Plasma Processing Using Radio Frequency (RF) Sources and Bias Signal Waveforms

Abstract

1. A method for plasma processing, comprising: maintaining a plasma in a plasma processing chamber, the plasma processing chamber comprising a first radio frequency (RF) electrode and a second RF electrode, wherein maintaining the plasma comprises: coupling an RF source signal to the first RF electrode; and coupling a bias signal between the first RF electrode and the second RF electrode, the bias signal having a B-DC waveform comprising a plurality of bipolar DC (B-DC) pulses, each of the B-DC pulses comprising a negative bias duration during which the pulse has a negative polarity with respect to a reference potential, a positive bias duration during which the pulse has a positive polarity with respect to the reference potential, and a neutral bias duration during which the pulse has a neutral polarity with respect to the reference potential.

Description

[Technical Field]

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS This application claims priority to U.S. Non-Provisional Patent Application No. 17 / 451,094, filed October 15, 2021, which is incorporated herein by reference.

[0002] The present invention relates generally to plasma processing, and in particular embodiments to systems and methods for plasma processing using a radio frequency (RF) source and a bipolar DC bias signal waveform. [Background technology]

[0003] Integrated circuits (ICs) are typically fabricated by sequentially depositing and patterning dielectric, metal, and semiconductor layers above a substrate to form a network of electronic components connected by metal lines and vias, all integrated within a single monolithic structure. Advances in photolithography and three-dimensional (3D) devices, such as gate-all-around field-effect transistors (GAAFETs) and 3D NAND memory, are enabling component density to double approximately every two years, reducing the cost per function of ICs. Plasma processes, such as plasma-enhanced atomic layer deposition (PEALD), high-aspect-ratio contact (HARC) etching, and selective deposition, are often used to fabricate 3D devices with nanoscale features, requiring challenging plasma technologies that offer atomic-scale control over a wide range of metrics, including edge profile, roughness, anisotropy, uniformity, conformality, and selective deposition selectivity. Plasma processing metrics depend on plasma properties such as ion energy and ion angular distribution, ion and radical density, and radical-to-ion flux ratio. The plasma characteristics and chemical environment at the surface and within deep trenches are affected by the RF signal used to sustain the plasma during processing, therefore further innovations in RF signal waveform generation would aid in the precision and control of plasma processing. Summary of the Invention [Means for solving the problem]

[0004] A method for plasma processing includes maintaining a plasma in a plasma processing chamber, the plasma processing chamber including a first radio frequency (RF) electrode and a second RF electrode, where maintaining the plasma includes coupling an RF source signal to the first RF electrode and coupling a bias signal between the first RF electrode and the second RF electrode, the bias signal having a B-DC waveform including a plurality of bipolar DC (B-DC) pulses, each of the B-DC pulses including a negative bias duration during which the pulse has a negative polarity relative to a reference potential, a positive bias duration during which the pulse has a positive polarity relative to the reference potential, and a neutral bias duration during which the pulse has a neutral polarity relative to the reference potential.

[0005] 1. A system for plasma processing comprising: a plasma processing chamber, the plasma processing chamber comprising: a first radio frequency (RF) electrode; a second RF electrode; and a substrate holder configured to hold a semiconductor substrate in the plasma processing chamber; a processor; and a non-transitory memory storing a program executed on the processor, the program comprising: instructions for coupling an RF source signal to the first RF electrode; and instructions for coupling a bias signal between the first RF electrode and the second RF electrode, the bias signal having a B-DC waveform comprising a plurality of bipolar DC (B-DC) pulses, each of the B-DC pulses comprising: a negative bias duration during which the pulse has a negative polarity relative to a reference potential; a positive bias duration during which the pulse has a positive polarity relative to the reference potential; and a neutral bias duration during which the pulse has a neutral polarity relative to the reference potential.

[0006] A method for plasma processing includes maintaining a plasma in a plasma processing chamber, the plasma processing chamber including a first radio frequency (RF) electrode and a second RF electrode, where maintaining the plasma includes coupling an RF source signal to the first RF electrode and coupling a bias signal between the first RF electrode and the second RF electrode, where the bias signal has a bipolar DC (B-DC) waveform that is the difference between a first unipolar DC (U-DC) waveform and a second unipolar DC waveform, and where the polarity of the first U-DC waveform is the same as the polarity of the second U-DC waveform, where coupling the bias signal includes coupling the first U-DC signal to the first RF electrode, where the first U-DC signal has a first U-DC waveform including a first plurality of U-DC pulses, each of the first plurality of U-DC pulses having a pulse width of 100 Hz. coupling a second U-DC signal to a second RF electrode, the second U-DC signal having a second U-DC waveform including a second plurality of U-DC pulses, each of the second plurality of U-DC pulses including a second U-DC pulse width, the pulse having a first bias polarity relative to a reference potential, and a second U-DC pulse separation time, the pulse having a neutral bias polarity substantially equal to the reference potential; and synchronizing the first U-DC signal with the second U-DC signal, the synchronizing delaying the U-DC pulses of the first U-DC signal from the U-DC pulses of the second U-DC signal by a fixed U-DC delay time.

[0007] For a more complete understanding of the present invention and its advantages, reference is now made to the following descriptions taken in conjunction with the accompanying drawings. [Brief explanation of the drawings]

[0008] [Figure 1] 1 is a schematic diagram of a plasma processing system according to some embodiments. [Figure 2A]10A-10C show plots of various parameterized waveforms of an RF source signal and a pulsed bipolar DC (B-DC) bias signal, according to some embodiments. [Figure 2B] 10A-10C show plots of various parameterized waveforms of an RF source signal and a pulsed bipolar DC (B-DC) bias signal, according to some embodiments. [Figure 2C] 10A-10C show plots of various parameterized waveforms of an RF source signal and a pulsed bipolar DC (B-DC) bias signal, according to some embodiments. [Figure 2D] 10A-10C show plots of various parameterized waveforms of an RF source signal and a pulsed bipolar DC (B-DC) bias signal, according to some embodiments. [Figure 3A] 10A-10C show plots of various B-DC waveforms of a bias signal used in conjunction with a continuous wave (CW) RF waveform of an RF source signal to maintain a plasma in a plasma processing chamber, according to some embodiments. [Figure 3B] 10A-10C show plots of various B-DC waveforms of a bias signal used in conjunction with a continuous wave (CW) RF waveform of an RF source signal to maintain a plasma in a plasma processing chamber, according to some embodiments. [Figure 4A] 10A-10C show plots of various B-DC waveforms of a bias signal used in conjunction with an RF burst waveform of an RF source signal to maintain a plasma in a plasma processing chamber, according to some embodiments. [Figure 4B] 10A-10C show plots of various B-DC waveforms of a bias signal used in conjunction with an RF burst waveform of an RF source signal to maintain a plasma in a plasma processing chamber, according to some embodiments. [Figure 4C] 10A-10C show plots of various B-DC waveforms of a bias signal used in conjunction with an RF burst waveform of an RF source signal to maintain a plasma in a plasma processing chamber, according to some embodiments. [Figure 5A] 1A-1C are schematic diagrams of two alternative configurations of a plasma processing system for coupling a B-DC bias signal between two electrodes of a plasma processing chamber, according to some embodiments. [Figure 5B] 1A-1C are schematic diagrams of two alternative configurations of a plasma processing system for coupling a B-DC bias signal between two electrodes of a plasma processing chamber, according to some embodiments. DETAILED DESCRIPTION OF THE INVENTION

[0009] This disclosure relates to systems and methods for plasma processing of semiconductor substrates. The characteristics of the plasma and plasma processing generally depend on the properties of two electrical signals, known as a radio frequency (RF) source signal and a bias signal, used to ignite and maintain the plasma in a plasma processing chamber. This disclosure describes exemplary embodiments of plasma processing systems and methods in which the plasma is ignited and maintained by coupling a parameterized RF source signal and a time-varying bias signal to an RF electrode of the plasma processing chamber. The plasma processing system includes a processor that executes a program having instructions for configuring the system, the program being stored in a memory included with the system. When the program is executed, the processor sends commands to a timing controller to provide selected timing parameters for adjusting and synchronizing the RF source and bias signal waveforms. Using the timing parameters, the timing controller can send appropriate control signals to synchronize and control the output signals of electronics that generate the RF source and bias signals coupled to the RF electrode of the plasma processing chamber. By adjusting and synchronizing the RF source and bias signal waveforms, the timing controller can adjust the plasma environment at the surface of the substrate, including the surface inside a deep trench.

