Embodiments of the present invention relate generally to the field of microelectronic fabrication, and, in particular to systems and structures for measuring the mechanical stress on dice.
Conventionally, finite element mechanical modeling is used for in-situ measurement of mechanical stresses imparted to a silicon die on commercial packaged microelectronic parts or on certification test vehicles. However, such modeling typically involves approximations that do not provide accurate enough determinations of those stresses.
Alternatively, Raman spectroscopy may be used for measuring stress applied to silicon. However to do so requires the use of precision spectroscopy tools and also requires that the silicon to be measured be exposed to the view of the beam, making it impossible to measure silicon encased on a microelectronic package.
For simplicity and clarity of illustration, elements in the drawings have not necessarily been drawn to scale. For example, the dimensions of some of the elements may be exaggerated relative to other elements for clarity. Where considered appropriate, reference numerals have been repeated among the drawings to indicate corresponding or analogous elements.
In the following detailed description, a microelectronic die including a CMOS stress sensor thereon, a method of using the sensor, and a system including the die are disclosed. Reference is made to the accompanying drawings within which are shown, by way of illustration, specific embodiments by which the present invention may be practiced. It is to be understood that other embodiments may exist and that other structural changes may be made without departing from the scope and spirit of the present invention.
The terms on, above, below, and adjacent as used herein refer to the position of one element relative to other elements. As such, a first element disposed on, above, or below a second element may be directly in contact with the second element or it may include one or more intervening elements. In addition, a first element disposed next to or adjacent a second element may be directly in contact with the second element or it may include one or more intervening elements. In addition, in the instant description, figures and/or elements may be referred to in the alternative. In such a case, for example where the description refers to Figs. X/Y showing an element A/B, what is meant is that FIG. X shows element A and FIG. Y shows element B.
Aspects of this and other embodiments will be discussed herein with respect to
Referring to
The ring oscillators of the exemplary embodiment of
By “CMOS ring oscillator,” what is meant in the context of the instant description is a ring oscillator, such as a conventional ring oscillator including CMOS inverters. As is well known, ring oscillators are devices commonly used in test chips as an indicator of transistor speed. A drawing of a conventional ring oscillator, such as a CMOS ring oscillator, is shown by way of example in
Thus, referring still to
In general, it is desirable to be able to measure all three principal components of the stress on the silicon. However, using a single CMOS ring oscillator on the die substrate provides only one output: a ring oscillator frequency. As noted above, inverters making up the ring oscillator device typically include both NMOS and PMOS transistors. While both NMOS and PMOS are necessary for a CMOS functioning inverter, the inverters can, according to one embodiment, be NMOS weighed and/or PMOS weighed by varying the width of the transistors making up each inverter. Since NMOS and PMOS carrier mobility have different dependences on silicon strain with respect to one another, an additional independent measurement can be made by using both PMOS and NMOS weighted ring oscillators at any given die location.
ΔfN=aN,1·Δs1+aN,2·Δs2+aN,3·Δs3
ΔfP=aP,1·Δs1+aP,2·Δs2+aP,3·Δs3
Where ΔfN and ΔfP represent the change in frequency for the N-weighted and P-weighted oscillators, respectively. Δs1 represents the change in stress state of the ith principle component of stress, and aN,i and aP,i represent the changes in frequency expected due to a change in the ith principle component of stress for the N or P-weighted oscillator.
This still allows only a measurement of two quantities (NMOS and PMOS mobility change), with three unknowns (three principal stress components). One solution to this problem according to an embodiment is to position the oscillators at strategic locations on the die where one or more of the principal components of the package-induced stress are known to be small, and where the third principle component of the packaged-induced stress is known to have a maximum value. For the die edge and corner principal locations, as the oscillator approaches the extreme edge of the die, one or more of the principal stress components will go to zero. Therefore, optimally an oscillator may be placed as close to the edge or corner as possible. In practice, if an oscillator can be placed close enough so that the package-induced principal stress components in question can be kept to less than 0-10% of the package-induced principal stress component of interest, then useful data can be obtained.
