Non-polished glass wafer, thinning system and method for using the non-polished glass wafer to thin a semiconductor wafer

Abstract

A non-polished glass wafer, a thinning system, and a method for using the non-polished glass wafer to thin a semiconductor wafer are described herein. In one embodiment, the glass wafer has a body (e.g., circular body) including a non-polished first surface and a non-polished second surface substantially parallel to each other. In addition, the circular body has a wafer quality index which is equal to a total thickness variation in micrometers plus one-tenth of a warp in micrometers that is less than 6.0.

Claims

1. A method for manufacturing non-polished glass wafers, the method comprising the steps of: melting batch materials and forming molten glass; providing a forming device for receiving the molten glass and forming a glass sheet; providing an apparatus comprising an enclosure disposed about the forming device, the enclosure comprising an opening below the forming body to allow a flow of molten glass descending from the forming device to pass from the enclosure, the apparatus further comprising thermal shields positioned below the forming device and spaced apart to form the opening, where the thermal shields are configured for minimizing radiative heat loss from the forming device, each thermal shield being movable relative to the flow of the molten glass, the apparatus further comprises cooling members positioned adjacent to the thermal shields, wherein the step of providing the enclosure further comprises steps of, controlling a pressure within the enclosure; controlling the cooling members; and minimizing the opening between the thermal shields to control heat transfer uniformity with respect to the glass sheet; drawing the glass sheet; separating the glass sheet into distinct glass sheets; and cutting at least one of the distinct glass sheets into a plurality of non-polished glass wafers, wherein each non-polished lass wafer comprises: a body including a non-polished first surface and a non-polished second surface which are substantially parallel to each other; the body has a wafer quality index which is less than about 6.0, wherein the wafer quality index is equal to a total thickness variation in micrometers plus one-tenth of a warp in micrometers, wherein the total thickness variation is a difference between a highest thickness elevation and a lowest thickness elevation across the body between the non-polished first surface and the non-polished second surface, and the warp is a sum of absolute values of a maximum distance between a highest point and a least squares focal plane applied to a shape of the body and a maximum distance between a lowest point and the least squares focal plane applied to the shape of the body, where the highest point and the lowest point are both with respect to same surface of the body.

2. The method of claim 1, wherein the non-polished first surface has a surface roughness less than about 5 RMS.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) A more complete understanding of the disclosure may be had by reference to the following detailed description when taken in conjunction with the accompanying drawings wherein:

(2) FIG. 1 (PRIOR ART) is a diagram illustrating how the packaging of integrated circuits has evolved over the years from wire bound, flip chip, stacked die, package-on-package to 3D IC;

(3) FIG. 2 (PRIOR ART) illustrates a 32 Gigabyte (GB) standard eight IC design (top), which is four inches long, located on a traditional circuit board and a commercially available 32 GB 3D memory stack (bottom) having eight IC's each being 55 m thick and 1 mm wide;

(4) FIG. 3 (PRIOR ART) is a schematic diagram illustrating a 3D IC structure which has a poor TTV that resulted in poor interconnects between a top IC layer and a bottom IC layer;

(5) FIG. 4 (PRIOR ART) is a schematic diagram of a glass wafer used to explain TTV which is defined to be the difference between a highest thickness (Tmax) elevation and a lowest thickness (Tmin) elevation on the entire surface of the unclamped (free state) glass wafer;

(6) FIG. 5 (PRIOR ART) is a schematic diagram of a glass wafer used to explain the definition of warp;

(7) FIG. 6 (PRIOR ART) is a schematic of a glass wafer that is polished to reduce warp but by reducing warp this will at the same time also increase the TTV;

(8) FIGS. 7A and 7B respectively show a perspective view and a side view of an exemplary non-polished glass wafer;

(9) FIGS. 8A-8D are several diagrams illustrating the warp (flatness) and TTV of an exemplary non-polished glass wafer;

(10) FIGS. 9A-9B (PRIOR ART) are two diagrams illustrating the warp (flatness) of a first side and a second side of a traditional polished glass wafer with a 200 mm outer diameter;

(11) FIGS. 10A-10C are three diagrams respectively illustrating the surface roughness of a portion of two traditional polished glass wafers and an exemplary non-polished glass wafer;

(12) FIG. 11 is a graph used to illustrate the strengths of three traditional silicon wafers compared to an exemplary non-polished glass wafer;

(13) FIG. 12 is a diagram illustrating the basic steps associated with a method for using the non-polished glass wafer to thin a silicon wafer;

(14) FIG. 13 is a block diagram of an exemplary thinning system which uses a thinning mechanism to thin a silicon wafer while it is bonded to the non-polished glass wafer;

(15) FIG. 14 is a schematic view of an exemplary glass manufacturing system that uses the fusion process and a forming body (e.g., isopipe) to manufacture a non-polished glass sheet which can be cut to form a non-polished glass wafer;

(16) FIG. 15 is a perspective view illustrating in greater detail the forming body (e.g., isopipe) shown in FIG. 14; and

(17) FIG. 16 is a cross sectional side view of an enhancement that can be made with respect to the forming body (e.g., isopipe) shown in FIGS. 14-15 which can help better control the final thickness of the non-polished glass sheet that can be cut to form the non-polished glass wafer.

