Non-Mass-Analyzed Ion Implantation

This application relates to the subject matter disclosed in U.S. patent application Ser. No. 375,583 assigned to Hitachi Ltd. and claiming priority on a Japanese patent application entitled "Ion Implanting Apparatus" filed on May 18, 1981, invented by one of the present inventors and others.

1. Field of the Invention

This invention relates to ion implantation and more particularly to a non-mass-analyzed ion implantation process wherein ions of non-selected species are accelerated and implanted into a target.

2. Description of Prior Art

Today, ion implantation is evaluated as one of important techniques, especially, in the semiconductor industry. Conventionally, most ion implantation apparatus for processing semiconductors have been based on mass-analyzed ion implantation wherein mass analysis is effected by utilizing a static magnetic field generated by a magnet and an ion of one selected species is used as an ion source. The mass-analyzed ion implantation has an advantage of high purity of implanted impurities but it requires a mass analyzer and suffers from a disadvantage that the apparatus is costly and running cost is high.

On the other hand, non-mass-analyzed ion implantation has been proposed especially for formation of a pn junction of solar batteries and it eliminates the above disadvantage (Japanese Patent Application Kokai (Laid-Open) No. 130975/78). In the non-mass-analyzed ion implantation, ions are accelerated and without being subjected to mass analysis, implanted into a target. This ion implantation process dispenses with the mass analyzer so that not only apparatus cost and running cost can be reduced but also travel path of ion beams can be reduced to reduce loss of ions due to collision and scattering of the ion beams at the inner wall of the apparatus. The reduction in loss leads to an increase in implantation current which in turn increases implantation processing speeds.

The conventional non-mass-analyzed ion implantation, however, simply implants all the ions of nonselected species into the target and has difficulties with the use of compound source materials. Further, even with an elemental source material such as for example red phosphorus, a phosphorus vaporizer is required and it is necessary to prevent solidification of vaporized phosphorus by heating an atmosphere between the phosphorus vaporizer and the ion source up to about 400.degree. C.

It is an object of this invention to provide an ion implantation process which can use compounds as a source material without selecting ion species in advance.

According to one aspect of this invention, ion beams containing ions of different species and electrically accelerated are scanned magnetically to obtain, on a target surface, a distribution profile of implanted ion doses which depends on ion species. As the source material, a hydrogen compound is most preferably used which has a large ion mass ratio among ion species and which is easy to handle. As the source material adapted for a silicon target, phosphine PH.sub.3 or arsine AsH.sub.3 is most preferably used. Diborane B.sub.2 H.sub.6 or the like may also be used.

For ion beam scanning, a magnetic scanning system suitable for handling large ion beam currents of a large spot size is preferably used.

Other objects, features and advantages of this invention will become apparent from the following detailed description taken in conjunction with the accompanying drawing.

FIG. 1 is a spectrum diagram of ion beams produced with a microwave ion source;

FIG. 2 is a graph showing influence of the dose of implanted H.sup.+ ion upon open-circuit voltage of a solar battery; and

FIG. 3 is a schematic diagram of a magnetic field scanning type non-mass-analyzed ion implantation apparatus used for implementing an ion implantation process according to the invention.

A brief description will first be given of ion implantation by way of doping a semiconductor. To dope a semiconductor of a group IV element such as silicon, a group III element such as boron B or aluminum Al is typically used as a p-conductivity dopant and a group V element such as phosphorus or arsenic As is typically used as an n-conductivity dopant. In the ion implantation, either an ion of any one of these elements or ions containing any of these elements are implanted into the semiconductor. The ion for this purpose is typically produced by ionization of an impurity source of a compound by heating or discharging.

For example, when a discharge gas of PH.sub.3 is used, ion species as shown in FIG. 1 are produced, where abscissa represents mass(m)/charge(q) (arbitrary scale). It is revealed in FIG. 1 that in addition to a P.sup.+ ion, P.sub.2.sup.+, P.sup.++, PH.sup.+, PH.sub.2.sup.+, PH.sub.3.sup.+, H.sup.+, H.sub.2.sup.+ and H.sub.3.sup.+ ions are produced. According to the mass-analyzed ion implantation, the P.sup.+ ion is selectively used, whereas according to the non-mass-analyzed ion implantation, all the ions are implanted into the semiconductor substrate simultaneously. Superposed implantation of other ions than the P.sup.+ ion upon P.sup.+ ion implantation was experimented through the mass-analyzed ion implantation process to examine what influence other ion species than the P.sup.+ ion have upon solar battery characteristics. It was revealed by the experiment that ions such as P.sub.2.sup.+, P.sup.++, PH.sup.+, PH.sub.2.sup.+ and PH.sub.3.sup.+ containing phosphorus would be used efficiently as the impurity source. However, experimental results showed that implantation of H.sup.+ ion, at an implantation energy of 33 KeV and at an annealing temperature of 850.degree. C. conducted for 30 minutes particularly, had a greatly adverse influence. In this connection, one may refer to Digest of Technical Papers of 2nd Photovoltanic Science Engineering Conference in Japan, 1980, IP-2, page 2 by H. Itoh et al. The adverse influence of the H.sup.+ ion implantation was observed in the solar battery characteristics, particularly, open-circuit voltage (Voc) and controlling the dose of implanted H.sup.+ ion was proven to be necessary.

