[1]United States Patent Office 3@463,977 3,463,977 OPTIMIZED DOUBLE-RING SEMICONDUCTOR DEVICE Andrew S. Grove, Palo Alto, Calif., Otto Leistiko' Trorod, Denmark, and Ronald J. Whittier, Los Altos, Calif., assignors to Fairchild Camera and Instrument Corporation, Syosset, N.Y., a corporation of Delaware Filed Apr. 21, 1966, Ser. No. 544,229 Int. Cl. HOll 11/00, 15100 U.S. Cl. 317-235 6 Claiins ABSTRACT OF THE DISCLOSURE The metal contact to one region of a planar device containing a dish-shaped PN junction extending to one surface of the device, is extended across, but insulated from, the intersection of the PN junction with the surface. By selecting the correct insulation thickness, the breakdown voltage of the semiconductor PN junction is made equal to the breakdown voltage of the PN junction induced by the extended contact and thus the device breakdown voltage is maximized. This invention relates to a semiconductor device wherein a junction is employed and an avalanche current occurs at aparticular reverse bias voltage referred to as the breakdown voltage. In particular, this invention relates to double-ring diode or transistor strticture such as described in U.S. patent application Ser. No. 506,750, filed Nov. 8, 1965, and @ssigned to the assignee of this invention wherein the breakdown voltage attainable is optimised by preventing the breakdown of a field-induced junction. In the aforementioned patent application, a diode and transistor structure were described wherein, in addition to the usual structure, a contact layer was included. This layer was connected to one of the semiconductor re-ions and with a portion atop the protective material overlying the depletion region associated with a PN junction. Also, a charged-particle sink was included laterally exterior to the portion of the contact layer atop the protective material. This structure is herein referred to as a double-ring device. The double-rin@ struettire increases the breakdown voltage while maintainin,- the stability of the device. However, it has now been discovered that this arrangement while advantageous does not permit optimum results unless certain parameters are controlled. Thus, this invention provides a double-ring structure wherein optimum results (e.g., breakdown voltages of 750 volts or greater) may be achieved. Briefly, one embodiment of a device incorporating the invention comprises a first semiconductor re@ion of a first conductivity type having a surface; a second semiconductor region of opposite conductivity type located adjacent said first region and forming a PN junction th-@rewith that extends to said surface; a layer of protective material overlying at least a portion of said surface including said PN junction; and, a contact layer located adjacent said protective niaterial overlying the depletion region associated with said PN junction, said layer of protective material having a thickness selected to improve the breakdown volta,-e of said device. The invention will now be described with reference to the drawings, wherein: FIG. I is a simplified cross-sectional view of a prior art semiconductor device; FIG. 2a is a simplified cross-sectional view of one embodiment of the double-ring structure showing the depletion region prior to the formation of an inversion layer; Patented Aug. 26, 1969 2 FIG. 2b is a simplified cross-sectional view of one embodiment of the double-ring structure showing the depletion region after the formation of the invention layer; FIG. 3 is a graph of reverse biased diode current versus reverse bias voltage for various voltages applied to portion 26 of the contact layer; FIG. 4 is a graph of the breakdown voltage, BV, versus the voltage, VIG, applied to the portion 26 of the contact layer; 10 FIG. 5 is a graph of operational breakdown voltage, BVop, versus protective material thickness; FIG. 6 is a graph of the breakdown voltage of the field-induced junction, BVFJ, versus surface concentration for devices havin- different protective material 15 thicknesses; and, FIG. 7 is a graph of BVF,J versus protective material thickness for devices with various surface concentrations. R.cferring to FIG. 1, a prior art passivated-surface semiconductor device, such as semiconductor device 10 20 of monocrystalline silicon, is shown. The upper surface 20 of the device is covered with an insulating and protective layer 12 (e.g., oxide of silicon) that passivates the surface. Tbe device has a first region 14 and a second region 16 located adjacent thereto and, in particular, 25 nested within first region 14. The first region 14, has a first conductivity type which may be lightly doped p-type (e.g., p-), while the second region 16 has an opposite conductivity type which is then heavily doped n-type (e.g., n+). The regions 14 and 16 form a PN jtinction 18 30 which is island shaped and extends to substantially flat surface 20. In addition, a channel stop region .111 of the same conductivity type as the first region 14 but having hi.