US 3750055 A
A phase shifter for microwaves is monolithically formed from a silicon panel with a grounded metallic bottom layer and a metallic top layer defining a central conductor strip longitudinally subdivided into several line segments by junctions with reactive shunts spaced a quarter wavelength apart. Each shunt path includes the series combination of a condenser and a diode both integral with the monobloc panel. Successive diodes are provided, in pairs, with common biasing leads for alternately blocking and unblocking same to modify the loading effect of the shunt reactances, thereby enabling the selective introduction of a predetermined phase-shift increment per line segment.
Claims available in
Description (OCR text may contain errors)
United States Patent 1 1 11 3,750,055 1 1 July 31, 1973 Funck  INTEGRATED PHASE-SHIFTING 3,597,706 8/1971 Kibler 317/235 D X MICROCIRCUIT 3,491,314 [/1970 White 333/31 R  inventor: Ronald Funck, Port Marly, France Primary Examiner paul L Gensler  Assignee: Thomas-CS1", Paris, France Attorney-Karl F. Ross  Filed: Dec. 14, 1970  ABSTRACT [211 App! 98077 A phase shifter for microwaves is monolithically formed from a silicon panel with a grounded metallic  Foreign Application Priority Data bottom layer and a metallic top layer defining a central Dec. 16, 1969 France 6943512 conductor Strip longitudinally subdivldfid into Several line segments by junctions with reactive shunts spaced 52 Us. or 333/31 R, 317/235 D, 333/7 a quarter Wavelength apart Each Shunt P includes 5 1 1 Int. Cl. H0lp 1/18 the Series wmbihalioh of a condenser and a dim? both  Field of Search 333/31 R, 81 A, 81 R, integral with the monobloc Panel Successive diodes 3 7; 7 35 D are provided, in pairs, with common biasing leads for alternately blocking and unblocking same to modify 5 References Cited .the loading effect of the shunt reactances, thereby en- UNITED STATES PATENTS abling the selective introduction of a predetermined phase-shift increment per line segment. 3,475,700 l0/l969 Ertel 333/7 3,454,906 7/1969 Hyltin 333/31 R 19 Claims, 8 Drawing Figures PAIENIEU 3. 750. 055
suwlnrs FIGQI II 0, I 0 T 0,, 5 0
S 2 J C J IL/ 5% 2 m- RONALD FUNCK INVENTOR.
g in RNEY PAIENIEL JUL 3 1 1973 saw 2 or 3 FIG.2
RONALD FUNCK INVENTOR.
BY 9M ATTORNEY INTEGRATED PI-IASE-SIIIFTING MICROCIRCUIT My present invention relates to an integrated circuit serving as a linear phase shifter for microwaves.
Selective phase shifts by different magnitudes may be produced in a transmission line by subdividing same into a plurality of line segments, each of these line segments being individually switchable to introduce a predetermined phase-shift increment. The switching means associated with each line segment may include a diode alternately blocked or unblocked by application of a suitable biasing potential.
A principal object of my present invention is to provide an improved phase shifter of this general type which is of compact construction, inexpensive to manufacture, and operative with great fidelity in a range of ultrahigh and superhigh frequencies.
This object is realized, pursuant to the present invention, by the provision of a dielectric panel of semiconductor material, preferably silicon, with a grounded first metallic layer on one major surface thereof (referred to hereinafter for convenience as the bottom layer) and a second metallic layer on an opposite major surface (referred to hereinafter as the top layer), the latter layer forming a conductor strip with longitudinally spaced branch points defining the aforementioned line segments. Each branch point represents a discontinuity by forming a junction with a reactive shunt path constituted in part by the top layer and including a diode integral with the body of the panel. This diode is of the PIN type, i.e., a junction diode with two heavily doped layers of opposite conductivity types (P and N) which are separated by an intermediate layer of substantially intrinsic conductivity (relatively free from conduction-determining impurities) and are portions of the panel body disposed adjacent the two major surfaces thereof. These diodes are provided, individually or preferably in pairs, with biasing means for selectively altering their conductivity to modify the magnitudeof the phase shift introduced by any of these line segments.
