Complementary bipolar SRAM

This application is a divisional application of U.S. patent application Ser. No. 14/793,561, filed Jul. 7, 2015 which is a continuation of U.S. patent application Ser. No. 14/717,218, filed May 20, 2015 the entire content and disclosure of which is incorporated herein by reference.

The present invention relates generally to transistor devices, and in particular a complementary bipolar SRAM, and a method of building and operating a complementary bipolar SRAM.

Semiconductor-on-Insulator (SOI) lateral bipolar transistors are ideally suitable for building complementary bipolar inverters, which is the basic building block for complementary bipolar circuits. The teaching of a complementary lateral bipolar inverter using SOI can be found in U.S. Pat. No. 8,531,001.

FIG. 1 shows a SOI lateral bipolar inverter device 10 according to prior art implementations. In FIG. 1 the SOI lateral bipolar junction transistor (BJT) device 10 includes a first bipolar (P-N-P) transistor 12 and a second bipolar (N-P-N) transistor 15 formed on an SOI substrate 11 . In the device 10 , the base terminal of the PNP bipolar junction transistor 12 is electrically connected with the base terminal of the NPN bipolar junction transistor 15 . The emitter terminal of the bipolar junction transistor 12 is tied to a supply voltage source V CC and the emitter terminal of the bipolar junction transistor 15 is tied to a ground reference. Further, the collector terminal of the PNP bipolar junction transistor 12 is electrically connected with the collector terminal of the NPN bipolar junction transistor 15 and form an SOI lateral bipolar inverter device 10 output.

FIG. 2 shows a pair of complementary bipolar inverters cross-coupled to form a bistable memory element (cell) 55 with relatively large noise margin. In the cross-coupled implementation shown, the memory element 55 includes a first set of lateral bipolar transistors 32 fabricated on the SOI substrate and a second set of lateral bipolar transistors 42 fabricated on the SOI substrate. As discussed above, the first set of lateral bipolar transistors 32 is configured to form a first inverter device 50 and the second set of lateral bipolar transistors 42 is configured to form a second inverter 60 . Furthermore, the first inverter 50 is cross-coupled to the second inverter 60 such that a first input terminal 51 to the first inverter 50 is electrically coupled to a second output terminal 62 of the second inverter 60 , and the second input terminal 61 of the second inverter 60 is electrically coupled to the output terminal 52 of the first inverter 50 . In this configuration, the cross-coupled inverters act as a memory element for storing a logic state, e.g., a logic “1” or logic “0”.

As shown in FIG. 3 , a memory cell 55 ′ includes the memory element 55 of FIG. 2 and further includes: first access transistor 74 controlling electrical impedance from the first inverter 50 to a bit line true (BLT) conductor 75 , and second access transistor 76 controlling electrical impedance from the second inverter 60 to bit line complement (BLC) conductor 85 . In one embodiment, the first and second access transistors 74 and 76 are field effect transistors (FETs). For example, the first and second access transistors 74 and 76 may be n-channel FETs or p-channel FETs. Additionally, each of the first and second access transistors 74 and 76 include a gate terminal coupled to a word line 95 . Such a configuration is taught and described in U.S. Pat. No. 8,526,220.

In the circuit 55 ′ of FIG. 3 , in a standby mode, the voltage at V DD is lowered (e.g., 0.5 Volts) to take advantage of the memory element's noise margin characteristics. Furthermore, a voltage V EE is set to 0 Volts. The wordline 95 is also set to 0 Volts, thereby turning off the access transistors 74 and 76 .

In the prior art circuit of FIG. 3 , during a memory write operation, V EE is set to 0 Volts. BLT 75 voltage is set to the desired binary value to be stored. Likewise, BLC 85 is set to the complement of the desired binary value to be stored. The wordline 95 is then raised to V DD so that the memory cell 55 is overwritten with the desired binary value.

During a memory read operation, V EE is pulled negative to avoid read disturb. In other words, the inverter circuits 50 and 60 must carry enough current to supply the FET current to maintain memory cell stability. In general, the larger the voltage difference between V CC and V EE , the more current the bipolar inventors can supply. The word line 95 corresponding to the memory cell 55 ′ is then set so that the access transistors 74 and 76 are activated. Sense amplifiers (not shown) coupled to the memory cell's BLT 75 and BLC 85 lines are then used to detect the logic value stored in the memory cell 55 ′.

