CN104793120B - The measurement structure of silicon hole electrical characteristics is measured based on De- embedding method - Google Patents

Abstract

The invention discloses a measurement structure for measuring the electrical characteristics of through-silicon holes based on a de-embedding method. For two-port and multi-port interconnection structures, an isolated TSV structure is arranged on the vertical bisector of the center connection line of the signal and ground TSV structure to be tested on the same side, and the isolated TSV structure is used to isolate the vertical TSV structure to be tested from the horizontal bottom surface RDL The electromagnetic coupling between conductors is mainly composed of metal through-silicon via columns arranged at intervals; the isolation TSV structure is an inner isolation TSV structure or a combination of an inner isolation TSV structure and an outer isolation TSV structure. The invention is fully compatible with the existing TSV production process, and is suitable for measuring the electromagnetic characteristics of microwave and millimeter-wave band TSVs through de-embedded experimental testing methods. Compared with traditional measurement methods, the measurement accuracy is greatly improved. , The field of measurement of three-dimensional structures in the millimeter wave band will have great application value.

Description

Measuring Structure for Measuring TSV Electrical Properties Based on De-embedding Method

technical field

The present invention relates to a measurement structure of integrated circuit through-silicon vias, in particular to a measurement structure for measuring the electrical characteristics of through-silicon vias based on the de-embedding method in the field of microwave and millimeter-wave device testing, which weakens the through-silicon vias and their connections The coupling noise of the horizontal interconnection lines can significantly improve the accuracy of the electrical characteristics measured by the de-embedding method.

Background technique

At present, the design of mainstream integrated circuits, including Intel's multi-chip architecture, continues the traditional two-dimensional flat system architecture. Limit constraints call for the emergence of a new integrated circuit system architecture in order to fully reflect its three-dimensional vertical scale—this is the three-dimensional integrated circuit. Three-dimensional integrated circuit technology has become an internationally recognized key frontier technology for the medium and long-term sustainable development of the microelectronics industry, and the through-silicon via structure, as its core technology, has become an international research hotspot.

The electrical characteristics of through-silicon vias in microwave and millimeter wave bands have an important impact on the performance of three-dimensional integrated circuits. As a vertical structure, the electrical characteristics of through-silicon vias cannot be directly measured by single-sided probes commonly used at present. The experimental work is carried out on the "vertical-horizontal-vertical" structure. By changing the length of the horizontal structure, the electrical characteristics of the entire structure are measured multiple times, or the electrical characteristics of the entire structure and the electrical characteristics of the horizontal structure are measured. The electrical characteristics of vertical TSVs are obtained by embedding method.

The present invention is based on a de-embedding method, which has the advantage that it only needs to be measured on the same plane, and does not need to measure a separate horizontal interconnection structure.

Assume that the transmission matrix of the electrical characteristics of the measured passive vertical structure TSV is [X] _T , and the matrix of the scattering coefficient is [X] _S ; the transmission matrix of the electrical characteristics of the horizontal connection line at the bottom is [R] _T , and the matrix of the scattering coefficient is [R] _S ; the transmission matrix of the electrical properties of the overall structure is [DUT] _T , and the scattering coefficient matrix is [DUT] _S . In the actual measurement, multiple sets of bottom horizontal connecting lines with different lengths are set, the lengths of which are 500 μm, 1000 μm and 2000 μm, etc. respectively. According to different lengths, the parameters of its electrical characteristics should have the following relationship:

[R] _T,2000μm ＝[R] _T,1000μm ² ＝[R] _T,500μm ⁴ (1)

For overall structures of different lengths, the following relationship should ideally be satisfied,

[DUT] _T = [X] _T × [R] _T × [X] _T (2)

The scattering coefficient matrix [DUT] _S of the overall structure can be directly measured by a probe station and a vector network analyzer. The corresponding relationship between the two-port scattering matrix [T] and the transmission matrix [S] is as follows (four ports are omitted),

Therefore the [DUT] _T of multiple group lengths is a quantity that can be obtained by direct measurement. The vertical structure TSVs to be tested remain unchanged in these overall structures, so formulas (1) (2) can be used to derive:

Multiple sets of lengths can be used to obtain multiple sets of measurement results, and the least square method can be used to obtain more accurate results. However, this method is based on the ideal condition of formula (2), that is, the vertical interconnection and the horizontal interconnection do not affect each other, and the electrical characteristics of the two remain unchanged after the connection, but in the microwave and millimeter wave frequency bands, the strong electromagnetic coupling between the two This makes them unable to keep their respective electrical characteristics unchanged, and this condition cannot be met, which in turn greatly increases the error of the entire method, as shown in Figure 1.

