System for controlling the deposition of thin layers, comprising a continuous light source and an optical cavity

EP4758401A1Pending Publication Date: 2026-06-17CENT NAT DE LA RECH SCI (C N R S) +4

Patent Information

Authority / Receiving Office
EP · EP
Patent Type
Applications
Current Assignee / Owner
CENT NAT DE LA RECH SCI (C N R S)
Filing Date
2024-08-06
Publication Date
2026-06-17

Smart Images

  • Figure EP2024072288_20022025_PF_FP_ABST
    Figure EP2024072288_20022025_PF_FP_ABST
Patent Text Reader

Abstract

The invention relates to a system for controlling the deposition of thin layers by gas phase growth in a deposition reactor (10), the gas phase corresponding to a gaseous state or a plasma state, the control system comprising: - a continuous light source (2); - an optical cavity (3) comprising at least two mirrors, and configured such that a light emitted by the light source is trapped in the optical cavity and interacts with the gas phase, resulting in the deposition of thin layers; and - a spectrometer (4) which is configured to analyze light emitted by the light source (2) and passing through the optical cavity (3), by measuring the optical absorbance of at least one species of the gas phase penetrated by said light.
Need to check novelty before this filing date? Find Prior Art

Description

THIN FILM DEPOSITION CONTROL SYSTEM COMPRISING A CONTINUOUS LIGHT SOURCE AND AN OPTICAL CAVITY

[0001] The invention relates to the deposition of thin layers. It relates more particularly to a system for controlling such deposition.

[0002] A thin film is an element whose thickness, one of its dimensions, is nanometric or micrometric, ranging from a few atomic planes to a few thousand atomic planes. Thin films are obtained by depositing a material on a substrate.

[0003] The deposition of thin layers is used in many industrial fields (optics, chemistry, mechanics, micro-optoelectronics, renewable energies).

[0004] Thin film deposition processes use known technologies, including: molecular beam epitaxy (MBE), laser ablation-type systems, cathode sputtering or electron beam evaporation, chemical vapor deposition (CVD) systems, etc.

[0005] The performance of many components / devices and their manufacturing costs depend directly on the level of control of thin film deposition processes.

[0006] Various systems exist for monitoring the growth of thin films. One principle is to determine the composition of the gas phase used for thin film deposition (directly linked to the composition and growth rate of the thin film) by measuring the optical absorbance of the species that compose it. In other words, it involves measuring the ratio of the intensity of an optical beam partially transmitted by the gas phase (denoted I in the following) to that of the incident optical beam (denoted I0 in the following). The species composing the gas phase exhibit a certain number of absorption lines in specific spectral ranges, as illustrated in the figure. This figure gives the line strengths of several species of interest for the fabrication of thin films, particularly for applications in micro-optoelectronics.Line strength characterizes the propensity of a chemical species to absorb optical radiation at a given wavelength: at equal concentration in the gas phase, optical absorption increases linearly with line strength.

[0007] OFM sensors for "optical flux monitoring" are based on this principle of measuring optical absorbance and include a light source and an atomic absorption spectrometer. For example, document US20150284851 discloses such an OFM sensor.

[0008] OFM sensors use hollow cathode lamps (HCLs) as their light source. An example of a prior art OFM sensor is shown in. These lamps emit a set of lines characteristic of a single element. Their emission spectrum is discrete: the lines are separated from each other by spectral regions where the emitted intensity is zero.

[0009] HCL lamps have certain disadvantages that strongly impact the performance of the sensor.

[0010] First, the discrete nature of the emission spectrum of HCL lamps does not allow the incident intensity to be deduced from a measurement of the unabsorbed background since this background is zero. However, measurement of the incident intensity is necessary to correct for drifts in the source intensity.

[0011] For an OFM sensor, the incident intensity measurement is generally carried out as illustrated in the figure. The light emitted by the HCL lamp is separated into two paths: a measuring beam Meas and a reference beam Ref. The measuring beam Meas makes one pass through the gas to be analyzed. The reference beam Ref passes, for example, outside the reactor. Both paths are then analyzed in a light analysis system. The intensity of the Meas path corresponds to I, that of the Ref path to I0. The figure shows two paths. This number of paths depends on the number of elements to be measured. There are as many Meas and Ref paths as there are elements to be measured.