[0010] Exemplary embodiments of the present disclosure use an RF source signal having either a continuous wave RF (CW-RF) waveform or an RF burst waveform including a train of RF burst pulses. Each RF burst pulse has an RF waveform (e.g., an RF sine wave) that is present during the RF burst duration of the RF burst pulse. The RF burst duration is followed by an RF burst separation time during which the RF source signal is not present. The sum of the RF burst duration and the RF burst separation time is one period of the RF burst waveform, referred to as the RF burst period. The length of the RF burst period can be from about 1 microsecond to about 10 milliseconds.

[0011] The time-varying bias signal has a pulsed bipolar DC waveform, referred to in this disclosure as a B-DC waveform. The B-DC waveform includes a radio-frequency train of B-DC pulses, including DC pulses of positive and negative polarities and a neutral polarity duration. The various polarities are relative to a reference potential of the RF system, defined as equal to 0 V. The reference potential is ground potential or earth. Each B-DC pulse in the pulse train is parameterized by three timing parameters: a negative bias duration (during which the pulse is negative polarity), a positive bias duration (during which the pulse is positive polarity), and a neutral bias duration (during which the pulse is neutral polarity). In some embodiments, a fourth timing parameter can be introduced by dividing the neutral bias duration into a first neutral bias pulse segment and a second neutral bias pulse segment to separate the negative and positive bias durations, as described in further detail below. One cycle of B-DC pulses, which is the sum of the three (or four) timing parameters, can be approximately 100 nanoseconds to approximately 400 microseconds long. The B-DC waveform of the bias signal may be a continuous B-DC waveform including a continuous train of B-DC pulses, or a B-DC burst waveform including a train of B-DC burst pulses. Each B-DC burst pulse includes multiple consecutive B-DC pulses present during a B-DC burst duration followed by a B-DC burst separation time during which the bias signal is absent, and the sum of the B-DC burst duration and the B-DC burst separation time is one B-DC burst period. As will be explained in more detail below, the RF burst period must be equal to the B-DC burst period and is referred to as the burst period in this disclosure.

[0012] First, one embodiment of a plasma processing system is described using Figure 1. Alternative embodiments of a plasma processing system are described using Figures 5A-5B. Alternative embodiments of an RF source signal and a pulsed bipolar DC (B-DC) bias signal waveform are described using Figures 2A-2D. Alternative embodiments of a B-DC waveform of a bias signal used with a continuous wave (CW) RF waveform of an RF source signal are described using Figures 3A-3D. Alternative embodiments of a B-DC waveform of a bias signal used with an RF burst waveform of an RF source signal are described using Figures 4A-4C.

[0013] 1 is a schematic diagram of a plasma processing system 100 including a plasma processing chamber 110, a CW-RF source signal source 120, a continuous B-DC bias signal source 130, a timing controller 140, and a processor 142. The CW-RF source signal source 120 and the continuous B-DC bias signal source 130 may be programmable. The timing controller 140 is configured to receive commands from the processor 142, which, when executed, programs the continuous B-DC bias signal source 130 to generate B-DC pulses using selected timing parameters received from the processor 142. The signals from the RF source signal source 120 and the continuous B-DC bias signal source 130 are routed through programmable first and second choppers 124 and 134, respectively. The first and second choppers 124 and 134 may be programmed by the timing controller 140 to transmit the input signal unmodulated or modulate the input signal with a pulsed gating signal to transmit a modulated signal. The gate signals generated in each of the first chopper 124 and the second chopper 134 periodically pass the input signal (directly couple the input to the output) during the burst duration and block the signal (disconnect the output from the input) during the burst separation time, where the sum of the burst time and the burst separation time is the burst period. The first gate signal in the first chopper 124 and the second gate signal in the second chopper 134 have the same burst period, and the two gate signals are synchronously shifted by a fixed burst delay. By synchronizing the B-DC burst waveform with the RF burst waveform, the B-DC burst pulse is delayed from the RF burst pulse by a fixed burst delay. This burst delay is used to synchronize the superposition of the RF burst waveform and the B-DC burst waveform within the plasma processing chamber 110. The two burst times, burst duration, and burst delay are specified in commands sent by the processor 142 of the plasma processing system 100 to the timing controller 140, as indicated by the block arrows pointing to the timing controller 140 in FIG.The timing controller executes instructions to synchronously control the output signal of the first chopper 124, the output signal of the second chopper 134, and the output signal of the continuous B-DC bias signal source 130, as indicated by the two block arrows directed from the timing controller 140 to each of the first chopper 124 and the second chopper 134 and the fourth angled block arrow directed from the timing controller 140 to the continuous B-DC bias signal source 130.

[0014] In some embodiments, the CW-RF signal source 120 may be a low-power source and may be combined with a signal generator, such as a programmable chopper 124, into one unit. The signal generator generates a low-power signal by switching the CW-RF signal using control signals received from a timing controller 140. The low-power signal is then amplified to full power by an RF power amplifier (not shown in FIG. 1). The amplifier may also be combined with the low-power RF source and the signal generator into one unit.

[0015] Typically, a plasma processing chamber includes two RF electrodes that can be configured to couple to an RF source signal and / or a bias signal transmitted from outside the plasma processing chamber. The outer wall of the plasma processing chamber typically has a conductive portion coupled to the reference potential (or ground) of the RF system and can be referred to as the ground electrode of the plasma processing chamber. The RF source signal, bias signal, and the reference potential for which the signals are generated are transmitted to the plasma processing chamber. The RF source signal and bias signal are coupled to the RF electrodes, and the reference potential is coupled to the ground electrode of the plasma processing chamber. For example, in FIG. 1, if two coaxial cables are used to couple the RF source signal and bias signal to the plasma chamber 110, each of the two inner conductors of the coaxial cables can be connected to either the upper RF electrode 116 or the lower RF electrode 118. The two outer conductors can be connected to the conductive portion 119 of the outer wall of the plasma chamber 110.

[0016] An RF source signal can be coupled to a first RF electrode, and a bias signal can be coupled between the first and second RF electrodes. As mentioned above, in the embodiments described in this disclosure, the bias signal has a B-DC waveform. There are several configurations that can be used to couple a bias signal between a first RF electrode and a second RF electrode, where an RF source is coupled to the first RF electrode. In one embodiment, a bias signal having a B-DC waveform is coupled to the first RF electrode, and the second RF electrode is coupled to a reference potential. In another embodiment, a bias signal having a B-DC waveform is coupled to the second RF electrode. In yet another embodiment, a bias signal having a B-DC waveform is the difference between a first unipolar DC (U-DC) signal and a second U-DC signal. In this embodiment, a first U-DC signal is coupled to the first RF electrode, and a second U-DC signal is coupled to the second RF electrode. As will be described in further detail below, the selected timing parameters of the B-DC waveform can be achieved by synchronizing the first U-DC signal with the second U-DC signal by selecting appropriate pulse widths and pulse separation times for the two U-DC signals and delaying the U-DC pulses of the first U-DC signal by a fixed U-DC delay time from the U-DC pulses of the second U-DC signal.

[0017] The plasma processing chamber 110 of FIG. 1 is a vacuum chamber configured as a capacitively coupled plasma (CCP) chamber. A CCP chamber may include two disk-shaped RF electrodes positioned on opposite sides of the interior of the plasma processing chamber, with each RF electrode having a flat surface facing the opposing RF electrode. In the exemplary configuration shown in FIG. 1, the upper RF electrode 116 is coupled to an RF source signal, and the lower RF electrode 118 is coupled to a bias signal. A reference potential (denoted GND in FIG. 1) is coupled to the ground electrode of the plasma processing chamber 110, which is a conductive portion 119 of the outer wall.

