6.

Yet another object of the present invention is to ionize B ₂ H ₆ feed gas and then implant boron clusters into the semiconductor substrate to achieve P-type doping, thereby satisfying 0 ≦ x at n = 2, 3, or 4. It is to provide a method for making boron cluster ions in the form of B _n H _x ⁺ ≦ 6.

  Yet another object of the present invention is to provide an ion implantation system for manufacturing a semiconductor device designed to form an ultra-shallow impurity doped region having N or P-type conductivity in a semiconductor substrate using cluster ions. Is to provide.

  According to one aspect of the present invention, preparing a supply of dopant atoms or molecules into the ionization chamber, combining the dopant atoms or molecules into a cluster containing a plurality of dopant atoms, and ionizing the dopant cluster to form a dopant A method of implanting cluster ions, comprising: a step of forming cluster ions; a step of extracting dopant cluster ions and accelerating with an electric field; a step of mass analyzing an ion beam; and a step of implanting dopant cluster ions into a semiconductor substrate. Is provided.

Another object of the present invention is to implant a cluster of n dopant atoms (n = 4 in the case of As ₄ H _x ⁺ ) rather than implanting one atom at a time, thereby reducing the energy of the low energy ion beam. It is to provide a method that allows a semiconductor device manufacturer to improve the difficulty in extracting. The cluster ion implantation method is equivalent to low energy single atom implantation because each atom of the cluster is implanted with an energy of E / n. Thus, the implanter is operated with an extraction voltage that is n times higher than the required implant energy, which allows higher ion beam currents, especially at the low implant energy required for USJ formation. Considering the ion extraction step, the relative improvement enabled by cluster ion implantation can be quantified by calculating the Child-Langmuir limit. It will be appreciated that this limit can be approximated by:
J _max = 1.72 (Q / A) ^1/2 V ^3/2 d ^-2 (1)
Where J _max is the unit of mA / cm ² , Q is the ion charge state, A is the ion mass in AMU units, V is the extraction voltage in kV units, and d is the gap width in cm units. FIG. 1 is a graph of the formula (1) in the case of ⁷⁵ As ⁺ with d = 1.27 cm. In practice, the extraction optics used by many ion implanters can be brought close to this limit value. By extending equation (1), the following merit number Δ can be defined to quantify the increase in throughput or implantation dose rate for cluster ion implantation versus single atom implantation.
Δ = n (U _n / U ₁ ) ^3/2 (m _n / m ₁ ) ^−1/2 (2)

Where Δ is the dose rate achieved by cluster implantation of n atoms of related dopants at energy U _n for single atom implantation of atoms of mass m ₁ at energy U ₁ when U _i = eV. Relative improvement (atomics / second). When U _n is adjusted to obtain the same dopant implantation depth as in the case of a single atom (n = 1), Equation (2) is as follows.
Δ = n ² (3)

Thus, implantation of clusters of n dopant atoms has the potential to provide a dose rate n ² higher than conventional single atom implantation. For As ₄ H _x for small x, this maximum dose rate improvement is about 16 times. To illustrate this point, a comparison of low energy As and As ₄ implants is shown in FIG.

  The use of clusters for ion implantation also addresses the transfer of low energy ion beams. It should be noted that the cluster ion implantation process requires only one charge per cluster, rather than having each dopant atom carry one charge as in the conventional case. Therefore, when the charge density is reduced, the dispersion Coulomb force is reduced, so that the transfer efficiency (beam transmission) is improved. In addition, the cluster has a larger mass than its monomer and is therefore less affected by the Coulomb force within the beam. Thus, implantation using clusters of n dopant atoms rather than single atoms improves the basic transport problem in low energy ion implantation and allows a significant increase in process productivity.

  In order to be able to implement this method, the formation of the cluster ions described above is necessary. Conventional ion sources used in commercial ion implanters simply produce only a very small proportion of primarily low-order (eg, n = 2) clusters relative to their monomer production, thus these In this implantation apparatus, the above-mentioned advantages of the low energy cluster beam implantation cannot be substantially realized. In fact, the very strong plasma produced by many conventional ion sources rather decomposes molecules and clusters into constituent elements. The new ion source described herein produces abundant cluster ions through the use of a “soft” ionization process, ie, electron impact ionization with strong primary electrons. The ion source of the present invention is designed for the purpose of explicitly generating and storing dopant cluster ions.

  These and other advantages of the present invention will be readily understood with reference to the following specification and attached drawings.

  Several embodiments of the present invention are provided. These embodiments relate to the generation of various N-type and P-type dopant cluster ions and negatively charged cluster ion beams. Both N-type and P-type dopant cluster ions and negatively charged cluster ion beams can be generated using the ion source shown in FIGS. 2 to 2e.

2 to 2e show conceptual diagrams of the cluster ion source 10 and its various components. First, referring to FIG. 2, a supply gas 11 such as AsH ₃ , PH ₃ , B ₂ H ₆ , or vaporized B ₁₀ H ₁₄ is supplied. The feed can be kept in the cylinder as a gas at room temperature, or introduced as a vapor sublimated from a heated solid, or evaporated from the liquid phase. The supply gas supplier 11 is connected to the ionization chamber 13 through the flow controller 12. The flow controller 12 can be as advanced as a computer-controlled mass flow controller, or can be as simple as a connecting tube having a predetermined gas conductivity. . In the latter case, the flow rate is changed by controlling the gas pressure in 11. A controlled flow rate of the gaseous feed containing the dopant provides a stable gas pressure in the ionization chamber 13 between, for example, about 3 × 10 ⁻⁴ Torr and 3 × 10 ⁻³ Torr. The implantation energy 14 is supplied in the form of an electronic control current having a well-defined energy or velocity. The temperature of the ionization chamber 13, and indeed all components of the ion source, is generally controlled to a desired value. By adjusting the ion source pressure, temperature, electron current, and electron energy, for example, AsH ₃ dopant atoms or molecules are combined and more than one of the desired dopant elements, eg, x is 0 and 4 The environment within the ionization chamber 13 is created to form cluster ions containing As ₄ H _x ⁺ , an integer between.

Through an opening 17 in the ionization chamber 13, ions can be extracted by a strong electric field between the ionization chamber 13 and the extraction electrode 15 and escape to the beam path. This extraction or accelerating electric field is generated by a high voltage power supply that biases the ionization chamber 13 to a voltage V with respect to the ground potential, and the extraction electrode 15 is substantially at ground potential. The accelerating field is established in the forward direction to attract positive ions from the ionization chamber 13 and in the reverse direction when negative ions are desired. The accelerated ions become an ion beam 16 by the extraction electrode 15. The kinetic energy E of the ion beam 16 is expressed by equation (4).
E = | qV | (4)
Where V is the source potential and q is the charge / ion. When V is expressed in volts and q is expressed in units of charge, E has units of electron volts (eV).

