JPH02257676A - Wavelength selective photodetector - Google Patents

Abstract

PURPOSE:To drastically miniaturize a photodetector, and to lower electrostatic driving voltage by forming a groove through etching in a region between a light-receiving layer forming region and a spacer disposing region on the rear of a semiconductor substrate. CONSTITUTION:When the spacer 30 of a silicon substrate 10 is faced so as to be brought into contact with a glass substrate 20, electrodes 27 are superposed to the spacer 30 through outer peripheral sections 27a. The Si substrate 10 is anisotropic-etched, and a groove 80 having an approximately trapezoid longitudinal section shape is formed in a region 200 between a light-receiving region 100 and a spacer disposing region 300 on the rear 10b of the Si substrate 10. Consequently, the thickness of a substrate 10 section can be thinned down only by an etching amount, thus easily curving the substrate. Accordingly, a photodetector is drastically miniaturized, and low-voltage driving can be attained.

Description

Detailed Description of the Invention (a) Industrial Application Field The present invention relates to a wavelength-selective photodetector based on a Fapley-Bello type variable interference filter in which two substrates are integrated by bonding. The present invention relates to a wavelength-selective light-receiving element that can be used as a color discriminator, a light-receiving element in wavelength multiplex communication, a color time-division detection type image sensor, and the like.

(b) Conventional technology Generally, in this type of variable interference filter, reflective films are formed on two glass substrates, they are placed facing each other, and both substrates are bonded via a thin film spacer.
A void is formed that causes Fabry-Perot interference. Light interference occurs depending on the spacing between the gaps, creating a filtering effect that transmits only specific wavelengths. Furthermore, by applying force to one substrate and bending it to change the gap distance, it is possible to change the transmission wavelength.

This variable interference filter can select wavelengths without bending the optical path, so there is no need for mechanically moving parts such as a rotating mechanism, and the light emitting element or light receiving element can be installed in contact with the variable interference filter. This makes it possible to downsize the entire system.

Here, an integrated wavelength-selective light-receiving element in which one glass substrate of the variable interference filter is replaced with a semiconductor photodetector has been proposed (Japanese Patent Application No. 61-102
(See No. 989).

In other words, by integrating the light receiving element and the Fabry-Perot variable filter, there is no need to align them, making handling easier, as well as reducing light loss by reducing the number of interfaces between the elements. has.

(c) Problems to be Solved by the Invention A light-receiving element and a light-transmitting substrate are bonded together, and an element is created that performs wavelength selection by utilizing Fapley Bellows interference that occurs in a microscopic cavity formed between the two substrates. ), the important points are whether the cavity spacing can be controlled with an appropriate driving force, whether the cavities have good parallelism, and whether the device can be miniaturized.

However, in the conventional element described above, the spacer interval cannot be narrowed beyond a certain level due to the following reasons.
The device had to be made significantly larger than its effective light-receiving area.

The first reason is to ensure parallelism of the cavity. In order to change the cavity spacing, one of the substrates is simply curved within the area surrounded by the spacers. In this case, in order to maintain good parallelism of the cavity on the light-receiving area, the curvature of the substrate must be set small to ensure flatness of the substrate, so the spacer interval must be wide, which inevitably results in In this case, the element becomes considerably large compared to its effective light-receiving area.

The second reason is to reduce the force required to control the cavity spacing. Particularly when controlling cavity spacing using electrostatic drive methods (described below), increasing force increases electric field strength, creating a risk of dielectric breakdown. Another way to reduce the force is to reduce the thickness of the substrate, but in order to maintain the flatness of the substrate, the substrate needs to be thick to a certain extent. Since the required force decreases in inverse proportion to the cube of the spacer interval, the only way to keep the force small is to widen the spacer interval.

(d) Means for Solving the Problems This invention provides a semiconductor substrate having a light receiving section on one surface;
A light-transmitting substrate that is bonded to the semiconductor substrate so as to face the above-mentioned one surface via a spacer, thereby forming a cavity that causes Fabry Bellows interference, and a cavity spacing that allows the cavity spacing between both substrates to be made variable. a wavelength-selective light-receiving element, characterized in that the semiconductor substrate has, on the other surface thereof, a groove formed by etching in a region between the light-receiving layer formation region and the spacer arrangement region. It is.