DETAILED DESCRIPTION

(18) Referring to FIGS. 7A and 7B, there are respectively shown a perspective view and a side view of an exemplary non-polished glass wafer 700. As shown, the exemplary glass wafer 700 includes a body 702 that has a non-polished first surface 704 and a non-polished second surface 706 which are substantially parallel to each other. Although the body 702 is shown and described herein as being a circular body 702 it should be appreciated that the body 702 can also have any other shape such as rectangular, square, oval, or the like. Furthermore, the body 700 has a physical attribute that is referred to herein as the wafer quality index which is less than about 6.0, or less than about 4.5, or less than about 3.0. The wafer quality index is equal to TTV (m) plus [warp (m)/10]. The TTV is a difference between a highest thickness elevation 708 and a lowest thickness elevation 710 across the body 702 between the non-polished first surface 704 and the non-polished second surface 706. The warp is a sum of absolute values of a maximum distance 712 between a highest point 714 and a least squares focal plane 716 applied to a shape of the body 702 and a maximum distance 718 between a lowest point 720 and the least squares focal plane 716. The highest point 714 and the lowest point 720 are both with respect to same surface 704 or 706 of the glass wafer 700. The side view of the glass wafer 702 is not to scale and the dimensions have been greatly exaggerated so one can readily see the various features 708, 710, 712, 714, 716, 718 and 720 associated with warp and TTV.

(19) The wafer quality index was developed because the individual attributes of TTV and warp are not only important by themselves, but the performance of the glass wafer 700 depends on the interrelationship of TTV and warp. Hence, the wafer quality index is characterized by both TTV and warp. As one example, many semiconductor customers today specify only TTV and have a cursory loose specification for warp. For instance, a semiconductor customer may specify that there be only a maximum warp of 100 m across the glass wafer. However, this loose specification for warp negatively affects the performance of the glass wafer. Hence, the wafer quality index was developed to provide a far more valuable metric for predicting the performance of the glass wafer rather than just specifying TTV alone with a cursory loose specification for warp. In particular, by defining the wafer quality index, the major factors can be accounted for and useful predictions can be made concerning the quality of a thinned semiconductor wafer.

(20) The glass wafer 700 can also have several physical and compositional attributes which are desirable so the glass wafer 700 can effectively be used to thin a semiconductor wafer. The various physical and compositional attributes can be as follows:

(21) 1. Coefficient of Thermal Expansion (CTE)the glass wafer 700 could have a CTE in a range of about 3.0 ppm/ C.-3.5 ppm/ C. In this case, the CTEs of the glass wafer 700 and the semiconductor wafer would match or at least closely match one another which is often desirable because during the thinning process the glass wafer 700 will be bonded to the semiconductor wafer and the composite structure will likely undergo thermal cycling. An exemplary composition which can be used to make the glass wafer 700 so it has a CTE in the range of about 3.0 ppm/ C. to about 3.5 ppm/ C. has a nominal composition of (in mole percent on an oxide basis): SiO.sub.2: 64.0-71.0; Al.sub.2O.sub.3: 9.0-12.0; B.sub.2O.sub.3: 7.0-12.0; MgO: 1.0-3.0; CaO: 6.0-11.5; SrO: 0-2.0; BaO: 0-0.1, wherein: (a) 1.00[RO]/[Al.sub.2O.sub.3]1.25, where [Al.sub.2O.sub.3] is the mole percent of Al.sub.2O.sub.3 and [RO] equals the sum of the mole percents of MgO, CaO, SrO, and BaO; and (b) the glass has at least one of the following compositional characteristics: (i) on an oxide basis, the glass comprises at most 0.05 mole percent Sb.sub.2O.sub.3; and (ii) on an oxide basis, the glass comprises at least 0.01 mole percent SnO.sub.2 (hereinafter this composition is referred to as composition no. 1).