[0018] The CCP configuration is chosen only as an example, and other configurations, such as inductively coupled plasma (ICP), can also be used. In an ICP configuration, an RF source signal is coupled to an RF electrode (called an ICP antenna) that is physically separated from the plasma by a thick dielectric window. RF source power is coupled to the plasma from the ICP antenna outside the plasma chamber. In some embodiments, a pulsed bipolar DC waveform can be used for plasmas formed using microwave power instead of RF power.

[0019] The plasma processing chamber 110 of FIG. 1 is configured with a gas inlet 112 and an exhaust 114 coupled to a gas flow system. As known to those skilled in the art, the gas flow system includes gas canisters, valves, and a vacuum pump that maintain a flow of a low-pressure gas mixture through the plasma processing chamber 110. A mixture of a process gas and a carrier gas, referred to as a feed gas, is introduced through the inlet 112 and flows through the plasma processing chamber 110. The ambient inside the plasma processing chamber 110 is typically maintained at low pressure and a temperature that is relatively low compared to typical furnace temperatures, e.g., room temperature. The pressure within the plasma processing chamber 110 is typically between about 1 mTorr and about 1 Torr, although some plasma processes can be performed at higher pressures (e.g., atmospheric pressure).

[0020] When the supply gas is exposed to the electromagnetic fields generated at the RF electrodes 116 and 118, a small portion of the gas molecules can be ionized to form a plasma (often called a weakly ionized plasma) between the electrodes.

[0021] In a CCP chamber (e.g., plasma processing chamber 110), energy is transferred from the RF source and bias signal to the plasma (e.g., plasma 150) by capacitive coupling between the plasma and the RF electrode. The plasma is also coupled to ground by a capacitive impedance between the plasma (e.g., plasma 150) and the plasma chamber's ground electrode (e.g., the outer wall conductive portion 119). As described above, the RF source signal can be a CW-RF or RF burst waveform transmitted from the first chopper 124. Similarly, the bias signal transmitted from the second chopper 134 can be a continuous B-DC pulse train or can have a B-DC burst waveform. The lower RF electrode 118 in the plasma processing chamber 110 can be included in a substrate holder (e.g., an electrostatic chuck) that holds the substrate 108, e.g., a semiconductor substrate. In some embodiments, the conductive bulk of the substrate 108 is physically separated from the lower RF electrode 118 by a dielectric layer. In some embodiments, the substrate holder (eg, an electrostatic chuck) may include a dielectric layer between the lower RF electrode 118 and the surface of the substrate holder that is in contact with the backside of the substrate 108 .

[0022] The feed gas is generally charge-neutral and in thermal equilibrium at a temperature T equal to the ambient temperature within the plasma processing chamber 110. A portion of the energy imparted to the feed gas dissociates some of the neutral gas molecules and atoms into a net-neutral, weakly ionized plasma containing positively charged ions (denoted by "+") and negatively charged free electrons (denoted by "e") in the plasma 150, as shown schematically in FIG. 1. Additional energy can be transferred from the electromagnetic field to the moving charged particles as kinetic energy. As explained further below, kinetic energy generally has a random component (due to randomizing collisions between various particles) and a directional component in the direction of the accelerating electric field. In regions where the electric field is low and the collision frequency is high, the random component can dominate, while in regions where the electric field is high and the collision frequency is low, the directional component (in the direction of the electric force) can dominate.

[0023] In addition to producing charged ions and free electrons, a portion of the energy imparted to the feed gas generates neutral radicals (designated "R" in FIG. 1). A neutral radical is an atom (or group of atoms) that has an unpaired electron and is neutral in charge. Due to the presence of the unpaired electron, the radical has high chemical reactivity.

[0024] Initially, the average kinetic energy per particle of all types of particles is its thermal equilibrium value, (3 / 2)k B is approximately equal to T, where T is the ambient temperature in Kelvin, and k B is Boltzmann's constant. For example, at room temperature, i.e., 300 K, the average kinetic energy of each particle in the feed gas is 39 meV. When ionized, charged particles are accelerated and gain kinetic energy from the electric field, and the gained energy is excess kinetic energy. Although the acceleration is directed parallel to the electric field, the momentum gained from the electric field is rapidly scattered in all possible directions by random collisions with various particles (mostly neutral particles in a weakly ionized plasma). Thus, much of the kinetic energy gained from the electric field in the bulk of the plasma is transferred in directions that increase the random component of the kinetic energy. For each type of charged particle (ion and electron), the average value of the random component of their kinetic energy is determined by their respective non-equilibrium effective temperatures: electron temperatures T e and ion temperature T ionDuring the initial transient state, the effective temperature rises from an initial thermal equilibrium value, T. As the energy increases beyond the thermal equilibrium value, the frequency of inelastic collisions increases until a steady state is established where the excess energy gained from the electric field is balanced by the loss of excess energy in inelastic collisions. The steady state value of the effective temperature depends on the frequency of inelastic collisions. The collision frequency depends on the probability of inelastic collisions (collisions that cause significant energy loss) averaged over an ensemble of particles of each species (e.g., free electrons or ions). The lower the frequency of inelastic collisions, the higher the effective temperature at which the steady state is established. The majority of collisions between free electrons and neutral particles are nearly elastic (negligible energy loss), whereas ions almost always lose energy in collisions (e.g., between ions and neutral particles). Thus, T e Generally, T ion is much higher. The preferential increase in the kinetic energy of free electrons is more pronounced at lower pressures. Thus, in the plasma 150 of FIG. 1, collisions may establish a non-equilibrium steady state where the random component of the average kinetic energy of ions is not much higher than that of gas molecules at T=300 K, but the randomized average kinetic energy of free electrons may be 100 times higher (e.g., about 1 eV to about 10 eV). As explained further below, in plasma processes such as sputtering and anisotropic reactive ion etching (RIE) processes that use ions to bombard a surface, ions can acquire high directional kinetic energies (e.g., about 50 eV to about 1000 eV) as they pass through a region of high electric field before striking a substrate.

[0025] Generally, the bulk of the plasma, where most of the ions and free electrons are concentrated, is in a large central quasi-neutral region where the electric field is low due to screening by mobile charges. The quasi-neutral region is surrounded by a narrow, space-charge-depleted region known as the plasma sheath, where the electric field is high. The mobility and diffusivity of free electrons are typically much higher than those of ions. Because positively charged ions are much slower than negatively charged free electrons, free electrons near the periphery are lost from the plasma more quickly, leaving behind a net positive charge that forms a sheath around the central quasi-neutral region. The charge distribution within the sheath establishes an electric field that pushes free electrons back toward the quasi-neutral region and accelerates ions outward toward the conductive portion 119, causing them to collide with the chamber walls and substrate 108. Generally, the charge density and electric field within the sheath are higher than those in the quasi-neutral region.

[0026] In the quasi-neutral region, a small electric field (small relative to the electric field in the sheath) called the ambipolar field exists to balance the flow of ions and free electrons to maintain a steady-state charge density distribution. The ions and free electrons gain momentum and kinetic energy from the ambipolar field. However, the directional components of the excess momentum and energy are small; that is, the ions and free electrons move randomly in all possible directions and collide relatively frequently with various particles, especially the numerous neutral particles in weakly ionized plasmas. In other words, the distribution of ions and free electrons with respect to angle is nearly uniform. Furthermore, because of the randomized collisions, the energy distribution can be approximated by a Maxwell-Boltzmann distribution function with increasing temperature, which represents the non-equilibrium average kinetic energy. Energy balance is largely achieved by charged particles (e.g., free electrons and ions) losing energy to neutral particles in inelastic collisions. Because neutral particles are not accelerated by the electric field, they remain at a lower temperature, very close to the thermal equilibrium value T. As explained above, both electrons and ions experience their own rates of energy loss due to inelastic collisions, resulting in their own elevated effective temperatures T e and T ion Typically, T e >>Tion and the ratio is (T e / T ion )>10.