The memory cell 55 ′ according to prior art implementation of FIG. 3 requires a BiCMOS process, i.e. a manufacturing process that integrates both SOI lateral bipolar inverters and MOSFETs on the same substrate, to build the SRAM which, as known, is not cost-effective.

In an aspect of the present disclosure, there is provided a complementary Bipolar SRAM memory cell.

In a further aspect of the present disclosure, there is provided a complementary Bipolar SRAM memory element formed of a cross-coupled bipolar inverter pair as memory element, and having NPN bipolar junction transistors as access devices, thereby avoiding BiCMOS processes and just requiring a complementary bipolar device manufacturing process.

Thus, in one embodiment, there is provided a complementary Static Random Access Memory (SRAM) device. The device comprises: a first set of lateral bipolar transistors fabricated on a semiconductor substrate, the first set of lateral bipolar transistors forming a first inverter device, and a second set of lateral bipolar transistors fabricated on the semiconductor substrate, the second set of lateral bipolar transistors forming a second inverter device, the first inverter device and second inverter device in a cross-coupled configuration to store a logic state; a first bipolar transistor of each the first set and second set being an PNP type bipolar transistor having a base terminal, an emitter terminal and a collector terminal, and a second bipolar transistor of each the first set and second set being an NPN type bipolar transistor having a base terminal, a first emitter terminal, a second emitter terminal, and a collector terminal, a first conductor electrically coupling an emitter terminal of the PNP type transistor of the first inverter device and the emitter terminal of the PNP type transistor of the second inverter device, the first conductor adapted to receive a first applied voltage; and a second conductor electrically coupling the first emitter terminal of the NPN transistor of the first inverter device and the first emitter terminal of the NPN transistor of the second inverter device, the second conductor adapted to receive a second applied voltage, wherein one NPN type transistor of either the first inverter device or second inverter device becomes turned on activated responsive to application of the first voltage and second voltage such that electrical current flows through the first emitter terminal of the activated NPN transistor device to the second conductor, and wherein the stored logic state is accessed via the second emitter terminals of both the NPN bipolar transistors of the first inverter and second inverter devices.

In this embodiment, the second emitter terminal of the NPN bipolar transistor of the first inverter device is electrically coupled to a bit line true conductor (BLT) for controlling electrical impedance from the first inverter to the BLT conductor, and the second emitter terminal of the NPN bipolar transistor of the second inverter device is electrically coupled to a bit line complement conductor (BLC) for controlling electrical impedance from the second inverter device to the BLC conductor, each the BLT and BLC conductors used to access the stored logic state.

In a further aspect, there is provided a method for manufacturing a memory cell comprising: forming a first set of lateral bipolar transistors on a semiconductor substrate, the first set of lateral bipolar transistors forming a first inverter device, and forming a second set of lateral bipolar transistors on the substrate, the second set of lateral bipolar transistors forming a second inverter device, wherein a first bipolar transistor of each the first set and second set is an PNP type bipolar transistor having a base terminal, an emitter terminal and a collector terminal, and a second bipolar transistor of each the first set and second set being an NPN type bipolar transistor having a base terminal, a first emitter terminal, a second emitter terminal, and a collector terminal, the first inverter device and second inverter device configured in a cross-coupled configuration to store a logic state; forming a first conductor layer that electrically couples the emitter terminal of the PNP type transistor of the first inverter device to the emitter terminal of the PNP type transistor of the second inverter device; forming a second conductor layer that electrically couples the first emitter terminal of the NPN transistor of the first inverter device to the first emitter terminal of the NPN transistor of the second inverter device; and forming a third conductor layer that electrically couples the second emitter terminal of the NPN bipolar transistor of the first inverter to a bit line true (BLT) conductor; and forming an fourth conductor layer that electrically couples the second emitter terminal of the NPN bipolar transistor of the second inverter device to a bit line complement (BLC) conductor.