[0012] Determining the intensities I and I0 thus requires the installation of additional measurement and reference channels for each species. This greatly complicates the sensor architecture. Furthermore, since the reference and measurement channels follow different optical paths, sensitivity to drift is high.

[0013] Furthermore, HCL lamps emit a low photon flux per unit area, and produce a wide and relatively divergent beam, a large part of which is lost in the low numerical aperture elements of the optical path (fibers, spectrometer, in particular). Consequently, the number of photons arriving at the detector is limited, which limits the signal-to-noise ratio of the measurement (SNR) with an impact on the accuracy and sensitivity of the sensor.

[0014] The performance of conventional OFM sensors is also intrinsically limited by the very principle of measurement: the interaction length between the light beam and the gas phase to be analyzed, limited to a few tens of centimeters, leads to relatively low absorbances, and therefore to relatively high measurement uncertainties. The best relative uncertainties on the control of composition and growth rates reported for OFM sensors are of the order of tens of % and at best a few % (cf. [Y. Du et al. 2014]), for growth rates of the order of a few monolayers per minute. This is well below the precision required for thin-film deposition processes (uncertainty on thickness and composition less than a %).

[0015] Finally, HCL lamps are bulky, relatively unstable, quite expensive and have a limited lifespan.

[0016] The invention aims to provide a thin film deposition control system that is more efficient and more precise in controlling the growth of thin films.

[0017] The invention proposes for this purpose a system for controlling the deposition of thin layers by gas phase growth in a deposition reactor, the gas phase corresponding to a gaseous state or a plasma state. The control system comprises: a continuous light source; an optical cavity comprising at least two mirrors, and configured so that light emitted by the light source is trapped in the optical cavity and interacts with the gas phase; and a spectrometer configured to analyze the light emitted by the light source and passing through the optical cavity by measuring the optical absorbance of at least one species of the gas phase passed through by said light.

[0018] Thanks to its configuration and in particular that of the optical cavity, the invention makes it possible to achieve considerable sensitivities, in particular for low growth speeds, and therefore very high precision in the control of growth processes.

[0019] Since the gas phase to be analyzed is located in the optical cavity, the light is reflected by the mirrors. Thus, instead of interacting with the gas only once as in existing OFM systems, the light makes a large number of round trips (up to 1000 passages). This considerably increases the light / gas interaction length, and consequently the absorbance and thus the signal-to-noise ratio of the measurement (SNR), thus leading to a significant reduction in measurement uncertainty. A growth rate measurement accuracy better than % is thus easily achieved, even at low growth rates.

[0020] Furthermore, the light source used is continuous. This type of source is more reliable, more stable, easier to integrate into an optical system and has a much longer lifespan than HCL lamps. The use of a continuous light source considerably simplifies and makes the reference intensity measurement procedure more reliable, making the control system simpler and allowing very effective drift control.

[0021] Particularly convenient preferred features of the thin film deposition control system according to the invention are presented below.

[0022] The optical cavity surrounds the gas phase.

[0023] The optical cavity is confocal.

[0024] The mirrors are assembled on optical ports of the reactor. The mirrors are arranged inside or outside the reactor.

[0025] The length of the optical cavity is chosen so that the spectral width of the absorption lines measured by the spectrometer covers a minimum number of resonance modes, said minimum number preferably being equal to 10.

[0026] The spectrometer has a dispersion stage, the dispersion stage being an echelle-type diffraction grating.

[0027] The dispersion stage is configured to disperse at diffraction orders between 10 and 100.

[0028] The light source is a light-emitting diode, a superluminescent diode, or an arc lamp.

[0029] The spectrometer is equipped with a camera, preferably a CCD or CMOS camera.

[0030] Other features and advantages of the invention will also appear in the description below with reference to the appended drawings, given by way of non-limiting example: represents examples of light sources and transition intensities of certain elements in the UV-visible range; represents an OFM sensor of the prior art; represents a thin film deposition control system according to an embodiment of the invention; and schematically represents signals measured on a camera of the control system as well as their intensity.

[0031] The aim of the invention is to determine the composition of a gas phase used for the deposition of thin layers by measuring the optical absorbance of species composing the gas phase.

[0032] The gas phase is understood to be a gaseous state or a plasma state. The gas can be in particular in the molecular regime.