[0027] Plasma processing is generally performed on the substrate material by physically bombarding the exposed surface of the substrate with energetic ions and / or chemically reacting radicals generated within the plasma that are present in the feed gas, depending on the plasma process. The physical and chemical interactions may produce volatile by-products (designated "B" in FIG. 1) that are pumped out of the plasma processing chamber 110 through the exhaust port 114.

[0028] In a weakly ionized plasma, only a small fraction of neutral feed gas particles are ionized, but ions and free electrons are essential for plasma processing. Collisions between neutral gas molecules and high-energy free electrons on the high-energy side of the Maxwell-Boltzmann distribution play a major role in generating radicals and ions in the quasi-neutral region of the plasma 150 by dissociating and ionizing the feed gas molecules. Low-energy "cold" ions from the quasi-neutral region are energized as they enter the narrow plasma sheath close to the substrate, gaining directional kinetic energy from the sheath's electric field. This ion energy, determined primarily by the potential difference across the sheath, helps radicals chemically react with other radicals in the gas and the substrate (e.g., in a chemical vapor deposition (CVD) process). Over time, the flow of ions to the substrate can result in the accumulation of undesirable positive charge. In addition to their role in generating radicals and ions, negatively charged free electrons can be used to neutralize the accumulation of positive charge, as further explained below.

[0029] This disclosure describes a method for plasma processing using innovative RF source signal and B-DC bias signal waveforms. The electromagnetic field, and therefore the RF source signal and B-DC bias signal waveforms, directly affect the flow and energy of charged particles (e.g., free electrons and ions) within the plasma. The influence of the RF source signal and B-DC bias signal waveforms on neutral species (e.g., neutral radicals and byproducts of chemical and physical interactions) is achieved indirectly through pulse timing and synchronization. For example, in some embodiments, the plasma 150 within the plasma chamber 110 can be periodically extinguished for a duration of about 1 microsecond to about 1 millisecond and reignited to prevent the accumulation and redeposition of reaction byproducts, as described in further detail below. In some embodiments, the neutral bias duration can be adjusted in the range of about 100 nanoseconds to about 1 millisecond to achieve a desired ratio of radical fluence to ion fluence to ensure a sufficient density of radicals is present to react with material sputtered from the substrate by ion bombardment. The neutral bias duration also allows time for ions extracted from the plasma by the electric field within the sheath to be replenished.

[0030] As defined above, the neutral bias duration refers to the duration within a single B-DC pulse during which the bias signal is neutral. It is understood that during the B-DC burst separation time, the bias signal is off. However, by definition, the neutral bias duration includes only the duration within each individual B-DC pulse when the bias polarity is neither positive nor negative relative to the reference potential.

[0031] For illustrative purposes, the schematic diagram of the plasma processing chamber 110 in FIG. 1 shows a snapshot of the plasma 150 in which positively charged ions (denoted by "+") are preferentially attracted toward the lower RF electrode 118 and the substrate 108, while negatively charged electrons (denoted by "e") are repelled in the opposite direction toward the upper RF electrode 116, possibly under the influence of a bias signal. Neutral radicals (denoted by "R") may drift with the gas flow or diffuse toward the RF electrode, but because their charge is neutral, their flux is not modulated by the electromagnetic fields within the plasma processing chamber 110. Reaction by-products (denoted by "B") are seen near the substrate 108. The positively charged ions and negatively charged electrons may be pushed in opposite directions, e.g., toward the bottom or the top, by the electric field, depending on the instantaneous polarity of the time-varying bias signal applied between the upper RF electrode 116 and the lower RF electrode 118. As described in further detail below, this dependency can be exploited in embodiments described in this disclosure to influence plasma processing by shaping the B-DC waveform of the bias signal in synchronization with the RF source signal.

[0032] The parameterization and synchronization of the RF source and bias signals is described with reference to Figures 2A-2D.

[0033] FIG. 2A shows an example plot of a CW-RF waveform 2A1 output from the CW-RF source signal source 120 and a continuous B-DC waveform 2A2 output from the continuous B-DC bias signal source 130 of the plasma processing system 100 of FIG.

[0034] In Figure 2A, the CW-RF waveform 2A1 is a continuous RF sine wave of fixed frequency and amplitude, although it will be appreciated that other continuous RF waveforms may also be used, such as an RF sawtooth waveform.

[0035] The dashed double arrow "A" indicates one B-DC pulse of the continuous B-DC waveform 2A2 in FIG. 2A . The exemplary B-DC waveform 2A2 is a pulse train in which each B-DC pulse has three pulse segments. The timing parameters are the durations of the three pulse segments, as indicated by the solid double arrows in the enlarged plot of a single B-DC pulse 2A2′. The parameters are a negative bias duration 212 during which the pulse has negative polarity, a positive bias duration 214 during which the pulse has positive polarity, and a neutral bias duration 216 during which the pulse has neutral polarity relative to the reference potential (0 V). As shown in the respective plots in FIG. 2A , in B-DC pulse 2A2′, the negative bias duration occurs before the positive bias duration, while the opposite is true for B-DC pulse 2A3. B-DC pulse 2A3 can be obtained by simply reversing the polarity of the negative and positive bias segments of B-DC pulse 2A2′. 1 , applying B-DC pulse 2A2′ to lower RF electrode 118 first attracts ions to substrate 108 (during negative bias duration 212) and then attracts free electrons to neutralize the buildup of positive charge during positive bias duration 214. In another embodiment, the connections between the RF source signal and the bias signal can be reversed, with upper RF electrode 116 coupled to the RF source signal and lower RF electrode 118 coupled to the bias signal. In this configuration, B-DC pulse 2A3 achieves the same effect as B-DC pulse 2A2′ in the configuration shown in FIG. 1 . Therefore, continuous B-DC bias signal source 130 can be programmed to generate B-DC pulse 2A3. The bias signal is near 0 V (reference potential) during neutral bias duration 216, allowing sufficient time for radicals and ions to diffuse to the surface before bombarding the surface with ions again. In some embodiments, the neutral bias duration 216 can be divided into a first neutral bias pulse segment 215 and a second neutral bias pulse segment 217, as shown by B-DC pulse 2A4 in FIG. 2A.Although rarely used, the first neutral bias pulse segment 215 may delay the neutralizing flux of free electrons during the transient ion flow following the end of the negative bias duration 212 .

[0036] Although the exemplary B-DC waveform 2A2 and B-DC pulses 2A2', 2A3, and 2A4 of FIG. 2A depict equal magnitudes of negative and positive bias, it will be understood that unequal magnitudes may also be used.

[0037] 2B shows a plot of a CW-RF waveform 2A1, e.g., an exemplary gating waveform 2B1 generated in the first programmable chopper 124 (see FIG. 1), and an RF burst waveform 2B2 generated by modulating the CW-RF waveform 2A1 with the gating waveform 2B1. As shown in FIG. 2B, one burst period 226 of the gating waveform 2B1 is the sum of one burst duration 222 and one burst separation time 224. A gating signal having the gating waveform 2B1 allows the CW-RF waveform 2A1 to pass during the burst duration 222 and block the CW-RF waveform 2A1 during the burst separation time 224, thereby generating the RF burst waveform 2B2. Generally, the primary power source that sustains a plasma (e.g., plasma 150) in a plasma processing chamber (e.g., plasma processing chamber 110) is an RF source signal. Therefore, the weakly ionized plasma can be extinguished during the burst separation time 224 and re-ignited after the start of the next RF burst pulse.