To study the reason of the adverse influence, transmission electron microscope observation and DLTS (Deep Level Transient Spectroscopy) measurement were conducted to find that the implanted H.sup.+ ions were vaporized from the crystalline substrate by approximately 100% at a temperature of 700.degree. C. or more but damage in the crystal caused by the H.sup.+ ion implantation created, after annealing, crystal lattice defects and trapping centers which were localized near the junction.

It will thus be appreciated that in some cases, the implantation of non-selected ion species into the target leads to unfavorable results. It has therefore been desired to solve the problems set forth above by taking advantage of the non-mass-analyzed ion implantation which is advantageous in that ions can be implanted into the target at a high dose within a short time with a simplified apparatus.

Generally, ions of different ion species have different charges or masses. When generated ions of different species are accelerated under the application of an electric field, ions of one polarity are selected to produce ion beams which contain various ion species as shown in FIG. 1.

When an electromagnetic field is applied to the ion beams, the different ion species are subjected to different influence by the electromagnetic field. Typically, when a magnetic field is applied in a direction transverse to a travelling direction of the ion beams, ions are acted on by Lorentz force in a direction which is vertical to both the travelling and magnetic field directions and they are dispersed in accordance with values of q/m. By scanning the magnetic field, different ion species can be scanned over different scanning lengths.

The ratio of scanning length between two ion species of equal charge can be so approximated as to be in inverse proportion to a square root of the mass ratio. Thus, percentage of an ion having a large mass increases at the central area of scanning whereas percentage of an ion having a small mass increases at the periphery of scanning. For example, when a univalent ion of 100 mass is scanned over a width of 1 cm, a univalent ion of 1 (one) mass is considered to be scanned over approximately 10 cm width.

It is now assumed that a compound containing boron B and fluorine F is ionized to produce a univalent boron ion and a univalent fluorine ion. Since boron B and fluorine F have an atomic weight of about 10.8 and an atomic weight of about 19.0, respectively, the ratio of scanning width will be about .sqroot.19.0/10.8.apprxeq..sqroot.1.76.apprxeq.1.33. With a target placed at the center of scanning width, the fluorine ion will be implanted at a greater dose than the boron ion but with the target placed at one end region of scanning width, the dose of the implanted ions will be reversed. For phosphor P and fluorine F, the mass ratio is about 1.63 and the ratio of scanning width will become about 1.28. In this manner, for the mass ratio being about 1.6 or more, a ratio of scanning width of about 30% or more can be obtained.

For arsenic and fluorine, the mass ratio is about 3.9 and the scanning width ratio is about 2, indicating that the mass ratio being about 4 or more will provide a scanning width ratio of about 2 or more. Predictably, the dose of implanted fluorine will be approximately halved with the target placed at the center of scanning width as compared to the dose obtained by the implantation without being subjected to scanning.

Halogen compounds are corrosive and difficult to handle. Hydrogen compounds are less corrosive and easy to handle and besides, they have a large ion mass ratio.

A univalent boron ion and a univalent hydrogen ion both generated from diborane B.sub.2 H.sub.6 will provide a mass ratio of about 10.8 and a scanning width ratio of about 3.29. Thus, with these ions, the dose of implanted hydrogen ion at the center can be about 1/3 or less.

A univalent phosphorus ion and a univalent hydrogen ion generated from phosphine PH.sub.3 will provide a mass ratio of about 31 and a scanning width ratio of about 5.6.

With arsine AsH.sub.3, univalent arsenic and hydrogen ions generated therefrom will provide a mass ratio of about 75 and a scanning width ratio of about 8.6.

With phosphine and arsine capable of providing a mass ratio of 25 or more, impurity ion may efficiently be implanted into a semiconductor target and the dose of implanted hydrogen ion may be made about 1/5 or less by matching the width of the target within the scanning width of a univalent phosphorus ion or of a univalent arsenic ion.