That is, the present invention provides a wavelength-selective light-receiving element comprising a semiconductor substrate having a light-receiving portion on one surface, and a light-transmitting substrate bonded to the semiconductor substrate to face the one surface via a spacer. A groove is formed by etching a portion of the semiconductor substrate on the other side between the light-receiving region and the spacer region.

At this time, since the thickness of the substrate portion can be reduced by the amount of etching, the substrate can be easily bent. In particular, by using silicon as the semiconductor substrate constituting the light-receiving element and anisotropically etching it, the shape of the groove obtained can be well controlled and the reproducibility can be improved.

(e) In the working light-receiving area, the substrate is used as it is without being etched to ensure flatness of the substrate and parallelism of the cavity;
By etching the etching layer between the spacer region and the light-receiving region, the substrate can be easily bent, making it possible to significantly downsize the device and drive it at a lower voltage.

(f) Examples The present invention will now be described in detail based on examples. Note that the invention is not limited thereby.

FIG. 1 shows a schematic configuration diagram of a wavelength selective light receiving element according to an embodiment of the present invention. The manufacturing method of the device and each component will be explained based on FIG. 2.3.4.

FIG. 2(a) is a sectional view of the element immediately before bonding.

In FIG. 2(a, ), the substrate 10 has a plane orientation of (10
0), specific resistivity is 4Ω'cm, thickness W is 300mm and 1m
This is an n-type silicon substrate. Front surface 10a and back surface tob+: thermally oxidized S t O2 layers 13, 14 are formed, respectively, and p-type silicon layers 16, 17 . are formed by patterning the 5iOz layer 13 and performing boron diffusion.
18 are formed as a light-receiving layer, an electrostatic drive/capacitance monitoring electrode layer, and an etching stop layer, respectively. Here p
The mold silicon layer 18 is not necessarily necessary, and if it is not present, the Stow layer 13 itself functions as a stop layer during Si etching. Furthermore, 5102 layer 1
A spacer film-cum-electrostatic bonding anode (A1 layer having a thickness of 2.0 μm) 30 is formed on the substrate 3. In addition, the electrode 35 connected to the spacer 30 and the p-type silicon layer 16.1
Connect each electrode to 7 (A with a thickness of 500
1 layer) 165.175 (see Figure 1).

The substrate 20 is made of glass with a thickness E of 0.5R. This glass was Pyrex glass, which has a coefficient of thermal expansion close to that of Si. An electrostatic bonding anode 40 of an aluminum thin film is formed on the back surface 20b of the surface 20a of the substrate 20 facing the silicon substrate 10. On the other hand, on the surface of the surface 20a facing the silicon substrate 10, an electrostatic drive/capacitance monitoring electrode 27 (A1 layer having a thickness of 500 layers) is formed. At this time, the electrode 27
When bonded, the outer peripheral edge 27a slightly overlaps the spacer 30, and after bonding, it is connected to the outside via a pad on the silicon substrate side.

That is, on the silicon substrate 10, pads are provided for electrodes 165 drawn out from the light-receiving layer 6, electrodes 175 drawn out from the p-type diffusion layer 17, and electrodes 35 connected to the spacers 30. [See Figure 1].

After bonding, a cavity capacitor is formed between the electrode 27 and the p-9i layer 17, so the spatial spacing can be controlled by applying a voltage between both layers 27 and 17 to generate electrostatic attraction. , and also monitors the spatial distance d from the capacitance between the two 'r4 poles 27 and 17.

In this example, the electrostatic drive/capacitance monitoring electrode 27 is not formed in the area facing the electrodes 165 and 175, and a space S is provided by partially cutting out the electrode 27 [see FIG. )], this can prevent problems such as short circuits.

Reflective films 11.21 are formed on the surface 50 of the light-receiving region 100 of the Si substrate IO and on the surface 51 of the glass substrate 20 facing the region 100, respectively.