Therefore, in the microwave and millimeter wave frequency bands, there is a strong electromagnetic coupling between the vertical TSVs and the horizontal interconnection structures, and they affect each other, which greatly increases the error of the above method.

Contents of the invention

Aiming at the above-mentioned problem of how to weaken the coupling between vertical vias and horizontal interconnection lines and improve the accuracy of the de-embedding measurement method, the present invention proposes a measurement structure based on the de-embedding method for measuring the electrical characteristics of through-silicon vias, which is suitable for two-port And multi-port, can greatly improve the accuracy of measurement.

The technical solution adopted by the present invention to solve its technical problems:

The invention includes a silicon substrate, a bottom RDL conductor, a TSV structure to be tested connected through the bottom RDL conductor, bumps and test pins, and the TSV structure to be tested is composed of two sets of TSV pillars to be tested symmetrically distributed on both sides. The TSV structure to be tested includes a signal TSV structure to be tested and a ground TSV structure to be tested, which is characterized in that: for two-port and multi-port interconnection structures, the vertical bisection of the central connection between the signal TSV structure to be tested and the ground TSV structure to be tested on the same side An isolated TSV structure is arranged on the line. The isolated TSV structure is used to isolate the electromagnetic coupling between the vertical TSV structure to be tested and the horizontal bottom RDL conductor. The isolated TSV structure is mainly composed of metal through-silicon via columns arranged at intervals; the isolated TSV structure The inner isolation TSV structure or the combination of the inner isolation TSV structure and the outer isolation TSV structure, the inner isolation TSV structure is located inside the interconnection, and the outer isolation TSV structure is located outside the interconnection.

The top surface of the silicon substrate is provided with a top substrate insulating layer, and the bottom substrate insulating layer is provided between the lower surface of the silicon substrate and the RDL conductor on the bottom surface, and the metal through-silicon via column for isolating the TSV structure penetrates the silicon substrate, without Into the upper and lower top and bottom substrate insulating layers of the silicon substrate.

Preferably, the distance between the midpoint of the connecting line between the signal TSV structure to be tested and the ground center of the TSV structure to be tested and the center of the TSV column closest to the TSV structure to be tested is greater than that of a single silicon chip to be tested in the TSV structure to be tested. The sum of the radii of the via pillar and the individual TSV pillars that isolate the TSV structure.

Preferably, the gap between the adjacent metal TSV columns is greater than or equal to the diameter of the metal TSV columns.

An oxide insulation layer is provided between the metal TSV column and the substrate for isolating the TSV structure.

The metal through-silicon via for isolating the TSV structure can use a metal whose work function is close to that of silicon, and an ohmic contact layer is formed between the isolating TSV structure and the silicon substrate.

Preferably, the metal through-silicon vias for isolating the TSV structure are made of metals such as gold and platinum, and an ohmic contact layer is formed between the metal through-silicon vias and the silicon substrate.

The silicon substrate is heavily doped around the isolation TSV structure, and an ohmic contact layer is formed between the metal TSV column and the silicon substrate.

The isolation TSV structure of the present invention has three structural situations, namely, an oxide insulating layer is provided, a metal with a work function similar to silicon is used, and an ordinary metal is used but a heavily doped layer is introduced. The latter two have no oxide layer. The three structures are generally Not used at the same time. The effect of removing the oxide layer is better, but the cost is higher.