[0038] 2C shows a plot of a continuous B-DC waveform 2A2, e.g., another gating waveform 2C1 generated in the second chopper 134 (see FIG. 1 ), and a B-DC burst waveform 2C2 generated by modulating the B-DC waveform 2A2 with the gating waveform 2C1. The gating waveform 2C1 allows the continuous B-DC waveform 2A2 to pass unchanged during a burst duration 232 and block the continuous B-DC waveform 2A2 during a burst separation time 234, thereby generating a B-DC burst waveform 2C2, where each B-DC burst pulse is a burst of three B-DC pulses, as shown in FIG. 2C. In various embodiments, the B-DC burst pulses can include a single B-DC pulse or a different number of multiple B-DC pulses.

[0039] 2B. As noted above, in embodiments of plasma processing systems (e.g., plasma processing system 100) described in this disclosure, both the first chopper 124 and the second chopper 134 are controlled by the timing controller 140 to use a single value for the burst duration of the B-DC burst waveform and the burst duration of each RF burst waveform. As described with reference to the waveform plots of FIG. 2D, the RF source signal and each bias signal are synchronized by a fixed burst delay 242, so that the two burst durations are set to the same value.

[0040] FIG. 2D shows plots of gated waveform 2D1 and respective B-DC burst waveform 2D2, along with gated waveform 2B1 and respective RF burst waveform 2B2, as also shown in FIG. 2B. In this example, the burst duration 232 of gated waveform 2D1 is different from the burst duration 222 of gated waveform 2B1. Similarly, the burst separation time 234 of gated waveform 2D1 is different from the burst separation time 224 of gated waveform 2B1. However, both gated waveforms 2B1 and 2D1 have the same burst period 226. In FIG. 2D, gated waveform 2D1 is delayed by a burst delay 242 relative to gated waveform 2B1. Accordingly, B-DC burst waveform 2D2 is also delayed by a burst delay 242 from RF burst waveform 2B2. Note that each B-DC burst pulse of B-DC burst waveform 2D2 begins at a fixed delay time after the start of each RF burst pulse of RF burst waveform 2B2. A fixed separation equals the burst delay, synchronizing the RF source signal having RF burst waveform 2B2 with the bias signal having B-DC burst waveform 2D2. The burst period 226 is common to the gate signal waveforms 2B1 and 2D1, allowing for synchronization with a fixed delay.

[0041] 3A and 3B show plots of pairs of RF source and bias signals used in an embodiment in which the RF source signal has a CW-RF waveform. For each pair, the RF source and bias signals are used in combination with each other to maintain a plasma for plasma processing.

[0042] In Figure 3A, a CW-RF waveform 3A1 is paired with a bias signal having a continuous B-DC waveform 3A3 that includes a train of continuous B-DC pulses. Each B-DC pulse in the train has three timing parameters: a negative bias duration 312, a positive bias duration 314, and a neutral bias duration 316, similar to B-DC pulse 2A2' described above with reference to Figure 2A.

[0043] In FIG. 3B, the same CW-RF waveform 3B1 is paired with a bias signal having a B-DC burst waveform 3B3 that includes a train of B-DC burst pulses, each of which is a burst of three B-DC pulses. In this example, each of the three B-DC pulses in a burst uses the same three timing parameters as used for the B-DC pulses in the continuous B-DC waveform 3A3 (see FIG. 3A). In addition, because the B-DC burst waveform 3B3 is a train of periodic bursts of three B-DC pulses, it has two additional timing parameters, burst duration 332 and burst separation time 334, that are derived from the gating signal waveform 3B2 used to generate the B-DC burst waveform 3B3. As shown in FIG. 3B, the sum of burst duration 332 and burst separation time 334 is burst period 326.

[0044] The three parameters for the B-DC pulses 3A3 and 3B3 and the two additional parameters for the B-DC burst pulse may be selected by, for example, a processor 142 and controlled by a timing controller 140 in the plasma processing system 100 (shown in FIG. 1). Because the RF source signal waveform is continuous, there is no burst delay and no synchronization is performed.

[0045] A plasma (e.g., plasma 150) resembles an engine that generates ions, free electrons, and radicals and is fueled to a large extent by RF power from the RF source signal. Thus, when the RF source signal has a CW-RF waveform, ions and radicals are continuously generated in a central quasi-neutral region. These ions and radicals slowly diffuse outward toward the plasma sheath at the edge of the quasi-neutral region. The electric field of the plasma sheath accelerates (or retards) charged particles (positive ions and negative free electrons), while neutral radicals continue to diffuse through the sheath toward the substrate (and chamber walls). A time-varying bias signal can have a strong effect on the electric field within the plasma sheath. In the embodiment described in this disclosure, the bias signal has a B-DC waveform. Different segments of the B-DC pulse affect the interaction of ions, radicals, and free electrons with the substrate in different ways, as further described below. The removal and / or redeposition of byproducts of these interactions is also affected by parameters of the B-DC waveforms 3A3 and 3B3, such as the burst separation time 334 in FIG. 3B.

[0046] When a substrate (e.g., substrate 108) is negatively biased, the electric field within the plasma sheath increases in a direction that accelerates ions from the quasi-neutral region toward the substrate. The ions gain very high directional kinetic energy, resulting in bursts of ions impacting the surface. To achieve a narrow spread in the energy distribution of the energetic ions, a rectangular pulse with very short rise and fall times may be preferred. The duration of this negative bias need not be too short due to the slow response time of the relatively low mobility ions.

[0047] On the other hand, there are several reasons why using a long duration of negative bias is undesirable. When a negative bias is present for a long time, there is a large fluence of ions bombarding the surface, and the low fluence of radicals provided by the slow diffusion of radicals from the plasma to the substrate physically removes a much larger amount of material than can chemically react. Because insufficient radicals are available to chemically react with the removed material, the etch rate is slow, some of the stripped material may randomly redeposit instead of being converted to volatile byproducts, and the etching mechanism is physical rather than chemical, resulting in low selectivity to other exposed materials. Furthermore, ion bombardment can become inefficient over time because the supply of ions may be depleted.

[0048] Another reason the duration of the negative bias cannot be too long is that in some processes, positive ions can cause charge buildup in the substrate. To neutralize the buildup of positive charge, a pulse segment in which a negative bias is applied to the substrate can be followed by a segment in which a positive bias is applied to the substrate. The positively biased substrate attracts negatively charged free electrons to neutralize the buildup of positive charge. Because electrons are light and highly mobile, the time the substrate has a positive polarity can be kept short.

[0049] The neutral bias duration (the duration within a single B-DC pulse during which the bias signal has neutral polarity) can be used to adjust the radical fluence relative to the ion fluence in each B-DC pulse. If the neutral bias duration is too short, i.e., if the ion bursts are too frequent, the ratio of radicals to ion fluence may become unacceptably low, and the radicals must diffuse over a longer distance, especially in deep trenches, as the surface is further removed from the plasma. Because the RF source signal (e.g., the RF source signal having the CW waveform 3B1 in FIG. 3B) is continuously present and therefore radicals are continuously generated, the burst separation time (e.g., the burst separation time 334 in FIG. 3B) also helps the radicals diffuse over a longer distance.

[0050] Another advantage provided by burst separation time 334 is that it allows volatile byproducts of the chemical and physical interactions to be removed by a vacuum pump through an exhaust port (e.g., exhaust port 114 in FIG. 1 ). In some embodiments, the volatile byproducts may be unstable and decompose in secondary chemical reactions to form solid byproducts that may deposit on exposed surfaces and distort the profile of, for example, the sidewalls of etched features. Note that in some embodiments, the solid byproducts may be used to passivate surfaces, for example, trench sidewalls.

[0051] 4A-4C show plots of exemplary pairs of RF source and bias signals, where both signals have burst pulse waveforms. Because the RF source signal has an RF burst waveform, when the RF signal is turned off during the RF burst separation time, the plasma can be extinguished, thereby "cooling" the free electrons and pausing the generation of ions and radicals. When the RF source signal is reapplied at the beginning of the next burst duration, the plasma can be reignited and the generation process resumed. The RF and B-DC burst waveforms of each pair are synchronized by a burst delay, similar to the synchronized waveform pair of FIG. 2D (RF burst waveform 2B2 and B-DC burst waveform 2D2). Each B-DC pulse of B-DC burst waveforms 4A4, 4B4, and 4C4 (of FIGS. 4A, 4B, and 4C, respectively) has the same negative pulse duration 412, positive pulse duration 414, and neutral pulse duration 416, for illustrative purposes only.

