DE112014001428T5 - Electrochemical deposition processes for semiconductor wafers - Google Patents

Abstract

A method of electroplating a wafer detects a failure of a plating bath based on a voltage change. The method is useful in plating wafers with TSV structures. The voltage of each anode of a plating tool can be monitored. A sudden drop in voltage signals a bad failure resulting from the conversion of an accelerator, such as SPS, into its by-product MPS. Bad failure can be delayed or prevented by power cycling or power up. An improved plating bath has a catholyte with a very low acid concentration.

Description

Technical area

 The invention relates to devices, systems and methods for electroplating substrates, such as semiconductor material wafers. In particular, the invention provides improved techniques that are particularly useful on wafers with silicon via (TSV) or similar structures.

Background of the invention

Microelectronic devices such as semiconductor devices are basically fabricated on and / or in substrates or wafers. In a typical manufacturing process, one or more layers of a metal or other conductive material are formed on a wafer in an electroplating apparatus. The device may have an electrolyte bath in a vessel or basin with one or more anodes in the basin. The wafer itself can be held in a rotor in a head which can be moved into the basin for processing and which can be moved away from the pool for loading and unloading. A contact ring on the rotor basically has a large number of contact fingers that make electrical contact with the wafer.

 Many advanced microelectronic devices have silicon via (TSV) contacts. A TSV is a vertical electrical connection that usually runs completely through the wafer or chip, which may or may not be silicon. TSVs are used to create three-dimensional electronic structures and assemblies. The use of TSVs enables very high density integrated circuits. The electrical properties of the connections are also improved since TSVs are generally shorter than alternative connections. This results in faster device operation and reduced effects of unwanted inductive or capacitive properties of the connections.

 TSVs tend to have higher aspect ratios because they are essentially tall, narrow microscale columns of metal, generally copper, formed in a hole in silicon or other substrate material. TSVs can be formed by electroplating copper from the bottom upwards. Achieving proper filling of the TSVs is technically challenging for several reasons, including the microscale dimensions of the TSVs, high aspect ratios, and other factors.

 Historically, the processes and chemistry used to plate the TSVs exhibit unusual instability as the plating bath ages, directly affecting the microelectronics manufacturing process. Since the plating bath is still within specification at its failure, the reason for the bad failure is not well understood. Improved techniques and an understanding of the plating of TSV structures are required.

Brief description of the drawings

In the drawings, in each of the views, the same reference numerals indicate the same elements.

1 Figure 12 is a graph of data from chronopotentiometric measurements (voltage vs. time) for a fresh plating bath and a plating bath that failed with corresponding X-ray images of wafers plated using the baths.

2 is a graph of data from chronopotentiometric measurements for a bath with a chemical composition different from the graphs of the 1 is.

3 is a graph of data similar to the 2 , where MPS was injected into the bath and a power was booted.

4A is a graph of voltage for a fresh electrolyte control bath.

4B is a graph of voltage for a bath that failed after about 30 minutes.

4C is a graph of voltage for a bath that has recovered after 70 hours of waiting.

5A is a representation of a tension for a fresh bath vs. a bath after ten passes.

5B are graphs of a voltage of a chronopotentiometric aging track on a laboratory scale.

6A to 6F are x-ray images of TSVs on wafers that have been processed as described.

6G is a graph of a tension of chronopotentiometric signs of aging.

7 is a graph of chronopotentiometry vs. a wafer rotation speed.

8A and 8B show X-ray images of wafers processed as described.

9 is a graph of chronopotentiometry vs. a bath age.

10 is a comparative table of a prior art electrolyte and a new electrolyte.

11 Fig. 15 is a perspective view of a Raider M apparatus used in performing tests represented in the data shown in the above figures.

12 is a sectional view of the in 11 shown device.

Detailed description of embodiments of the invention

I. detection of a bad failure

A. Detection of a laboratory scale bad failure

Detecting bath failure was a challenge with TSV plating baths. Bad failure can be defined by underfill deposition, seam cavities, and pinch cavities in the structures. There is a general trend that fresh baths perform well, but the bath fails with continuous reductive plating (up to 0.45 A Hr / L).