The beneficial effects that the present invention has are:

The present invention greatly weakens the electromagnetic coupling between the vertical through-silicon vias and the horizontal interconnection lines in the basic measurement structure "vertical-horizontal-vertical" by adding isolation columns to the silicon, and at the same time controls the isolation from the isolation columns to the through-silicon vias that transmit signals. The distance between the holes and the density of the isolation columns make the isolation columns have very little influence on the electromagnetic characteristics of the vertical structure and the horizontal structure itself, and then the electromagnetic characteristics of the TSVs can be obtained more accurately through the de-embedding test method.

In the present invention, considering the ideal conditions of the de-embedding method, the isolation through-silicon vias are appropriately added, so that the conditions are closer to the ideal situation, and then the error of the de-embedding method for the transmission characteristics of the two-port network is reduced by tens of percentage points, while for For a four-port network, the errors of its transmission characteristics and remote coupling are reduced by tens of percentage points.

Description of drawings

Fig. 1 is a schematic diagram of the principle of parasitic parameters of the prior art problem addressed by the present invention.

Fig. 2 is a schematic diagram of the principle of the measurement structure of the present invention.

Fig. 3 is a schematic diagram of the structure of the present invention for dual ports.

Fig. 4 is a schematic diagram of the structure of the present invention for four ports.

Fig. 5 is a top view of the present invention using one-sided shielding.

Figure 6 is a front view of the present invention using a single-sided shield.

Fig. 7 is a top view of the present invention using shielding on both sides.

FIG. 8 is a partially enlarged view of the isolated TSV structure with an oxidation protection layer in FIG. 5 .

FIG. 9 is a partially enlarged view of the isolated TSV structure with a heavily doped layer in FIG. 5 .

Fig. 10 is a schematic diagram of the near-end comparison results using the four-port structure of Fig. 4 in the embodiment.

FIG. 11 is a schematic diagram of remote comparison results using the four-port structure in FIG. 4 in the embodiment.

In the figure: 1. Silicon substrate, 2. RDL conductor on the bottom surface, 3. TSV structure for signals to be tested, 4. Via, 5. Internal isolation TSV structure, 6. Test pins (Pads), 7. Oxidation Insulation layer, 8, top substrate insulation layer, 9, bottom substrate insulation layer, 10, outer isolation TSV structure, 11, ohmic contact layer, 12, ground TSV structure to be tested.

detailed description

The present invention will be further described below in conjunction with drawings and embodiments.

As shown in Figure 2, the measurement structure of the present invention includes a silicon substrate 1, a bottom RDL (Redistribution Layer, rewiring layer) conductor 2, a TSV (Through Silicon Via, TSV) structure to be tested connected through the bottom RDL conductor 2, a bump 4 and test pin 6, the TSV structure to be tested is composed of two groups of TSV pillars to be tested symmetrically distributed on both sides, the TSV structure to be tested includes a signal TSV structure to be tested 3 and a ground TSV structure to be tested 12, these structures are It is made of metal, and it is isolated from the silicon substrate 1 by insulating materials such as a structural oxide insulating layer 7 , a top substrate insulating layer 8 , and a bottom substrate insulating layer 9 .

As shown in Figures 3 and 4, for two-port and multi-port interconnection structures, an isolated TSV structure is arranged on the vertical bisector of the central connection line of the signal TSV structure 3 to be tested and the ground TSV structure 12 to be tested on the same side, and the isolated TSV structure It is used to isolate the electromagnetic coupling between the vertical TSV structure to be tested and the horizontal bottom RDL conductor 2. The isolation TSV structure is mainly composed of metal through-silicon via columns arranged at intervals, and all metal through-silicon via columns are symmetrical on both sides (left and right sides The center line between the two groups of TSV pillars to be tested on both sides is symmetrical, that is, the metal TSV pillars on both sides of the inner isolation TSV structure 5 are symmetrical or the metal TSV pillars on both sides of the outer isolation TSV structure 10 are symmetrical, The inner and outer TSVs of the TSV structure to be tested need not be symmetrical); the isolation TSV structure is the combination of the inner isolation TSV structure 5 or the inner isolation TSV structure 5 and the outer isolation TSV structure 10, and the inner isolation TSV structure 5 is located inside the interconnection, and the outer Isolation TSV structures 10 are located outside the interconnects.