[0052] In FIG. 4A , an RF source signal having an RF burst waveform 4A2 is used in combination with a bias signal having a B-DC burst waveform 4A4. In this example, a gating waveform 4A1 is used to generate the RF burst waveform 4A2, and the same waveform is used to generate the B-DC burst waveform 4A4. A burst duration 432 and a burst separation time 434 are used, added to the burst period 426. In this example embodiment, the burst delay is zero. Using the same gating waveform (e.g., gating waveform 4A1) and zero burst delay generates an in-phase burst waveform pair: RF burst waveform 4A2 and B-DC burst waveform 4A4. The example in FIG. 4A is a special case of an in-phase burst waveform. Generally, with an in-phase burst waveform, if there is no RF source signal, the B-DC burst waveform is not present. Therefore, the RF burst duration is equal to or greater than the B-DC burst duration, and the burst delay can be between zero and the difference between the RF burst duration and the B-DC burst duration. Having a pulse segment where neither the RF source signal nor the bias signal is present creates a duration during which no by-products can be generated. Because the gas flow system operates continuously, the by-products can be efficiently removed from the plasma processing chamber (e.g., plasma processing chamber 110 of FIG. 1) by the gas flow system's vacuum pump.

[0053] FIG. 4B shows an RF burst waveform 4B2 generated using a gating waveform 4B1. The RF burst waveform 4B2 and the gating waveform 4B1 are identical to the respective RF burst waveforms 4A2 and 4A1 shown in FIG. 4A for illustrative purposes only. FIG. 4B also shows a B-DC burst waveform 4B4 (and respective gating waveform 4B3) used in combination with the RF burst waveform 4B2. The burst duration 432 and burst separation time 434 (and thus the burst period 426) of the RF burst pulse waveform 4B2 are selected to be identical to the B-DC burst waveform 4B4, also for illustrative purposes only. Unlike the exemplary waveforms of FIG. 4A, in FIG. 4B, the burst delay 442 is not zero. Note that, as shown in FIG. 4B, there is no overlap of the B-DC burst pulse with the RF burst pulse. Because there is no overlap of the B-DC burst pulse with the RF burst pulse, this exemplary waveform pair is referred to as an out-of-phase burst waveform. Generally, in an in-phase burst waveform, the B-DC burst waveform and the RF source signal are never simultaneously present, so the B-DC burst duration is less than or equal to the RF burst separation time, and the burst delay is greater than or equal to the RF burst duration and less than or equal to the B-DC separation time.

[0054] For example, consider the configuration of the plasma processing system 100 shown in FIG. 1 , where an RF source signal having an RF burst waveform 4B2 and a bias signal having a B-DC burst waveform 4B4 are coupled to the upper RF electrode 116 and the lower RF electrode 118, respectively. Because the burst delay 442 is greater than the RF burst duration 432, the following sequence of events may occur periodically: Plasma 150 is ignited at the beginning of each RF burst pulse and persists throughout the burst duration 432, generating ions, free electrons, and radicals. However, without a bias signal, some ions accelerate through the plasma sheath and impact the surface, but bursts that cause very high-energy ions to impact the substrate surface are infrequent. However, without a bias signal, the energy gained by accelerating through the sheath is relatively low. As explained above, diffusion of radicals and other particle species in the plasma processing chamber 110 occurs continuously, largely independent of the presence or absence of a bias signal. At the end of the RF burst pulse, the electron temperature T e decreases and the plasma 150 is extinguished.

[0055] Because the burst delay 442 is greater than the RF burst duration 432, there is a short wait time during which no RF or bias signals are present. After this short wait, the B-DC burst pulse begins. Typically, the relaxation time for ions and radicals to reach thermal equilibrium concentrations is relatively long compared to the short wait time. Therefore, ions and radicals exist in non-equilibrium concentrations within the plasma processing chamber 110 during the B-DC burst duration. A burst of several B-DC pulses at this point accelerates ions, causing energetic ions to impact the substrate, and radicals may chemically react with the substrate material and other radicals to form volatile by-products. eOne benefit of accelerating ions when the ion flux is reduced is that it helps reduce the random component of the ion kinetic energy. Reducing the random component of the ion kinetic energy (or equivalently reducing the ion temperature) reduces the component of ion velocity parallel to the substrate surface, thereby shifting the impact angle closer to the normal to the surface. This results in a narrower spread in the angular distribution of energetic ions. In other words, more energetic ions are directed perpendicular to the surface. A near-perpendicular ion flux helps achieve high anisotropy for RIE processes (often referred to as HARC etching), which can be used to form high-aspect-ratio contact holes. A narrow angular spread is beneficial in avoiding undesired collisions with the sidewalls of deep trenches during the HARC etching process.

[0056] FIG. 4C shows another pair of plots of an RF source signal having an RF burst waveform 4C2 generated using a gating waveform 4C1 combined with a bias signal having a B-DC burst waveform 4C4 generated using a gating waveform 4C3. For illustrative purposes, all timing parameters of waveforms 4C1, 4C2, 4C3, and 4C4, except for burst delay 444, are set equal to their respective parameters in the respective waveforms of FIGS. 4A and 4B. In FIG. 4C, burst delay 444 is greater than zero and less than or equal to RF burst duration 432. Thus, a portion of each B-DC burst pulse overlaps with the RF burst pulse, and the remaining duration of the B-DC burst pulse extends beyond the end of the RF burst pulse. Generally, in overlapping burst waveforms, a portion of the B-DC burst waveform exists while the RF source signal is present, and the remaining duration of the B-DC waveform exists while the RF source signal is absent. As mentioned above, the burst delay is greater than zero but less than or equal to the RF burst duration to ensure that there is some overlap. To ensure that the B-DC burst pulse extends within the RF burst separation time, the burst delay (e.g., burst delay 444) is selected so that the sum of the burst delay and the B-DC burst duration is greater than the RF burst duration. To prevent the B-DC burst waveform from extending beyond the RF burst separation and overlapping with the next RF burst pulse, the burst delay should not exceed the B-DC burst separation time.

[0057] Now consider the sequence of events that occur when an RF source signal (RF burst waveform 4C2) and a bias signal (B-DC burst waveform 4C4) are applied to the upper RF electrode 116 and the lower RF electrode 118, respectively, of the plasma processing chamber 110. When the RF burst pulse begins, a plasma 150 is ignited. In the absence of a bias signal, ions, free electrons, and radicals are generated. During this time, radicals diffuse to the substrate, but because there is no bias signal to accelerate the ions to very high directional kinetic energies, the number of energetic ions striking the substrate 108 is small.

[0058] As shown in FIG. 4C, at the end of the burst delay 444, a B-DC burst pulse of the bias signal begins. The negative polarity segment of the B-DC pulse accelerates ions. The energetic ions, along with radicals, physically and chemically interact with the substrate, generating volatile by-products. The RF sine wave and B-DC pulse are both present until the RF source signal is turned off at the end of the RF burst duration 432. The presence of the RF source signal increases the electron temperature T e and ion temperature T ion As soon as the RF source signal is turned off, T e and T ion Both the ion kinetic energy and the bias voltage begin to decrease. As explained above, the random component of the ion kinetic energy is undesirable for anisotropic plasma etching, e.g., a HARC etching process. Therefore, the RF source signal is turned off, but the duration that the bias signal is present provides the advantage of accelerating the ions to achieve very high directional kinetic energy without excessively increasing the random component.

[0059] As shown in Figure 4C, by continuing the B-DC burst duration beyond the end of the RF burst pulse, the electron temperature T e This allows the etching process to continue during a reduced time period, which, as explained above, is advantageous for highly anisotropic etching of high aspect ratio trenches (e.g., HARC etch processes).