The ion source that forms part of the ion implantation system according to the invention is an electron impact ion source. FIG. 2 a is a schematic cross-sectional view of an ion source according to the present invention showing the structure and functionality of the components that make up the ion source 10. This cross section is cut along a plane including the propagation direction of the ion beam, and the ion source is separated into two. The ion source 10 includes a vaporizer 28 and a beam forming region 12 joined together at a mounting flange 36. The ion source 10 is designed to connect through a mounting flange 36 to a degassed vacuum chamber or other processing tool of the ion implantation system. Accordingly, the portion of the ion source 10 on the right side of the flange 36 in FIG. 2a has a high degree of vacuum (pressure <1 × 10 ⁻⁴ Torr). Gaseous material is introduced into the ionization chamber 44 where gas molecules are caused by electron bombardment from one or more electron beams 70a and 70b that enter the ionization chamber 44 through a pair of opposing electron beam entrance openings 71a and 71b. Ionized. In such a configuration, ions are generated in the ion extraction aperture plate 80 near the ion extraction aperture 81. These ions are extracted by an extraction electrode (not shown) located in front of the ion extraction aperture plate 80 to become a powerful ion beam.

A variety of vaporizers 28 are suitable for use with the present invention. An exemplary vaporizer 28 is shown in FIG. The vaporizer 28 is exemplary and can be formed with a vaporizer body 30 and a crucible 31 for carrying a solid source feed 29, eg, decaborane B ₁₀ H ₁₄ . A resistance heater can be incorporated into the vaporizer body 30. The water cooling channel 26 and the convection type gas cooling channel 27 are configured to be in close contact with the vaporizer body 30 and can be used to supply a uniform operating temperature exceeding room temperature to the crucible 31. Heat conduction between the crucible 31 and the temperature controlled vaporizer body 30 can be effected by pressurized gas introduced by the gas supply 41 into the crucible-vaporizer body connection 34, while the crucible 31. The temperature of is monitored through a thermocouple. The evaporated decaborane B ₁₀ H ₁₄ or other evaporated material 50 collects in the crucible ballast volume 51, passes through the vaporizer outlet hole 39, passes through a pair of shutoff valves 100 and 110, and is contained in the source block 35. After passing through the trough 32, it enters the ionization chamber 44 through the vapor inlet opening 33. Also, the shut-off valves 100 and 110, the mounting flange 36, and the source block 35 can be temperature controlled near or above the vaporizer temperature to prevent vapor condensation.

  The ion source gas delivery device can include two leads for supplying gas to the ionization chamber 44 from two separate sources. The first supply source can be a small diameter, low conductivity path that supplies gaseous material, such as a gas cylinder (not shown), to the pressurized gas supply. The second source can be a highly conductive path from a low temperature vaporizer that evaporates solid material. Regardless of the source, the gas delivery device maintains a gas pressure of, for example, a few millitorr in the ionization chamber 44. The vaporizer 28 maintains tight temperature control of the surface in contact with the solid material in order to maintain a stable flow of gas entering the ionization chamber and thus a stable pressure in the ionization chamber.

  Prior to operating the vaporizer 28, the isolation valve 110 can be closed to maintain the ion source and ion implantation system in a vacuum. The shut-off valve 100 can also be closed to maintain confinement of the vapor 50 in the crucible 31. The vaporizer 28 can then be safely transferred to a chemical hood where the crucible 31 can be refilled or cleaned. Prior to opening the valve 100, the vent valve 111 welded to the valve 100 body can be opened to bring the crucible volume to atmospheric pressure. Once the operation is complete, the valve 100 can be closed again and the vaporizer 28 is mounted on the ion source 10 by attaching the valve 100 to the valve 110, and then dead between the crucible 31 and the valve 100 and valve 110. In order to exhaust the volume, this vent valve 111 is connected to a rough line. Thereafter, if necessary, the shutoff valve 110 can be opened without impairing the vacuum environment of the ion source and ion implantation system.

  The vaporizer assembly 30a is formed by a heating and cooling body 30 and a removable crucible 31. Access to the crucible 31 is possible by removing the end plate (not shown) on the back of the vaporizer 28. When the crucible 31 is removed from the vaporizer 28, the crucible 31 can be refilled by removing the cover 34b, which is elastically sealed at the end of the crucible, and raising the grid 34a isolating the solid 29. After refilling, the crucible 31 is inserted into the vaporizer body 30 and the crucible ballast volume 51 is placed in front of the vaporizer body 30 in order to isolate it from the heat transfer gas in the connection 34 between the crucible 31 and the vaporizer body 30. A certain outlet hole 39 is vacuum-sealed. The hole 39 is used as an outlet for the vaporized gas. In order to achieve temperature uniformity of the crucible 31, the mechanical fit between the crucible 31 and the vaporizer body 30 is hermetically sealed. To help heat transfer between the two sources, if there is a gap between the crucible 31 and the vaporizer body 30, it can be filled with gas. The heat transfer gas enters the gap through the end plate fixture 28a and can be at or near atmospheric pressure.

  The temperature control can be performed by using, for example, proportional integral derivative (PID) closed loop control of a resistance element that can be incorporated in the vaporizer body 30. FIG. 2e shows a block diagram of a preferred embodiment in which three temperature zones are defined: the vaporizer body 30 as zone 1, the shutoff valves 100 and 110 as zone 2, and the source block 35 as zone 3. Each section can have a dedicated controller. For example, an OMRON E5CK digital controller. In the simplest case, only heating elements are used to actively control temperatures above room temperature, for example between 18 ° C. and 200 ° C. Therefore, a resistance cartridge heater can be incorporated into the vaporizer body 30 (heater 1) and the source block 35, while the valves 100 and 110 have a silicon strip heater (heating element) whose resistance element is a wire or foil strip. A device 2) can be wound. The three thermocouples labeled TC1, TC2, and TC3 in FIG. 2e can be incorporated into the three components 30, 35, and 100 (110) and read continuously by three dedicated temperature controllers. . The temperature controllers 1, 2, and 3 are programmed by the user with temperature setpoints SP1, SP2, and SP3, respectively. In one embodiment, the temperature setpoint is such that SP3> SP3> SP1. For example, when the vaporizer temperature is desirably 30 ° C., SP2 can be 50 ° C. and SP3 can be 70 ° C. The controller generally operates such that when the TC readback does not match the setpoint, the controller's comparator is cooled or heated as necessary. For example, if only heating is used to change the temperature, the comparator output is zero unless TC1 <SP1. The controller includes a look-up table of output power as a non-linear function of the temperature difference SP1-TC1, and provides an appropriate signal to the heater power supply of the controller to smoothly adjust the temperature to a programmed setting value. it can. A common means of changing the heater power is by pulse width modulation of the power supply. Using this approach, the power can be adjusted between 1% and 100% of the total. Such a PID controller can generally maintain a temperature set value within 0.2 ° C.