(22) If desired, the glass wafer 700 can have a CTE that matches or at least closely matches that of the composite bonded structure including the silicon wafer with features plus the bonding adhesive (See FIG. 12). In this case, the glass wafer 700 would likely have a CTE different than 3.0 ppm/ C.-3.5 ppm/ C. which is not possible with a silicon carrier wafer which has a CTE of about 3.2 ppm/ C. To accomplish this, the glass wafer 700 can be made from different glass compositions so it would have different CTEs while all the other physical attributes such as the WQI, TTV, warp, size, thickness, and surface roughness would remain the same as the glass wafer 700 described above which has a CTE in the range of about 3.0 ppm/ C. to about 3.5 ppm/ C. For instance, an exemplary nominal glass composition which results in the glass wafer 700 having a CTE of about 6.0 ppm/ C. is 69.2 mol % SiO.sub.2, 8.5 mol % Al.sub.2O.sub.3, 13.9 mol % Na.sub.2O, 1.2 mol % K.sub.2O, 6.5 mol % MgO, 0.5 mol % CaO, and 0.2 mol % SnO.sub.2 (hereinafter this composition is referred to as composition no. 2). And, an exemplary nominal glass composition which results in the glass wafer 700 having a CTE of about 8.5 ppm/ C. is 61 mol %SiO.sub.275 mol %; 7 mol %Al.sub.2O.sub.315 mol %; 0 mol %B.sub.2O.sub.312 mol %; 9 mol %Na.sub.2O21 mol %; 0 mol %K.sub.2O4 mol %; 0 mol %MgO7 mol %; and 0 mol %CaO3 mol %. The glass, in some embodiments, further comprises at least one fining agent such as, but not limited to, a halogen or a multivalent fining agent such as As.sub.2O.sub.3, Sb.sub.2O.sub.3, SnO.sub.2, or Fe.sub.2O.sub.3. In a particular embodiment, the total concentration of As.sub.2O.sub.3 and Sb.sub.2O.sub.3, if either or both are present, is less than 500 ppm by weight (hereinafter this composition is referred to as composition no. 3). Another, an exemplary nominal composition which results in the glass wafer 700 having a CTE of about 10.0 ppm/ C. is a glass nominally comprising in mole percent on an oxide basis: SiO.sub.2 64.0-72.0; Al.sub.2O.sub.3 9.0-16.0, B.sub.2O.sub.3 1.0-5.0; MgO+La.sub.2O.sub.3 1.0-7.5; CaO 2.0-7.5; SrO 0.0-4.5; BaO 1.0-7.0, wherein (MgO+CaO+SrO+BaO+3La.sub.2O.sub.3)/(Al.sub.2O.sub.3)1.15, where Al.sub.2O.sub.3, MgO, CaO, SrO, BaO, and La.sub.2O.sub.3 represent the mole percents of the respective oxide components (hereinafter this composition is referred to as composition no. 4). In fact, depending on the composition, the glass wafer 700 can have any desired CTE which is in a range of about 3.0 ppm/ C. to about 12.0 ppm/ C.

(23) 2. TTVthe TTV of the semiconductor wafer can only be as good as the TTV of the glass wafer 700. Plus, as the semiconductor wafer's thickness requirements become smaller, the TTV of the glass wafer 700 must be smaller. Hence, in exemplary embodiments, the glass wafer 700 should have a TTV that is below about 2.0 m or below about 1.0 m.

(24) 3. Compositionmetals, most notably sodium, can cause failures in IC's. Although the glass wafer 700 is not a final part of the IC structure, the degree of migration of sodium and other alkali from the glass wafer 700 to the semiconductor wafer is not yet known when the glass wafer 700 is bonded and then de-bonded from the semiconductor wafer during the thinning process. Therefore, the glass wafer 700 should avoid having metals altogether or at least to have them only at some trace level. For instance, the glass wafer should have less than about 0.05% weight percent of Na.sub.2O and K.sub.2O.

(25) The glass wafer 700 can be obtained by using any one of a variety of glass compositions. For example, the glass composition chosen can be any of a wide range of silicate, borosilicate, aluminosilicate, or boroaluminosilicate glass compositions, which optionally can comprise one or more alkali and/or alkaline earth modifiers. By way of illustration, one such glass composition includes the following constituents: 58-72 mole percent (mol %) SiO.sub.2; 9-17 mol % Al.sub.2O.sub.3; 2-12 mol % B.sub.2O.sub.3; 8-16 mol % Na.sub.2O; and 0-4 mol % K.sub.2O, wherein the ratio

(26) $\frac{{\mathrm{Al}}_2{O}_3\left(\mathrm{mol}\%\right)+B_2{O}_3\left(\mathrm{mol}\%\right)}{.Math.\mathrm{modifiers}\left(\mathrm{mol}\%\right)}>1,$
where the modifiers comprise alkali metal oxides. Another glass composition includes the following constituents: 61-75 mol % SiO.sub.2; 7-15 mol % Al.sub.2O.sub.3; 0-12 mol % B.sub.2O.sub.3; 9-21 mol % Na.sub.2O; 0-4 mol % K.sub.2O; 0-7 mol % MgO; and 0-3 mol % CaO. Yet another illustrative glass composition includes the following constituents: 60-70 mol % SiO.sub.2; 6-14 mol % Al.sub.2O.sub.3; 0-15 mol % B.sub.2O.sub.3; 0-15 mol % Li.sub.2O; 0-20 mol % Na.sub.2O; 0-10 mol % K.sub.2O; 0-8 mol % MgO; 0-10 mol % CaO; 0-5 mol % ZrO.sub.2; 0-1 mol % SnO.sub.2; 0-1 mol % CeO.sub.2; less than 50 parts per million (ppm) As.sub.2O.sub.3; and less than 50 ppm Sb.sub.2O.sub.3; wherein 12 mol %Li.sub.2O+Na.sub.2O+K.sub.2O20 mol % and 0 mol % MgO+CaO10 mol %. Still another illustrative glass composition includes the following constituents: 55-75 mol % SiO.sub.2, 8-15 mol % Al.sub.2O.sub.3, 10-20 mol % B.sub.2O.sub.3; 0-8% MgO, 0-8 mol % CaO, 0-8 mol % SrO and 0-8 mol % BaO. Further, the substrate optionally can be annealed and/or strengthened (e.g., by thermal tempering, chemical ion-exchange, or like processes).