In still another aspect, there is provided a method of operating a memory cell. The memory cell comprises: a first set of lateral bipolar transistors fabricated on a semiconductor substrate, the first set of lateral bipolar transistors forming a first inverter device, and a second set of lateral bipolar transistors fabricated on the semiconductor substrate, the second set of lateral bipolar transistors forming a second inverter device, the first inverter device and second inverter device in a cross-coupled configuration to store a logic state; a first bipolar transistor of each the first set and second set being an PNP type bipolar transistor having a base terminal, an emitter terminal and a collector terminal, and a second bipolar transistor of each the first set and second set being an NPN type bipolar transistor having a base terminal, a first emitter terminal, a second emitter terminal, and a collector terminal, a first conductor electrically coupling an emitter terminal of the PNP type transistor of the first inverter device and the emitter terminal of the PNP type transistor of the second inverter device, the first conductor adapted to receive a first applied voltage; and a second conductor electrically coupling the first emitter terminal of the NPN transistor of the first inverter device and the first emitter terminal of the NPN transistor of the second inverter device, the second conductor adapted to receive a second applied voltage, wherein the second emitter terminal of the NPN bipolar transistor of the first inverter device is electrically coupled to a bit line true conductor (BLT) for controlling electrical impedance from the first inverter to the BLT conductor, and the second emitter terminal of the NPN bipolar transistor of the second inverter device is electrically coupled to a bit line complement conductor (BLC) for controlling electrical impedance from the second inverter device to the BLC conductor, each the BLT and BLC conductors used to access the stored logic state, wherein the method comprises: applying a first voltage to the first conductor; applying a second voltage to the second conductor, wherein one PNP type transistor of either the first inverter device or second inverter device becomes activated responsive to application of the first voltage and second voltage such that electrical current flows through the first emitter terminal of one NPN type transistor of either the first inverter device or second inverter device to the second conductor, and applying a further voltage to each the respective the BLT conductor and BLC conductor to write a logic state value to or read a logic state value from the memory cell.

These and other objects, features and advantages of the present invention will become apparent from the following detailed description of illustrative embodiments thereof, which is to be read in connection with the accompanying drawings. The various features of the drawings are not to scale as the illustrations are for clarity in facilitating one skilled in the art in understanding the invention in conjunction with the detailed description. In the drawings:

FIG. 1 shows a SOI lateral bipolar inverter device 10 according to a prior art implementation;

FIG. 2 shows a pair of complementary bipolar inverters cross-coupled to form a bistable memory element (cell) according to the prior art;

FIG. 3 shows the prior art memory cell of FIG. 2 including access FET transistors for controlling electrical impedance from the inverters to respective bit line true (BLT) and bit line complement (BLC) conductors;

FIG. 4 depicts the Complementary Bipolar SRAM cell that avoids BiCMOS processing according to one embodiment;

FIG. 5 shows in greater detail a multi-emitter BJT device that functions as the two-emitter bipolar transistor (labeled Q 1 ) of a first inverter, and the two-emitter bipolar transistor (labeled Q 2 ) of a second inverter;

FIG. 6A shows, in a method using standard lithography and semiconductor manufacturing technique, a layout of the transistors in the Complementary Bipolar SRAM cell of FIG. 4 ;

FIG. 6B shows results of semiconductor processing 210 to form a first level metal layer (M 1 ) layout for the Complementary Bipolar SRAM cell of FIG. 4 ;

FIG. 6C shows results of semiconductor processing 250 to form a second level metal layer (M 2 ) layout for the Complementary Bipolar SRAM cell of FIG. 4 ;

FIG. 6D shows results of semiconductor processing 300 to form a third level metal layer (M 3 ) layout for the Complementary Bipolar SRAM cell of FIG. 4 ; and

FIG. 7 illustrates a circuit depiction of a memory system 400 that includes an SOI substrate and an array of memory cells 100 1 , 100 2 , . . . , 100 n .

Aspects of the present disclosure are now described in detail with accompanying figures. It is noted that like reference numerals refer to like elements across different embodiments. The drawings are not necessarily drawn to scale. As used herein, ordinals such as “first,” “second,” and “third,” etc. are employed to distinguish similar elements, and a same element may be labeled with different ordinals across the specification and the claims.

The present disclosure is directed to a Complementary Bipolar SRAM that avoids BiCMOS processing.

The current invention is an all bipolar SRAM cell, using cross-coupled complementary semiconductor-on-insulator (SOI lateral bipolar transistors as SRAM memory element, and NPN bipolar transistors as access transistors. No BiCMOS processing is required in its manufacture-only SOI lateral bipolar transistors are formed at a lower cost.