  The vaporizer body material can be selected to have a high heat transfer rate to maintain temperature uniformity. Intentionally adding a small heat leak to the vaporizer body 30 (FIG. 2a) to improve the stability of the control system and shorten the settling time using an air passage located on the outer surface of the vaporizer body 30. it can. The air passage 27 surrounds the carburetor body 30 and is covered with a plate (not shown). In order to provide moderate continuous convection cooling, air can be ducted into a channel in the manifold integrated with the vaporizer endplate 38. Air is supplied from the inlet after passing through a throttle valve used for flow control. Air is expelled from the air assembly to the housing exhaust.

  In addition to air cooling, provision is also made for liquid cooling of the vaporizer body 30. For example, the coolant can be supplied in a duct manner through a 1 m long and 6 mm diameter hole that is transferred back and forth through the vaporizer body 30. A connection can be made to the body port 26 through the fixture. Rapid cooling of the vaporizer assembly is accomplished by liquid cooling in order to provide a quick turnaround when needed.

  The gas can be supplied to the ionization chamber 44 through a gas conduit 33, for example, from a pressurized gas cylinder. The solid feed can be vaporized in vaporizer 28 and vapor can be supplied to ionization chamber 44 through vapor conduit 32 as described above. The solid feed material 29 located under the perforated separation barrier 34a is maintained at a uniform temperature by controlling the temperature of the vaporizer body 30 as described above. The vapor 50 accumulated in the ballast volume 31 is supplied to the ionization chamber 44 by the vapor guide 32 located in the source block 38 as it passes through the shutoff valves 100 and 110 through the holes 39. Thus, both gaseous and solid dopant support materials can be ionized by this ion source.

  FIG. 2b is a cross-sectional side view showing the basic optical design of a multiple electron beam ion source configuration according to the present invention. In one embodiment of the present invention, a pair of space-separated electron beams 70a and 70b are emitted from a pair of space-separated heating filaments 110a and 110b, and a beam steerer or static magnetic field B 135a and 135b (paper plane as shown). Into the ionization chamber 44 following a 90 ° trajectory and first through a pair of bottom plate openings 106a and 106b and a pair of spaced apart bottom plates 105a and 105b, and thereafter an electron entrance opening. It passes through 71a and 71b. Electrons that pass through the entire ionization chamber 44 (i.e., through both electron entrance openings 71a and 71b) are bent toward the pair of emitter shields 102a and 102b by a beam steerer or static magnetic field 135a and 135b. As the electron beams 70a and 70b propagate through the bottom plate openings 106a and 106b, the voltage Va is applied to the bottom plates 105a and 105b (performed by the positive power supply 115) and the filaments 135a and 135b before entering the ionization chamber 44. Is decelerated by the application of voltage Ve to (which is done by the negative power supply 116). It is important to maintain the electron beam energy at a position substantially higher than the position generally desired for ionization within the beam forming and transfer region, ie, outside the ionization chamber 44. This is due to the space charge effect that severely reduces the beam current to increase the electron beam diameter with low energy. Therefore, it is desirable to maintain the electron beam energy between about 1.5 keV and 5 keV in this region.

All voltages are for the ionization chamber 44. For example, when Ve = −0.5 kV and Va = 1.5 kV, the energy of the electron beam is expressed by e (Va−Ve), where e is an electron charge (6.02 × 10 ⁻¹⁹ coulomb). is there. Thus, in this example, the electron beam 70a or 70b is formed and deflected with an energy of 2 keV, but when entering the electron entrance openings 71a and 71b, the energy is only 0.5 keV.

  The table below shows approximate values of the magnetic field B required to bend the electron beam 90 ° with energy E.

(Table 1)

  Other elements shown in FIG. 2b include an extracted ion beam 120, a source electrostatic shield 101, and a pair of emitter shields 102a and 102b. These emitter shields 102a and 102b serve two purposes: shielding from electromagnetic fields and shielding from stray electrons or electron beams. For example, the emitter shields 102a and 102b protect the electron beams 70a and 70b from the magnetic field associated with the potential difference between the bottom plates 105a and 105b and the source shield 101, and as a stray electron beam dump from the opposing electron emitters. Work. The source shield 101 protects the ion beam 120 from the magnetic field generated by the potential difference between the bottom plates 105a and 105b and the ionization chamber 44, and functions to absorb stray electrons and ions that collide with the ion source element. . For this reason, both the emitter shields 102a and 102b and the source shield 101 are made of a refractory metal such as molybdenum or graphite. Alternatively, more complete shielding of the ion beam 120 from the magnetic fields B 135a and 135b can be achieved by making the source shield 101 with a ferromagnetic material such as electromagnetic stainless steel.

  FIG. 2c is a cutaway view showing mechanical details and exemplarily showing how the contents of FIG. 2b are incorporated into the ion source of FIG. 2a. Electrons are thermally ionized from one or more of filaments 110a and 110b and are accelerated to a pair of corresponding anodes 140a and 140b forming electron beams 70a and 70b. Such a configuration has several advantages. First, the filaments 110a and 110b can be actuated separately or in concert. Second, since the electron beam is generated outside the ionization chamber, the lifetime of the emitter is increased for known configurations, which is that the emitter is in the implanter vacuum housing where the ion source is located. This is because the emitter is substantially protected from ion bombardment.

  The magnetic flux from the permanent magnets 130a and 130b and the pair of pole assemblies 125a and 125b form a beam steerer that is used to establish a uniform magnetic field across the air gap between the ends of the pole assembly through which the electron beam propagates. To do. The electron beam energies of the magnetic fields 135a and 135b and the electron beams 70a and 70b are adapted so that the electron beams 70a and 70b are deflected 90 ° and enter the ionization chamber 44 as shown. By deflecting the electron beams 70a and 70b, for example by 90 °, there is no line of sight between the emitter 110 and the ionization chamber containing the ions, so that the impact of the emitter by strong charged particles is prevented.

  Since Va is a positive voltage with respect to the ionization chamber 44, the electron beams 70a and 70b are decelerated as they pass through the gap formed by the bottom plate openings 106a and 106b and the electron entrance openings 71a and 71b. Accordingly, the combination of the bottom plate opening 106a and the electron entrance opening 71a, the bottom plate opening 106b and the electron entrance opening 71b, and the gap therebetween each form an electrostatic lens, in this case a deceleration lens. By using a deceleration lens, the ionization energy of the electron beam can be adjusted without substantially affecting the electron beam generation and deflection.