(27) 4. Warpthe warp of the glass wafer 700 can greatly affect the performance/properties of the thinned semiconductor wafer. The glass wafer 700 should have a warp that is less than about 60 m, or less than 30 m, or less than 20 m.

(28) 5. Sizethe size of the quality area of the glass wafer 700 with a circular body 702 should be able to reach about 300 mm outer diameter, and scale to about 450 mm to match the larger semiconductor wafers in the coming years. In particular, the glass wafer 700 with a circular body 702 should have an outer diameter in the range of about 150 mm to about 450 mm.

(29) 6. Thicknessthe glass wafer 700 should have a thickness that is nominally about 0.7 mm but no less than about 0.4 mm and no greater than about 1.1 mm.

(30) 7. Surface roughnessthe glass wafer 700 should have a very low surface roughness to improve surface adhesion of the bonding agent used to attach the glass wafer 700 to the semiconductor wafer. In addition, the glass wafer 700 should have a very low surface roughness to improve number of re-cycles or number of times the glass wafer 700 can be used in a thinning process. In particular, the glass wafer 700 should have a very low surface roughness because microcracking and microcorrugation can lead to accelerated damage during the cleaning cycle, which frequently uses a hydrogen fluoride (HF) solution. In certain embodiments, the glass wafer 700 has a RMS (Root Mean Square) surface roughness that is less than or equal to about 5 Angstroms ().

(31) Referring to FIGS. 8A-8D, there are several diagrams illustrating the warp (flatness) and TTV of an exemplary non-polished glass wafer 700. FIGS. 8A-8B are respectively a top view and an isometric view illustrating the flatness/warp (Zernike fit map) @ 1000 of the exemplary glass wafer 700. FIGS. 8C-8D are respectively a top view and an isometric view illustrating the TTV thickness (thickness/parallelism) @ 1000 of the exemplary glass wafer 700. This exemplary glass wafer 700 had a 300 mm outer diameter, a nominal thickness of about 0.7 mm. This particular glass wafer 700 and 87 other glass wafers 700 were made using the aforementioned composition no. 1 in a glass manufacturing system, which uses a fusion process as discussed below.

(32) TABLE 1 represents the average TTV and the average warp of all 88 glass wafers 700 which had the aforementioned composition no. 1 to highlight the repeatability and capability of the glass manufacturing system.

(33) TABLE-US-00001 TABLE 1 TTV Warp Target 2.0 m 30.0 m Average 1.4 m 17.4 m Maximum 2.0 m 29.1 m Minimum 1.1 m 6.5 m Standard Deviation 0.2 m 5.3 m Cp 1.9 0.9

(34) The data that was generated for these measurements involved over 3 million data points for each exemplary glass wafer 700 and has an accuracy of 0.10 m for TTV and 0.5 m for warp.

(35) Referring to FIGS. 9A-9B (PRIOR ART), there are two diagrams illustrating the warp (flatness) of a first side 902 and a second side 904 of a traditional polished glass wafer 900 with a 200 mm outer diameter. In particular, FIG. 9A shows a top view (upper) and an isometric view (lower) of the first side 902 of the traditional polished glass wafer 900 where the first side 902 had a warp of about 41 m. FIG. 9B shows a top view (upper) and an isometric view (lower) of the second side 904 of the traditional polished glass wafer 900 where the second side 904 had a warp of about 105 m. The traditional polished glass wafer 900 was made from 0.7 mm thick commercially-available borosilicate glass material. The traditional polished glass wafer 900 had substantially more warp than the exemplary glass wafers 700 associated with FIGS. 8A-8D even when the exemplary glass wafers 700 had a 300 mm outer diameter while the traditional polished glass wafer 900 has a 200 mm outer diameter.

(36) Referring to FIGS. 10A-10C, there are three diagrams respectively illustrating the surface roughness of a portion of two traditional polished glass wafers 1000a and 1000b and an exemplary non-polished glass wafer 700. The first traditional glass wafer 1000a had a first side 1002a with a surface roughness of about 5.8 . The second traditional glass wafer 1000b had a first side 1002b with a surface roughness of about 6.1 . In contrast, the exemplary glass wafer 700 had a non-polished first surface 702 with a surface roughness of about 2.5 . The traditional polished glass wafers 1000a and 1000b were made from 0.7 mm thick commercially-available borosilicate glass while the exemplary glass wafer 700 was made from 0.7 mm thick glass based on the aforementioned composition no. 1. The exemplary glass wafer 700 had a surface roughness that is substantially less than the surface roughness of the two traditional polished glass wafers 1000a and 1000b. The smaller surface roughness (less microcracks) of the exemplary glass wafer 700 is desirable for several reasons such as increasing the strength of the glass wafer 700 and increasing the number of times the glass wafer 700 can be re-used in a thinning process. The strength of the glass wafer 700 was found to be dependent more on the surface quality rather than the bulk properties of the glass wafer 700.