Advantageously, in accordance with a further feature of my invention, each diode is connected in series with a capacitance which is also integral with the monobloc panel of semiconductor material. With each metal layer separated from the panel body by a respective nonconductive layer, preferably consisting of an oxide of the semiconductor material, the capacitance may be a condenser of the MOS (metal-oxide/semiconductor) type constituted by one of these metal layers, the adjoining nonconductive layer and a doped portion of the panel body forming part of the associated diode. The biasing potential may be applied to the last-mentioned doped portion constituting the junction between the diode and its series capacitance.
Such a diode may be connected either in series or in parallel with a transverse conductor, defined by the top metal layer, which extends from the corresponding branch point and is in conductive contact with one of the doped layers of the associated diode. The transverse conductor, representing a significant inductance at a contemplated operating frequency, may be returned to ground (for high frequencies) at the bottom metal layer by way of the diode and series condenser or through a separate bypass condenser also integral with the panel body. If the loop length is about a quarter wavelength, it will represent an open circuit as seen from the main line in which case the transverse conductor may serve merely as a biasing lead, the diode and its series capacitance being then disposed close to the main-line conductors for capacitively bridging or effectively short-circuiting them to establish a nodal point in the conductive state of the diode. With the diode serially included in the loop, the latter may measure about one-eighth of a wavelength to enable selective switching between capacitive and inductive shunts over a wide frequency range.
The above and other features of my invention will be described in detail hereinafter with reference to the accompanying drawing in which:
FIG. 1 is a diagrammatic view of a phase which the invention is applicable FIG. 2 is a fragmentary plan view of a physical realization of the phase shifter of FIG. 1 in accordance with my invention;
FIG. 3 is a cross-sectional view taken on the line III III of FIG. 2;
FIG. 4 is a fragmentary sectional view taken on the line IV IV of FIG. 2, showing a modification;
FIG. 5 is a view similar to FIG. 4, illustrating a further variant;
FIG. 6 is a plan view similar to FIG. 2, representing another embodiment;
FIG. 7 is a cross-sectional view taken on the line VII VII of FIG. 6; and
FIG. 8 is a view similar to FIG. 6, illustrating still another modification.
FIG. 1 shows a transmission line with an ungrounded conductor L and a grounded conductor M energized, from a source not further illustrated, with highfrequency oscillations O which may fall within the gigacycle range. At regular intervals (even though suchregularity is not always necessary) the line is subdivided into relatively short segments defined by junctions P, Q, R between the main conductor L and several transverse conductors T T T T lying in pairs on opposite sides of this main conductor. Each of these transverse conductors forms a shunt path by being returned to ground at conductor M in a loop including a respective diode D D D D in series with an associated condenser C C C C A biasing voltage is applied shifter to to the two diodes D D in the two loops bounding the line segment P-Q, by a common control lead S terminating at the junctions J J of their cathodes with the associated condensers C C In an analogous manner, the junctions J J of the diodes D D with condensers C C are connected to a control lead S" for the segment QR. If the corresponding diode pair is blocked by the application of a sufficiently positive bi asing voltage, the loading effect of the corresponding branch upon the main line is substantially eliminated. Conversely, the presence of an unblocking (here negative) biasing potential places the inductance of the loopas a shunt impedance across the two main line conduc" tors. Thus, the overall phase delay of the two-wire line can be altered at will, within a predetermined range (e.g., of and by a number of steps determined by the number of available line segments, through the selective open-circuiting and short-circuiting of the associated diodes. With the length of the loop defined by any of the conductors T T, etc. equal to about oneeighth of the free-space wavelength of the impressed oscillations O, which substantially corresponds to the propagation wavelength along the unloaded line, the
shunt impedance at points P, R etc. will be either capacitive or inductive, according to the state of the corresponding diodes. It will be understood that the remainder of the line L, M, indicated in dotted lines in FIG. 1, may contain any desired number of further branch points and reactive loops.