 The conventional way to detect bad failure is to plate a wafer in the tool and perform x-ray imaging / cross-sectional imaging using a focused ion beam (FIB) to detect voids. However, availability of wafers for imaging is usually limited. This is an expensive and time-consuming process. So far, there is no real and practical method available to detect a bad failure.

 As described below, a chronopotentiometric method of detecting a bad malfunction has now been invented. The inventors have now determined that the reason for this is that the bath is dominated by the plating time by the accelerator and loses some inhibition. This results in uniform growth and voids in the via contacts or trenches.

 The method can be used in a laboratory electrochemical arrangement or in a tool or system level arrangement.

In one form of laboratory procedure, a chronopotentiometric measurement with a long time scale (3600 seconds) is used to detect a bad failure. Referring to 1 a one-hour time scale allows full consideration of the attachment kinetics for organic additives during the plating step. Generally, plating takes between 10-180 minutes for TSVs (for example, for 3x50 to SOx 150 structures). As in 1 3, a fresh bath is highly inhibitory by -240 mV immediately after immersing the copper-plated Pt electrode with a terminal potential at 3600 seconds.

 Conventionally, organic additives are added to the plating bath to improve the results in TSV plating. A suppressor additive (usually a high molecular weight polyalkene glycol such as PEG) strongly attaches to the surface of the copper cathodes in the presence of chloride ions to form a film that sharply increases the overpotential for copper deposition. An accelerator additive counteracts the inhibiting effect of the suppressor to provide the accelerated deposition in trenches and via contacts required for bottom fill. SPS (sodium sulfopropyl disulfide) was used as accelerator. MPS (3-mercaptopropyl sulfonic acid) is a known by-product or degradation product of the SPS. A leveler such as amine and heterocyclic are also used in TSV plating. The leveling agent is also a strong suppressor.

The chronopotentiometric measurements of bath samples in 1 show rapid attachment of the suppressor and the leveling agent to the electrode surface, followed by replacement of the suppressor and leveling agent by the accelerator. With progressive reductive plating, up to 0.347 A Hr / L, the bath becomes less inhibitory (about 25 mV less) than the fresh bath. This is because the SPS (accelerator) is reduced to MPS or copper (I) thiolate, which simplifies copper deposition. As the bath ages to 0.45 A Hr / L, the test data show an increasing rate of replacement of the suppressor and leveling agent by the accelerator and also a competitive build up or oscillatory behavior. This oscillation behavior correlates directly with the failure of the bath.

1 shows that the fresh bath is highly suppressor-dominated at 0 A Hr / L and less suppressor-dominated over time with the reductive plating. The lower x-ray image in 1 is from a wafer that has been plated in a device at 0.34 A Hr / L and shows no voids. The upper radiograph in 1 is from a wafer clad at 0.45 A Hr / L, showing the voids at the slightly gray areas towards the top of the via contacts. At about -110 mV, either formed copper (I) thiolate is incorporated into the copper film or it dissolves from the copper surface. The bath is then again suppressor-dominated. The potential oscillations and the failure mode in a laboratory test were confirmed in a tool test.

Key chemical reactions that describe the oxidative thiol-disulfide relationship at the root of instability are: 2Cu (II) + 2MPS ^- → PLC ^2- + 2Cu (I) + 2H ⁺ [1] 4Cu (I) + SPS ^2- → 2Cu (I) (MPS ^2- ) + 2Cu (II) [2] 4Cu (I) (MPS ^2- ) _n + O ₂ + (4 + 4n) H ⁺ → 4Cu (II) + 4nMPS ^- + 2H ₂ O [3]

In a laboratory procedure, a bath sample of 200 ml was taken from the bath of the device with a total electrolyte volume of about 80 liters. A three-electrode potentiostat was used to pass a constant current through the sample while the potential was monitored over time. Referring to the upper curve in FIG 1 For example, the potential increases gradually from about -250 mV to about -180 mV until about 2000 seconds when the potential goes up to about -110 mV, and then drops rapidly down to about -250 mV at about 2400 seconds. A TSV test wafer plated with this bath showed voids after 2400 seconds.