FIG. 4 shows an embodiment of the SOI lateral bipolar junction transistor (BJT) memory cell 100 that employs only lateral bipolar technology in its production. This memory cell 100 is an SOI lateral bipolar inverter device including a pair of complementary bipolar inverters cross-coupled to form a bistable memory element (cell). In the cross-coupled implementation shown, the memory cell 100 includes a first set of lateral bipolar transistors 102 , 202 fabricated on the SOI substrate and a second set of lateral bipolar transistors 104 , 204 fabricated on the SOI substrate. The first set of lateral bipolar transistors 102 , 202 are configured to form a first inverter device 150 and the second set of lateral bipolar transistors 104 , 204 are configured to form a second inverter 160 . Furthermore, the first inverter 150 is cross coupled to the second inverter 160 such that a first input terminal 155 is electrically coupled to a second output terminal 162 and the second input terminal 165 is electrically coupled to the first output terminal 152 .

In the first inverter device 150 , BJT transistor 102 is an PNP type and BJT transistor 104 in the second inverter device 160 is also an PNP type. However, in the embodiment of FIG. 4 , the paired lateral bipolar transistor set of inverter 150 includes a two-emitter NPN type transistor Q 1 202 , having emitters labeled E 1 and E 3 . Likewise, in the embodiment of FIG. 4 , the paired lateral bipolar transistor set of inverter 160 includes a two-emitter NPN type transistor Q 2 204 , having emitters labeled E 2 and E 4 . Thus, as compared to the prior art device 55 ′ of FIG. 3 , the FET access transistor 74 is replaced by an NPN device 202 having a second emitter terminal connected to respective BLT line, and the FET access transistor 76 is replaced by an NPN device 204 having a second emitter terminal connected to respective BLC line.

FIG. 5 shows in greater detail a multi-emitter BJT device 175 that functions as the two-emitter NPN bipolar transistor 202 (labeled Q 1 ) of inverter 150 , and the two-emitter NPN bipolar transistor 204 (labeled Q 2 ) of inverter 160 . As shown in FIG. 5 , the multi-emitter BJT device 175 forming inverter elements 202 and 204 , includes two-BJT transistors 176 , 177 connected in parallel, with a common collector 180 and a common base terminal 190 . With device 175 of FIG. 5 employed as multi-emitter transistor 202 of the first inverter 150 , the emitter terminals 185 A, 185 B of respective parallel transistors 176 , 177 , form respective emitters E 3 and E 1 of transistor 202 . Likewise, with device 175 employed as multi-emitter transistor 204 of the second inverter 160 , the emitter terminals 185 A, 185 B of respective parallel transistors 176 , 177 , form respective emitters E 2 and E 4 of transistor 204 .

Thus, returning to FIG. 4 , in inverter device 150 , the emitter terminal E 3 (of first transistor Q 1 ), e.g., corresponding to emitter terminal 185 B of device 175 shown in FIG. 5 , is electrically coupled to the BLT bit line 75 , while the emitter terminal E 1 (of first transistor Q 1 ) of inverter 150 , e.g., corresponding to emitter terminal 185 A of device 175 shown in FIG. 5 , is electrically coupled to the V EE voltage supply line. The emitter terminal E 1 (of first transistor Q 1 ) of inverter 150 is additionally electrically coupled to the corresponding emitter terminal E 2 (of second transistor Q 2 ) of cross-coupled inverter 160 .

Further, in view of FIG. 4 , in inverter device 160 , the emitter terminal E 4 (of second transistor Q 2 ), e.g., corresponding to emitter terminal 185 B of device 175 shown in FIG. 5 , is electrically coupled to the BLC bit line 85 , while the emitter terminal E 2 (of transistor Q 2 ), e.g., corresponding to emitter terminal 185 A of device 175 shown in FIG. 5 , is electrically coupled to the V EE voltage supply line. The emitter terminal E 2 (of transistor Q 2 ) of inverter 160 is additionally electrically coupled to the corresponding emitter terminal E 1 (of first transistor Q 1 ) of cross-coupled inverter 150 .

