WO2001006553A1

WO2001006553A1 - Polishing mixture and process for reducing the incorporation of copper into silicon wafers

Info

Publication number: WO2001006553A1
Application number: PCT/US2000/019872
Authority: WO
Inventors: Andjelko Basic; Lois Illig; James C. Jose, Jr.
Original assignee: Memc Electronic Materials, Inc.
Priority date: 1999-07-19
Filing date: 2000-07-18
Publication date: 2001-01-25
Also published as: EP1212789A1; KR20020015715A; JP2003505858A; CN1361923A

Abstract

A process for reducing the incorporation of copper into a semiconductor single-crystal silicon wafer during polishing includes the steps of: a) adding a copper-controlling additive to a polishing mixture containing copper, the copper-controlling additive reacting with the copper in the polishing mixture to form a copper compound having a solubility product (Ksp) less than about 10-20; and b) thereafter contacting a surface of the wafer with a polishing material and the polishing mixture as the wafer moves relative to the polishing material to polish the surface of the wafer. Novel polishing mixtures for use in polishing semiconductor single-crystal silicon wafers and reducing the incorporation of copper therein are also disclosed.

Description

POLISHING MIXTURE AND PROCESS

FOR REDUCING THE INCORPORATION OF

COPPER INTO SILICON WAFERS

Background of the Invention

This invention relates generally to a polishing mixture and process for reducing the incorporation of copper into silicon wafers and, more particularly, to such a polishing mixture and process for reducing the amount of copper that enters silicon wafers during polishing without affecting polishing performance adversely.

Conventional polishing mixtures contain an inorganic base and colloidal silica. Such mixtures and other polishing chemicals present in the mixture contain trace metal impurities. Of these, copper presents a particular problem because it is incorporated into wafers during surface polishing. It is important to reduce such copper contamination because a large fraction of random semiconductor device and gate-oxide failures can be traced to copper suicide precipitates. Furthermore, high concentrations of copper contamination cause an undesired resistivity increase and promotes a degradation of the wafer surface known as chemical haze. For these reasons, integrated circuit manufacturers generally require that the concentration of copper on the surface of a silicon wafer be no more than lxlO¹⁰ atoms/cm² to lxl 0ⁿ atoms/cm², as determined by methods standard in the art. It is foreseeable that this requirement will be decreased to a value of lxl 0⁹ atoms/cm² to 5x10⁹ atoms/cm², or less.

One way of reducing wafer contamination by copper during surface polishing is by using polishing mixtures and other polishing chemicals with higher purity. However, specifications for trace metals on wafers are becoming so stringent that to meet them by simply using purer chemicals is becoming increasingly expensive and inefficient.

German patent no. DE 39393661 and an article by Prigge et al., Journal of the Electrochemical Society, vol. 138, (1991), pp. 1385-1389, describe an attempt to solve this problem without removing the trace copper impurity from the polishing mixture. According to the disclosure in this German patent, copper incorporation into silicon wafers during surface polishing may be reduced by the addition to the polishing mixture of certain chemicals which form copper-complexes with a certain coordination arrangement or geometry. A favorable geometry is any that differs sufficiently from the square planar arrangement, a tetrahedral geometry or arrangement being preferred. Ligands that form such favorable complexes with copper are identified as alcohols and hydroxycarboxylic acids, for example, ethylene glycol and tartaric acid, respectively. In contrast to this, compounds that promote copper incorporation into silicon during polishing were identified as any nitrogen compounds that act as Lewis bases, such as ammonia and amines. It is customary in polishing practice to add amines and/or ammonia to polishing mixtures in order to increase the silicon removal rate during polishing. As a result, amines and ammonia are present in the polishing mixtures regardless of the concern regarding copper contamination. Thus, while the above-noted prior art teaches that amines and ammonia aggravate copper contamination, it also teaches that copper contamination is to be controlled by using additives that form favorable copper-complexes without any indication as to whether such additives are effective enough to counteract the detrimental effect of amines or ammonia. Moreover, the above- noted prior art teaches controlling copper contamination by adding the complexing ligands in an amount 6-7 orders of magnitude greater than the copper present in the polishing mixture. Such addition is not only inefficient, but may shift the pH and other conditions of the polishing mixture away from the optimum required by the polishing process.