  This gap can be achieved by one or more ceramic spacers 132a and 132b that support each bottom plate 105a and 105b and act to isolate it from the source block 35 at the potential of the ionization chamber. Ceramic spacers 132a and 132b provide both electrical isolation and mechanical support. For the sake of clarity, the emitter shield 102 and the source shield 101 are not shown in FIG.

  Since the electron entrance openings 106a and 106b can limit the amount of transmission of the electron beams 70a and 70b, the bottom plates 105a and 105b can capture a portion of the powerful electron beams 70a and 70b. Therefore, the bottom plates 105a and 105b should be actively cooled or passively cooled. Active cooling can be achieved by passing a liquid coolant such as water through the bottom plates 105a and 105b or by forcing compressed air through the bottom plates 105a and 105b. Alternatively, passive cooling is achieved by allowing the bottom plates 105a and 105b to reach the cooling temperature by dissipating heat to the surroundings. This steady state temperature depends on the captured beam power, the surface area and emissivity of the bottom plate, and the temperature of the surrounding components. Operating the bottom plates 105a and 105b at a high temperature, eg, 250 ° C., may be advantageous when flowing a condensable gas such as decaborane vapor that may form a contamination and particle forming film on the cold surface.

  FIG. 2d is a simplified top view of the electron beam forming region of the source. The filament 110b has a potential Ve with respect to the ionization chamber 44, eg, −0.5 keV, and the anode 140b, the magnetic pole assembly 125b, the bottom plate 105b, and the emitter shield 102b all have an anode potential Va, eg, 1.5 keV. Therefore, the electron beam energy is 2 keV. The electron beam 70b is deflected by the magnetic field 135b in the gap between the poles of the pole assembly 125b such that the electron beam 70b passes through the bottom plate opening 106b. A typical value for the bottom plate openings 106a and 106b and the electron entrance openings 71a and 71b is 1 cm in diameter, respectively.

FIG. 3 shows the ion source associated with an important downstream element including the proposed cluster ion implantation system. Configurations other than those shown in FIG. 3 are possible. The ion source 21 is coupled with the extraction electrode 22 to generate an ion beam containing cluster ions. The ion beam 20 typically includes all of the species from which many different mass ions, i.e. ions of a given charge polarity, are generated in the ion source 21. The ion beam 20 then enters the analyzer magnet 23. The analyzer magnet 23 generates a dipole magnetic field in the ion beam transfer path in response to the current in the magnet coil. The direction of the magnetic field is perpendicular to the plane of FIG. The function of the analyzer magnet 23 is to spatially separate the ion beam into a set of constituent beamlets by bending the ion beam into an arc whose radius depends on the mass-to-charge ratio of the individual ions. It is. Such an arc is shown in FIG. 3 with the beam component 24 selected as the ion beam. The magnet 23 bends the specific beam along the radius obtained by the following equation (5).
R = (2 mU) ^1/2 / qB (5)
However, R is a bending radius, B is a magnetic flux density, m is ion mass, U is ion kinetic energy, and q is an ion charge state.

  The selected ion beam consists of a narrow range of mass-energy product ion masses such that the bending radius of the ion beam by the magnet sends the beam through the mass resolving aperture 27. The unselected beam components do not pass through the mass resolving aperture 27 but are captured elsewhere. For example, for a beam composed of hydrogen ions having a mass of 1 or 2 atomic mass units and having a mass to charge ratio m / q smaller than the selected beam 25, the magnetic field induces a small bend and the beam Captures the inner diameter wall 30 of the magnet chamber or elsewhere. For a beam with a mass-to-charge ratio greater than the selected beam 26, the magnetic field induces a large bend radius, and the beam strikes the outer diameter wall 29 of the magnet chamber or elsewhere. As is well established in the art, the combination of analyzer magnet 23 and mass resolving aperture 27 constitutes a mass analysis system that selects ion beam 24 from multiple species beam 20 extracted from an ion source. Thereafter, the selected beam 24 can pass through the post-analysis acceleration / deceleration electrode 31. This step 31 can adjust the beam energy to the desired final energy value required for a particular implantation process. The post-analysis acceleration / deceleration stage 31 can take the form of, for example, an electrostatic lens, or alternatively a LINAC (linear accelerator). To prevent ions that have undergone a charge exchange or neutralization reaction between the resolving aperture and the wafer (and therefore do not have the correct energy) from propagating to the wafer, a “neutral beam filter” or “energy” A "filter" can be incorporated in this beam path. For example, the post-analysis acceleration / deceleration stage 31 can incorporate a “dog leg” or small angle swing that is limited to follow the selected electromagnetic field through the applied DC electromagnetic field. Beam components that are neutral or multiply charged will never follow this path. Thereafter, the energy conditioned beam enters the beam scanning system 32 in the implantation system shown in FIG. The beam scanning system 32 scans the beam so that the entire target 28 is uniformly injected. Various configurations are possible, for example, one-dimensional or two-dimensional scanning and electrostatic versus magnetic scanning systems.

  The beam then enters the wafer processing chamber 33, which is also maintained in a high vacuum environment, where it strikes the target 28. Various configurations of wafer processing chambers and wafer processing systems are possible, with the main categories being continuous (one wafer at a time) or batch (many batches of wafers on a rotating disk). In a continuous processing chamber, one direction (lateral or vertical) is typically mechanical across a beam that is electromagnetically scanned in an orthogonal direction to ensure good spatial uniformity of the implant. Scanned. In the batch apparatus, mechanical scanning is performed in the radial direction by rotating the disk, and scanning in the vertical direction or horizontal direction of the rotating disk is also performed at the same time, and the ion beam remains fixed.

For accurate dopant placement by cluster ion implantation, each of the n atoms contained in the cluster must penetrate the substrate with the same kinetic energy, and the molecular ion is in the form of An ^+. In the simplest case (ie it is uniquely composed of n dopant atoms A), each of the n dopant atoms must receive the same proportion of 1 / n cluster energy when entering the semiconductor substrate. Don't be. It has been demonstrated, for example, by Sze, “VLSI Technology”, McGraw Hill, pages 253-254 (1983), that this equal division of energy occurs whenever a polyatomic molecule strikes a solid target surface. Furthermore, the electrical result of such implantation needs to be the same as an equivalent implant using a single atomic ion implantation. These results show that in the case of implantation using decaborane, Jacobson et al., “Alternative method of ultra-low energy ion implantation called decaborane”, IEEE Proceedings of the 13th International Conference on Ion Implantation Technology, Alpsbach, Austria, 300-303 (2000), shown in detail, in fact the Applicant expects similar results for all dopant clusters.

