Negative charge pump architecture with self-generated boosted phases

[0002] The present invention relates to electronic circuits and, more particularly, to a charge pump circuit with self-generated boosted phases.

[0003] As is well known in semiconductor integrated non-volatile memory devices, it is often necessary to perform programming operations thus modifying the information content of the memory device. These programming operations may be a writing phase or an erasing phase. Generally, the first writing operation determines a logic “0” in the memory cells of the memory device, while the second erasing operation determines a logic “1”.

[0004] During the writing and erasing operations a voltage higher than the supply may be required. In many cases, negative voltage values are required and produced by charge pumps. For example, for the erasing phase of a single supply voltage flash memory device, all the driving terminals of the transistors forming the memory arrays, that is, all the terminals of the gate arrays, are driven by a voltage value of about −9 V. A ramp potential is applied on the bulk or substrate terminal of these transistors and at the end of the erasing phase, when the last erasing pulse is generated, a voltage value of 8.75 V will be present.

[0005] The technical processes implemented to manufacture flash memory devices of the last generation are triple well processes allowing complete isolation of the semiconductor regions to which different voltage values are applied. The triple well technology allows realizing negative charge pumps by using N-channel MOS transistors.

[0006] The electrical schematic of an N-channel MOS transistor obtained by a triple well structure is shown in FIG. 1 as being part of a negative charge pump. As may be noted in FIG. 6, wherein the same transistor is shown in its real implementation, a suitable biasing of the buried N-doped region at the highest available voltage value, for instance the supply voltage value Vdd, allows a perfect isolation between the well region and the substrate. Such an isolation allows biasing the well region to a negative voltage value, thus, avoiding a direct bias between the well region and the substrate, and as a consequence, between the source and drain regions of the pass transistor.

[0007] The ideal condition for the well region is obtained by biasing it to the lowest possible voltage value available in the circuit. This causes a high body effect on the transistors performing the charge transfer. This could cause a low charge transfer efficiency between the pump stages at low supply voltage values.

[0008] A schematic view of a negative charge pump for semiconductor integrated memory devices having a single power supply and manufactured by a triple well process is shown in FIG. 1. There is also visible an enlarged scale particularly concerning a charge pass transistor. As shown in this FIG. 1, in the first stages of the charge pump the bulk terminals are biased with a voltage value that is available only two stages downstream. The last two stages have the bulk terminals simply connected to the output terminal. This charge pump architecture is a good compromise between the need to avoid a direct biasing on the parasitic diodes between the junctions, and the other need to limit the body effect that is always there.

[0009] As an example, with this architecture the voltages available to the terminals of the second stage pass transistor, in the conduction state, are shown in FIG. 2. The transistor shown in FIG. 2 sees a bulk voltage corresponding to the voltage value reached by the voltage two stages downstream, while the gate terminal sees a voltage corresponding to the voltage on the node 1 plus a voltage swing equal to the supply voltage and due to the inverters that work as phase drivers for the A, B, C and D phases. In this example, the charge transfer from the load capacitance toward ground is penalized by an undesired start-up of those transistors since they receive just the single supply voltage Vdd. In this manner the charge transfer from one stage to the other subsequent stage could be difficult when increasing the working frequency which is fixed by the required performance in terms of drive capability and setting time (Trise).

[0010] An imperfect charge transfer is immediately transformed into lower pump efficiency, or, in other words, poorer performance parameters as previously mentioned. Poorer performance is mainly noted in low voltage devices, for instance, those working with a supply voltage Vdd=1.8/2.5V.

[0011] An object of the present invention is provide a negative charge pump architecture having structural and functional features allowing a better charge transfer between the pump stages through a better voltage overdrive on the gates of the pass transistors.

[0012] A first embodiment of the invention provides a higher voltage swing on the node 1, shown in FIG. 2, for instance, by doubling the voltage overdrive on such a node passing from a common voltage swing corresponding to the supply voltage value to a double voltage swing corresponding to a double supply voltage value, as shown in FIGS. 3A and 3B.

[0013] To achieve this object, a negative voltage value Vdd is picked up by the same charge pump changing its architecture. Thus, the invention relates to a negative charge pump architecture comprising: a plurality of circuit stages cascade connected one to the other between an input stage, coupled to an input reference potential, and an output stage including an output terminal producing a pumped potential. The architecture further includes a corresponding plurality of a pass transistors, one for each stage, each driven by a corresponding phase signal through a decoupling capacitor, and a corresponding plurality of charge capacitances, one for each stage. In addition, the negative charge pump architecture may include a further output stage connected downstream of the input stage and including a further output terminal producing a further raised or pumped potential.

[0014] The features and advantages of the charge pump architecture according to the invention will be apparent from the following detailed description of embodiments thereof, given as non-limiting examples with reference to the accompanying drawings, in which:

[0015] FIG. 1 is a schematic diagram of a negative charge pump architecture according to the prior art;

[0016] FIG. 2 is a schematic diagram of a pass transistor incorporated in a charge pump architecture according to the prior art;

[0017] FIG. 3A is a schematic plot of the pattern of one of the phases used for the charge transfer in the charge pump architecture of the prior art;

[0018] FIG. 3B is a schematic plot of the pattern of one of the phases used for the charge transfer in the negative charge pump architecture of the present invention;

[0019] FIG. 4 is a schematic diagram of a negative charge pump architecture according to the present invention;

[0020] FIG. 5 is a schematic diagram of a possible circuit implementation of a stage of the negative charge pump architecture of FIG. 4;

[0021] FIG. 6 is a schematic cross-sectional view, on enlarged scale, of an N-Channel MOS transistor incorporated into a stage of the negative charge pump architecture according to the present invention;

[0022] FIG. 7 is a comparative plot of voltage vs. time showing the pattern of a pumped signal produced by the inventive charge pump circuit and by a known charge pump circuit, respectively;

[0023] FIG. 8 is a comparative plot of voltage vs. time showing the pattern of signals produced by the inventive charge pump circuit and by the known charge pump circuit of FIG. 1, respectively; and

[0024] FIG. 9 is a comparative plot of voltage vs. time showing the pattern of further signals produced by the inventive charge pump circuit and by the known charge pump circuit of FIG. 1, respectively.