There remains a need, therefore, for an improved polishing mixture and process for reducing copper incorporation into wafers during polishing and which are effective in the presence of ammonia or amines and do not adversely affect the polishing rate or process.

Summary of the Invention

Among the several objects of the invention may be noted the provision of a polishing process and polishing mixture for reducing the incorporation of copper into silicon wafers during polishing; the provision of such a process and polishing mixture which are effective in the presence of ammonia or amines; and the provision of such a process and polishing mixture which employ additives in such concentrations that the polishing rate and performance are not materially affected. Other objects and features will be in part apparent and in part pointed out hereinafter.

Briefly, the present invention is directed to a process for reducing the incorporation of copper into a semiconductor single-crystal silicon wafer during polishing which comprises the steps of: a) adding a copper-controlling additive to a polishing mixture containing copper, the copper-controlling additive reacting with the copper in the polishing mixture to form a copper compound having a solubility product (K^) less than about 10^"20; and b) thereafter contacting a surface of the wafer with a polishing material and the polishing mixture as the wafer moves relative to the polishing material to polish the surface of the wafer.

Additionally, the present invention is directed to a polishing mixture for use in polishing a single-crystal silicon wafer and reducing the incorporation of copper in the wafer during polishing, the polishing mixture comprising an inorganic base, colloidal silica, and a copper compound having a solubility product (K^) less than about 10^"20 formed by reacting a copper-controlling additive added to the polishing mixture with copper present in the polishing mixture.

Brief Description of the Drawings Fig. 1 is a graph showing the effect of adding potassium monohydrogen phosphate to a polishing mixture in reducing the incorporation of copper into semiconductor single-crystal silicon wafers; and

Fig. 2 is a graph showing the effect of adding hydrogen sulfide to a polishing mixture in reducing the incorporation of copper into semiconductor single-crystal silicon wafers.

Description of Preferred Embodiments

In accordance with the present invention, it has now been found that the incorporation of copper into a semiconductor single-crystal silicon wafer during polishing may be reduced by adding to a conventional polishing mixture a copper-controlling additive which reacts with copper impurity in the mixture to form a copper compound having a solubility product (K_._p) less than about 10^"20, and thereafter polishing the wafer using the resulting polishing mixture. As is known, polishing mixtures conventionally employed in polishing semiconductor single-crystal silicon wafers typically contain colloidal silica, commercially available as silica-slurry, and an inorganic base (e.g., alkali metal hydroxide) as the principal ingredients. In such polishing mixtures, the copper impurity is typically present in concentrations on the order of 1 to 10 ppb by mass, and rarely up to 100 ppb by mass, depending upon the purity grade of the silica and other ingredients of the polishing mixture.

In the practice of the invention, the copper-controlling additive is any compound which reacts with the copper impurity in the polishing mixture to form a copper compound having a solubility product (K^) less than about 10^"20. The copper-containing additive preferably comprises an anion selected from the group consisting of phosphate, sulfide, selenide and arsenate and is chemically stable in alkaline aqueous media having a pH in the range usual for the polishing mixture. The additive may be dissociated in solution prior to being added to the polishing mixture. The amount of copper-controlling additive added to the polishing mixture is preferably at least stoichiometrically equivalent to the copper content of the polishing mixture, more preferably, at least about 100 times the stoichiometric equivalent of the copper content of the polishing mixture, still more preferably about 100 to about 10,000 times the stoichiometric equivalent of the copper content of the polishing mixture. Examples of compounds which may be employed in the practice of the invention include potassium monohydrogen phosphate, potassium phosphate, hydrogen sulfide, ammonium sulfide, and potassium selenide. It will be understood that other compounds which react with the copper impurity to form a sufficiently insoluble copper compound such as copper (II) phosphate, copper (II) sulfide, copper (II) selenide or copper (JJ) arsenate may also be utilized. It has been observed that adding the copper-controlling additive to the polishing mixture at least about 24 hours before beginning the polishing operation maximizes the reduction of copper incorporated into the wafer during polishing. Without being held to a particular theory, it is believed that a period of at least about 24 hours allows the polishing mixture to reach chemical equilibrium. While not being bound to any particular mechanism for the present invention, it is believed that the copper-controlling additive added to the polishing mixture precipitates copper as a very nearly insoluble compound, thereby rendering it less available for incorporation into silicon wafers during polishing. Accordingly, any compound which reacts with the copper impurity to form a precipitate with a very low solubility, i.e. a precipitate having a solubility product (K^) of less than about 10^"20, is effective for use in controlling copper contamination. The threshold maximum solubility product (K_sp) of less than about 10^" ²⁰ was determined quantitatively using a simplified single phase solution model — a liquid solution comprising an inorganic base and copper impurity. The hydroxide ions are present in a high concentration and, therefore, copper(II)-hydroxide, Cu(OH)₂, is formed before the addition of any copper-controlling additive. Because copper(II)-hydroxide is a stable compound with a K^ of about 10^"20, and because undesirable copper contamination of wafers takes place when polishing with the standard polishing mixture, this value of K_sp is not low enough. Thus, an effective copper-controlling additive should form a copper compound with a K_._p value less than about 10^"20. Preferably, the copper-controlling additive forms a copper compound with a K._p value less than about 10^"23, more preferably less than about 10^"26, still more preferably less than about 10^"29, yet more preferably less than about 10^"32, and even more preferably less than about 10^'35.