Thus, due to the scanning of ion beams according to the present invention, damage in the semiconductor due to H.sup.+ ion implantation as described hereinbefore can be mitigated greatly and also distribution of implanted ions in the target can be made uniform.

The non-mass-analyzed ion implantation can be used with a large ion current. To make use of this advantage efficiently, such an ion source as a microwave ion source capable of readily providing a large ion current may preferably be used but the type of the ion source is not limited thereto.

The center of the scanning width is matched to preferential implantation of heavy ions and the periphery of the scanning width is matched to preferential implantation of light ions.

Scanning of ion beams can readily be effected by means of an electromagnet. The scanning can be done in various manners by changing current in an alternating fashion, superimposing a DC bias or using a sawtooth or sinusoidal wave current.

Experimentally, ion beams were scanned by a magnetic field in the non-mass-analyzed ion implantation process to reveal that based on the difference in mass between a P.sup.+ ion and a H.sup.+ ion, the scanning width for a P.sup.+ ion beam could be made larger than for a H.sup.+ ion beam and the ratio between doses of implanted P.sup.+ ion and H.sup.+ ion could be a fraction or less. At the total dose of P being 5.times.10.sup.15 cm.sup.-2, it is possible to reduce the dose of H.sup.+ ion to 5.times.10.sup.14 cm.sup.-2 which is so minimal that adverse influence of the H.sup.+ ion implantation upon output characteristics of the solar battery can be neglected. Thus, utilizing the present invention, when ion implanting using PH.sub.3 as the source gas the total dose of H.sup.+ ion can be 1.times.10.sup.15 cm.sup.-2 or less when the total dose of P ion is 1.times.10.sup. 15 to 3.times.10.sup.16 cm.sup.-2.

When implanting ions into a semiconductor substrate having a larger area than the beam spot size, the beam is scanned on the substrate and the substrate may be moved continuously in a direction vertical to the beam scanning direction, thereby making it possible to make uniform the dose in the substrate surface and to assure continuous processing.

The invention will now be described by way of example with reference to FIG. 3.

A target 8 was a p-conductivity silicon substrate of a resistivity of 3 .OMEGA..multidot.cm and a non-mass-analyzed ion implantation apparatus comprised a microwave ion source 1, a beam scanning magnet section 2, and an implantation chamber 3. The ion source 1 includes an ion generating chamber 4 and an accelerating section 5. Acceleration energy was normally of the order of KeV or more. Magnets are disposed above and below the sheet of drawing and the target is moved vertically to the sheet of drawing. A discharge gas of PH.sub.3 was used, ion beam was scanned so that the ratio between a beam current density not subject to the scanning and that subject to the scanning is about 0.6 to 0.7, and the total dose was set to 5.times.10.sup.15 cm.sup.-2. As diagrammatically shown in FIG. 3, the scanning width of P.sup.+ ion is designated by reference numeral 6 and that of H.sup.+ ion by 7. The ratio between the total dose and the dose of H.sup.+ ion when not subjected to the beam scanning was 0.2 but it became 0.1 or less when subjected to the beam scanning. After the ion implantation, the target was annealed in an atmosphere of dry N.sub.2 gas at a temperature of 850.degree. C. to form an n.sup.+ p junction. Electrodes were formed on front and rear surfaces of the target by printing, and an anti-reflection coating film was formed, thereby completing a solar battery. The thus prepared solar battery exhibits a conversion efficiency of 14 to 14.5% under a condition of AM1 solar simulator, which conversion efficiency is comparable to that of a solar battery prepared by the mass-analyzed ion implantation. This means that the adverse influence of H.sup.+ ion implantation can substantially be eliminated. The term AM1 represents solar spectrum at earth's surface for optimum conditions at sea level, sun at Zenith.

For ion implantation into a 100 mm.phi. wafer, it usually takes 20 to several of tens of seconds in the case of the mass-analyzed ion implantation but several seconds in the case of the non-mass-analyzed ion implantation. This means that the wafer processing speed can be increased and the pn junction processing cost can be reduced. The discharge gas of a hydrogen compound used in this embodiment is easy to handle. Further, the scanning of ion beams by the magnetic field reduces the dose of H.sup.+ ion as compared to the dose of dopant ion so that the adverse influence of H.sup.+ ion implantation can be neglected. In addition, the beam scanning is effective to make uniform the ion implantation into the substrate.

The ion implantation apparatus which dispenses with the mass analyzer can be simplified and made inexpensive, and the elimination of a magnetic field generator for mass analysis can reduce the running cost. Further, the reduced ion beam travelling path can suppress reduction of beam current in the course of the travelling path, thus increasing the amount of beam current. Advantageously, the ion implantation processing speed can be increased and the processing cost can be reduced.