The method of the present invention considers that a large electromagnetic coupling is generated between the vertical TSV and the horizontal interconnection line, which leads to an increase in the error measured by the de-embedding method, and if this part of the coupling is weakened by isolation, it can be increased. Great measurement accuracy.

The isolated TSV structure has an obvious isolation effect on the coupling between the vertical "signal" TSV and the horizontal "ground" RDL. In addition, because there is a certain distance between the isolated TSV and the center of the structure to be tested, for the structure to be tested, only It is equivalent to making a slight change to the characteristics of the substrate 1, which has little influence on the electrical characteristics of the structure to be tested; at the same time, the top surface of the silicon substrate 1 is provided with a top substrate insulating layer 8, and the lower surface of the silicon substrate 1 A bottom substrate insulating layer 9 is provided between the bottom surface and the bottom RDL conductor 2, and the isolation TSV structure penetrates the silicon substrate 1, and does not enter the top substrate insulating layer 8 and the bottom substrate insulating layer 9 above and below the silicon substrate 1. Therefore, the influence on the electromagnetic characteristics of the horizontal RDL is also very small.

The metal through-silicon vias of the TSV structure of the present invention do not require uniform distribution. In fact, the closer the area of the "signal-ground" or "signal-ground-signal" TSV to be tested is, the stronger the electromagnetic coupling is, without affecting the TSV to be tested. Under the premise of its own characteristics, the higher the density of the isolation column in this area, the better, and the density can be reduced if it is far away from this area. Of course, the design of the density needs to comprehensively consider the influence of the TSV to be measured and the medium.

In the present invention, the distance between the midpoint of the connecting line between the center of the signal TSV structure 3 to be tested and the ground TSV structure 12 to be tested and the center of the TSV column closest to the TSV structure to be tested is preferably such that it does not affect the electrical characteristics of the TSV to be tested. The density should not affect the electrical properties of the substrate 1. It should be at least greater than the sum of the radii of the single TSV column to be tested and the single metal TSV column that isolates the TSV structure. Preferably, it should be close to the 1.5 times the sum of the radii of a single TSV pillar to be tested and a single metal TSV pillar isolating the TSV structure.

The gap between the edges of the adjacent metal TSV pillars is greater than or equal to the diameter of the metal TSV pillars. Under the premise of not affecting the electrical characteristics of the structure 3 to be tested, the diameter of the TSV pillars isolating the TSV structure can be as large as possible, and the spacing can be as small as possible, but there is no requirement for whether they are evenly distributed.

As shown in Figure 3, the isolation TSV can only be arranged inside the interconnection, but if conditions permit, the isolation effect of the combination of the inner and outer sides of the interconnection will be better, as shown in Figure 4.

As shown in FIG. 6 , an oxide insulating layer 7 should be provided around the cylinder surface of the TSV cylinder.

The preferred metal TSV column uses a metal whose work function is close to that of silicon, such as gold, platinum, etc., so that the inner metal forms an ohmic contact with the peripheral silicon, and an ohmic contact layer 11 is formed between the metal TSV column and the silicon substrate 1 .

Preferably, the silicon substrate 1 is heavily doped around the metal TSV column, so that the inner metal and the peripheral silicon form an ohmic contact, and an ohmic contact layer 11 is formed between the metal TSV column and the silicon substrate 1 .

For the inner isolation TSV structure 5 and the outer isolation TSV structure 10, in terms of size, it should be as small as possible under the condition of process compatibility; for density, it is better to make the edge-to-edge distance close to the diameter; in terms of position, as shown in Figure 5 and Figure 7 As shown, a certain number of inner isolation TSV structures 5 and outer isolation TSV structures 10 can be arranged on both sides of the vertical structure to be tested, or they can only be arranged inside the connection line as shown in FIG. The vias should be as close to the signal TSVs as possible under the premise that the influence of the electrical characteristics of the vias can be ignored, and the distance between them and the signal TSV should be as small as possible without significantly affecting the signal TSV in the frequency band of interest. The inner isolation TSV structure 5 can be isolated from the silicon substrate 1 by using the oxidation insulating layer 7 shown in FIG. 8 , or by using the ohmic contact layer 11 in FIG. touch.