During ion implantation, dopant atoms can penetrate deeper into the semiconductor substrate by channeling, that is, by entering the crystal lattice of the substrate along a symmetric direction that includes low density lattice atoms or “channels”. If the ion trajectory coincides with the direction of the channel in the semiconductor crystal lattice, the ions substantially avoid collision with the substrate atoms and widen the scope of the dopant projectile. An effective means of limiting or even preventing channeling consists of forming an amorphous layer on the substrate surface. One means of generating such a layer is that the crystal damage caused by the implantation process is sufficient to eliminate the layer crystal structure at the semiconductor surface without substantially changing the electrical properties of the substrate during the activation stage. As such, the substrate is implanted using ions of the same element that the substrate is composed of, or ions that have the same electrical properties (ie, from the same column of the periodic table). For example, silicon and germanium ions are implanted into a silicon substrate at an energy of 20 keV and a dose of 5 × 10 ¹⁴ cm ⁻² to form such an amorphous layer in the silicon substrate, and then a shallow dopant layer by cluster ion implantation. Infusion continues.

An important application of this method is the use of cluster ion implantation for N-type and P-type shallow junction formation as part of a CMOS manufacturing sequence. CMOS is the dominant digital integrated circuit technology in current use, its name is N-channel and P-channel MOS transistor on the same chip: the (complementary MOS, C Omplementary MOS both N and P) formed of Represents. The success of CMOS is that circuit designers can take advantage of the complementary nature of opposing transistors to create better circuits, which draw less active power than alternative technologies. The terms N and P, negative (n egative) and positive (p ositive) (N-type semiconductor has negative majority carriers, and vice versa correct) is based on, N-channel and P-channel It should be noted that the transistors are duplicates of each other with the type (polarity) of each region reversed. Fabrication of both types of transistors on the same substrate requires the continuous implantation of N-type impurities and then P-type impurities while protecting the other type of element with a photoresist shielding layer. . It should be noted that for each transistor type, both polar regions need to operate correctly, but the implant that forms the shallow coupling is of the same type as the transistor. N-type shallow ones are injected into N-channel transistors, and P-type shallow ones are injected into P-channel transistors. An example of this process is shown in FIGS. 4a and 4b. In particular, FIG. 4 a shows a method of forming an N-channel drain extension 89 with an N-type cluster implant 88, and FIG. 4 b shows the formation of a P-channel drain extension 90 with a P-type cluster implant 91. N-type and P-type transistors require shallow junctions with similar geometries, and thus having both N-type and P-type cluster implants is advantageous for the formation of advanced CMOS structures. It should be noted.

An example of the use of this method is shown in FIG. This figure shows a semiconductor substrate 41 that undergoes some of the front end processing steps of semiconductor device fabrication. This structure consists of an N-type semiconductor substrate 41 processed through the P well 43, trench isolation 42, and gate stack formation 44 and 45 steps. P-well 43 forms a junction with N-type substrate 42 that provides junction isolation for the transistors in the well. Trench isolation 42 provides lateral dielectric isolation between the N and P wells (ie, in the entire CMOS structure). The gate stack is then configured to include a gate oxide layer 44 and a polysilicon gate electrode 45 that are patterned to form a transistor gate stack. Also, a photoresist 46 is added and patterned so that the area for the NMOS transistor is opened while the other areas of the substrate are shielded by the photoresist layer 46. At this point in the process flow, the substrate is ready for the drain extension implant, which is the shallowest doping layer required by the device fabrication process. The general processing requirements for the state-of-the-art devices at the 0.13 μm technology node are an arsenic implant energy between 1 keV and 2 keV and an arsenic dose of 5 × 10 ¹⁴ cm ⁻² . The cluster ion beam 47, in this case As ₄ H _x ^+, is directed to the semiconductor element so that the ion beam propagation direction is perpendicular to the substrate in order to avoid blockage by the gate stack. The energy of the As ₄ H _x ⁺ cluster should be 4 times the desired As ⁺ implantation energy, eg between 4 keV and 8 keV. When the clusters separate on impact to the substrate, the dopant atoms become present in a shallow layer near the surface of the semiconductor substrate that forms the drain extension region 48. It should be noted that the same implant enters the surface layer of the gate electrode 49 and further doping of the gate electrode takes place. Thus, the process described in FIG. 5 is one important application of the proposed invention.

Yet another example of the application of the method is shown in FIG. 5a, which is the formation of deep source / drain regions. This figure shows the semiconductor substrate 41 of FIG. 5 after execution of further processing steps of semiconductor device manufacturing. Additional processing steps include pad oxide 51 formation and spacer 52 formation on the gate stack sidewalls. At this point, a photoresist layer 53 is added and patterned to expose the transistor to be implanted, in this case the NMOS transistor. Next, ion implantation for forming the source and drain regions 55 is performed. This implantation is a suitable application of the proposed cluster implantation process because it requires a high dose at low energy. The typical implantation parameters for the 0.13 μm technology node are about 6 keV / arsenic atoms (54) at an arsenic dose of 5 × 10 ¹⁵ cm ⁻² , and therefore a 24 keV 1.25 × 10 ¹⁵ cm ⁻² As ₄ Hx ⁺ implant, A 12 keV 2.5 × 10 ¹⁵ cm ⁻² As ₂ H _x ⁺ implant or a 6 keV 5 × 10 ¹⁵ cm ⁻² As ⁺ implant is required. As shown in FIG. 5, source and drain regions 55 are formed by this implant. These regions provide a high conductivity connection between the circuit interconnect (formed later in the process) and the intrinsic transistor formed by the drain extension 48 along with the channel region 56 and the gate stacks 44 and 45. . It can be seen that the gate electrode 45 can be exposed to this implant (as shown), and if this is possible, the source / drain implant is the primary doping source of the gate electrode. This is shown in FIG.

Detailed views illustrating the formation of the PMOS drain extension 148 and the PMOS source and drain regions 155 are shown in FIGS. 5b and 5c, respectively. The structure and processing are the same as in FIGS. 5a and 5b, with the dopant type being reversed. In FIG. 5 b, PMOS drain extension 148 is formed by implantation of boron cluster implant 147. Typical parameters for this implant are implant energy of 500 eV / boron atom and a dose of 5 × 10 ¹⁴ cm ⁻² for a 0.13 μm technology node. That is, the B ₁₀ H _x implant is considered to be at 5 keV and an octadecaborane dose of 5 × 10 ¹³ cm ⁻² . FIG. 5c shows the formation of the PMOS source and drain region 148 by implantation of a P-type cluster ion beam 154, also decaborane. The general parameters of this implant would be approximately 1 keV / boron atom and a boron dose of 5 × 10 ¹⁵ cm ⁻² (ie, 20 keV decaborane at 5 × 10 ¹⁴ cm ⁻² ) for a 0.13 μm technology node.