A prior knowledge of regions of the die where one of the principal stress components may be assumed to have a maximum value may guide a strategic positioning of one or more ring oscillators as described above in order to allow a measurement of such stress component, and further to aid in de-convolution of the individual principal stress components from one another. Typically, for a die that is either rectangular or square shaped in a top or bottom plan view thereof, strategic positions where one of the package-induced principal stress components may be assumed to have a maximum value and where one or more of the package-induced principal stress components may have a small value correspond to the following four locations: a center region of the die substrate, a horizontal edge region of the die substrate, a vertical edge region of the die substrate, a corner region of the die substrate. Thus, as seen in
Since the stress field on a flip chip part is also strongly influenced by its location with respect to the nearest C4 bump, according to one embodiment, at least two sensors may be placed at each location where a ring oscillator is needed: one at a location, with respect to the bump, known to experience the maximum stress effect due to the bump (such as the edge of the bump facing either toward or away from the die center, where maximum out-of-plane stress is expected), and another at a proximity to the bump but removed to an extent that the stress at its location is not affected by the bump. In this case, equations (previously given) may be used to deduce the additional mechanical stress induced by the bump.
In view of the above, an embodiment may include placing one ring oscillator at each location where stress is to be determined, two ring oscillators at that same location (one ring oscillator being NMOS weighed and the other being PMOS weighed), four ring oscillator at that same location (one ring oscillator being NMOS weighed and being placed at a die bump edge or other strategically chosen location on the bump, one ring oscillator being PMOS weighed and being placed at the die bump edge or other strategically chosen location on the bump, one ring oscillator being NMOS weighed and being placed at a proximity of the die bump edge, and one ring oscillator being PMOS weighed and being placed at the proximity of the die bump edge), or any combination of the above. Thus, according to an embodiment, any die location where frequency shift measurements may be taken may include anywhere from one to four ring oscillators. Embodiments further include the provision of multiple ring oscillators on a die at regions where differing strain responses are expected to package induced stress. It is noted that embodiments are not necessarily limited to the use of ring oscillators at the given differing regions, but include within their scope the use of transistors in the same locations as described in detail above for the ring oscillators. In such a case, a measurement of electrical characteristics of the respective transistors would yield various stress values at the locations of the transistors.
Advantageously, embodiments provide a microelectronic die and method that allow an in-situ measurement of the mechanical stress imparted to a silicon die as a result of packaging using electrical tests. Embodiments bring about the above advantage by providing a microelectronic die which includes: a die substrate and a plurality of CMOS ring oscillators on the die substrate, the ring oscillators being disposed at regions of the die substrate that are adapted to exhibit differing strain responses to package-induced stress with respect to one another. Oscillators according to embodiments may advantageously be of standard design, or preferably include custom oscillators that are NMOS and PMOS weighed to advantageously allow the differentiation of principal stress components. Advantageously, using ring oscillators allow an easy multiplexing of the oscillators and require only measurement of frequency. Moreover, advantageously, embodiments allow an in-situ measurement of package-induced stress on a die without the use of modeling and of the approximations associated with such modeling.
Referring to
For the embodiment depicted by
The various embodiments described above have been presented by way of example and not by way of limitation. Having thus described in detail embodiments of the present invention, it is understood that the invention defined by the appended claims is not to be limited by particular details set forth in the above description, as many variations thereof are possible without departing from the spirit or scope thereof.
Number | Name | Date | Kind |
---|---|---|---|
6657504 | Deal et al. | Dec 2003 | B1 |
6724214 | Manna et al. | Apr 2004 | B2 |
6903564 | Suzuki | Jun 2005 | B1 |
7190233 | Bhushan et al. | Mar 2007 | B2 |
7242093 | Ueda | Jul 2007 | B2 |
7268399 | Bowen et al. | Sep 2007 | B2 |
20030134496 | Lee et al. | Jul 2003 | A1 |
20050040397 | Hui et al. | Feb 2005 | A1 |
20050118758 | Yuan et al. | Jun 2005 | A1 |
20050139929 | Rost | Jun 2005 | A1 |
20050140418 | Muniandy et al. | Jun 2005 | A1 |
20050218871 | Kang et al. | Oct 2005 | A1 |
20050225415 | Mahony et al. | Oct 2005 | A1 |
20050273290 | Asano et al. | Dec 2005 | A1 |
20060043579 | He et al. | Mar 2006 | A1 |
20070013452 | Bhushan et al. | Jan 2007 | A1 |
20080136400 | Chi et al. | Jun 2008 | A1 |
20080258700 | Block et al. | Oct 2008 | A1 |
20080270049 | Kim et al. | Oct 2008 | A1 |
Number | Date | Country | |
---|---|---|---|
20090058540 A1 | Mar 2009 | US |