(37) Referring to FIG. 11, there is a graph used to illustrate the strengths as measured of three traditional silicon carrier wafers 1102, 1104, and 1106 compared to the exemplary non-polished glass wafer 700. In this graph, the y-axis represents the probability of failure (%) and the x-axis represents the load in megaPascals (MPa). The first traditional silicon carrier wafer 1102 had a 100 mm outer diameter polished surface and was 520 m thick. The second traditional silicon carrier wafer 1104 had a 100 mm outer diameter ground surface and was 520 m thick. The third traditional silicon carrier wafer 1106 had a 300 mm outer diameter polished surface and was 780 m thick. The exemplary glass wafer 700 was made from the aforementioned composition no. 1 and had a 300 mm outer diameter with non-polished surfaces 704 and 706 and was about 0.5 mm thick. The load was 0.5 inches in dimension, and a 1 inch base was used to support the traditional silicon wafers 1102, 1104 and 1106 and the exemplary glass wafer 700.

(38) Referring to FIG. 12, there is a diagram illustrating the basic steps associated with a method 1200 for using the non-polished glass wafer 700 to thin a silicon wafer 1202. At step 1204, the silicon wafer 1202 is provided which has a first surface 1206 with a circuit die 1208 formed therein and a second surface 1210 where the first surface 1206 and the second surface 1210 are substantially parallel to each other. The circuit die 1208 can be formed in the silicon wafer 1202 by a photolithography process.

(39) At step 1212, the glass wafer 700 is provided which in this example has the circular body 702 with the non-polished first surface 704 and the non-polished second surface 706 that are substantially parallel to each other (see description above for details about the wafer quality index and other exemplary attributes of the glass wafer 700). In this example, the glass wafer's non-polished first surface 702 has a bonding agent 1214 that was applied thereto by, for instance, a spin coating process. Alternatively, the silicon wafer's first surface 1206 could have the bonding agent 1214 applied thereto for instance by a spin coating process.

(40) At step 1216, the silicon wafer 1202 is placed next to the glass wafer 700 so the bonding agent 1214 bonds the silicon wafer's first surface 1206 to the glass wafer's non-polished first surface 704. In one example, an ultraviolet light may be directed through the glass wafer's non-polished second surface 706 to cure the bonding agent 1214 such that the silicon wafer's first surface 1206 is bonded to the glass wafer's non-polished first surface 704. In another example, a thermally cured temporary adhesive bonding agent 1214 might be used to bond the silicon wafer 1202 to the glass wafer 700. Alternatively, the bonding agent 1214 could utilize other mechanisms (e.g., polymerization, exposure to air, chemical reaction between components of the bonding agent 114, or the like) to facilitate bond formation between the silicon wafer 1202 to the glass wafer 700.

(41) At step 1218, the silicon wafer's second surface 1210 is thinned while the silicon wafer 1202 is bonded to the glass wafer 700. There are several ways that the silicon wafer 1202 can be thinned including using, for example, a mechanical grinding device, a chemical mechanical polishing device, an etching device (e.g., within a solvent bath) or any combination of these techniques. The silicon wafer 1202 can have an original thickness of around 700 m and be thinned to a thickness of around 50 m-60 m. In any case, regardless of the thinning technique used the quality of the glass wafer 700 ultimately leads to the quality of the thinned silicon wafer 1202. An exemplary thinning system 1300 which uses a thinning mechanism 1306 to thin the silicon wafer 1202 while it is bonded to the glass wafer 700 is described below with respect to FIG. 13.

(42) At step 1220, the thinned silicon wafer 1202 can be further processed if needed for instance by forming precision holes 1222 (two shown) therein which will function as future vertical interconnects when making the 3D IC. After the precision holes 1222 are formed then they can be metalized.

(43) At step 1224, the thinned silicon wafer 1202 is separated (de-bonded) from the glass wafer 700. In one case, dicing tape 1226 (laminating tape 1226 etc.) can be applied to the silicon wafer's second surface 1210 and then used to help pull and separate the thinned silicon wafer 1202 from the bonding agent 1214 and glass wafer 700. Alternatively, a laser can emit a laser beam through the glass wafer's non-polished second surface 706 to selectively remove or destroy the bonding agent 1214. In another example, if a thermally cured temporary adhesive bonding agent 1214 is used then this bonding agent 1214 could be heated to release the thinned silicon wafer 1202 from the glass wafer 700. Alternatively, a solvent could be used to release the thinned silicon wafer 1202 from the glass wafer 700.