-her doping concentration is formed such as to completeiy surround the PN junction 18. A contact layer 24 3r, (e.g., Al) is electrically connected to second region 16 that has a portion 22 of its surface exposed by an opening in the insulating layer 12. A contact 27 is also electrically connected to first region 14. This contact may be attached to either surface 20 or surface 28. In this 40 prior art device, when a reverse bias is applied to regions 14 and 16 via the contacts, a depletion region 40 is created with an electric field existing across the depletion region. The intensity of the electric field across depletion region 40 is particularly high in the vicinity of cor45 ners 42 and surface 20, thus resulting in a relatively low breakdown voltage of the device. This br6akdown Voltage in commercially available diffused planar devices is typically in the range of 30-200 volts. In order to overcome the low breakdown voltages en50 countered in prior art diffused planar devices, the doublering strticture was invented. This structure is described in detail in the above-referenced U.S. patent application. A semiconductor device 10 incorporating one embodiment of the double-ring structure is shown in FIGS. Za, 55 and 2b, wherein structural elements that are similar to the ones shown in FIG. I bear the same numerical designation. Briefly, the double-ring structi-ire includes first region 14, second region 16, insulating layer 12, surface 20 and contact 27, which are described above with ref60 erence to FIG. 1. This structure is changed by including a contact layer 24 that has an extended portion 26 which lies atop the insulating layer 12 and extends laterally beyond PN junction 18 to overlie depletion region 40 all around PN junction 18. It should be understood that ex65 tended portion 26 may be either continuous with the portion of the contact layer 24 that makes electrical contact with second region 16 or it may be discontinuous as shown in FIG. 2a. The contact layer 24 and its portion 26 are preferably in direct electrical connection either 70 via the leads connected thereto or by poition 26 being a direct extension of the part of contact layer 24 that cont,,icts second region 16. Portion 26 may be connected
[2]3 to a potential or a source different than the potential or source that is connected to the part of contact layer 24 that contacts region 16. In addition to extended portion 26, a charged-particle sink 30 fornied @as a metal layer or film is added to the prior art structure shown in FIG. 1. Sink 30 (e.g., layer of Al) is adapted to collect char.-ed particles present in or on top of insulating layer 12 and is located over insulating laver 12 laterally exterior to contact layer 24 and portion 26 and completely surrounds portion 26. Sink 30 is preferably attached to the heavily doped region 11 formed in first region 14. This may be accomplished by extending sink 30 to make direct electrical contact with region 11 via an opening in layer 12. Alternatively, sink 30 may be attached to a different potential or via a conductor not continuous with the layer forming sink 30 to a portion of region 14 not -adjacent or in the vicinity of second region 16. Thus, extended portion 26 and sink 30 form a double-ring -,iving rise to the name double-ring structure which this type of structure is designated. The portion 26 and sink 30 may also be referred to as the inner and outer rings or gates, respectively. The portion 26 and sink 30 have enabled higher breakdown voltages to be achieved while stability is at least maintained and in many instances is improved. This improvement in breakdown voltage characteristic is achieved by the application of a I>otential to portion 26 which alters the electric field associated with depletion region 40. The electric field associated with the depletion region is altered in such a manner that a lower intensity field exists in the vicinity of corner 42 and near surface 20 than normally existed in planar diffused devices. In exploring the limits of breakdown volta@e that may be obtained with the double-ring structure, it has been discovered that when a votta@e eqtial to or larger than a particular voltage, VT, is applied to portion 26, an inversion layer 50 (shown in FIG. 2b) is created in first region 14 beneath portion 26 and extending outwardly from junction 18. The inversion layer 50 has a conductivity type the same as second region 16, thereby forming in effect an extension of this region. The inversion layer 50 forms a field-induced junction 52 substantially continuous with junction 18, which like junction 19 has a depletion region and electric field associated therewith. At the corner 54 of fieldinduced junction 52 and near the surface 20, the electric field is at a high intensity, whereby breakdown of the field-induced junction occurs. This breakdown of the fieldinduced junction may occur at a voltage substantially lower than the breakdown voltage of junction 18. This phenomenon is illustrated by the ,@raphs of FIGS. 3 and 4. The symbol BV herein employed refers to the breakdown voltage of the fabricated junction, while the symbol BVpjj refers to the breakdown voltage of the field-induced junction. FIG. 3 shows a current-versus-voltage graph with each curve representative of a difi.erent voltage, VIG, applied to portion 26. As the voltage, VIG, applied to the portion 26 is increased (VG=l and YIG=2, etc.) the breakdown voltage is also increased. Ibis corresponds to the formation of a depletion region under portion 26 which is shaped by portion 26 to reduce the fields in the vicinity of the corner of the junction and near surface 20. However, when the voltage, VIG, applied to portion 26 is large enough (e.g., YG=3), an inversion layer 50 is formed and biased to a potential which exceeds the breakdown voltage of the field-induced junction, BVFJ. As a result, an avalanche current will start flowing between the end of inversion layer 50 and region 14. This avalanche current must go through the series resistance of the inversion layer 50, thereby giving rise to a saturated currentvoltage characteristic as indicated in FIG. 3. The breakdown voltage BV occurs at a higher voltage Vn. This effect is summarized in the graph of FIG. 4, wherein curve A shows that as the voltage, Vjr,, apphed to portion 26 is increased, the voltage at which the reverse current ex3,463,977 4 ceeds some low value, e.g., I microamp, will increase to a point X, at which point this voltage experiences a sharp decline caused by the breakdown of the fieldinduced junction. It has been found that the creation of an inversion layer in a semiconductor material occurs at a particular voltage, VT, applied to portion 26. The magnitude of VT depends on the reverse bias applied to the junction 18. In particular, when the reverse bias is large enough that 10 breakdown of the field-induced junctic>n can take place, i.e., equal to BVF,J, the value of this voltage is given by the Equation 1: VT =- Q.