B. detection of a bad failure at the tool or system level

In existing plating equipment designed for TSV applications, the plating process tends to be unstable, with underfilled and / or voids occurring in TSVs, even after a relatively small number of wafers have passed through a fresh bath. The inventors have determined that the instability is associated with the accelerator SPS and its by-product MPS, resulting in field depolarization or loss of inhibition, with electrical current shifting from the via contacts or trenches to the field or top surface of the wafer becomes. The inhibition refers to a combined inhibitory effect of the suppressor and the leveling agent.

 In a tool-level or system-level setup, a test wafer with a copper seed layer can be loaded into the device. The potential of each anode in the device can be monitored to detect changes in bath chemistry, and the onset of voids or underfill can be detected. Oscillation or drop in cell tension will occur if surface inhibition is lost or reduced. If this occurs while a TSV structure is still filling, it will create a void or underfill. A cavity is the primary form of failure. Overfilling and underfilling can occur as minor forms of failure, especially if the failure occurs near the end of the process when the structure is already largely completed. In this case, slight underfilling may occur.

 Smaller structures fill up faster than larger structures. The number of wafers that can be plated before a predicted bad failure occurs may be affected by the plating time for each wafer, which is determined at least in part by the feature size. A cumulative plating time is identified as a key factor in the prediction of bad failure, as opposed to the number of plated wafers.

4A to 4C show voltage waveforms for a four anode device (an Applied Materials Raider S Plating tool). The voltage measurement for each of the anodes is labeled A1, A2, A3 and A4, where A1 is the inner anode and A4 is the outer anode. The TSV is 10 μm × 100 μm. 4A shows data for a fresh bath with an expected stable behavior. The radiograph in 4A shows no voids in the TSV structure. 4B shows a bad failure at 30 minutes, the failure being indicated by the sudden change in potential. The radiograph of the 4B shows cavities at the slightly greyish area at the bottom end of the TSV structure. 4C shows voltage runs after 70 hours of waiting time. Compared with the 4a to 4c It can be seen that the wait recovery can return the bath to its original fresh state, but only after a long recovery period.

5A Figure 14 shows data from a tool or device level test (an Applied Materials Raptor M instrument) using a 300 mm wafer with top copper seed layer running at 2 mA / cm ² for 60 minutes. The bottom line is passage # 1 using a fresh bath. The top line is passage no. 10. The sudden drop of the top line (at about 95 mV) at 30 minutes indicates the bath failure.

5B shows similar data from a corresponding test at the laboratory or cup level.

• A.] 15–20 mV Depolarisierung zwischen 0 und 0,5 Ahr/L Badalter.
• B.] Solides 10 × 100 Füllverhalten bis zu 2,6 A Hr/L.
• C.] Stabile (+/–5 mV) Hemmung zwischen 0,5 und 2,5 Ahr/L.
• D.] Leichte Unterfüllung in der Mitte des Wafers zusammen mit einem zusätzlichen Verlust der Hemmung und einer Spannungsoszillation bei 3,2 Ahr/L.
• E.] Effektives B&F der Probenentnahme ist < 3%.
• F.] Durch das Betreiben bei einer kleineren DO-Konzentration (3–5 ppm vs. Sättigung) kann die Badlebensdauer um > 300% verlängert werden.

6A to 6G show test data with different concentrations of dissolved Oxygen (DO) in the bath. Existing devices generally work with 7-8ppm dissolved oxygen baths, which is the saturation level. Reducing the dissolved oxygen in the bath to 3-5 ppm can increase the effective bath life. The following was observed:

• A.] 15-20 mV depolarization between 0 and 0.5 Ahr / L bath age.
• B.] Solid 10 × 100 filling behavior up to 2.6 A Hr / L.
• C.] Stable (+/- 5 mV) inhibition between 0.5 and 2.5 Ahr / L.
• D.] Slight underfilling in the center of the wafer along with an additional loss of inhibition and stress oscillation at 3.2 Ahr / L.
• E.] Effective B & F sampling is <3%.
• F.] By operating at a lower DO concentration (3-5 ppm vs. saturation), the bath life can be extended by> 300%.