(44) Referring to FIG. 13, there is a block diagram of an exemplary thinning system 1300 which uses a thinning mechanism 1306 to thin the silicon wafer 1202 while it is bonded to the non-polished glass wafer 700. The thinning system 1300 includes an enclosure 1302 within which there is located at least a support 1304 (e.g., vacuum 1304, electrostatic chuck 1304), and the thinning mechanism 1306. The glass wafer's second surface 706 is placed on and secured to the support 1304. Prior to this step, the glass wafer 700 would have already been bonded to the silicon wafer 1202 by the bonding agent 1214 (see also description above for details about the wafer quality index and other exemplary attributes of the glass wafer 700). The thinning mechanism 1306 (e.g., mechanical grinding device (shown), chemical mechanical polishing device, etching device) then operates to thin the silicon wafer's second surface 1210 while the silicon wafer 1202 is bonded to the glass wafer 700. The silicon wafer 1202 can have an original thickness of around 700 m and be thinned to a thickness of around 50 m-60 m. In any case, regardless of the type of thinning mechanism 1306 used the quality of the glass wafer 700 ultimately leads to the quality of the thinned silicon wafer 1202.

(45) From the foregoing, one skilled in the art will appreciate that the present disclosure includes the non-polished glass wafer 700, the thinning system 1300, and the method 1200 for using the glass wafer 700 to thin a semiconductor wafer 1202. As described above, glass wafers 700 have been cost effectively manufactured which have an outer diameter of about 300 mm and are about 0.7 mm thick with a TTV less than about 2.0 m and warp less than about 30 m without polishing. This is very significant for the 3D IC packaging technology not just from a cost standpoint but also from an improved attribute standpoint, an improved cleaning standpoint, an improved re-use standpoint, as well as from an improved bond/de-bond standpoint. The following is a list of some of the potential advantageous features associated with the non-polished glass wafer 700:

(46) 1. Exceptional Low Costno polishing steps are needed when a fusion process is used, to make the glass wafers 700.

(47) 2. Simultaneous low TTV and low warp can improve yields and capabilities for thinned silicon wafers 1202.

(48) 3. The glass wafers 700 can have exceptionally low surface roughness and reduced subsurface damage when compared to traditional polished glass wafers.

(49) 4. The glass wafers 700 can be made, as one example, from the aforementioned composition no. 1 that has desirable features such as: a. CTE match to silicon. b. Low alkali metalsNa concentrations are about 2 orders of magnitude less than traditional polished glass wafers made from borosilicate glass. c. Strengthtesting has demonstrated that the glass wafers 700 are as strong as silicon wafers 1202, and stronger than traditional polished glass wafers made from borosilicate glass.

(50) The following is a discussion about an exemplary glass manufacturing system that can make the glass sheets from which are cut multiple glass wafers 700 that have the non-polished first and second surfaces 704 and 706, the wafer quality index less than 6.0, and one or more of the aforementioned physical and compositional attributes which make them very useful in thinning a semiconductor wafer. The exemplary glass manufacturing system utilizes a glass forming process known as the overflow down-draw or fusion process where the two surfaces of the glass sheet are formed in air so no polishing of the two surfaces is required. One exemplary glass manufacturing system that uses the overflow down-draw or fusion process to make a glass sheet that can be cut to form multiple glass wafers 700 is described below with respect to FIGS. 14 and 15.

(51) Referring to FIG. 14, there is a schematic view of an exemplary glass manufacturing system 1400 which uses a fusion draw process to manufacture a non-polished glass sheet 1402 that can be cut and further processed to form the glass wafer 700. The exemplary glass manufacturing system 1400 includes a melting vessel 1410, a fining vessel 1415, a mixing vessel 1420 (e.g., stir chamber 1420), a delivery vessel 1425 (e.g., bowl 1425), a fusion draw machine 1418 (which includes an inlet 1428, an isopipe 1430, and a pull roll assembly 1435), and a traveling anvil machine 1440. The melting vessel 1410 is where glass batch materials are introduced, as shown by arrow 1412, and melted to form molten glass 1426. The fining vessel 1415 (i.e., finer tube 1415) has a high temperature processing area that receives the molten glass 1426 (not shown at this point) via a refractory tube 1413 from the melting vessel 1410 and in which bubbles are removed from the molten glass 1426. The fining vessel 1415 is connected to the mixing vessel 1420 (i.e., stir chamber 1420) by a finer to stir chamber connecting tube 1422. The mixing vessel 1420 is connected to the delivery vessel 1425 by a stir chamber to bowl connecting tube 1427. The delivery vessel 1425 delivers the molten glass 1426 through a downcomer 1429 to the inlet 1428 and into the isopipe 1430 (forming body 1430). The isopipe 1430 includes an inlet 1436 that receives the molten glass 1426 which flows into a trough 1437 and then overflows and runs down two sides 1438 and 1438 before fusing together at what is known as a root 1439 (see FIG. 15). The root 1439 is where the two sides 1438 and 1438 come together and where the two overflow walls of the molten glass 1426 rejoin (i.e., re-fuse) before being drawn downward between two rolls in the pull roll assembly 1435 to form the glass sheet 1402. The traveling anvil machine 1440 then cuts the glass sheet 1402 into distinct glass sheets 1402 which are later cut into the desired shape for the glass wafer 700. A more detailed discussion about an exemplary configuration of the isopipe 1430 is provided next with respect to FIG. 15.