  In general, ion implantation alone is not sufficient to form an effective semiconductor junction, and heat treatment is required to electrically activate the implanted dopant. After the implantation, the crystal structure of the semiconductor substrate is significantly damaged (the substrate atoms are moved from the crystal lattice position), and the implanted dopant is only weakly bonded to the substrate atoms, and as a result The electrical properties of the implanted layer are poor. A heat treatment at high temperature (greater than 900 ° C.), ie, annealing, is generally performed to repair the semiconductor crystal structure and position the dopant atom in a substitution, ie, one position of the substrate atom in the crystal structure. This substitution allows the dopant to bond with the substrate atoms and become electrically active. That is, the conductivity of the semiconductor layer can be changed. However, this heat treatment adversely affects the formation of shallow junctions because diffusion of the implanted dopant occurs during the heat treatment. In fact, boron diffusion during heat treatment is a limiting factor in achieving USJ in the region below 0.1 microns. In order to minimize the diffusion of shallow implanted dopants, advanced methods for this heat treatment such as “spike annealing” have been developed. Spike annealing is an abrupt heat treatment in which the residence time at the maximum temperature approaches zero, and the temperature goes up and down as quickly as possible. Thus, there is a high temperature required to activate the implanted dopant while diffusion of the implanted dopant is minimized. The use of such advanced heat treatments with the present invention is envisaged in order to maximize its advantages in the manufacture of final semiconductor devices.

FIG. 6 shows the generation of phosphorus cluster ions and the formation of a mass-resolved phosphorus cluster ion beam. This mass spectrum shows data collected during operation of the ion source of the present invention using phosphorus (PH ₃ ) as the source feed gas. The mass spectrum shows the intensity of the ion current on the vertical scale 61 while the analyzer magnetic field determines the mass-to-charge ratio of the ions on the horizontal scale 62. The current was measured in a Faraday cup where secondary electrons were substantially suppressed. The horizontal scale 62 is linear with respect to the magnetic field, but is non-linear with respect to the mass-to-charge ratio, because for a given extraction voltage V, the two quantities are related by m / q = aB ² where a is a constant. It is. Accordingly, the higher the mass peak on the horizontal scale 62, the closer to each other. Phosphorus clusters are observed as signals 65, 66, and 67 with 2, 3, and 4 phosphorus atoms per cluster, respectively. From the results of this spectral analysis, it can be seen that the ion source of the present invention supports cluster formation and maintenance during operation. The classification of the signal 63 on the left of the graph is hydrogen ions, and the mass numbers are 1 and 2. The hydrogen peak is relatively small and much smaller than the peak containing phosphorus. A second classification of signal 64 is performed between masses 31 and 32 and corresponds to an ion containing one phosphorus atom. During a conventional injection process, one, some, or all of these peaks can be injected depending on the selection of the selected mass resolving aperture 27 (see FIG. 2). For some applications, if the process is sensitive to H, only the ³¹ P ⁺ peak needs to be selected. In this case, a narrow mass resolving aperture can be implemented to eliminate the hydrogen peak, ie PH _x ⁺ . However, x = 1, 2, 3, or 4. In other processes, it may be necessary to inject all of the peaks in this group to increase productivity. The next group of signals near the right 65 consists of phosphorus dimers, each of which contains two phosphorus atoms. The left most important signal corresponds to P ₂ ⁺ with mass number 62. The signal on the right is the P ₂ H _x ⁺ signal, where x is between 1 and 6. Applicants have also noted that the intensity of these signals is small compared to the monomer peak 64, but the observed intensity depends on the entire set of source input settings, for example, Note that the body can be optimized for the desired beam conditions in order to maximize the relative height of the P ₂ ⁺ peak when needed. The choice of the mass resolving aperture determines the number of these beams that are injected during the implantation process. The next signal classification 66 to the right corresponds to a phosphorus cluster ion containing _three phosphorus atoms (P ₃ ⁺ ). The next signal 67 to the right corresponds to a phosphorus cluster ion containing four phosphorus atoms. The intensity of this cluster is higher than that of the P ₃ H _x ⁺ cluster, and the net dose rate using P ₄ ⁺ (4x observed intensity) is higher than that of injecting P ⁺ or P ₂ ^+. It is interesting to note that the energy per phosphorus atom made is only a quarter of the nominal ion beam energy.

FIG. 7 shows the mass spectrum of AsH ₃ using the present invention. Since the ion beam energy was 19 keV, the effective As implant energy of As ₄ H _x ⁺ is 4.75 keV. Since the beam current of As ₄ H _x ^{+ in} FIG. 7 was about 0.25 A, the equivalent As dopant current is about 1 mA. FIG. 7 also shows that particle currents between 0.5 mA and 1.0 mA result from implantation of an As, As ₂ , As ₃ , or As ₄ containing ion beam, and select different portions of the FIG. 7 spectrum. However, it is shown that an effective implant energy range between about 20 and 5 keV can be obtained by simply adjusting the analyzer magnet current.

FIG. 8 shows As ₄ H _x ⁺ current as a function of As implant energy. The angular divergence of the ion beam was limited to a lateral or dispersion half-angle of 11 mR or about 0.6 degrees by an opening between the mass resolving aperture (see, eg, 27 in FIG. 3) and the Faraday cup. 1 keV / atom is the lower limit that semiconductor processing will require for arsenic implantation into the USJ element.

FIG. 9 shows a comparison of the beam current of FIG. 8 converted to units of beam brightness and a “typical” modern medium current injector. The improvement is about 30 times (the applicant's assumed medium current injection device specification was 40 mrad half-angle acceptance and 200 μA beam current at 10 keV). Stephens, J. et al. F. The luminance B is defined as follows in Ziegler's “Ion Implantation Technology Handbook”, North Holland, pages 455-499 (1992).
B = 2I / π ² ε ² (μA−mm ⁻² −mrad ⁻² ) (6)
Where I is the effective dopant beam current in microamperes, and ε is the beam emissivity expressed in squares (milliradians-millimeters). The emissivity is calculated by:
ε = δa (7)
Where δ is the beam half width on the dispersion surface, and a is the half pencil angle, both of which are measured at the image plane, ie, the resolving aperture position.

Beam brightness is an important figure of merit that quantifies the amount of beam current that can be transmitted in a particular acceptance, for example, through a tube of a particular diameter and length. Since ion implantation system beam lines have a well-defined acceptance, brightness is an important measure of productivity for emissivity limited beams. Emissivity is usually a limiting factor in the transport of low energy beams. As can be seen from the formulas (1) to (3), this is generally a benefit of using cluster ions for monomer ions. In the case of As ₄ injection, a 16-fold increase in throughput, ie Δ = n ^2, is predicted by equation (3).