Focusing on each cross-coupled inverter, e.g., inverters 150 , 160 , in each respective Q 1 , and Q 2 , the two emitters are separate, i.e., each passes current separately. Typically current carried by the emitter increases exponentially with qV/kT (a dimensionless ratio) where k is the Boltzmann constant, T is a temperature value, q is an electrical charge and V is a voltage across the emitter-base diode of the transistor. Here the base voltage of multi-emitter BJT device Q 1 is common to both emitters E 1 and E 3 , however, the base-emitter voltages of Q 1 are separately controllable, by modifying either the BLT voltage on BLT line 75 relative to the V EE voltage and/or modify the V EE voltage relative to the BLT voltage. Thus, given E 1 =0 and E 3 =0, then the transistor Q 1 will have the same base-emitter voltage (V BE ) for emitter E 1 and emitter E 3 and equal currents will pass through E 1 and E 3 . If emitter voltage at E 3 >>E 1 then the voltage V BE 3 becomes larger than V BE 1 and more current will pass through E 3 of Q 1 as compared to current passing through E 1 of Q 1 . Similarly, the base voltage of multi-emitter BJT device Q 2 is common to both emitters E 2 and E 4 , however, the base-emitter voltages of Q 2 are separately controllable, by modifying either the BLC voltage on BLC line 85 relative to the V EE voltage and/or modify the V EE voltage relative to the BLC voltage. Thus, given E 2 =0 and E 4 =0, then the transistor Q 2 will have the same base-emitter voltage (V BE ) for emitter E 2 and emitter E 4 and equal currents will pass through E 2 and E 4 . If emitter voltage at E 4 >>E 2 then the voltage V BE 4 becomes larger than V BE 2 and more current will pass through E 4 of Q 2 as compared to current passing through E 2 of Q 2 . For example, at room temperature (T), a delta voltage increase of about 60 millivolts (60 mV) at a VBE will cause approximately a ten-fold increase in current flowing through the transistor's corresponding emitter terminal. Similarly, a delta voltage decrease of about 60 millivolts (60 mV) at a VBE will cause approximately a ten-fold decrease in the amount of current flowing through the transistor's corresponding emitter terminal.

Thus, assuming that each of the dual BJT transistors 175 of FIG. 5 have identical geometries, given 1 milliamp passing through the collector, the proportion of current through each emitter will depend on the respective voltage across the respective base—emitter terminal. If the VBE at each terminal is the same (E 1 and E 3 at same voltage), then roughly equal current amounts will pass through each emitter (e.g., 0.5 milliamp). To provide current in an emitter terminal to 90% of the 1 milliamp, e.g., V BE 3 for emitter E 3 of FIG. 4 is brought much larger than V BE 1 for emitter E 1 (V BE 3 >>V BE 1 ), e.g., V BE 3 ≥V BE 1 +60 mV. Similarly, if the VBE at each terminal is the same (E 2 and E 4 at same voltage), then roughly equal current amounts will pass through each emitter (e.g., 0.5 milliamp). To provide current in an emitter terminal to 90% of 1 milliamp, e.g., V BE 4 for emitter E 4 of FIG. 4 is brought much larger than V BE 2 for emitter E 2 (V BE 4 >>V BE 2 ), e.g., V BE 4 ≥V BE 2 +60 mV. The dependence of current flow upon the geometry of the BJT is relatively small.

Operations employing the complementary Bipolar SRAM cell 100 of FIG. 4 , are now described. Typically, in operation of the memory cell 100 , either the cross-coupled inverter 150 side (Q 1 ) is “on” and the other inverter 160 (Q 2 ) is “off” to program one bit value; or alternately, the cross-coupled inverter 160 side (Q 2 ) is on and the other inverter 150 (Q 1 ) is off to program a complement bit value.

In one embodiment, to place the memory cell 100 in a standby mode, with BJT transistor Q 1 being turned “on,” the WL voltage applied is programmed at some hold voltage above the voltage V EE , while bitlines BLT 75 and BLC 85 are kept at a voltage of at least several kT/q volts above the V EE . For example: grounding or programming supply line voltage line V EE =0 V, and programming the WL voltage 95 (i.e., V WL ) at approximately 0.5 V, and programming the voltage at the BLT (V BLT ) at approximately 1 V and at the BLC (V BLC ) at approximately 1 V renders device 100 in a stand-by mode of operation. That is, given these input voltages, the voltage at the base of Q 1 , i.e., VB(Q 1 ), is equal to the voltage at the WL, i.e., V BE 1 =V WL =0.5 V; the voltage at the base of Q 2 , i.e., VB(Q 2 ) is equal to the voltage V EE =0 V. Thus, with Q 1 transistor “on,” current flows through E 1 to V EE , no current flows through E 3 as the V BE 3 <<V BE 1 . With transistor Q 2 off, no current is flowing through E 2 and E 4 . In one embodiment, a lower range between V WL and V EE (voltage difference) is about 0.25 volts, e.g., V EE is at 0 V and V WL is about 0.2-0.25 Volts. In one embodiment, a maximum value for V WL may be about 1.0 volts.