(52) Referring to FIG. 15, there is a perspective view of the exemplary isopipe 1430 (forming body 1430) that can be used to form the glass sheet 1402. The isopipe 1430 includes a feed pipe 1502 that provides the molten glass 1426 through the inlet 1436 to the trough 1437. The trough 1437 is bounded by interior side-walls 1504 and 1504 that are shown to have a substantially perpendicular relationship, but could have any type of relationship to a bottom surface 1506. In this example, the isopipe 1430 has a bottom surface 1506 which has a sharp decreasing height contour near the end 1508 farthest from the inlet 1436. If desired, the isopipe 1430 can have a bottom surface 1506, which has located thereon an embedded object (embedded plow) near the end 1408 farthest from the inlet 1436.

(53) The exemplary isopipe 1430 has a cuneiform/wedge-shaped body 1510 with the oppositely disposed converging side-walls 1438 and 1438. The trough 1437 having the bottom surface 1506, and possibly the embedded object (not shown), is longitudinally located on the upper surface of the wedge-shaped body 1510. The bottom surface 1506 and embedded object (if used) both have mathematically described patterns that become shallow at end 1508, which is the end the farthest from the inlet 1436. As shown, the height between the bottom surface 1506 and the top surfaces 1512 and 1512 of the trough 1437 decreases as one moves away from the inlet 1436 towards the end 1508. However, it should be appreciated that the height can vary in any manner between the bottom surface 1506 and the top surfaces 1512 and 1512. It should also be appreciated that the cuneiform/wedge-shaped body 1510 may be pivotally adjusted by a device such as an adjustable roller, wedge, cam or other device (not shown) to provide a desired tilt angle shown as which is the angular variation from the horizontal of the parallel top surfaces 1512 and 1512.

(54) In operation, the molten glass 1426 enters the trough 1437 through the feed pipe 1502 and inlet 1436. Then the molten glass 1426 wells over the parallel top surfaces 1512 and 1512 of the trough 1437, divides, and flows down each side of the oppositely disposed converging sidewalls 1438 and 1438 of the wedge-shaped body 1510. At the bottom of the wedge portion, or root 1439, the divided molten glass 1426 rejoins to form the glass sheet 1402 which has very flat and smooth non-polished surfaces. The high surface quality of the glass sheet 1402 results from a free surface of molten glass 1426 that divides and flows down the oppositely disposed converging side-walls 1438 and 1438 and forming the exterior surfaces of the glass sheet 1402 without coming into contact with the outside of the isopipe 1430.

(55) For a more detailed discussion of the general parameters of the overflow down-draw process for making thin glass sheets, reference is made to one or more of the following co-assigned, co-owned patents and/or applications, the relevant contents thereof are incorporated herein by reference in their entirety:

(56) WO03/014032 entitled SHEET GLASS FORMING APPARATUS;

(57) WO05/081888 entitled SHEET WIDTH CONTROL FOR OVERFLOW DOWNDRAW SHEET GLASS FORMING APPARATUS;

(58) WO09/070262 entitled METHOD FOR DETECTING CHANGE IN SHAPE OF A MOVING SUBSTRATE;

(59) US-2010-0212360A1 entitled APPARATUS AND METHOD FOR DRAWING A RIBBON OF GLASS; and

(60) US-2011-0289969A1 entitled APPARATUS AND METHOD FOR CONTROLLING THICKNESS OF A FLOWING RIBBON OF MOLTEN GLASS.

(61) The exemplary glass manufacturing system 1400 or any other down-draw process glass manufacturing system can have one or more enhancements made thereto to better control the final thickness of the non-polished glass sheet 1402 which can be cut to form the glass wafer 700. An example of one such enhancement that was made to the exemplary glass manufacturing system 1400 (but can be made to any other down-draw process glass manufacturing system) to make non-polished glass sheets 1402 which can be cut to form the glass wafers 700 that have the desired wafer quality index requirements specified herein is described next with respect to FIG. 16.

(62) Referring to FIG. 16, there is a cross sectional side view of an enhancement that can be made with respect to the isopipe 1430 (e.g., forming body 1430) shown in FIGS. 14-15 that can help better control the final thickness of the non-polished glass sheet 1402 which can be cut to form the non-polished glass wafer 700. The isopipe 1430 is comprised of a ceramic refractory material, such as zircon or alumina and housed in enclosure 1602. The enclosure 1602 comprises heating elements 1604 arranged behind interior walls 1606 (muffle 1606). The heating elements 1604 are used to control the temperature of the molten glass 1426 on the forming surfaces 1438 and 1438 of the isopipe 1430, and hence the viscosity of the molten glass 1426, and may be arranged throughout the enclosure 1602 as needed. The heating elements 1604 are in banks arranged vertically so that the temperature within the enclosure 1602 can be controlled as a function of vertical position in the enclosure 1602.

(63) To provide control of the thermal environment within the enclosure 1602, and in particular the temperature at the root 1439 of the isopipe 1430, thermal shields 1610 are positioned below the enclosure 1602 and are movable in a horizontal direction as indicated by arrows 1612. The thermal shields 1610 are used to control radiative heat loss from the isopipe 1430, and in particular radiative heat loss from the root 1439 region of the isopipe 1430. The thermal shields 1610 are independently movable in the horizontal direction as indicated by arrows 1612, wherein the thermal shields 1610 translate to increase or decrease the gap (opening) between the thermal shields 1610. In particular, one thermal shield 1610 is movable independently from the opposite thermal shield 1610 (on the other side of the glass sheet 1402). More specifically, the thermal shields 1610 are capable of movement, toward the glass sheet 1402, and outward, away from the glass sheet 1402. The movement toward or away from the glass sheet 1402 can be provided for in several different ways.