7 is a chronopotentiometric graph, where the left-hand plot uses a 1500 rpm wafer rotation speed and the right-hand display uses 500 rpm, again with the Raptor M device. All other parameters were the same. The higher rpm provide higher mass transfer, and are also shown with a previous bath failure. Reducing mass transfer can extend bath life. The test was performed on a 200 ml sample at 2 mA / cm ² and 3.2 A Hr./L.

The results discussed above generally apply to all types of devices. Some devices use a membrane that separates the anodes from the wafer, the electrolyte across the membrane being called the catholyte, and the electrolyte below the membrane being called the anolyte. 8A and 8B show results of laboratory tests of a low acid and accelerator arms catholyte compared to a moderately acid and moderately accelerated catholyte. 8A Figure 12 shows X-ray images of wafers plated using low acid (10 g / L sulfuric acid) and low accelerator (5 mL / L) at 0, 5, 10 and 15 A Hr / L with no voids present. 8B Figure 10 shows X-ray images of wafers plated using a moderate acid (50 g / L sulfuric acid) and a moderate accelerator (10 mL / L) at 0 and 1 A Hr / L, with voids present at 1 A Hr / L.

9 Figure 1 shows chronopotentiometric curves of laboratory data from baths containing those in the 8A showed results. No potential oscillations were observed for bath samples aged to 24 A Hr / L with low acid and low accelerator concentrations. By reducing the sulfuric acid concentration, H ⁺ ion availability is reduced. This can affect the SPS decay rate. Both the H ⁺ availability and the SPS effectively reduce the equilibrium concentration of the MPS through the chemical reaction: 2Cu ⁺ + PLC + 2H ⁺ ↔ 2CU ²⁺ + 2MPS

10 compares a new electrolyte for plating copper TSV wafers in a membrane device compared to the existing design. The catholyte VMS of 63.5 / 10/80 is 63.5 g / l copper, 10 gm / l sulfuric acid and 80 ppm chloride concentration. The bath life is extended from less than 2.5 A Hr / L to over 20 A Hr / L. In the 10 listed new parameters were determined experimentally. Initial theories of the improvements were created. These theories were then tested by verifying variables. This identified the key variables of bath stability as sulfuric acid concentration, accelerator concentration and plating process design. Based on this, optimized specifications were determined. Production simulations were then carried out, which the in 10 demonstrated results. The reduction of acidity is a big factor in the longer bath life.

II. Recovery from Badversagen

Bath instability correlates with the formation of MPS, a strong accelerator, during reductive plating. This leads to a poor filling from the ground, a bad inhibition in the field and in the trenches. It is difficult or impossible to maintain a constant concentration of MPS during the plating process. However, MPS can be moderated in some ways.

 MPS can be minimized with unloading and feeding (30%), whereby the bath is constantly renewed. This continuously removes MPS from the bath, so the MPS concentration generally remains stable. However, relieving and feeding increases costs and complications in the plating process.

 MPS can also be controlled by a wait time recovery. By allowing the bath to rest, MPS is oxidized and transformed back to SPS. However, this can take hours or days. It is highly time consuming and naturally delays processing.

 A rinse of the bath also removes MPS. This can be done by bubbling clean dry air up through the bath. Deplating or performing the reversed polarity plating process also removes MPS. These techniques are generally inefficient and also time consuming.

A. Power clocking

An improved technique for delaying or avoiding bath failure resulting from the MPS is current cycling. In standard plating processes the current is continuous. This results in the continuous formation of MPS or copper (I) thiolate, a complex group, by combining copper (I) ions with MPS thiolate groups. As a result, the bath becomes highly accelerated over time, resulting in underfilling due to the decreasing suppression on the fields.

 By cycling the current during the plating process, formation of MPS can be controlled using short pulses or long pulses. The timing may be negative, i. Current may be clocked from a positive or plating current to a negative current, or the timing may be positive, i. a timing of the plating current or towards an open circuit potential. A changeover timing may also be used via a constant current, constant voltage timing. The timing can be performed at even intervals in a POR process. This can help maintain bath stability by loosening MPS from the copper surface and increasing bath inhibition.