The amount of copper-controlling additive added to the polishing mixture should be sufficient to potentially precipitate all the copper impurity, i.e. the amount of the copper- controlling additive should at least be equal to the stoichiometric equivalent of the copper content of the polishing mixture. For example, if the molar concentration of copper is denoted by [Cu], when hydrogen sulfide is used as the additive, its molar concentration, denoted by [H₂S], should be at least equal to [Cu] because the precipitate formed is CuS. Similarly, when potassium monohydrogen phosphate is used as the additive, its molar concentration, denoted by [K₂HPO₄], should be at least equal to (%)[Cu], because the precipitate is Cu₃(PO₄)₂, assuming that the monohydrogen phosphate ion is dissociated completely. In practice, however, it has been observed that the precipitation of copper is not maximized by a stoichiometric amount of additive, therefore, the amount of copper- controlling additive added to the polishing mixture is preferably in excess of the stoichiometric amount. Since typical values of [Cu] are extremely small (usually on the order of 10^"7M), the preferable amount of additive, 100 to 10,000 times the copper content of the polishing mixture, is still low enough not to shift the pH of the mixture and adversely affect polishing performance.

The process of the present invention is useful in any polishing operation in which silicon is removed from the surface of a wafer by contacting the wafer with a polishing material and the polishing mixture as the wafer moves relative to the polishing material to polish the surface of the wafer and encompasses all types of silicon polishers such as, for example, but not limited to single-side polishers, double-side polishers and edge-polishers.

The present invention thus overcomes the shortcomings of the prior art and provides an effective means for reducing copper incorporation into silicon wafers during polishing. The improved process and polishing mixture of the present invention are effective in the presence of ammonia or amines and permit the use of such low concentrations of the additive compound that the chemical composition of the final polishing mixture is not substantially changed and the polishing rate is unaffected.

The following examples illustrate the practice of the invention.

Example 1

In the following tests, wafers of a Czochralski-grown boron-doped p-type silicon, 200 mm diameter, (100) crystallographic orientation and resistivity between 0.071 and 0.084 Ω^»cm were used. In one test, potassium monohydrogen phosphate was added to a conventional polishing mixture in concentration of 10^"4 mole/L as fed to the polishing pad. In the second test, hydrogen sulfide was added to a conventional polishing mixture in concentration of 1.5 X 10^"3 mole/L as fed to the polishing pad. The standard ingredients of the polishing mixture were colloidal silica stabilized by an alkali hydroxide, and an amine. The two resulting polishing mixtures were used to polish the wafers for 245 seconds with an applied pressure of 7 psig and a pH of 12.6. The results are shown in Figs. 1 and 2. The dependent variable was surface concentration of copper measured on wafers after heat- treating at 120° C for 90 minutes. The heat-treatment step allows the copper incorporated into the wafer bulk to diffuse back onto the wafer surface (see, Prigge et al., Journal of the Electrochemical Society, vol. 138, (1991), pp. 1385-1389). The measurement method consisted of acid drop extraction of the surface and analysis of the extract for trace copper by ICP-MS (inductively coupled plasma/mass spectroscopy). The result was converted to surface copper concentration denoted by {Cu} and expressed in number of atoms per square centimeter of the wafer surface. In both tests, a control group of wafers was polished using the same polishing mixture but without the copper-controlling additive. As can be seen from Figs. 1 and 2, the present invention is effective in reducing the incorporation of copper into semiconductor single-crystal silicon wafers during polishing.