In this example, this reflective film is T i 01. Si
A nine-layer alternating multilayer film of Ot, M gF t, ZnS
Dielectric multilayer film consisting of Ag5Al, Au, etc.
A metal film such as may also be used.

The method for manufacturing the element will be described below.

(1) Both layers tfiElO, 20 are joined as follows.

First, as shown in FIG. 2(b), the spacer 30 of the silicon substrate 10 is opposed to the glass substrate 20 so as to be in contact with it. At this time, as described above, the electrode 27 overlaps the spacer 30 via its outer peripheral edge 27a. Subsequently, the spacer 30 and the substrate 1O are entirely grounded in a heated atmosphere (400° C.), and -300 V is applied to the electrostatic bonding cathode 40 for 10 minutes to perform electrostatic bonding. A cavity 99 that rises above this causes fapley bellow interference between the two layered plates 10° and 20°.
is formed.

(11) Next, the SF base film 10 is anisotropically etched.

That is, on the back surface tab of the Si substrate 10, a region 200 between the light receiving region 100 and the spacer arrangement region 300
A groove 80 having a substantially trapezoidal longitudinal section is formed in the groove 80 .

In preparation for this, patterning of a layer [4] of SrO is performed, which serves as a mask layer for Si anisotropic etching. Two squares are drawn inside the spacer 30 and outside the light receiving section, and the 5iOz layer 14 in the area surrounded by these small squares and large squares is removed. Each side of the square is <
It was set to face in the <ito> or <lot> direction. The shape of this patterning is not limited to a square, but may be circular. Note that this work may be performed before joining.

As another preparation, it is preferable to apply a sealing material to the end faces of the two-layer plates to, 20 that have been bonded, in order to prevent the sealing liquid from entering the gap 99 formed by the bonding.

Subsequently, both layers bonded as described above [1020 are Si
Soak in anisotropic etching solution. However, anisotropic etching means that the etching rate is at least 2 depending on the crystal orientation.
In general, the etching rate is faster on the (100) plane and slower on the (lit) plane, so the (111) plane that is formed once acts as the etching stop layer 18. It is characterized by the ability to obtain highly accurate etched shapes with almost no dependence on etching time.

Here, EPW (ethylenediamine/pyrocatechol aqueous solution) was used as an anisotropic etching solution, and
Etching was performed for 00 minutes.

Note that as the anisotropic etching solution, a KOH aqueous solution, a hydrazine aqueous solution, or the like may be used. When KOH is used as the etching solution, Stow is slightly etched, so it is desirable to form the sio and film i4 somewhat thickly. Furthermore, if the etched shape is not required to change so precisely, an isotropic Si etching solution (such as a mixed solution of butic acid and acidic acid) may be used.

Although this embodiment shows a method in which St etching is performed after bonding, the order may be reversed and bonding may be performed after S+ etching.

(iii) Next, form electrodes.

That is, after the Si etching is completed, the 5ift layer 14 on the back surface tob of the Si substrate 10 is removed, and the electrode 19 is formed on that surface. Note that even if the 5ift layer 14 is not removed, the electrode 19 can make electrical contact with the Si substrate IO at the etched portion of Si, so this removal is not an essential condition.

(iv) Furthermore, the substrate is diced.

That is, the glass substrate 20 is diced to divide the substrate 20, and each element is taken out. As shown in FIG. 3 (1), the device during dicing is made by forming a glass plate on the pad portion 91 for connecting the electrodes 165, 175, a, and s to the outside, in addition to the dicing grooves 81 necessary for dividing the device. Dicing grooves 82 are also formed for removing.

FIG. 3(b) shows a case in which dicing grooves 61 and 62 are formed from both sides so as not to penetrate through the gaps in order to prevent water from entering the gaps during dicing. The groove 62 can also be formed simultaneously with the anisotropic etching of the Si substrate IO.

(v) attaching the element to the stem;

The stem 83 on which the element is mounted is shown in FIG. The center of the element mounting part of this stem is hollowed out as shown in the figure, and S
It is designed not to come into contact with the area below the light receiving section 16 of the i-board 10. Only the Si substrate 10 below the spacer 30 is attached to the stem 83 using conductive paste.