[0033] The gas phase corresponds to the atmosphere containing the precursors in which a substrate is placed for growth. This can in fact be a gas or a plasma whose pressure is between 10 -5 Torr and atmospheric pressure, for growth processes such as evaporation, sputtering, CVD or laser ablation, but also a set of molecular jets whose pressure can drop to 10 -10 Torr or less, for molecular beam epitaxy type growth processes.

[0034] The invention uses for the determination of the composition of the gas phase, a control system 1 for the deposition of thin layers shown schematically in. The control system 1 is used in a gas phase deposition reactor 10.

[0035] The layer deposition control system 1 comprises at least one light source 2. In the exemplary embodiment according to the invention shown in , the control system 1 comprises two light sources 2.

[0036] Light source 2 is spectrally continuous. A continuous light source is understood to be a source whose intensity varies continuously over a spectral range that is large compared to the spectral width of the absorbance lines to be measured. In other words, the light source can be any continuous source whose intensity varies continuously over a spectral range from a few hundred picometers to several hundred nanometers wide.

[0037] For example, the light source is an LED (Light Emitting Diode). The typical spectral width of the LED is about 10 nm. Alternatively, the light source 2 is a SLED (Superluminescent Diode). The typical spectral width of the SLED varies between a few nm and 10 nm. The light source can also be an arc lamp, e.g., Xe lamp, mercury lamp, deuterium lamp, halogen lamps, spectral Neon lamps, LDLS sources, etc. The spectral width of an arc lamp is typically several hundred nm.

[0038] Of course, light source 2 can be any other type of continuous source.

[0039] Ladonne, in the upper part, the spectral radiance of some light sources as a function of wavelength. In particular, the spectral radiance of LEDs, three SLEDs and different arc lamps (xenon Xe lamp, and LDLS for "laser driven light sources"). The spectral radiance of an HCL lamp is also shown for comparison.

[0040] As can be seen in this figure, the spectral width of LED / SLED diodes, of the order of tens of nm, makes them much less specific than HCL lamps. Thus, a single continuous light source of the invention covers several lines of several species.

[0041] Several light sources 2 of different nature can be used in the control system 1. A single combination of three or four continuous light sources (e.g. three or four LED / SLED diodes) allows coverage of many species and is suitable for a large number of materials. The very wide spectral range covered by a single arc lamp also allows measurement of the absorbance of all species of interest.

[0042] The control system 1 thus configured (having one or more continuous light sources) is more flexible than prior art control systems or sensors using HCL lamps. In addition, thanks to the invention, it is less often necessary to add or change a light source to change the deposited material. Furthermore, unlike HCL lamps limited to only monatomic elements, continuous sources allow any atomic or molecular species to be measured, and make the control system applicable to many growth techniques.

[0043] Furthermore, continuous light sources, especially SLEDs, are brighter than HCL lamps. They have a greater spectral radiance than HCL lamps. This helps to limit optical losses in the optical path and helps to improve the signal-to-noise ratio (SNR) of the control system 1.

[0044] Continuous light sources are also stable and easy to integrate into an optical system. They are easier to change if necessary because they are available in fiber versions.

[0045] More specifically, LED / SLED diodes are compact and have a longer lifespan than HCL lamps. LED / SLED diodes are also available in a wide spectral range and can cover the entire spectrum of interest, typically between 350 and 700 nm. Their cost is also reasonable, even low for LEDs.

[0046] Arc lamps also have the advantage of covering a very wide spectral range. They are of particular interest for the most distant lines in the ultraviolet.

[0047] The use of continuous light sources also allows the use of the unabsorbed continuous background for measuring the reference beam intensity. The measurement beam intensity and the reference beam intensity are measured on the same optical path. They are therefore subject to the same drifts and are intrinsically compensated in real time. This is optimal for the stability and reproducibility of the control system. Furthermore, this scheme considerably simplifies the control system architecture, which no longer requires reference paths.

[0048] The layer deposition control system 1 further comprises an optical cavity 3 and a light analysis system 4.

[0049] The optical cavity 3 surrounds the gas phase, resulting in the deposition of thin layers. For this purpose, the optical cavity 3 is configured so that light emitted by the light source 2 is trapped in the optical cavity and interacts with the gas phase.

[0050] The optical cavity 3 comprises at least two mirrors 31.