Example 2 In the following tests, wafers of a Czochralski-grown boron-doped p-type silicon, 200 mm diameter, (100) crystallographic orientation and resistivity between 0.071 and 0.084 Ω*cm were polished to measure the effectiveness of various additives at reducing the copper incorporation into the wafers. Other conditions were as described in Example 1.

Specifically, each test was dedicated to determining the effectiveness of a single polishing additive; the first six additives listed in Table 1 , infra, are known in the art as copper- complexing additives, the remaining additives are believed to control copper by forming a copper compound/precipitate. Each test consisted of: a) polishing 7 or 8 wafers using a polishing mixture containing the additive being tested; b) polishing 7 or 8 wafers in the same polishing mixture without the additive; c) heat treating all the foregoing polished wafers, as well as 5 to 8 unpolished wafers; d) extracting copper from the surface of the heat treated wafers by the acid drop method and analyzing the extract for trace copper by the ICP-MS method (inductively coupled plasma/mass spectroscopy) and converting the result into surface copper concentration expressed in number of atoms per square centimeter of the wafer surface. The heat-treatment step allows the copper incorporated into the wafer bulk to diffuse back onto the wafer surface (see, Prigge et al., Journal of the Electrochemical Society, vol. 138, (1991), pp. 1385-1389).

The numerical result of the tests depend on the characteristics of the wafers used and on the regime of heat- treatment, however, a quantity of identical wafers sufficient for all the tests could not be obtained. Despite the variability manifested in the test wafers and the heat- treatment procedures between the different tests, a common basis for comparison of numerical results was obtained by: a) using randomized wafers that had been cut from the same crystal section, and b) introducing a special measure of surface copper concentration change due to the additive. This quantity is denoted by δ {Cu} and is defined by the following formula:

{CuJ^and {Cu}_st denote average surface copper concentrations on wafers polished using a polishing mixture with the additive and without the additive, respectively. {Cu}_uπp denotes the average surface concentration on all unpolished wafers used in the same test. Thus, Equation (1) defines δ{Cu} as the difference in surface copper concentration to be expected between wafers polished with and without the additive, and expressed in percent of the change due to polishing that would be observed when polishing without the additive. The isolation of δ{Cu} from the variability manifested in the test wafers and the heat-treatment procedures between the different tests is based on the following assumptions: a) the rate of copper incorporation during polishing is independent of the average surface concentration of copper before polishing, {Cu}_uπp, and b) the copper incorporated in the wafer during the polishing process completely diffuses out of the wafer during the heat-treatment. A negative value of δ{Cu} indicates that a decrease of copper contamination during polishing was achieved by using the additive, and a positive value indicates an increase. Furthermore, if there were no statistical uncertainty associated with the numerical results, a negative δ{Cu} value of one hundred would represent complete suppression of copper contamination during polishing, and a value of zero would indicate no effect. Lastly, additives which produce copper compounds that have the same δ{Cu} value when tested under identical conditions should be considered equally effective at reducing the incorporation of copper into a wafer during polishing.

Table 1

The first column in Table 1 contains the additive which was tested. The overall molar concentration of the additive was 3x10^"4 mole/L in all cases except for K₂HPO₄, whose molar concentration was 10^"4 mole/L, and for H₂S whose molar concentration was 1.5xl0^"3 mole/L. The latter concentration was higher to compensate for the low dissociation constant of hydrogen sulfide and ensure a high enough concentration of dissociated sulfide ions in the polishing mixture. The overall molar concentration of copper in the polishing mixtures was between 6xl0^"8 and 10^"7 mole/L (3.8 and 6.4 ppb by mass, respectively). Thus, the stoichiometric excess of the additive with respect to overall copper was between the orders of 10³ and 10⁴ in all cases. The second column in Table 1 contains the compound which is believed to be precipitated by the reaction between the copper impurity and the additive, as well as that compound's solubility product, K^. The third column provides the difference in surface copper concentration between wafers that have been polished without and those polished with the additive, expressed by the quantity δ{Cu}, defined in Eq. (1). The fourth column indicates whether the difference in surface copper concentration between the wafers polished with the additive and those polished without was significant at a 95% confidence level as determined using the analysis of variance (ANOVA). Finally, the fifth column contains the/?-value obtained in the ANOVA.