- xo+BVrjj+20r15 Koco Xo 2K@coq(NA7ND) (BVFIJ+2g5F) wherein: 20 Q,,=Effective surface-state charge density Ko=dielectric constant of protective insulating material eo=permittivity of free space, 8.85XJO-14 fd/cm. xo=thickness of the protective insulating material K,,=dielectric constant of the semiconductor 25 q=unit of electron charge, 1.6021 X 10-19 coulombs NA7ND=substrate net impurity concentration BVF,j=breakdown voltage of the field-induced junction OF=potential of the Fermi level of the substrate majority carrier measured from the intrinsic Fermi level in the 30 substrate. All applied voltages are referred to the potential of the substrate. The positive sign is taken for an n+p diode, the negative sign for a p+n diode. In the rest of this dis35 cussion, we restrict ourselves to the n+p case, unless otherwise indicated. The discussion, however, is equally valid for the p+n case with proper changes in polarities. The creation of the inversion layer 50 in itself is not problem. It is the creation of the inversion layer 50 40 cotipled with the application of a reverse bias lar.-e enou.-h that the breakdown of the field-induced junction can take place, i.e., equal to BVpjj, that results in premature breakdown. Thiis, in order to optimize the operation of the double-ring structure and solve this problem, the 45 voltage VT should be of such value that the potential applied to portion 26 will not exceed this value, since in that case a field-induced junction biased to a potential greater than BVF,J cannot exist. It has been found that the value of VT may be increased by controlling the 50 thickness xo of protective material 12, by controlling the value of the surface-state charge density, Q,@, or by controlling the value of the breakdown voltage of the fieldinduced junction, BVFJ. According to this invention, it is preferred that the thickness of protective material 12 6,5 be increased to increase the value of VT. However, it is within the broad scope of the invention to control any one or all of these variables (Q,,, BVF,J and xo) to achieve a desired value of the volta,-e VT. The otber parameters of Equation I depend on the materials se60 lected for a particular device. It is preferred that the value of voltage, VT, be increased by varying the thickness of protective material 12 because the thickness of the protective material may readily be varied, measured and precisely controlled by 65 present techniques. In addition, the variations in the thickness of protective material 12 have a substantial effect on VT because an increase in this thickness also brings about an increase in BVFJ (as will be discussed later), thereby resulting in a greater effect on VT than ex70 plicitly indicated by Equation 1. The only drawback with varying the thickness of the protective material to vary the value of the voltage, VT, is that too large an increase in thickness will result in removing the portion 26 too far from the surface 20 and 75 thereby decrease its influence on the junction breakd(>wn
[3]3,463,977 5 voltage. This is illustrated in FIG. 5, wherein the device tested had portion 26 connected to contact 24. It is shown in FIG. 5 that the breakdown voltage (which may specifically be referred to as operational breakdown voltage, BVop) attainable is increased along curve D as the 5 thickness of the protective material is increased until optimum point Y is reached. At this point further increase of the thickness of the protective material will result in a decrease of the breakdown voltage of the particular device in accordance with curve E. It seems that after the 10 optimum point the breakdown of the field-induced junction 52 will no longer present a problem, but that after this point the effectiveness of the field resulting ftom the potential applied to portion 26 will diminish and breakdown will occur at the corner of junction 18 near 42. 15 Thus, along curve E breakdown is due primarily to semiconductor junction 18, while alon.- curve D breakdown volta-e is due primarily to the field-induced junction 52. From a study of FIG. 51 it can be seen that the slope of curve E is much less than that of curve D. Thus, devices 20 havin.- a protective material thickness which falls along curve E permit greater thickness tolerance. Generally, if protective material 12 is silicon dioxide, it should have a thickness in excess of approximately 1.25 microns and, more specifically, between approximately 1.20-2.5 mi25 crons. For many devices the optimum point has been 1.6+0.1 microns. The minimum thickness of the protective material for a particular value of voltage VT may be determined from the Equation 2: VT-BVF,j20F 30 zo KoEo - -@2K.EoO(NA7ND) (BVFIJ+20F) -Q.. wherein: xomw is the minimum thickness of the protective material f(>r which breakdown of the field-induced junction is prevented, and all other symbols and sign con35 vention are as given following Equation 1. In summary, the layer of protective material 12 should hdve a thickness that enables the breakdown voltage to be increased by increasing the potential applied to por40 tion 26, btit not subsentially in excess of the thickness existin at the optimum point for the particle device. Al9 ternatively, the thickness of the protective material 12 should be such that the breakdown voltage of the device is attributable to the breakdown of the semiconductor junction rather than the field-induced junction. 