[0060] The RF source signal and the bias signal are both turned off from the end of the B-DC burst duration to the beginning of the next RF burst duration. As described above with reference to Figures 3B and 4A, the duration during which both the RF source and the bias signals are absent can be adjusted to remove volatile byproducts through the exhaust port 114 of the plasma processing chamber 110 (see Figure 1).

[0061] 5A and 5B show two configurations for applying the same bias signal between a first electrode and a second electrode, where the bias signal has a B-DC waveform 5A1. The bias signal having the B-DC waveform 5A1 can be provided by the difference between a first unipolar DC (U-DC) signal having a first U-DC waveform 5B1 and a second U-DC signal having a second U-DC waveform 5B2. In FIGS. 5A and 5B, the first electrode is the lower RF electrode 118 of the plasma processing chamber 110, and the second electrode is the upper RF electrode 116 of the plasma processing chamber 110. The plasma chamber 110 shown in FIGS. 5A and 5B is similar to the processing chamber 110 shown in FIG. 1. The lower RF electrode 118 is included in a substrate holder that holds a substrate 108. A conductive portion 119 of the wall of the plasma chamber 110 is coupled to a reference potential.

[0062] In the configuration shown schematically in Figure 5A, the lower RF electrode 118 (first electrode) is coupled to a bias signal having a B-DC waveform 5A1 that includes a train of B-DC pulses. The upper RF electrode 116 (second electrode) is coupled to a reference potential equal to 0V, shown as GND. For illustrative purposes, a positive bias is +V B V, and negative bias is -V B V, and 0V is a neutral bias. It is understood that biases of unequal magnitude may also be used. Each B-DC pulse of B-DC waveform 5A1 is similar to B-DC pulse 2A2' described above with reference to FIG. 2A. It is understood that other B-DC pulses, such as those similar to 2A3 or 2A4 described with reference to FIG. 2A, may also be used.

[0063] 5A, the B-DC pulse of B-DC waveform 5A1 has a negative bias duration 512, during which the pulse has a negative polarity, a positive bias duration 514, during which the pulse has a positive polarity, and a neutral bias duration 516, during which the pulse has a neutral polarity relative to a reference potential. The sum of the three bias durations 512, 514, and 516 is the pulse period 526 of the B-DC pulse of B-DC waveform 5A1. Similar to neutral bias duration 216 of B-DC pulse 2A2′ in FIG. 2A, neutral bias duration 516 is a continuous pulse segment, with no separation between the end of negative bias duration 512 and the beginning of positive bias duration 514.

[0064] In the configuration shown schematically in Figure 5B, the lower RF electrode 118 is coupled to a first U-DC signal having a first U-DC waveform 5B1 that includes a train of negative polarity unipolar pulses, and the upper RF electrode 116 is coupled to a second U-DC signal having a second U-DC waveform 5B2 that also includes a train of negative polarity unipolar pulses.

[0065] The pulse width of each U-DC pulse of the first U-DC waveform 5B1 is selected to be equal to the negative bias duration 512, and the pulse separation time 536 is selected so that the sum of the pulse width (negative bias duration 512) and the pulse separation time 536 is equal to the pulse period 526 of the B-DC pulse of the B-DC waveform 5A1.

[0066] The pulse width of each U-DC pulse of the second U-DC waveform 5B2 is selected to be equal to the positive bias duration 514, and the pulse separation time 546 is selected so that the sum of the pulse width (positive bias duration 514) and the pulse separation time 546 is also equal to the pulse period 526 of the B-DC pulse of the B-DC waveform 5A1. Note that the B-DC waveform 5A1 and the first and second U-DC waveforms 5B2, 5B3 all have equal pulse periods.

[0067] A first U-DC signal having a first U-DC waveform 5B1 is synchronized with a second U-DC signal having a second U-DC waveform 5B2. This synchronization is achieved by delaying the U-DC pulse of the first U-DC waveform 5B1 from the U-DC pulse of the second U-DC waveform 5B2 by a fixed U-DC delay time 544, as shown in FIG. 5B. In this example, the fixed U-DC delay time 544 is selected to be equal to the negative bias duration 512 of the B-DC pulse of the B-DC waveform 5A1. This selection causes the timing (or phase) of the difference signal to match the timing of the B-DC waveform 5A1. Here, the difference signal has a difference waveform, which is defined by the difference between the first U-DC waveform 5B1 and the second U-DC waveform 5B2. The difference waveform exactly matches the B-DC waveform 5A1. This exact match results from the particular choice of constant U-DC delay time 544 (the choice being that U-DC delay time 544 in FIG. 5B is equal to negative bias duration 512 in FIG. 5A) and the absence of a separation between the end of negative bias duration 512 and the start of positive bias duration 514 in B-DC waveform 5A1 (the neutral bias duration is a continuous pulse segment).

[0068] The exemplary U-DC waveforms 5B1 and 5B2 are negative polarity, but it will be appreciated that the B-DC waveform 5A1 can also be achieved using the difference of two U-DC waveforms having positive polarity U-DC pulses.

[0069] Using the embodiments described in this disclosure, the timing of the pulsed signals can be adjusted and synchronized to, for example, adjust the plasma characteristics, chemical environment, and charging and neutralization of the surface of the substrate by positively charged ions and negatively charged electrons, respectively. The plasma processing method can be performed using commands from a processor of the plasma processing system, including commands sent to a timing controller configured to synchronize and adjust various pulse waveforms of an RF source signal and a pulsed B-DC bias signal coupled to electrodes of the plasma processing chamber.

[0070] Example 1. A method for plasma processing, comprising maintaining a plasma in a plasma processing chamber, the plasma processing chamber comprising a first radio frequency (RF) electrode and a second RF electrode, wherein maintaining the plasma comprises coupling an RF source signal to the first RF electrode; and coupling a bias signal between the first RF electrode and the second RF electrode, the bias signal having a B-DC waveform comprising a plurality of bipolar DC (B-DC) pulses, each of the B-DC pulses comprising a negative bias duration during which the pulse has a negative polarity relative to a reference potential, a positive bias duration during which the pulse has a positive polarity relative to the reference potential, and a neutral bias duration during which the pulse has a neutral polarity relative to the reference potential.

[0071] Example 2. The method of Example 1, further comprising coupling a conductive portion of a wall of the plasma processing chamber to a reference potential.

[0072] Example 3. The method of Example 1 or 2, further comprising coupling a second RF electrode to a reference potential and coupling a bias signal to the first RF electrode.

[0073] Example 4. The method of any one of Examples 1-3, further comprising coupling a bias signal to a second RF electrode.

[0074] Example 5. The method of any one of Examples 1-4, wherein the RF source signal has a continuous wave (CW) RF waveform and the bias signal has a continuous B-DC waveform, the continuous B-DC waveform including a continuous train of B-DC pulses.

[0075] Example 6. The method of any one of Examples 1-5, wherein each of the B-DC pulses has a single pulse segment defined to have a neutral polarity relative to a reference potential for a neutral bias duration, and the neutral bias durations are contiguous in time.

[0076] Example 7. The method of any one of Examples 1-6, wherein the neutral bias duration is divided into a first neutral bias pulse segment and a second neutral bias pulse segment, the first neutral bias pulse segment separating the end of the negative bias duration from the start of the next positive bias duration, and the second neutral bias pulse segment separating the end of the positive bias duration from the start of the next negative bias duration.

[0077] Example 8. The method of any one of Examples 1-7, wherein the bias signal has a B-DC burst waveform including a train of B-DC burst pulses, each of the B-DC burst pulses having a plurality of consecutive B-DC pulses present during a B-DC burst duration followed by a B-DC burst separation time in which the bias signal is absent, the sum of the B-DC burst duration and the B-DC burst separation time being defined as the burst period.

[0078] Example 9. The method of any one of Examples 1-8, wherein the RF source signal has a continuous wave (CW) RF waveform.