The concrete implementation test process of the present invention is as follows:

(a) The electrical characteristics of the TSV structure to be tested on the silicon substrate 1 cannot be directly measured by an ordinary single-sided test probe station, so a TSV-RDL-TSV multi-segment structure is used to place the test pin 6 on the silicon substrate 1 side indirect measurement.

(b) As shown in Figure 1, in the measurement of the traditional de-embedding method, there is no shielding effect of structures 5 and 10, resulting in coupling noise between the vertical structure and the horizontal structure, so that in the vertical-horizontal-vertical cascaded structure The introduction of cross-coupling noise, this parasitic parameter, will cause large calculation errors in the de-embedding method.

(c) Isolating TSVs as shown in Figure 2 serves to block cross-coupling noise. Since this structure is made of metal and penetrates through the silicon substrate, it isolates the intersecting part of the three-dimensional coupling noise and blocks the noise propagation path to a large extent. At the same time, the isolated TSV is physically insulated from the transmission cascade structure, so the electrical characteristics of the structure to be tested are less affected. When the measurement structure in the form of Fig. 9 is adopted, the shielding effect is better improved, and the improved frequency band is also larger.

(d) In the case of not introducing an isolated TSV, the measurement results of the signal-ground-signal or multi-port structure using indirect measurement are quite different from the direct calculation results. After the introduction of the isolated TSV, the relative error between the measurement result and the direct calculation result is greatly reduced.

The present invention is illustrated by taking the full-wave simulation results of the four-port interconnection structure shown in FIGS. 1 and 2 as an example. The isolated TSV selected here uses copper, so an insulating layer needs to be added.

In the embodiment of the present invention, the four-port structure shown in FIG. 4 is adopted, and all TSV column structures 3, 5, 10, and 12 have a silicon dioxide oxide layer 7 with a thickness of 0.5 microns, a diameter of 40 microns, and a height of 100 microns; a silicon substrate The thickness of structure 1 is 100 microns, and the doped silicon material with the conductivity of 10 Siemens per meter is used. The thickness of the upper and lower insulating silicon dioxide oxide layers of the silicon substrate is 0.5 microns; the test pin 6 of the upper probe is 2 microns. Thick metal copper with a width of 50 microns, the thickness of the RDL conductor 2 on the bottom surface is also 2 microns, and the width is 50 microns; the diameter of the protrusion 4 is 35 microns, and the thickness is 0.5 microns the same as the silicon dioxide insulating layer; the signal to be tested TSV3 and the ground The distance between the centers of the TSVs 12 to be tested is 100 microns; the distance between the midpoint of the line connecting the centers of the signal TSV structures 3 to be tested and the ground TSV structures 12 to be tested and the center of the TSV pillar 5 or 10 closest to the TSV structure to be tested is 60 μm. micron; the center-to-center spacing of adjacent inner isolation TSV pillars 5 is 80 microns, and the center-to-center spacing of adjacent outer isolation metal TSV columns 10 is 80 microns.

In the embodiment of the present invention, the comparison results of the near-end coupling characteristics obtained by using the four-port structure in Figure 4 and the near-end coupling characteristics obtained by the traditional method and the ideal near-end coupling characteristics are shown in Figure 10, and the far-end coupling characteristics obtained by the traditional method The comparison results of the far-end coupling characteristics and the ideal far-end coupling characteristics are shown in Figure 11. Based on the simulation results of the "signal-ground-signal" structure to be tested, for the traditional structure, due to the large vertical-horizontal —Vertical coupling noise, the traditional measurement results of its transmission characteristics and far-end coupling characteristics are poor, and the accuracy of the de-embedding algorithm is low. After adding isolated TSVs, it has a greater impact on the electrical characteristics of the TSV structure to be tested and the horizontal RDL structure itself Under the premise of small, the coupling between them is weakened, thereby greatly improving the calculation accuracy of transmission characteristics and remote coupling characteristics. Since the isolation TSV introduces an additional insulating layer, it needs to be in a higher frequency band, such as the 10-30 GHz frequency band concerned in this example, to better reflect the isolation effect of the structure.