In FIGS. 2 I have shown several closely related structural embodiments of the system of FIG. 1. In each instance, the main line L, M and branches T,, T, etc. are constituted by a monobloc panel 1 of semiconductor material, specifically silicon, providedwith oxide coatings 2, 3 on its lower and upper major surface. A metallic bottom layer, extending over the entire oxide layer 2, represents the grounded conductor M whereas a similar but discontinuous top layer above oxide layer 3 constitutes the main conductor L and its branches. The body of silicon panel 1 is heavily doped at P+ and N+ to form the anodes and cathodes of the several diodes D,, D, etc. These doped silicon portions, it will be noted, are separated by regions I of substantially intrinsic conductivity. Each zone P+, i.e., the anode of each diode, is in direct contact with an extremity of the associated branch conductor T, etc. through a small aperture provided in the upper oxide layer 3; the zones H- are small blobs centered on these apertures, whereas the opposite zones N+ are relatively wide and are common to respective diode pairs such as the diodes D,, D, or D,,, D v
The top view of FIG. 2 is the same for the several modes of realization according to FIGS. 3, 4 and 5 described hereinafter. In the basic embodiment, illustrated in FIG. 3, the biasing voltage is fed directly to the N+ layers of respective diode pairs via incisions 4 formed along the edges of the panel 1, each incision extending into the region of a respective zone N-land being provided with a metal lining Sa connected to a source of control voltage not illustrated in these Figures.
In the modification illustrated in FIG. 4, each zone N+ common to a pair of diodes such as D,, D, is extended at 6 into the immediate vicinity of the upper oxide layer 3 which is traversed by a metallic terminal Sb (of the same outline in plan view as the incisions Sa of FIGS. 2 and 3) connected to the source of control voltage. A bottom recess 7 in the silicon body 1, underneath the extension 6 of the N+ layer, serves to reduce the thickness of that body so as to minimize the depth of the zone of semiconductor material into which the necessary impurities must be diffused.
As shown in FIG. 5, which is particularly applicable I to panels of considerable overall thickness, the recess 7 of FIG. 4 may be replaced by a wider cutout l7 underlying the corresponding diodes D,, D, so as to minimize also the depth of diffusion in the region of these diodes. x
Naturally, the conductivity types of the illustrated P and N zones could be interchanged with corresponding reversal of the polarity of the applied biasing potential.
The entire assembly may be enclosed in a protective housing, not shown, with the necessary biasing leads as well as input and output connections, e.g., in the form of coaxial cables. These connections could also include microstrip lines of the two-conductor or threeconductor kind, i.e., with one grounded and one ungrounded metal facing or with two grounded facings and an ungrounded central metal strip.
Even though the body of panel 1 may be lightly doped with N-type or P-type impurities, to control its resistivity which preferably should be greater than 1,000 ohm-cm, it may be regarded as substantially free from impurities and therefore intrinsic" as compared with the active zones N+ and P+. These active zones are produced in conventional manner by diffusion of the proper impurities from the respective panel surfaces. Recesses as shown at 4, 7 and 17 may be formed by selective chemical erosion, the conductive coatings on the body surfaces or their oxide layers being produced by known techniques such as vapor deposition of metal through a mask or over the entire surface with subsequent partial removal by photolithographic means.
The condensers C,, C, etc., which insulate the metallic top layer from the grounded bottom layer for direct current, may have such large capacitances as to constitute virtual short circuits for the high-frequency incident waves. If desired, however, these capacitances as well as the effective capacitances of the associated diodes could be designed as significant impedances in series with the loop inductance. These capacitances, therefore, as well as the dimensions of the loop conductors can be selected to establish a desired response characteristic. By virtue of their monolithic character, the diodes in their blocked state may be considered as having a capacitance equal to zero if their actual capacitance per unit length of the associated line conductors matches the distributed shunt capacitance of the line. This is true whether the diodes are disposed at theremote extremities of the branchconductors as seen from central strip L, in the manner illustrated in FIGS. 1 5, or are located next to this central strip as shown in FIGS. 6 8 described hereinafter.