[0025] With reference to the drawing views, in particular to the example of FIG. 4, reference numeral 1 globally and schematically denotes a negative charge pump circuit realized according to the present invention. More specifically, but not exclusively, the inventive circuit 1 may be validly used during the writing and erasing phases of memory cells. The following description is done with reference to this specific field of application just for clarity of explanation and should not be considered as limiting thereof.

[0026] For memory cells means a plurality of memory elements incorporated into a conventional semiconductor integrated electronic memory device comprising a matrix of those cells structured in rows and columns. Corresponding row and column decoder portions are associated with the memory matrix, as well as a reading or sense amplifier.

[0027] The memory cells may be of any kind. More specifically, however, non-volatile memory cells are preferred because of their specific characteristic to retain the information content even if the memory device is not electrically supplied. After this brief explanation, we will disclose hereinafter in greater detail the structure and behavior of the inventive circuit 1.

[0028] The circuit 1 according to the invention has been modified as compared to a common charge pump circuit like the one shown in FIG. 1. This kind of circuit comprises a plurality of stages cascade connected one after the other and each including at least one pass transistor and a charge capacitance.

[0029] The first transistor of the series has a conduction terminal connected to a supply voltage reference, for instance a ground reference GND. The other conduction terminal of the first transistor is connected to the conduction terminal of a second pass transistor and so on in a cascade connection.

[0030] The driving terminal of each pass transistor receives a driving signal through a decoupling capacitor. The first transistor receives a signal AA, while the second transistor of the series receives a different signal CC in phase opposition as compared to the previous one. The odd transistors: first, third, fifth, etc, all receive the signal AA, while the even transistors: second, fourth, sixth, etc., all receive the signal CC.

[0031] Since the charge pump disclosed here is a negative type, the output of the charge pump circuit 1 will be an output terminal producing a negative voltage value Vneg as compared to ground potential. Such an output terminal is connected to ground through a charge capacitor Cload. The output stage may be identified by the bulk terminal of the corresponding pass transistor being connected to the driving terminal.

[0032] Advantageously, the circuit 1 is modified as compared to the known circuits in the following respects. First, a further capacitance 2 has been added to the to the first charge pump stage (1st). This further capacitance is connected in parallel to the classic or conventional charge capacitance. In this manner the circuit is enabled to receive a higher charge quantity from both sides of the pump.

[0033] Second, a further output stage 3 has been associated with the first stage of the charge pump. Accordingly, the charge pump according to the present invention presents a double output stage—one at the end of the cascade connection of the pump stages, and the other located near the first stage (1^st) and connected downstream directly to the first stage.

[0034] These features present independent advantages and they may be included independently one from the other in the inventive charge pump, although these may also be used together in combination in some embodiments of the charge pump circuit 1.

[0035] The further output stage 3 is driven by the signal CC, since it is connected downstream to the first stage driven by the signal AA. Advantageously, the negative potential −Vdd is pumped or increased and picked up from the output terminal of the output stage 3 and it is used as a second supply for a pair of inverters 4, 5 that are also fed by the positive reference supply voltage Vdd.

[0036] The input of the first inverter 4 receives a signal A_N and produces the signal AA to be applied to the odd stages. The input of the second inverter 5 receives a signal C_N and produces the signal CC to be applied to the even stages.

[0037] In substance, the further or second output stage 3 allows generating a negative supply voltage to feed the inverters 4 and 5 that are the drivers of the driving phases of the charge pump circuit 1. Thus, the invention suggests picking up a negative voltage generated by the charge pump itself and using this negative voltage to feed the inverters 4 and 5 thereby generating the driving phases for all the pass transistors.

[0038] For example, using the voltage values previously indicated, each inverter 4, 5 is supplied by the voltage values Vdd and −Vdd and provides a phase (AA, CC) having a voltage swing corresponding to 2*Vdd instead of a single supply voltage value Vdd as in node 1 of FIG. 2 on the driving terminal of a pass transistor.

[0039] This simple approach allows improving the performance of a negative charge pump circuit. More particularly, the better performance is shown in the enclosed figures in terms of reduction of the so-called setting time Trise, that changes in the Vdd=2V case, from a setting time of 66 μs to 52 μs. The diagram in FIG. 7 illustrates this improvement. The better performance is also in terms of improvement in the charge transfer between the pump stages passing from an imperfect closure between the stages in the prior art circuit (as shown in FIG. 8B) to a complete charge transfer in accordance with the circuit of the invention as seen in the diagram of FIG. 8A with Tclock=30 ns.

[0040] FIGS. 9A and 9B are comparative voltage vs. time diagrams showing the pattern of the phase voltages produced by the inventive charge pump circuit 1 (FIG. 9A) and by the known charge pump circuit of FIG. 1 (FIG. 9B). The different voltage swing of the normal phases (Vdd) and the pumped phases (2Vdd) can be noted. The inventive circuit provides many advantages, perhaps the most important of which is that the pumped phases allow suppling the stages wiht phase signals generated internal to the charge pump circuit.