[0051] Since the gas phase to be analyzed is located in the optical cavity, the light is reflected by the mirrors. Thus, instead of interacting with the gas only once as in existing OFM systems, the light makes a large number of round trips (up to 1000 passages). This considerably increases the light / gas interaction length, and consequently the absorbance and thus the signal-to-noise ratio of the measurement (SNR), thus leading to a significant reduction in measurement uncertainty. A growth rate measurement accuracy better than % is thus easily achieved, even at low growth rates.

[0052] The mirrors 31 may be arranged inside the reactor 10 or outside the reactor 10. The mirrors 31 are assembled on optical accesses (not shown) of the reactor 10. For example, the mirrors 31 may be assembled to the reactor 10 by means of mounting flanges.

[0053] The mounting flanges are preferably protected from parasitic deposits that may form on the mirrors 31 and which would alter the optical properties of the optical cavity 3. The mounting flanges are for example placed back from the gas phase. Alternatively or in addition, differential pumping systems near the mirrors may be installed to avoid parasitic deposits. Alternatively or in addition, the mounting flanges may be heated to avoid parasitic deposits.

[0054] The reflectivities R of the mirrors 31 may be different. Preferably, the mirrors 31 have the same reflectivity value R.

[0055] The optimal reflectivity of mirrors 31 depends on the order of magnitude of the absorbance to be measured. If several lines with very different absorbances are to be measured simultaneously, a compromise must be made on the reflectivity R.

[0056] The radii of curvature RC mirrors 31 may be different. Preferably, the mirrors 31 have the same value of radius of curvature R C.

[0057] The radius of curvature R C is chosen so that the optical cavity 3 is optically stable.

[0058] The optical cavity 3 has a length L. The length L of the optical cavity 3 is between a few tens of cm and 3 m.

[0059] Optical cavity 3 has a free spectral interval ISL cav depending on the length L of the cavity. The free spectral interval ISL cav decreases when the length L increases.

[0060] The length L of the optical cavity 3 and therefore the free spectral interval ISL cav are configured so that the spectral width of the absorption lines measured by the light analysis system 4 covers a minimum number of resonance modes. Preferably, the free spectral interval ISL cavmust be small enough so that the spectral width of the absorption line to be measured contains at least 10 resonance modes of the cavity. This makes it possible to average the noise linked to the discretization of the absorption line. The spectral width of the lines to be measured is between 2 and 10 GHz.

[0061] According to an advantageous example, the optical cavity 3 is confocal. The radius of curvature R C is equal to the length L of the optical cavity 3.

[0062] For a confocal cavity, the free spectral interval ISL cav is given by the expression c / 2L, where c is the speed of light. An optical cavity 3 of length L greater than 30 cm allows minimizing the free spectral interval ISL cav and thus to respect the above-mentioned condition of at least 10 resonance modes of the cavity. An optimal spectral distribution of the modes of the optical cavity 3 is notably obtained.

[0063] As seen in the figure, the Meas1, Meas2 signals are sent into reactor 10. The Meas1, Meas2 signals pass through the gas phase to be analyzed.

[0064] The light analysis system 4 is configured to analyze the light emitted by each light source 2 and passing through the optical cavity 3 by measuring the optical absorbance of at least one species of the gas phase passed through by the light.

[0065] The light analysis system 4 is a spectrometer. The spectrometer comprises a monochromator 40. The signals Meas1, Meas2 are spectrally separated by the monochromator 40.

[0066] The monochromator 40 comprises a ladder-type diffraction grating 400. The diffraction grating 400 is coupled to a secondary dispersion stage or diffraction order separation stage 401. The order separation stage 401 makes it possible to separate the different diffraction orders at the output of the monochromator 40. The diffraction order separation stage 401 is, for example, a prism or a grating.

[0067] In the example shown, the order separation stage 401 is arranged after (i.e. at the output of) the diffraction grating 400. Alternatively, the order separation stage 401 can be arranged before (i.e. at the input of) the diffraction grating 400.

[0068] Echelle-type diffraction gratings allow spectral resolutions of the order of pm to be achieved, comparable to the spectral widths of absorption lines. They are in fact optimized for very high diffraction orders, typically between 10 and 100.

[0069] The monochromator 40 has a focal length F with an entrance slit of width s. The resolution of the light analysis system 4, or instrumental resolution Δi, is as close as possible to the spectral width of the line to be measured. Preferably, the instrumental resolution Δi is less than the spectral width of the line to be measured. The instrumental resolution Δi is in particular of the order of a few tenths of pm to a few pm depending on the chemical species and the nature of the gas phase. This makes it possible to maintain a good signal-to-noise ratio SNR of the measurement.

