This application claims the priority of Japanese Patent Application No. 2013-059285 filed Mar. 22, 2013, which is incorporated herein by reference in its entirety.
The present invention relates to a semiconductor device. More specifically, the present invention relates to a semiconductor device having a MOS transistor in the field of analog LSI technology, to a method of manufacturing a semiconductor device having a MOS transistor that is required to be relatively accurate, and to a variation reduction layout structure of a current mirror circuit.
In a current mirror circuit used in an analog circuit, it is generally demanded that the electrical characteristics of elements of the current mirror circuit exhibit small relative variation. Further, in the field of a power IC, it is demanded that enhanced control be exercised because the heat generation effect of the IC is a cause of variation.
In a driver IC 201 depicted in
In order to minimize an IC loss, the current value 208 (Im) of the driver IC 201 is detected, by a current detection circuit 205, in accordance with the current value 209 (Is) of a sense MOS transistor 207 and controlled by an external controller IC 202.
The current ratio (Is/Im) between the sense MOS transistor 207 and the main MOS transistor 206 is substantially set to be between 1:100 and 1:1000. This current ratio is substantially equal to the area ratio between the transistors 207, 206.
Further, as the current 208 flowing in the main MOS transistor 206 is much larger than a current flowing, for instance, in a digital circuit, the area of the driver IC 201 is mostly occupied by the main MOS transistor 206.
The accuracy of the current value 208 of the main MOS transistor 206, which determines the performance of the driver IC 201, is directly affected by the variation in the current ratio (Is/Im) between the current value 208 of the main MOS transistor 206 and the current value 209 of the sense MOS transistor 207. It is therefore demanded that relative variations and relative changes in the two MOS transistors 206, 207 be minimized.
Causes of relative variations in the above-mentioned MOS transistors used for power ICs are described below in comparison with relative variation in a common low-voltage MOS transistor operating, for instance, at a power supply voltage of 5 V or lower.
The above is known as the so-called Pelgrom plot (refer to “Matching properties of MOS transistors” written by M. Pelgrom, A. Duinmaijer, and A. Welbers (IEEE Journal of Solid-State Circuits, Vol. 24, No. 5, pp. 1433-1439, October 1989)). According to the Pelgrom plot, the amount of variation is proportional to the reciprocal of the square root of the area of the element.
As regards the relative variations in the above-mentioned MOS transistors used for power ICs, the area of the main MOS transistor is sufficiently large. Therefore, the variation in the current ratio is determined by the area of the sense MOS transistor.
However, the area of the above-mentioned power ICs is mostly occupied by the main MOS transistor. In other words, the power ICs have substantially the same size as the main MOS transistor. Therefore, the main MOS transistor significantly contributes to the variations.
As the regards the local fluctuation in the manufacturing process, which is mentioned firstly, the amount of variation may be decreased by increasing the area of the MOS transistors. However, a trade-off arises with respect to the cost of a chip. Therefore, the degree of freedom in IC design is low.
As regards the comprehensive changes in the manufacturing process, which is mentioned secondly, the use of a common-centroid layout is effective (refer to Japanese Patent No. 3179424). When the common-centroid layout is used, the two MOS transistors to be compared have the same center of gravity when they are disposed.
As regards the variation caused by an arrangement pattern, which is mentioned thirdly, the disposition of a dummy cell and the use of a common diffusion layer or the like are effective (refer to the disclosure in Japanese Unexamined Patent Application Publication No. 2010-27842).
Even when the first to third conventional measures against variations ((1) to (3) above) are taken in relation to a circuit applicable to the present invention, the fourth variation, that is, the temperature variation, occurs significantly. Thus, the accuracy of a control IC is low.
A main MOS transistor region 302, a sense MOS transistor region 303, a control circuit region 304 of the driver IC 301 are depicted in
Here, it is assumed that the package temperature is sufficiently heat-exchanged with an external environment and maintained constant without regard to an IC operation. The channel temperature rises from a package boundary face (the outermost circumference of the IC) toward the center. The characteristic curves 401, 402, which differ in the IC package temperature, also differ in the rate of temperature rise at the center. The reason is that the higher the chip temperature, the lower the possibility of heat dissipation.
The vertical axis of
In this document, the Ratio is defined as the value obtained when the current of the main MOS transistor is divided by the current of the sense MOS transistor.
The variation in the Ratio (or the temperature fluctuation), that is, the ΔRatio, is a main factor that determines the control accuracy of the IC covered by this document.
As indicated by
It is demanded that the driver IC deliver constant performance within an operating temperature range. Therefore, it is demanded that the ΔRatio be tolerated as a variation. Hence, the task to be achieved by an IC to be covered by the present invention is to minimize the ΔRatio.
In, the plan layout of a semiconductor device provided by the present invention is such that a main MOS transistor element section is paired with a sense MOS transistor element section, and that the pair of the main and sense MOS transistor element sections is cyclically disposed. Two or more main MOS transistor element sections and two or more sense MOS transistor element sections are electrically connected in parallel while source, drain, and gate electrode terminals are shared. Hence, the total gate length of a main MOS transistor element formed of the two or more main MOS transistor element sections is the sum of the gate lengths of the individual main MOS transistor element sections. Similarly, the total gate length of a sense MOS transistor element formed of the two or more sense MOS transistor element sections is the sum of the gate lengths of the individual sense MOS transistor element sections.