It is noted that, in a standby operation, the programming of the E 3 is such that the emitter-base diode for E 3 is reverse biased, i.e., no current flow through the E 3 emitter of Q 1 ; and likewise, no current flows through the E 2 because V BE 2 =0, and no current flows through E 4 because the emitter-base diode for E 4 is reverse biased.

A further operation employing the complementary Bipolar SRAM cell 100 of FIG. 4 is to select all cells connected to a wordline, WL. In an example operation to select a word line, the WL voltage at selected wordline, e.g., WL 95 is placed at about 1 V and the VEE is placed at about 0.6 V. Other non-selected WL voltages of other wordlines having connected cells (shown in FIG. 7 ) are brought to about 0.5 volts and the V EE voltage supplied to the transistors coupled to the non-selected WL are placed at 0.0 Volts. In this embodiment, for each of the memory cells 100 connected to the selected wordline, the corresponding BLT and BLC voltages are programmed to remain in their standby mode voltages (e.g., the BLT (VBLT)=1 V and the BLC (VBLC)=1 V) such that no current flows in the BLT 75 and BLC 85 lines for the selected cell. Here, the BLC and BLT voltages (in stand by mode) are caused to be equal to or larger than the voltage at the wordline (V WL ) and greater than V EE such that Q 1 is on with current only flowing through emitter terminal E 1 and BJT transistor Q 2 is off, no current flows through E 2 , E 3 , E 4 .

In this embodiment, the voltage across the cross-coupled inverter of the selected cells is about 0.4 volts, i.e., V WL −V EE =1.0 V−0.6 V=0.4 V. However, the voltage across the cross-coupled inverter of the non-selected cells is about 0.5 volts, i.e., V WL −V EE =0.5 V−0 V=0.5 V. Thus, to select cells of a wordline, the difference between wordline voltage V WL and V EE is smaller than the difference between the wordline voltage V WL and V EE of the non-selected WL cells. In this example, the V WL −V EE difference (e.g., 0.4 V) across the latch of cells coupled to a selected wordline is about 100 mV less than the V WL −V EE difference (e.g., 0.5 V) across the latch of cells coupled to a non-selected wordline. This 100 mV reduced voltage difference for the selected wordline cells is exemplary, however, this reduced difference may range between 50 mV to 200 mV.

In this embodiment, while cell 100 is in the standby mode, the wordline voltage is raised to 1 V, i.e., V WL =1.0 V and the V EE line is at about 0.6 V. In doing this, the base voltage VB(Q 1 ) of the on transistor Q 1 follows V WL voltage and reach about 1.0 V while the base voltage VB(Q 2 ) of the off transistor Q 2 follows V EE and reach about 0.6 V. As a result, the V BE2 voltage is zero volts, i.e., VB (Q 2 )−VEE=0.6 V−0.6 V=0 V (i.e., Q 2 is off).

A further operation employing the complementary Bipolar SRAM cell 100 of FIG. 4 is to read a stored logic state (bit) value from the selected cell. For purposes of description, it is assumed that a read operation is performed at cells of a selected wordline WL 95 , with BJT transistor Q 1 turned on, and BJT transistor Q 2 is turned off. Here, both applied wordline voltage V WL and V EE supply voltages have been raised (e.g., when the cell was selected). For example, wordline voltage has been raised to 1 V, i.e., V WL =1.0 V and the V EE =is about 0.6 V. In doing this, the “on” transistor Q 1 has a base voltage VB(Q 1 )=1 V, while the “off” transistor Q 2 has a base voltage VB(Q 2 )=0.6 V.

In this embodiment, to read a selected cell, BLT voltage 75 and BLC voltage 85 are both lowered to a value slightly less than their respective standby mode WL voltage value, e.g., to 0.4 V. From their standby values, a reduction ranging anywhere between 0.08 V to 0.120 V would be an adequate voltage reduction for reading a cell bit value from a standby state. Thus, the V BE 3 (i.e., base-emitter voltage at Q 1 emitter E 3 ) now has a value of 0.6 V, i.e., VB(Q 1 )−V BLT =1.0−0.4=0.6 volts causing a current to flow in BLT, while the V BE 4 (i.e., base-emitter voltage at Q 2 emitter E 4 ) now has a value of only 0.2 V, i.e., VB(Q 2 )−V BLC =0.6−0.4=0.2 volts, resulting in negligible current flow in BLC. In this embodiment, the V BE 1 (i.e., base-emitter voltage at Q 1 emitter E 1 ) is a value 0.4 V, i.e., VB(Q 1 )−V EE =1.0−0.6=0.4 volts (Q 1 is turned on). Further, the V BE 4 (i.e., base-emitter voltage at Q 2 at emitter E 4 ) now has a value of 0.2 V, i.e., VB(Q 2 )−V BLC =0.6−0.4=0.2 volts, hence having negligible current flowing to BLC, and V BE 2 is 0.0 V, i.e., VB(Q 2 )−V EE =0.6−0.6=0.0 volts (Q 2 is turned off). Thus, base-emitter voltage results here are 0.6 volts across the BLT transistor Q 1 and only 0.2 V across the BLC transistor Q 2 resulting in the current through the BLT line 75 to be about a million times greater than the current through the BLC 85 . There is negligible current flow in the bitlines of the non-selected cells.