FIG. 10 shows secondary ion mass spectrometer (SIMS) results for silicon samples implanted with AsH _x ⁺ and As ₄ H _x ⁺ ions at 4.75 keV and 19 keV, respectively. The atomic dose was about 1 × 10 ¹⁶ xm ⁻² . These data are compared to TRIM, a fully dynamic scattering model commonly used in the industry to simulate ion implantation into silicon. These results show that the applicant is actually injecting As and As ₄ with the specified energy.

FIG. 11 shows a mass spectrum of commercially available diborane B ₂ H ₆ , which is a gaseous material not commonly used in conventional ion implantation. FIG. 11 shows H (H ⁺ , H ₂ ⁺ , H ₃ ⁺ ), B (B, BH ⁺ , BH ₂ ⁺ ), B ₂ (B ₂ ⁺ , B ₂ H ⁺ , B ₂ H ₂ ⁺ , B ₂ H ₃ ⁺ , B ₂ H ₄ ⁺ ), B ₃ (B ₃ , B ₃ H ⁺ , B ₃ H ₂ ⁺ , B ₃ H ₃ ⁺ , B ₃ H ₄ ⁺ ), B ₄ (B ₄ , B ₄ H ⁺ , B ₄ H ₂ ⁺ , B ₄ H ₃ ⁺ , B ₄ H ₄ ⁺ ), and B ₅ groups. The spectrum of FIG. 11 is interpreted because there are two naturally occurring isotopes ¹⁰ B and ¹¹ B of boron represented by a ratio of about 4: 1 of ¹¹ B to ¹⁰ B reflecting natural abundance. It is a little complicated to do. For example, both ¹¹ B and ¹⁰ BH are present at the ¹¹ amu peak.

FIG. 12 shows the generation of borohydride clusters and positive cluster ions in the present invention. This mass spectrum shows data collected during ion source operation of the present invention using vaporized decaborane B ₁₀ H ₁₄ as the ion source feed. Boron hydride clusters of the form B _y H _x ⁺ with 1 ≦ y ≦ 10 and 0 ≦ x ≦ 14 are shown separated by 1 amu from 1 amu to about 124 amu. The maximum signal B ₁₀ H _x ⁺ observed corresponds to the decaborane molecular ion formed by direct ionization of the decaborane parent molecule.

FIG. 13 shows an anion spectrum of decaborane produced by an ion source of the present invention similar to that of FIG. Since towards state formed by decaborane anions much less, most of the ions (about 90%) of the parent B ₁₀ H _x ^- included in the peak. The use of anions for semiconductor ion implantation is very useful because it substantially eliminates the wafer charge observed with positive ion implantation. It is unusual for an ion source to produce both abundant amounts of cations and anions for a particular material, and the peak ion currents in FIGS. 12 and 13 are the same within a twofold range. . This is shown dramatically in FIG. 14 for the extended mass range. These data are collected using the ion implantation system of the present invention on the same sheet of paper over the same mass range, collecting positive ion mass spectra as shown, reversing the polarity of the ion implantation system power supply. It is collected by. In order to collect the data of FIG. 14, a Faraday cup current was supplied to the xy paper recorder. In the case of decaborane, great advantages are evident in the implantation of anions rather than cations. That is, 1) the more useful ionic current is in the relevant peak, resulting in a larger useful dopant flux, and 2) the parent peak is nearly halved in mass (9 amu for cations) The full width at half maximum of 5 amu for negative ions), and 3) wafer charges are eliminated when negative ions are used instead of positive ions, as is generally accepted in the art. The

  FIG. 15 shows SIMS profiles for both decaborane cations and decaborane anions implanted in a silicon sample with a decaborane energy of 20 keV. These profiles are approximately the same as would be expected if each ion had the same number of boron atoms, and therefore injected into the same predicted range.

FIG. 16 shows SIMS data for a negative decaborane implant that also shows H concentration. H dose is 0.9 times the boron dose, which, B ₁₀ H ₉ ^- suggests an average chemical formula for negative decaborane called.

FIG. 17 shows the dependence of ionization probability on electron energy for electron impact ionization. Ammonia (NH ₃ ) is used illustratively. The probability is expressed as a cross section σ in units of 10 ⁻¹⁶ cm ² . The electron energy (T) is eV, that is, an electron volt unit. Two sets of theoretical curves marked BEB (Vertical IP) and BEB (Adiabatic IP) based on the first principle and two sets of experimental data by Djuric et al. (1981) and Rao and Srivastava (1992) are shown. ing. FIG. 17 shows that ionization is more actively performed in a specific range of electron energy than in other ranges. These data apply to the production of cations, but similar considerations apply to the production of anions. That is, a strong energy dependence is evident. In general, the cross-section for cation generation is greatest for electron impact energy between about 50 eV and 500 eV, with a peak at 100 eV. Accordingly, since the energy with which the electron beam enters the ionization chamber 44 is an important parameter that affects the operation of the ion source of the present invention, Applicants have determined that the energy of electrons entering the ionization chamber is from about 0 eV to about 0 eV. The electron beam transfer was designed to vary up to 5000 eV. The features shown in FIGS. 2b to 2d show how an electron optical element that allows a wide control of electron impact ionization energy while operating in a substantially constant state in the electron beam forming and deflection regions of the ion source in the present invention. Indicates whether to be included.

FIG. 18 is a mass spectrum of a decaborane cation generated by the ion source of the present invention. The individual ions that make up the mass spectrum are labeled. In general, ions are in the form of B _n H _x ⁺ , where 0 ≦ n ≦ 10 and 0 ≦ x ≦ 14. The remarkably large peak is the parent B ₁₀ H _x ⁺ ion, and the majority of the peak intensity is within about 8 amu (atomic weight units). This parent ion is a likely choice for cation implantation.

FIG. 19 is a mass spectrum of both decaborane anion and decaborane cation generated with the ion source of the present invention. The individual ions that make up the mass spectrum are labeled. The anion spectrum is much simpler than the cation spectrum. In particular, there are no obvious hydrogen ions or boron ions of lower order than hydrogen, while about 90% of the spectrum is composed of parent B ₁₀ H _x ⁺ ions. Similar to B ₁₀ H _m ⁺ , the majority of the parent anion peak intensity is within about 8 amu. This parent ion is a likely choice for anion implantation.

  There are several related elements that are used in the formation of shallow junctions in semiconductors. For silicon applications, the main dopants are boron, phosphorus, arsenic, and antimony, so these elements are most likely to be applied in the formation of shallow junctions. In addition, since silicon and germanium implants are used to form amorphous regions in silicon, clusters of these elements are considered useful for forming shallow amorphous regions. In the case of composite semiconductors, the relevant elements for shallow junctions include silicon, germanium, tin, zinc, cadmium, and beryllium, so clusters of these elements have the opportunity to be used to form shallow junctions in composite semiconductor manufacturing. is there.