45 The effect of increasing the thickness of the - protective material can be further understood by reference to FIG. 4, wherein curve A represents a device having a first protective material thickness and curve B represents a device having a thinner prolective material. At a first potenial 50 applied to portion 26, such as 100 volts, the device of curve B with its thinner protective material thickness displays a slightly greater breakdown voltage. However, with 250 volts applied to portion 26, the device of curve ' A 55 haying a greater oxide thickness displays the higher breakdown voltage. Thus, increasing the oxide thickness may lower the breakdown voltage attainable with lower potentials applied to portion 26, while increasing the maximum breakdown voltage attainable. The breakdown Voltage with the lower potentials applied to portion 26 is not 60 particularly important, since in the preferred embodiment of the invention, the voltage applied across junction 18 and voltage applied to portion 26 are substantially the same because of the interconnection between the portions of 6,5 contact layer 24. Thus, a large voltage Vjr, will be applied to portion 26 when a large breakdown voltage is required. The breakdown voltage attainable may also be increased by increasing the value of the breakdown volta.-e of the field-induced junction, BVplj. The value of the breakdown 70 voltage, BVFJ, is largely determined by the surface concentration Cs of the impurity, the thickness of protective material 12, and the field existinbetween the particle sink 30 and the inversion layer 50. With regard to the latter factor, it is preferred that sink 30 is maintained at approxi- 75 6 mately the same potential as region 14 by directly connecting sink 30 thereto. Consequently, the field between inversion layer 50 and sink 30 is usually fixed. In some instances, it may be desirable to increase the breakdown voltage BVFJ by altering the field between inversion layer 50 and sink 30 by connecting sink 30 to a different potential. Thus, by connecting sink 30 to some external source of voltage of the same polarity as that applied to contact layer 24, the electric field at the end of the fieldinduced junction 54 will be reduced whereby the breakdown voltage BVF,J is increased. The manner in which the breakdown voltage of the field-induced junction BVFJ varies with the surface concentration of the dopant or impurity is shown in FIG. 6 wherein the various curves are for devices having different protective rnaterial thicknesses. In general, the highest value of the breakdown voltage of the fieldinduced junction is attained at the lowest surface concentrations and the largest protective material thickness. This is clearly shown in FIG. 7 wherein the fieldinduced junction breakdown voltage BVF,J is plotted as a function of protective material thickness for devices with various surface concentrations. It can be seen that a device with a surface concentration in the vicinity of 1014 atoms/cm.3 and a protective material thickness of 1.5 microns has an extremely high BVF,J. It can be seen from FIG. 7 that essentially the same value for BVFjj may be obtained at higher surface concentrations by increasing the thickness of the protective material. The gnly limitation on controlling BVFJ by increasing the protective material thickness is the effect that increasing the protective material has on the effectiveness of the potential applied to portion 26 as depicted in FIG. 5. In any event, it is preferred that the breakdown voltage of the field-induced junction, BVplj, be controlled primarly by altering the oxide thickness and, secondly, by controlling the surface concentration and maintaining this concentration as low as possible cosistent with other processing and device requirements. In summary, to maximize the breakdown voltage attainable by the double-ring structure, the value of the voltage VT at which an inversion layer 50 capable of breaking down is formed, should be increased and made as large as possible consistent with maintaining the effectiveness of portion 26 and the other device requirements. Th6 value of the voltage VT may be most readily increased by increasing the thickness of the protective material. When the polarity of the surface state charge density, Qss, is.proper, an increase in the protective material thickness will increase every term in Equation I with the exception of the term 2OF- The oxide thickness should not be increas@-d much beyond the optimum point nor should it be substantially less than the value at the optimum point in or r to achieve maximum breakdown voltage. In tests it has been shown that by increasing the protective material thickness from 0.75 micron to 1.75 microns, the breakdown voltage of devices using the double-ring structure may be increased from 350 volts to 720 volts. Although this invention has been disclosed and illustrated with reference. to a particular embodiment, the princi es involved are susceptible of numerous other embodimeiats which will be apparent t(> persons skilled in the art. The invention is, therefore, to be limited only as indicated by the scope of the appended claims. What is