In most tests no statistically significant difference was observed between polishing removal rates on wafers polished with the additive and those polished without the additive.

The only statistically significant differences observed were: 2.5% higher rate with ammonium citrate than without; 1.4 % higher rate with gluconic acid than without; and 4.6% higher rate with hydrogen sulfide than without. These removal rate differences are defined as percent of the removal rate without the additive.

Each of the tested copper-controlling additives that react with copper to form a copper compound/precipitate produced a statistically significant change in the final concentration of copper for the wafers, whether positive or negative. In contrast, only one of the copper- complexing additives used in the art to reduce copper contamination produced a statistically significant difference in final copper contamination — 1 ,2-benzenediol (o-catechol), and its effect was to increase copper contamination. Table 1 shows that hydrogen sulfide, ammonium sulfide, potassium monohydrogen phosphate and ammonium monohydrogen phosphate when added to a polishing mixture produce copper compounds which have a solubility product K_._p below 10^"20 and reduce the copper contamination of polished wafers by 21.6 %, 13.9 %, 18.6 % and 12.3 %, respectively. Table 1 also shows, however, that not all tested compounds which satisfy the defined solubility criterion are effective in reducing copper contamination of wafers during polishing. Specifically, two of the additives that were expected to reduce copper contamination, namely potassium carbonate and sodium monohydrogen phosphate, actually caused an increase in wafer contamination by copper.

It is believed that additives containing the carbonate anion, react with the hydroxide ion and copper impurity present in the polishing mixture to form a simple copper carbonate, CuCO₃, having a K_._p value of 10^"96 and basic copper carbonates, Cu₂(OH)₂CO₃ and Cu₃(OH)₂(CO3)₂, having K^ values of 10^{"33 8} and 10^"46, respectively (the forgoing K^ values from Smith, R. M. and Martell, A. E., Critical Stability Constants, Plenum, vol. 4,(1976)). The basic copper carbonates have a K^ value below 10^"20 and were expected to be effective in reducing copper contamination. Without being held to a particular theory, the observed negative results are considered to be due to the inability of the assumed single-phase model (i.e., liquid only) to adequately represent the two-phase colloidal suspension of an actual polishing mixture. Specifically, basic copper carbonates may preferentially precipitate on the surface of silica particles where the concentration of OH^" ions is significantly higher than the liquid phase. Furthermore, as silica particles participate in the chemomechanical reaction with silicon during the polishing process (see, Fussstetter, H. et al, "Impact of Chemomechanical Polishing on the Chemical Composition and Morphology of the Silicon Surface," Materials Research Society Symposium Proceedings, vol. 386, (1995), p. 97), it is possible that the basic copper precipitates are activated and react with the silicon to release copper. If the foregoing hypothesis is correct, any form of carbonate will be ineffective at reducing copper contamination from the polishing mixture. The undesirable effect of sodium phosphate may be due to the way in which the sodium and the lithium cations interact with colloidal silica particles at pH levels above 11. Namely, unlike all other cations, which have such effect only at pH values below 11, sodium and lithium cause flocculation of silica particles at pH values above 11 (Iler, R., Chemistry of Silica, Wiley, (1979), p. 375). Possibly, this behavior together with an interaction of the copper impurity and silica particles was responsible for the increase of copper contamination during polishing when sodium phosphate was used as the additive. If so, the sodium and the lithium cation will have the same effect when used as additives in combination with any other anion.

In view of the above, it will be seen that the several objects of the invention are achieved and other advantageous results attained.

As various changes could be made in the above process or method and product without departing from the scope of the invention, it is intended that all matter contained in the above description and shown in the accompanying drawings shall be interpreted as illustrative and not in a limiting sense.