  One aspect of the method is to provide a suitable environment within the ionization chamber for cluster ion formation. Since each of the various elements described has different chemical properties, the optimal environment is different for each element. Each element and each selected cluster requires a different set of input parameters to achieve optimal performance. Available parameters for optimization include source pressure as controlled by feed flow, temperature in the ionization chamber as controlled by a temperature control system, and electron beam current when the ionization energy is an electron beam. And ionization energy intensity and properties such as electron energy. These basic parameters work together to create an appropriate environment in the source ionization chamber for dopant cluster formation and ionization.

  As described above, both N-type and P-type dopants can be implanted at a shallow depth with high efficiency by ion implantation of a cluster of dopant atoms compared to ion implantation of a single dopant atom. become.

  The invention has been described with several embodiments. The present invention is not limited to these embodiments. For example, it will be apparent to those skilled in the art that various modifications, substitutions, improvements, and combinations thereof are possible.

  Obviously, many modifications and variations of the present invention are possible in light of the above teachings. In other words, it should be understood that the present invention can be practiced other than as specifically described above in the claims.

  What is claimed and what is desired to be included in the US Patent is set forth in the appended claims.

It is a graph which shows the extraction energy with respect to the maximum ⁷⁵ As ⁺ beam current by Child-Langmuir law. 6 is a graph showing a comparison of maximum extraction currents achievable through tetrameric arsenic and monomeric arsenic. 1 is a simplified diagram of a cluster ion source according to the present invention. 1 is a perspective view of an exemplary embodiment of a cluster ion source according to the present invention. FIG. 2b is a side view of a portion of the ion source shown in FIG. 2a, shown in cutaway view with an electron beam and a magnetic field superimposed thereon. 2 is a perspective view of a portion of an ion source shown in cutaway view showing a magnetic field and electron beam source according to the present invention. FIG. It is a schematic top view of the electron beam formation area | region of the ion source by this invention. 1 is a block diagram of a temperature control system that can be used with the present invention. 1 is a schematic diagram of an exemplary cluster ion implantation system according to the present invention. FIG. It is a figure of a CMOS manufacturing sequence at the time of NMOS drain extension part formation. It is a figure of the CMOS manufacturing sequence at the time of PMOS drain extension part formation. It is a figure of the semiconductor substrate in the manufacturing process of the NMOS semiconductor element in the N type drain extension part injection | pouring step. It is a figure of the semiconductor substrate in the manufacturing process of the NMOS semiconductor element in a source / drain injection | pouring stage. It is a figure of the semiconductor substrate in the manufacturing process of the PMOS semiconductor element in the P-type drain extension part injection | pouring step. It is a figure of the semiconductor substrate in the manufacturing process of the PMOS semiconductor element in a source / drain injection | pouring stage. It is a graph of mass spectrum of PH ₃ generated in the ion source of the present invention. It is a graph of mass spectrum of AsH ₃ generated in the ion source of the present invention. FIG. 5 is a graph showing on-wafer As ₄ H _x ⁺ ion current in the low energy range. FIG. FIG. 7 is a graph of the data shown in FIG. 6 converted into a unit of beam luminance. ₄ is a graph of comparison of SIMS profiles and TRIM calculation results during implantation of arsenic concentrations from AsH _x ⁺ and As ₄ H _x ⁺ ion beams implanted into silicon wafers using the present invention. It is a graph of mass spectrum of B ₂ H ₆ produced in the ion source of the present invention. 4 is a graph of recorded cation mass spectra for the present invention operating with a decaborane feed. 4 is a graph of recorded anion mass spectra for the present invention operating with a decaborane feed. FIG. 2 is a graph showing recorded mass spectra of both anion and cation decaborane taken continuously and dimer B ₂₀ H _x . FIG. 6 is a graph of the SISM profile during implantation of Yin and Yang B ₁₀ H _x ions using the present invention with a decaborane implantation energy of 20 keV. FIG. 5 is a graph of the SISM profile during implantation of 20 keV decaborane implanted in silicon showing B and H concentrations. It is a graph of the ionization cross-section σ as a function of electron energy T of ammonia (NH _3). It is a figure of the mass spectrum of the positive decaborane ion produced | generated with the ion source of this invention. It is a figure of the mass spectrum of the negative decaborane ion produced | generated with the ion source of this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 Ion source 12 Beam formation area 28 Vaporizer 36 Mounting flange 44 Ionization chamber 71a, 71b Electron beam entrance opening 70a, 70b Electron beam

Claims (13)

(A) generating N-type molecular cluster ions from the first molecular species;
(B) generating a P-type molecular cluster ion from the second molecular species;
(C) a pre-Symbol N type molecular cluster ions comprising the steps of injecting in a first region of the substrate,
(D) is pre-Symbol P type molecular cluster ions comprising the steps of implanting in a second area on the substrate,
A method of implanting a dopant material into a substrate having a first region and a second region. The method of claim 1, wherein step (a) includes generating the N-type molecular cluster ions from an arsine (AsH ₃ ) gas. 3. The method of claim 2, wherein step (a) comprises generating As ₄ H _x ⁺ , wherein x is an integer and 1 ≦ x ≦ 6. _3. The method of claim 2, wherein step (a) comprises generating As ₃ H _x ⁺ , wherein x is an integer and 1 ≦ x ≦ 5. Step (a), (wherein, x is a 1 ≦ x ≦ 4 integer.) As ₂ H _x ⁺ A method according to claim 2, characterized in that it comprises the step of generating a cluster ion. The method of claim 1, wherein step (a) comprises generating the N-type molecular cluster ions from phosphine (PH ₃ ) gas. Step (a), (wherein, x is a 1 ≦ x ≦ 6 at integer.) P ₄ H _x ⁺ A method according to claim 6, characterized in that it comprises the step of generating a cluster ion. The method of claim 6 , wherein step (a) comprises generating P ₃ H _x ⁺ , wherein x is an integer and 1 ≦ x ≦ 5. Step (a), (wherein, x is a 1 ≦ x ≦ 4 integer.) P ₂ H _x ⁺ A method according to claim 6, characterized in that it comprises the step of generating a cluster ion. The method according to claim 1, wherein step (b) comprises generating the P-type molecular cluster ions from decaborane (B ₁₀ H ₁₄ ) gas. Step (b), when n and x is 2 ≦ n ≦ 10 and 1 ≦ x ≦ 14 at integer claim 10, characterized in that it comprises the step of generating a B _n H _x ⁺ cluster ions the method of. The method of claim 10 , wherein step (b) comprises generating B ₁₀ H _x ⁺ , wherein x is an integer and 1 ≦ x ≦ 14. Step (b), a negative B ₁₀ H _x ^- (. Wherein, x is a 1 ≦ x ≦ 14 at integer) The method according to claim 1, characterized in that it comprises the step of generating a cluster ion .

Images