To be precise, the above-mentioned gate length is not an actual dimension, but an effective size based on electrical characteristics.
Using the above-described configuration makes it possible to obtain a semiconductor device that provides constant relative temperature dependence of accuracy when uneven temperature distribution occurs in the plane of circuitry due to self-heating of a main MOS transistor and of a sense MOS transistor.
Further, an advantageous effect of the present invention may be maximized by dividing a sense MOS transistor element and a main MOS transistor element by the number of divisions that is determined by the minimum permissible gate length of the sense MOS transistor element sections.
The minimum permissible gate length is a minimum gate length that maintains MOS transistor characteristics determined by limitations imposed on a manufacturing process. Let us assume, for example, that the total gate length of a sense MOS transistor element is Ws, and that the minimum size of a sensor MOS transistor element section is Wm, and further that the number of divisions is X. If the following equation is established when the number of divisions is X, the sense MOS transistor element is distributed in the most unbiased manner in the plane of circuitry while the minimum permissible gate size is employed:
Wm≦Ws/X<2×Wm.
Further, in order to minimize the cost of manufacture, it is preferred that the distance between individual element sections be minimized within the limits imposed on the manufacturing process. In other words, it is preferred that the individual element sections be positioned nearest to each other.
It is also preferred that the main MOS transistor element and the sense MOS transistor element have the same center of gravity when they are laid out.
The present invention produces a great advantageous effect particularly when it is applied to an LDMOS transistor that is prepared on an SOI substrate and isolated by an embedded oxide film.
As described above, the present invention makes it possible to easily obtain a semiconductor device having a MOS transistor that acts as a power IC generating a large amount of heat by itself and is expected to exhibit relative accuracy.
Embodiments of the present invention will be described in detail based on the following figures, in which:
First of all, an embodiment of the conventional semiconductor device will be described. The first embodiment will be described later.
The conventional semiconductor device 601 depicted in
The LDMOS transistors 603, 604 are identical with “MOS transistor elements” described in conjunction with the present invention.
The reference numeral 603 denotes a sense MOS transistor element. The reference numeral 604 denotes a main MOS transistor element.
Referring to
The individual MOS transistor element sections in the sense MOS transistor element 603 and main MOS transistor element 604 are configured so that their gate, source, and drain terminals are electrically connected in parallel. As an example, the drain terminal is made common as depicted in
The drain current ratio between the sense MOS transistor element 603 and the main MOS transistor element 604 is referred to as the current mirror ratio.
The number of LDMOS transistors formed of the main MOS transistor element 604 and sense MOS transistor element 603 is indicated as an example. In reality, the number of LDMOS transistors varies with a demanded current mirror ratio. The actually used current mirror ratio substantially ranges from 1:100 to 1:1000.
An LDMOS transistor that forms each transistor section is insulated and isolated by an isolating oxide film 607 and provided with a drain section 605 and a source section 606. The cross-sectional structure of the LDMOS transistor is depicted in
Referring to
The conductivity type of the LDMOS transistor 701 may be either N-type or P-type. Here, it is assumed that an N-type LDMOS transistor is used. In a region of the above-mentioned isolated SOI substrate, a low-concentration N-type drift layer 712, an N-type drain layer 705, and a source layer 706 are formed. A body layer 709 and a contact layer 710 for the body layer 709 are also formed. The body layer 709 is formed by using a gate oxide film 707 and a polysilicon cap film 708 as a photomask. The concentration of the drift layer 712 and body layer 709 may be adjusted to exhibit a high breakdown voltage so that even when a high voltage is applied to the drain layer 705 and source layer 706, the resulting electric field does not concentrate on the drift layer 712 and body layer 709, but is uniformly distributed between the drain and source. The LDMOS transistor 701 represents an exemplary configuration provided by the present invention. The present invention is not limited to the structure of the LDMOS transistor 701.
In the example of a conventional structure depicted in
Meanwhile, in the first embodiment of the present invention, a sense MOS transistor element section 103 is divided wherever possible as depicted in
The first embodiment is configured so that the sense MOS transistor and the main MOS transistor are uniformly distributed in an IC chip. Therefore, even when a temperature gradient depicted in
In other words, when the graph of
The second embodiment has the same electrical connections as the first embodiment. Further, the second embodiment is similar to the first embodiment in that the sense MOS transistor element is divided into as many sections as possible.
The second embodiment differs from the first embodiment in that individual units 805 formed of a sense MOS transistor element section 803 and a main MOS transistor element section 804 are disposed point-symmetrically to each other. Hence, the center of gravity of all the disposed main MOS transistor elements and sense MOS transistor elements can be positioned at the center of the IC chip. This makes it possible to reduce variation. Further, the required area can be rendered smaller than that is required when the configuration depicted in
Number | Date | Country | Kind |
---|---|---|---|
2013-059285 | Mar 2013 | JP | national |