A further operation employing the complementary Bipolar SRAM cell 100 of FIG. 4 is to write a bit value to a selected cell. For purposes of description, it is assumed that the cell is selected and BJT transistor Q 1 is turned on, and BJT transistor Q 2 is turned off. Here, the wordline voltage 95 has been raised to 1 V, i.e., V WL =1.0 V and the V EE =is about 0.6 V. In doing this, the “on” transistor Q 1 has base voltage VB(Q 1 )=1 V, while the “off” transistor Q 2 has base voltage VB(Q 2 )=0.6 V. As before, the cell 100 is assumed at a state where the V BE 1 (i.e., base-emitter voltage at Q 1 emitter E 1 ) is a value 0.4 V, i.e., VB(Q 1 )−V EE =1.0−0.6=0.4 volts (Q 1 is turned on). Further, the V BE 2 value is 0.0 V, i.e., VB(Q 2 )−V EE =0.6−0.6=0.0 volts (Q 2 is turned off).

In this embodiment, to write a selected cell requires turning BJT transistor Q 1 off and BJT transistor Q 2 on. Thus, the bit line truth voltage V BLT 75 is set to remain at its standby value, e.g., V BLT is at a standby value=1.0 V. Then, to perform the write, the BLC voltage 85 is lowered to the standby mode value of V EE =0.0 V. In this manner, the V BE 4 (i.e., base-emitter voltage at Q 2 emitter E 4 ) now has a value of 0.6 V, i.e., VB(Q 2 )−V BLC =0.6−0.0=0.6 volts. Thus, VBE(E 4 ) is now larger than VBE(E 1 ), forcing Q 2 to turn on and Q 1 to turn off. Here, transistor Q 2 is forced to turn on and carries about a million times more current than the BJT transistor Q 1 which is now turned off. To complete the writing, the voltage at BLC 85 is increased back to its standby value (1 V), and V WL and V EE are returned to their standby voltage values.

FIGS. 6A-6C shows a methodology 200 for building the SOI lateral bipolar junction transistor (BJT) memory cell 100 of FIG. 4 in one embodiment. In FIG. 6A , the method includes forming, using standard lithography and semiconductor manufacturing technique, a layout of the transistors in the cell 100 , e.g., on a semiconductor (SOI) substrate, forming the device 100 . As shown in FIG. 6A , prior to putting on metal layers, the method includes forming the first PNP device 102 and second PNP device 104 , and forming the multi-emitter NPN BJT transistor device Q 1 175 A having emitter terminals labeled E 3 , E 1 and forming multi-emitter NPN BJT transistor device Q 2 175 B having emitter terminals labeled E 4 , E 2 .

FIG. 6B shows results of semiconductor processing 210 to form a first level metal layer (M 1 ) layout. First, a metal M 1 conductor connector 215 is formed according to processes known in the art to electrically connect the base terminal of the PNP transistor 102 to the common base terminals of the NPN transistor Q 1 ; and a further M 1 metal conductor connection 220 is formed according to processes known in the art to electrically couple the collector terminal of the PNP transistor 102 to the common collector terminals of the NPN transistor Q 1 . Likewise, an M 1 metal conductor connector 216 is formed according to processes known in the art to electrically connect the base terminal of the PNP transistor 104 to the common base terminals of the NPN transistor Q 2 ; and a further M 1 metal conductor connection 221 is formed according to processes known in the art to electrically couple the collector terminal of the PNP transistor 104 to the common collector terminals of the NPN transistor Q 2 . For clarity of illustration, it is understood that underlying contact vias are not shown in FIG. 6B .