Claims

CLAIMSWHAT IS CLAIMED IS:

1. A process for reducing the incorporation of copper into a single-crystal silicon wafer during polishing which comprises the steps of: a) adding a copper-controlling additive to a polishing mixture containing copper, the copper-controlling additive reacting with the copper in the polishing mixture to form a copper compound having a solubility product (K_sp) less than about 10^"20; and b) thereafter contacting a surface of the wafer with a polishing material and the polishing mixture as the wafer moves relative to the polishing material to polish the surface of the wafer.

2. A process as set forth in claim 1 wherein the copper-controlling additive comprises an anion selected from the group consisting of phosphate, sulfide, selenide and arsenate, the anion reacting with the copper to form the copper compound.

3. A process as set forth in claim 1 wherein the copper-controlling additive is selected from the group consisting of potassium monohydrogen phosphate, potassium phosphate, hydrogen sulfide, ammonium sulfide, potassium selenide, ammonium monohydrogen phosphate.

4. A process as set forth in claim 3 wherein the copper-controlling additive is potassium monohydrogen phosphate.

5. A process as set forth in claim 3 wherein the copper-controlling additive is hydrogen sulfide.

6. A process as set forth in claim 1 wherein the formed copper compound is selected from the group consisting of copper (II) phosphate, copper (II) sulfide, copper (II) selenide and copper (II) arsenate.

7. A process as set forth in claim 1 wherein the formed copper compound has a solubility product (K_sp) less than about 10^"23.

8. A process as set forth in claim 7 wherein the formed copper compound has a solubility product (K ) less than about 10^"26.

9. A process as set forth in claim 1 wherein the amount of copper-controlling additive added to the polishing mixture is at least stoichiometrically equivalent to the copper content of the polishing mixture.

10. A process as set forth in claim 9 wherein the amount of copper-controlling additive added to the polishing mixture is at least about 100 times the stoichiometric equivalent of the copper content of the polishing mixture.

11. A process as set forth in claim 10 wherein the amount of copper-controlling additive added to the polishing mixture is about 100 to about 10,000 times the stoichiometric equivalent of the copper content of the polishing mixture.

12. A process as set forth in claim 1 wherein the surface of the wafer is not polished until at least about 24 hours after the copper-controlling additive is added to the polishing mixture.

13. A polishing mixture for use in polishing a single-crystal silicon wafer and reducing the incorporation of copper in the wafer during polishing, the polishing mixture comprising an inorganic base, colloidal silica, and a copper compound having a solubility product (K_sp) less than about 10^"20 formed by reacting a copper-controlling additive added to the polishing mixture with copper present in the polishing mixture.

14. A polishing mixture as set forth in claim 13 wherein the copper-controlling additive comprises an anion selected from the group consisting of phosphate, sulfide, selenide and arsenate, the anion reacting with the copper to form the copper compound.

15. A polishing mixture as set forth in claim 13 wherein the copper-controlling additive is selected from the group consisting of potassium monohydrogen phosphate, potassium phosphate, hydrogen sulfide, ammonium sulfide, potassium selenide, ammonium monohydrogen phosphate.

16. A polishing mixture as set forth in claim 15 wherein the copper-controlling additive is potassium monohydrogen phosphate.

17. A polishing mixture as set forth in claim 15 wherein the copper-controlling additive is hydrogen sulfide.

18. A polishing mixture as set forth in claim 13 wherein the copper compound is selected from the group consisting of copper (II) phosphate, copper (II) sulfide, copper (J) selenide, copper (TJ) arsenate.

19. A polishing mixture as set forth in claim 13 wherein the copper compound has a solubility product (K_sp) less than about 10^"23.

20. A polishing mixture as set forth in claim 19 wherein the copper compound has a solubility product (K_sp) less than about 10^"26.

21. A polishing mixture as set forth in claim 13 wherein the amount of the copper- controlling additive added to the polishing mixture is at least stoichiometrically equivalent to the copper content of the polishing mixture.

22. A polishing mixture as set forth in claim 21 wherein the amount of the copper- controlling additive added to the polishing mixture is at least about 100 times the stoichiometric equivalent of the copper content of the polishing mixture.

23. A polishing mixture as set forth in claim 22 wherein the amount of copper- controlling additive added to the polishing mixture is about 100 to about 10,000 times the stoichiometric equivalent of the copper content of the polishing mixture.