The distance d (FIG. 2) separating consecutive branch points P, Q, R may equal 'y/4'where 'y is the mean wavelength of the incident oscillation 0 (FIG. 1). The length of each branch conductor T, etc., and therefore of the corresponding shunt path, may be on the order of 7/8. The impedance of the main line L, M should be low compared with that of the shunt path to provide a large selective phase-shift increment per line segment. I
As illustrated in FIGS. 6 8, the reactive loops formed by branch conductors T, etc. in series with diodes D, and condensers C, etc. may be replaced by similar loops constituted by conductors S,, 8,, S,,, S,,, also part of the top metal layer, which are galvanically separated from central strip L and also serve as biasing leads for the associated diodes D,, D D,,, D,,,, the loops being completed by coupling or bypass condensers CP integral with the panel body 1 serving to ground the remote ends of these conductors for high frequencies. The diodes D, etc. are here shown disposed in the immediate vicinity of central strip L from which their anodes P+ are insulated by the upper oxide layer 3 while being capacitively coupled thereto. In the embodiment of FIGS. 6 and 7, strip L has a reduced section at each branch point constituting the upper plate of the corresponding condenser C,, C,, C,,, C,,; the lower condenser plate, i.e., the doped zone P+ forming part of the corresponding diode, is directly connected to the associated lead S, etc. through an aperture in the layer 3. The branch leads, whose length may range between about 7/ 8 andry/4, are interconnected in pairs 5,, S, and S,,, 8,, by conductive links 9 in the form of relareduced resistance. Recesses 8 and 11 on the underside of panel 1 again serve to minimize the depth of diffusion in the region of the diodes and the coupling condensers, respectively. The biasing potential for the diodes is supplied by control leads, not shown, soldered or otherwise connected to the metallic links 9.
In the embodiment of FIGS. 6 and 7, in which the blocking condensers C, etc. lie athwart the main line L, M, the branches extending to opposite sides of central strip L must be relatively staggered, as shown. Such a staggering is unnecessary in the modified arrangement of FIG. 8 where these blocking condensers are shifted sideways with reference to strip L, their upper plates being constituted by lateral projections of that strip. In either case, the diodes D etc. are now connected in parallel with the corresponding line loops which may thus be selectively shorted out or, if desired, resonated by the capacitances C, etc. unless, by being designed as quarter-wavelength lines, they are permanently removed from the high-frequency path.
In the construction of FIGS. 6 8 the N+ zones are individual to the several diodes so as not to introduce a conductive shunt for the associated line segments.
The conductive zones 10 may be produced by diffusing a given type of impurity into the panel body from the top and the bottom thereof.
The microcircuit according to my invention, enabling selective phase shifts up to, say, 90", may be coupled to a reflex-type phase shifter connected across the output end of the transmission line L, M. Such a reflex phase shifter, switchable between 0 and 180, serves to double the phase-shift increment selectively introduced by each line segment P-Q, Q-R, etc. and may comprise another diode, similar to those described above, adapted to be alternately biased into a blocking or an unblocking condition. As disclosed in commonly owned application Ser. No. 97,750 filed by Claude Vergnolle on even date herewith, now U.S. Pat. No. 3,705,366 the terminal diode of the reflex phase shifter may be connected to the main line L, M by way of a thin wire or ribbon in series with strip L having a length equal to a small fraction of 'y and constituting a significant inductance at the operating frequencies considered, an intermediate point of this inductance being connected to ground through a condenser whose capacitance is of the same order of magnitude as the effective capacitance of the terminal diode.
The system described and illustrated has particular utility, for example, in a sweep circuit for an antenna array of a radar transmitter as disclosed in commonly owned U.S. Pat. No. 3,448,450.
It will thus be seen that I have provided a linear phase shifter of compact construction, in the form of an integrated monolithic microcircuit, which can be conveniently manufactured by'the use of current technology with elimination of all assembly work and all separate handling to complete the internal connections. With frequencies in the gigahertz range, the panel structure will have small dimensions and may be mounted on associated components of a similarly compact nature.