[0070] As seen in Figure 4, the monochromator 40 has a free spectral interval ISL res corresponding to the spectral width of a diffraction order. If the spectral width of the light injected into the monochromator 40 containing the echelle grating 400 corresponds to , then the light is scattered over k diffraction orders. The free spectral interval ISL res increases when the wavelength λ of the incident light increases. For a blazed echelle grating at 63.43° with 100 lines / mm, the free spectral interval ISL res is of the order of 10 nm when the wavelength λ of the incident light is 400 nm.

[0071] The spectrometer is equipped with a camera 41. The camera 41 is configured to measure the signals at the output of the monochromator 40. Figure 4 shows, just like Figure 3, an example with two measured signals Meas1, Meas2. The wavelength of the incident light for each light source 2 is respectively denoted , . For the spectrum, only that for a first channel corresponding to one of the light sources 2 is represented. The reference signal for the first channel is noted I R1 and the measurement signal I M1 .

[0072] At each spectral position corresponding to an absorption line, the signal is reduced relative to the baseline due to vapor phase absorption. No reference channel is needed here. The reference intensity (I R1for the first channel for example) is directly deduced from the measured spectrum by adjusting the baseline. This greatly simplifies the assembly compared to HCL lamp sensors, and makes the reference signal more reliable, which by construction undergoes the same drifts as the measurement signal (I M1 for the first route for example).

[0073] The camera 41 may be a CCD camera or a CMOS camera. The camera 41 preferably has a read noise of less than a few e- RMS, and a low dark current (< 0.5 e- / pix / s) to maximize the signal-to-noise ratio of the SNR measurement. The camera 41 is preferably large to be able to measure as wide a spectral range as possible.

[0074] The control system according to the invention allows, thanks to the insertion of an optical cavity around the vapor or gas phase to be measured, the light to make numerous passages through the vapor. The number of passages of the light in the gas phase depends in particular on the reflectivity of the mirrors. The signal-to-noise ratio of the measurement and the accuracy of the measurement are thus improved, in particular for low absorbances.

[0075] The light analysis system allows high resolution and a wide spectral range to be combined in one acquisition on a camera.

[0076] This analysis system allows HCL lamps to be replaced by continuous spectrum sources that are more reliable, more stable and easier to integrate into an optical system.

[0077] The combination of all these elements results in a simpler, more precise and more efficient control system.

[0078] Bibliography

[0079] [Y. Du et al. 2014] : Du, Yingge, et al. "Self-corrected sensors based on atomic absorption spectroscopy for atom flux measurements in molecular beam epitaxy."Applied Physics Letters104.16 (2014): 163110.

Claims

System for controlling the deposition of thin layers by gas phase growth in a deposition reactor (10), the gas phase corresponding to a gaseous state or a plasma state, the control system comprising: a continuous light source (2); an optical cavity (3) comprising at least two mirrors, and configured so that light emitted by the light source is trapped in the optical cavity and interacts with the gas phase giving rise to the deposition of thin layers; and a spectrometer (4) configured to analyze light emitted by the light source (2) and passing through the optical cavity (3) by measuring the optical absorbance of at least one species of the gas phase passed through by said light. A control system according to claim 1, wherein the optical cavity (3) surrounds the gas phase. A control system according to claim 1 or claim 2, wherein the optical cavity (3) is confocal. Control system according to one of claims 1 to 3, in which the mirrors are assembled on optical accesses of the reactor (10), said mirrors being arranged inside or outside the reactor (10). Control system according to one of claims 1 to 4, in which the length (L) of the optical cavity (3) is chosen so that the spectral width of the absorption lines measured by the spectrometer (4) covers a minimum number of resonance modes, said minimum number preferably being equal to 10. Control system according to one of claims 1 to 5, in which the spectrometer (4) comprises a dispersion stage (400), the dispersion stage being an echelle-type diffraction grating. The control system of claim 6, wherein the dispersion stage (400) is configured to disperse at diffraction orders between 10 and 100. Control system according to one of claims 1 to 7, wherein the light source (2) is a light-emitting diode, a superluminescent diode, or an arc lamp. Control system according to one of claims 1 to 8, in which the spectrometer is provided with a camera (41), preferably a CCD or CMOS camera.