 The timing can also be performed if there is no wafer.

B. power up

Ramping can be used to reduce the effect of the MPS and restore bath stability. 2 Figure 3 shows chronopotentiometric measurements of a fresh JCU bath (63/50 / 80-10 / 5/15) and a fresh JCU bath (63/50 / 80-10 / 5/15) injected with 0.02 ppm MPS at a constant current density of -2 mA / sq.cm and a rotation at 500 rpm. As shown in the graphs, the bath provided with 0.02 ppm MPS shows potential oscillation associated with bad failure becomes. This behavior has also been observed in tool scale experiments in which underfill and / or cavities have been associated with potential oscillations.

3 Figure 3 shows a chronopotentiometric measurement of a fresh JCU (63/50 / 80-10 / 5/15) bath injected with 0.02 ppm MPS at an increasing current density of 2-3.2 mA / sq.cm over 3600 With a rotation at 500 rpm. The potential oscillation observed under a constant current density was attenuated and part of the bath inhibition was restored. The ramping current density increases the bath suppression and stabilizes the bath, either due to the increased chloride coverage on the copper surface or a negative current density suppressor / levelant attachment dependency.

III. Devices and systems

11 and 12 show an example of a device 20 which may be provided with a bad badge detection system. In this example, the device points 20 a rotor 24 in a head 22 on. The head can be lowered to a wafer 40 on the rotor 24 in contact with a catholyte in the vessel 26 over a membrane 32 to position. An anolyte and one or more anodes 28 are in the vessel 26 below the membrane 32 arranged. An agitator or stirrer 36 can be optional in the vessel 26 or at the top of the vessel 26 be provided.

In this version, the device control monitors 50 the voltage of each anode 28 , Upon detection of a sudden change in voltage, the controller determines that a bad failure has occurred. The controller may then issue a warning or alarm, and optionally turn it off. Basically, most devices of this type already have the electrical connections required to perform this function, so this function can be added to the device via software used in programming the controller. The methods described above can be used in devices with membrane or without a membrane.

 As described, an electroplating system for processing a wafer having TSV structures may include a vessel for holding a bath of electrolyte and one or more anodes in the vessel. A wafer holder has a contact ring that makes electrical contact with the wafer, with a cathode electrically connected to the contacts. A voltage monitor monitors a voltage between one or more of the anodes and the contact ring. A controller is connected to the voltage monitor, wherein the controller detects a bad failure based on a change in the voltage.

Claims (9)

A method of electroplating a wafer comprising: Placing the wafer in contact with a bath of electrolyte comprising an accelerator, a leveling agent and a suppressor; Passing an electric current from one or more anodes through the electrolyte and through a conductive layer on the wafer; Monitoring the voltage of the one or more anodes; Detecting a failure of the bath from a change in voltage.  The method of claim 1, further comprising detecting a failure of the bath based on a drop in voltage of at least 100 mV.  The method of claim 1, wherein the wafer comprises TSV structures.  A method of electroplating a wafer comprising: Placing the wafer in contact with a bath of an electrolyte having an accelerator and a suppressor; Passing an electric current from one or more anodes through the electrolyte and through a conductive layer on the wafer; and Electric current beats to control formation of MPS in the bath.  The method of claim 4, further comprising monitoring the voltage of the one or more anodes to detect a failure of the bath based on a drop or oscillation of the voltage.  The method of claim 4, wherein the wafer comprises TSV structures.  A method of electroplating a wafer comprising: Placing the wafer in contact with a bath of an electrolyte having an accelerator and a suppressor; Passing an electric current from one or more anodes through the electrolyte and through a conductive layer on the wafer; and Run-up of electric current to control formation of MPS in the bath.  The method of claim 7, further comprising monitoring the voltage of the one or more anodes to detect a failure of the bath based on a drop or oscillation of the voltage.  The method of claim 7, wherein the wafer comprises TSV structures.

Images