. rial with a grounded first metallic layer on one major surface thereof, a first nonconductive layer separating said first metallic layer from said panel, a second metallic layer on an opposite major surface, and a second nonconductive layer separating said second metallic layer from said panel, said second metallic layer forming a conductor strip with branch points longitudinally subdividing same into a plurality of line segments; a plurality of reactive shunt paths between said strip and said first metallic layer formed in part by said second metallic layer and coupled to said strip at said branch points, each of said shunt paths including a junction diode constituted by heavily doped portions of said semiconductor material adjacent said major surfaces, of opposite conductivity type, separated by an intervening portion of substantially intrinsic conductivity, each of said shunt paths further including a capacitance integral with said panel in series with said diode and constituted by one of said metallic layers, the adjoining nonconductive layer and one of said heavily doped portions proximal to the last-mentioned layer whereby said one of said heavily doped portions is common to said diode and said capacitance; and biasing means for selectively blocking and unblocking any of said diodes, thereby modifying the magnitude of phase-shift increments introduced by the corresponding line segments, said biasing means including a conductive external connection contacting said one of said heavily doped portions while being separated from said grounded first metallic layer and from said strip by portions of said nonconductive layers.
2. A phase shifter as defined in claim 1 wherein said semiconductor material is silicon and said nonconductive layers consist of silicon oxide.
3. A phase shifter as defined in claim 1 wherein said one of said heavily doped portions is common to two diodes forming part of a pair of consecutive shunt paths.
4. A phase shifter as defined in claim 1 wherein said one of said heavily doped portions is disposed adjacent said one major surface of said panel, the latter being provided with an incision extending to said one of said doped portions from said opposite major surface, said conductive connection including a metallic lining of said incision.
5. A phase shifter as defined in claim 1, in combination with a source of highfrequency oscillations connected across said metallic layers, the length of each line segment being substantially equal to a quarter wavelength of said oscillations.
6. A phase shifter as defined in claim 1 wherein said one of said metallic layers is said first metallic layer, said conductive connection including part of said second metallic layer.
7. A phase shifter as defined in claim 7 wherein said one of said heavily doped portions has an extension protruding through the body of said panel into the immediate vicinity of said second nonconductive layer, said part of said second metallic layer penetrating said second nonconductive layer and contacting said exten- 8. A phase shifter as defined in claim 8 wherein said panel is formed on the side of said first metallic layer with a recess substantially reducing the thickness of said body in the region of said extension.
9. A phase shifter as defined in claim 9 wherein said recess extends beyond the region of said extension into a section of said body containing said diode.
10. A phase shifter as defined in claim 1 wherein said biasing means includes a bypass condenser grounding said conductive connection for high frequencies, said bypass condenser being integral with said panel.
11. A phase shifter as defined in claim 11 wherein said one of said doped portions is disposed adjacent said first nonconductive layer and said conductive connection includes part of said second metallic layer, said bypass condenser including a uniformly doped portion of said semiconductor material extending from the region of said first metallic layer through the body of said panel into the immediate vicinity of said second nonconductive layer adjacent said part of said second metallic layer.
12. A phase shifter as defined in claim 1 wherein each of said shunt paths includes at least one transverse conductor forming part of said second metallic layer and extending laterally from said strip at a respective branch point, an extremity of said transverse conductor making conductive contact with one of said heavily doped portions of an associated diode.
13. A phase shifter as defined in claim 13 wherein said transverse conductor is integral with said strip at said respective branch point and forms an inductance in series with said associated diode.
14. A phase shifter as defined in claim 13 in combination with a source of high-frequency oscillations connected across said metallic layers, said transverse conductor having a length substantially equal to one-eighth of the free-space wavelength of said oscillations.
15. A phase shifter as defined in claim 13 wherein the transverse conductors of consecutive shunt paths are disposed in alternate pairs on opposite sides of said stri 12. A phase shifter as defined in claim 13 wherein said transverse conductor approaches said strip by said extremity at said respective branch point, said associated diode being capacitively coupled to said'strip at said branch point.
17. A phase shifter as defined in claim 17 wherein the transverse conductors of adjoining branch points are provided with a conductive link at a location remote from said strip, said biasing means including said conductive link and the transverse conductors interconnected thereby.
18. A phase shifter as defined in claim 18 wherein said conductive link is part of said second metallic layer and is provided with a bypass condenser grounding same for high frequencies.
19. A phase shifter as definedin claim 16 wherein said biasing means includes a common control lead for the diodes associated with each of said pairs of transverse conductors.
* t i i t