FIG. 6C shows results of semiconductor processing 250 to form a second level metal layer (M 2 ) layout. The M 2 layers provide for the cross coupling of the left half inverter 150 and right half inverter 160 as shown in FIG. 4 . In this processing, a metal M 2 conductor layer 255 is formed according to processes known in the art to electrically connect only the M 1 conductor line 215 (coupled to base terminal of the PNP transistor 102 and the common base terminals of the NPN transistor Q 1 ) to the M 1 conductor layer 221 (coupled to the collector terminal of the transistor 104 and the common collector terminals of the NPN transistor Q 2 ). Likewise, in this processing, a metal M 2 conductor layer 260 is formed according to processes known in the art to electrically only connect the M 1 conductor line 220 (coupled to the collector terminal of the PNP transistor 102 and the common collector terminals of the NPN transistor Q 1 ) to only the M 1 conductor layer 216 (coupled to the base terminal of the transistor 104 and the common base terminals of the NPN transistor Q 2 ).

FIG. 6C shows further results of semiconductor processing 250 to form the metal layer (M 2 ) layout for wordline conductor line WL 95 and V EE voltage line of circuit 100 of FIG. 4 . This further processing includes: forming the wordline WL as a single metal M 2 conductor layer 265 that only connects to each emitter terminal of the formed respective PNP transistors 102 and 104 at respective contact conductors 270 A, 270 B; and forming the V EE voltage supply line as a single metal M 2 conductor layer 272 that only connects to each E 1 , E 2 emitter terminals of the formed respective NPN transistors 175 A and 175 B at respective metal contact conductors 275 A, 275 B.

FIG. 6D shows results of semiconductor processing 300 to form the metal layer (M 3 ) layout for BLT 75 and BLC 85 conductor lines of the circuit 100 of FIG. 4 . This further processing includes: forming the BLT bitline 75 as a metal M 3 conductor layer 290 that connects only to emitter terminal E 3 of the formed NPN transistor Q 1 at a contact conductor 295 A; and forming the BLC bitline 85 as a metal M 3 conductor layer 292 that connects only to emitter terminal E 4 of the formed NPN transistor Q 2 at a contact conductor 295 B.

In a further embodiment, as shown in FIG. 7 , there is provided a circuit depiction of a memory system 400 that includes an SOI substrate and an array of memory cells 100 1 , 100 2 , . . . , 100 n . Each of the memory cells 100 1 , 100 2 , . . . , 100 n comprising the complementary SOI lateral bipolar SRAM structure 100 such as shown in FIGS. 4-6 (with V EE voltage line connections not shown for clarity). The array 400 includes a plurality of wordlines, 95 1 , 95 2 , . . . , 95 n having plural cells connected thereto, and a plurality of bitline true (BLT) lines 75 1 , 75 2 , . . . , 75 m and corresponding bitline complement (BLC) lines 85 1 , 85 2 , . . . , 85 m . Each cell 100 is connected to a respective bitline true line (BLT) and bitline complement (BLC) line in the manner as described. For example, memory cells 100 1 , 100 2 , . . . 100 m are shown selectable with a wordline WL 1 . As shown in FIG. 7 , the memory cell array of 400 is coupled to a write circuit 425 for applying the WL, V EE , BLT and BLC values in a manner as described herein to program a logic value into the selected memory cell 100 based on a data input, e.g., D IN . When a wordline WL is selected for reading, each coupled cell provides a differential output signal at respective coupled Bitline True (BLT) and Bitline Complement (BLC) lines. For example, when read, memory cell 100 1 provides the stored memory cell value at respective BLT and BLC lines 75 1 and 85 1 ; memory cell 100 2 provides the stored bit value of that memory cell at respective BLT and BLC lines 75 2 and 85 2 , etc., including a last stored bit value of the memory cell 100 m at respective BLT and BLC lines 75 m and 85 m . In one embodiment, a sense amplifier circuit 455 , i.e., a cross coupled sense amplifier, connects the array for receiving the differential signal output signals from respective BLT and BLC lines to provide a word output 440 , for example.

While various embodiments are described herein, it will be appreciated from the specification that various combinations of elements, variations or improvements therein may be made by those skilled in the art, and are within the scope of the invention. In addition, many modifications may be made to adapt a particular situation or material to the teachings of the invention without departing from essential scope thereof. Therefore, it is intended that the invention not be limited to the particular embodiment disclosed as the best mode contemplated for carrying out this invention, but that the invention will include all embodiments falling within the scope of the appended claims.