It can be seen that, adopting the measurement structure of the present invention, the introduction of the isolated TSV structure has little influence on the characteristics of the TSV structure to be measured and the horizontal RDL structure itself, but greatly weakens the coupling of the two, thus greatly improving the performance of the de-embedding method. Precision, with prominent and significant technical effects.

Claims (8)

1. a kind of measurement structure that silicon hole electrical characteristics are measured based on De- embedding method, including silicon substrate（1）, bottom surface RDL conductors （2）, by bottom surface RDL conductors（2）TSV structure to be measured, the projection of connection（4）And test pin（6）, TSV structure to be measured is by right Two groups of silicon hole posts to be measured that title is distributed in both sides are constituted, and TSV structure to be measured includes signal TSV structure to be measured（3）It is to be measured with ground TSV structure（12）, it is characterised in that：

For Two-port netwerk and multiport interconnection architecture, the signal TSV structure to be measured of the same side（3）With ground TSV structure to be measured（12） Isolation TSV structure is disposed with the perpendicular bisector of the line of centres, isolation TSV structure is used to isolate vertical TSV structure to be measured With the bottom surface RDL conductors of level（2）Between electromagnetic coupled, isolation TSV structure is main by spaced apart metal silicon hole post Constitute, all metal silicon hole post both sides are symmetrical；Isolation TSV structure is that inner side isolates TSV structure（5）Or inner side isolation TSV Structure（5）Isolate TSV structure with outside（10）Both combinations, inner side isolation TSV structure（5）Positioned at interconnection inner side, lateral septal From TSV structure（10）Positioned at interconnection outside.

2. a kind of measurement structure that silicon hole electrical characteristics are measured based on De- embedding method according to claim 1, its feature exists In：Described silicon substrate（1）Upper top surface is provided with top substrate insulating barrier（8）, silicon substrate（1）Bottom surface and bottom surface RDL conductors（2） Between be provided with base substrate insulating barrier（9）, isolate the metal silicon hole post insertion silicon substrate of TSV structure（1）, do not enter silicon substrate （1）Upper and lower top substrate insulating barrier（8）With base substrate insulating barrier（9）In.

3. it is according to claim 1 and 2 it is a kind of based on De- embedding method measure silicon hole electrical characteristics measurement structure, its feature It is：Described signal TSV structure to be measured（3）With ground TSV structure to be measured（12）Line of centres midpoint with from TSV structure to be measured most The distance between near metal silicon hole post center more than TSV structure to be measured silicon hole post single to be measured with isolate TSV structure Single metal silicon hole post radius sum.

4. it is according to claim 1 and 2 it is a kind of based on De- embedding method measure silicon hole electrical characteristics measurement structure, its feature It is：Diameter of the gap more than or equal to metal silicon hole post between the adjacent metal silicon hole post.

5. it is according to claim 1 and 2 it is a kind of based on De- embedding method measure silicon hole electrical characteristics measurement structure, its feature It is：The metal silicon hole post and silicon substrate of described isolation TSV structure（1）Between be provided with oxidation insulating layer（7）.

6. it is according to claim 1 and 2 it is a kind of based on De- embedding method measure silicon hole electrical characteristics measurement structure, its feature It is：The metal silicon hole post of described isolation TSV structure isolates TSV structure and silicon using work function and the close metal of silicon Substrate（1）Between form ohmic contact layer（11）.

7. a kind of measurement structure that silicon hole electrical characteristics are measured based on De- embedding method according to claim 6, its feature exists In：Described work function is gold or platinum, metal silicon hole post and silicon substrate with the close metal of silicon（1）Between formed ohm connect Contact layer（11）.

8. it is according to claim 1 and 2 it is a kind of based on De- embedding method measure silicon hole electrical characteristics measurement structure, its feature It is：To silicon substrate around the isolation TSV structure（1）Carry out heavy doping, metal silicon hole post and silicon substrate（1）Between Form ohmic contact layer（11）.