[0079] Example 10. The method of any one of Examples 1-9, wherein the RF source signal has an RF burst waveform including a train of RF burst pulses, each of the RF burst pulses being present during an RF burst duration followed by an RF burst separation time during which the RF source signal is not present, the sum of the RF burst duration and the RF burst separation time being equal to a burst period, and the method further includes synchronizing a B-DC burst waveform with the RF burst waveform, wherein the synchronizing delays the B-DC burst pulses from the RF burst pulses by a fixed burst delay.

[0080] Example 11. The method of any one of Examples 1-10, wherein the RF burst duration is greater than or equal to the B-DC burst duration, and the burst delay is greater than or equal to zero and less than or equal to the difference between the RF burst duration and the B-DC burst duration.

[0081] Example 12. The method of any one of Examples 1-11, wherein the B-DC burst duration is less than or equal to the RF burst separation time, and the burst delay is greater than or equal to the RF burst duration and less than or equal to the B-DC burst separation time.

[0082] Example 13. The method of any one of Examples 1-12, wherein the burst delay is greater than zero and less than or equal to the smaller of the RF burst duration and the B-DC burst separation time, and the sum of the burst delay and the B-DC burst duration is greater than the RF burst duration.

[0083] Example 14. A system for plasma processing, comprising: a plasma processing chamber including a first radio frequency (RF) electrode, a second RF electrode, and a substrate holder configured to hold a semiconductor substrate within the plasma processing chamber; a processor; a non-transitory memory storing a program executed on the processor, the program comprising: instructions for coupling an RF source signal to the first RF electrode; and instructions for coupling a bias signal between the first RF electrode and the second RF electrode, the bias signal having a B-DC waveform comprising a plurality of bipolar DC (B-DC) pulses, each of the B-DC pulses comprising a negative bias duration during which the pulse has a negative polarity relative to a reference potential, a positive bias duration during which the pulse has a positive polarity relative to the reference potential, and a neutral bias duration during which the pulse has a neutral polarity relative to the reference potential.

[0084] Example 15. The system of Example 14, further including: a continuous wave (CW) RF source signal source; a programmable continuous B-DC bias signal source; a programmable first chopper coupled to the CW-RF source signal source; a programmable second chopper coupled to the continuous B-DC bias signal source; and a timing controller coupled to the continuous B-DC bias signal source, the first chopper, and the second chopper, the timing controller configured to receive commands from the processor that, when executed, synchronously control an output signal of the first chopper, an output signal of the second chopper, and an output signal of the continuous B-DC bias signal source.

[0085] Example 16. The system of Example 14 or 15, wherein the system configuration includes a first RF electrode coupled to the output signal from the first chopper and the output signal from the second chopper, and a second RF electrode coupled to a reference potential.

[0086] Example 17. The system of any one of Examples 14-16, wherein the system configuration includes a first RF electrode coupled to the output signal from the first chopper and a second RF electrode coupled to the output signal from the second chopper.

[0087] Example 18. A method for plasma processing, comprising maintaining a plasma in a plasma processing chamber, the plasma processing chamber comprising a first radio frequency (RF) electrode and a second RF electrode, wherein maintaining the plasma comprises coupling an RF source signal to the first RF electrode; and coupling a bias signal between the first RF electrode and the second RF electrode, the bias signal having a bipolar DC (B-DC) waveform that is the difference between a first unipolar DC (U-DC) waveform and a second unipolar DC waveform, the polarity of the first U-DC waveform being the same as the polarity of the second U-DC waveform, wherein coupling the bias signal comprises coupling the first U-DC signal to the first RF electrode, the first U-DC signal having a first U-DC waveform comprising a first plurality of U-DC pulses, each of the first plurality of U-DC pulses being a first U-DC pulse width, the pulses having a first bias polarity relative to a reference potential, and a first U-DC pulse separation time, the pulses having a neutral bias polarity substantially equal to the reference potential; coupling a second U-DC signal to a second RF electrode, the second U-DC signal having a second U-DC waveform including a second plurality of U-DC pulses, each of the second plurality of U-DC pulses including a second U-DC pulse width, the pulses having a first bias polarity relative to the reference potential, and a second U-DC pulse separation time, the pulses having a neutral bias polarity substantially equal to the reference potential; and synchronizing the first U-DC signal with the second U-DC signal, wherein the synchronizing delays the U-DC pulses of the first U-DC signal from the U-DC pulses of the second U-DC signal by a fixed U-DC delay time.

[0088] Example 19. The method of Example 18, further comprising coupling a conductive portion of a wall of the plasma processing chamber to a reference potential.

[0089] Example 20. The method of example 18 or 19, wherein the first bias polarity is a positive bias polarity with respect to a reference potential.

[0090] While the present invention has been described with reference to several exemplary embodiments, it is not intended to be construed in a limiting sense. Various modifications and combinations of the exemplary embodiments, as well as other embodiments of the invention, will become apparent to those skilled in the art upon reference to the above description. It is therefore intended that the appended claims cover any such modifications or embodiments.

Claims

1. 1. A method for plasma processing, comprising: maintaining a plasma in a plasma processing chamber including a first radio frequency (RF) electrode and a second RF electrode; maintaining the plasma coupling an RF source signal to the first RF electrode; applying a bias signal between the first RF electrode and the second RF electrode; the bias signal has a bipolar DC (B-DC) waveform including a plurality of B-DC pulses, each of the B-DC pulses including a negative bias duration during which the pulse has a negative polarity relative to a reference potential, a positive bias duration during which the pulse has a positive polarity relative to the reference potential, and a neutral bias duration during which the pulse has a neutral polarity relative to the reference potential; the bias signal has a B-DC burst waveform including a train of B-DC burst pulses, each of the B-DC burst pulses having a plurality of consecutive B-DC pulses present during a B-DC burst duration followed by a B-DC burst separation time during which no bias signal is present; and the sum of the B-DC burst duration and the B-DC burst separation time is defined as a burst duration.

2. The method of claim 1 , further comprising coupling a conductive portion of a wall of the plasma processing chamber to the reference potential.

3. coupling the second RF electrode to the reference potential; coupling a bias signal source to the first RF electrode, the bias signal source applying the bias signal; The method of claim 1 further comprising:

4. The method of claim 1 , further comprising coupling a bias signal source to the second RF electrode, the bias signal source applying the bias signal.

5. 2. The method of claim 1, wherein the RF source signal generates a continuous wave (CW) RF waveform, and the bias signal has a continuous B-DC waveform, the continuous B-DC waveform including a continuous train of B-DC pulses.

6. 2. The method of claim 1, wherein the neutral bias duration is divided into a first neutral bias pulse segment and a second neutral bias pulse segment, the first neutral bias pulse segment separating an end of the negative bias duration from a start of a next positive bias duration, and the second neutral bias pulse segment separating an end of the positive bias duration from a start of a next negative bias duration.

7. The method of claim 1 , wherein the RF signal source generates a continuous wave (CW) RF waveform.

8. an RF signal source generating an RF burst waveform including a train of RF burst pulses, each of said RF burst pulses having an RF waveform present during an RF burst duration followed by an RF burst separation time during which there is no RF source signal; the sum of the RF burst duration and the RF burst separation time is equal to said burst period; 2. The method of claim 1, further comprising: synchronizing the B-DC burst waveform with the RF burst waveform, wherein the synchronizing delays the B-DC burst pulse by a fixed burst delay from the RF burst pulse.

9. the RF burst duration is equal to or greater than the B-DC burst duration; 9. The method of claim 8, wherein the burst delay is greater than or equal to zero and less than or equal to the difference between the RF burst duration and the B-DC burst duration.

10. the B-DC burst duration is less than or equal to the RF burst separation time; 9. The method of claim 8, wherein the burst delay is greater than or equal to the RF burst duration and less than or equal to the B-DC burst separation time.

11. the burst delay is greater than zero and less than or equal to the smaller of the RF burst duration and the B-DC burst separation time; The method of claim 8 , wherein the sum of the burst delay and the B-DC burst duration is greater than the RF burst duration.

Images