(64) In addition, to further help control the thermal environment within the enclosure 1602, and in particular the temperature at the root 1439 of the isopipe 1430, cooling members 1614 are located below the thermal shields 1610. If desired, the cooling members 1614 could be located above the thermal shields 1610. The cooling members 1614 may be movable so that the cooling members 1614 can be positioned an appropriate distance from descending glass sheet 1402. The cooling members 1614 contain cooling equipment that cools the surfaces of the cooling members 1614 and in particular the faces 1616 of the cooling members 1614. The cooling of the cooling member's faces 1616 in turn controls the temperature and therefore the viscosity of the glass descending from the forming body 1430 along the width of the glass (e.g. horizontally). For example, the cooling members 1614 may contain one or more coolant supply lines 1618 which have outlets that extend along the length of the cooling members 1614. Each outlet emits a coolant (e.g., air) that cools a portion of each cooling member's face 1616 adjacent to the outlet. The volume of coolant emitted by each outlet may be individually controlled so that the temperature of each cooling member's face 1616 can be controlled as a function of location on the face 1616 (e.g. horizontal location). In some embodiments, a single supply line 1618 may feed a header comprising a plurality of outlets, each outlet being controlled by a remotely controlled valve.

(65) The cooling arrangement described above allows the cooling member's faces 1616 to vary the temperature and viscosity of the glass descending from the forming body 1430 as a function of location across the width of the glass sheet 1402, and can be used, for example, to control the across-the-sheet thickness of the glass sheet 1402. While the cooling members 1614 are capable of horizontal translation (represented by arrows 1620) to enable positioning the cooling members 1614 relative to the major surfaces of the glass sheet 1402, once an optimum position is set, the cooling members 1614 are seldom moved, since such movement can affect sheet attributes (e.g. shape, thickness, and the like). Rather, the functionality of the cooling members 1614 is derived largely by controlling the flow of coolant to the cooling members 1614 and therefore temperature. The optimum position depends on the particular draw setup, and may vary from draw to draw.

(66) As can be appreciated by the above description, both cooling and heating can occur simultaneously in regions quite close to each other. In this regard, the thermal shields 1610 minimize radiant heat loss from the bottom of the forming body 1430 to prevent cooling of the molten glass 1426 at the root 1439 of the forming body 1430, while the cooling members 1614 are used to actively cool the glass across a width of the descending glass sheet 1402 as an aid to thickness control. Indeed, the operation of the thermal shields 1610 and the cooling members 1614 can be coordinated to maintain a specific thermal environment. For a more detailed description about the enclosure 1602, the thermal shields 610, the cooling members 614 and the associated components reference is made to EP Published Patent Application No. EP2253598, entitled APPARATUS FOR REDUCING RADIATIVE HEAT LOSS FROM A FORMING BODY IN A GLASS FORMING PROCESS the contents of which are incorporated by reference herein.

(67) Furthermore, to help manufacture glass sheets 1402 that are cut into glass wafers 700 which not only achieve the desired wafer quality index requirements but also result in a 2 m TTV or less one can further fine-tune or monitor certain components of the set-up shown in FIG. 16 to provide more extensive control of heat transfer since even the slightest irregularity can manifest itself as a thickness variation in the glass sheet 1402. In achieving a TTV of less than or equal to about 2 m, the following heat transfer improvements could be made:

(68) A. Control of pressure in enclosure 1602 (opening between thermal shields 1610): It is possible to control and minimize any pressure variations or significant pressure differences relative to the interior and exterior of the enclosure 1602 by adjusting the opening between the thermal shields 1610, where if there are pressure variations this could lead to uneven heat transfer which in turn leads to increased warpage and thickness variation in the glass sheet 1402.

(69) B. Control of cooling members 1614: It is possible to provide material and air flow control units for precision control and velocity of the cooling air to control the cooling as a function of the location on the cooling member's faces 1616.

(70) C. Muffle Door Slide Gate: In addition to controlling pressure within the enclosure 1602 through exhaust as mentioned above, it is possible to minimize the opening between the thermal shields 1610 as the glass exits to effect better control of heat transfer uniformity with respect to the glass sheet 1402

(71) It is noted that the glass wafers 700 discussed with respect to TABLE 1 were made in a glass manufacturing system utilizing the heat transfer improvements associated with A, B and C.

(72) Although several embodiments of the glass wafer, thinning system, and methods have been illustrated in the accompanying Drawings and described in the foregoing Detailed Description, it should be understood that the glass wafer, thinning system, and methods are not limited to the disclosed embodiments, but are capable of numerous rearrangements, modifications and substitutions without departing from the spirit of the glass wafer, thinning system, and methods as set forth and defined by the following claims.