Note that the shape of the stem is similar to that of the Si substrate 10 at the bottom of the light receiving section 16.
may be attached to the stem 83, and the St substrate immediately below the spacer 30 may not contact the stem 83.

Thereafter, the pads of the electrodes 35, 165, 175 and the lead wire pad 84 on the stem are bonded with a wire 86, and the transparent window 85 is attached to the stem 83 for hermetic sealing.

At this time, the inside of the package is vacuumed or filled with dry air, nitrogen, or discharge suppressing gas (SF, C2Fe, etc.). Is it desirable to perform hermetic sealing for the purpose of ensuring reliability and eliminating the influence of humidity?
Not a necessary condition. It should be noted that the device may be glued to the top of the stem 83 so that the glass substrate 20 of the device itself takes the place of the window 85 of the package, which has the advantage of reducing reflections at the interface. In this case, it is necessary to devise measures such as wiring to the pad and then bonding to the stem.

In this way, the element is created.

This embodiment can be further modified as follows.

First, a diffused type (PN type) silicon photodiode was used for the substrate IO, but low volume diffusion type, PNN" type, PI
You can use N-type, avalanche-type, etc. Even if the silicon substrate is p-type rather than n-type, anisotropic etching can be performed without any particular problem as long as the impurity concentration is normal. Furthermore, the substrate IO is Ge, SiC, or GaAs.

GaAlAs, GaAsP, GaP, InP.

Semiconductors such as [nAsP, [nGaAIP, etc.] are preferable.

Next, the substrate 20 is made of a light-transmitting material other than glass, such as quartz, sapphire, or transparent alumina, L i N b O3, L iT a O3, M
g O, acrylic, polycarbonate, etc. may be used. However, in the case of materials that cannot be electrostatically bonded, it is necessary to change the bonding method.

Further, it is desirable to coat the outside of the substrate 20 with an antireflection film. A dielectric multilayer film similar to the reflective films 11 and 21 can be used for this purpose.

Furthermore, the bonding method is not limited to electrostatic bonding.

For example, a low melting point glass or a low melting point metal such as lead, tin, or solder may be coated on the substrate 1O or the substrate 20, and may be bonded by heating it above the melting point.

The driving method is not limited to the electrostatic driving method. For example, a piezoelectric element or a coil may be installed outside the element, and the force generated therein may be applied to the light-receiving portion of the substrate 10 to make the gap interval variable. Further, capacitance monitoring does not necessarily have to be performed. If neither electrostatic driving nor capacitance monitoring is performed, the N2 27 and the p-type spreading layer 17 are unnecessary.

On the other hand, in order to suppress discharge between the electrode 27 and the p-type diffusion layer 17, an insulating film such as 5iOy may be formed on at least one of them. However, when an insulating film is formed on the electrode 17, in order to maintain electrical continuity between the spacer 30 and the electrode 27, the insulating film must not be formed in a region overlapping with the spacer 30.

In this embodiment, the electrode 27 and the p-type diffusion layer 17 perform different functions of electrostatic drive and capacitance monitoring at the same time, but they may be separated. For example, the electrode 27 may be shared, and the p-type diffusion layer may be provided separately for the electrostatic drive and capacitance monitor.

Further, the pads taken out from the electrodes 35 and 165 and 175 may be provided on the substrate 20 instead of on the substrate 10.

However, in order to do this, the spacer 30 is divided into several electrically independent regions after device formation, and each region is separated from the substrate i.
It is necessary to relay electricity from each electrical element (diffusion layer 16, 17, etc.) on the substrate 20 to the substrate 20.

The wavelength range used will be described later, but it is 700 to 160.
It may be other than near-infrared light of 0 nm (short wavelength region for optical communication). When using Si for the light-receiving element substrate and glass for the light-transmitting substrate, wavelength identification and detection of light in the range of 0.3 to 1.0 μm is possible simply by slightly changing the spacer spacing and reflective film thickness. . In addition, in order to use it in the wavelength region of 1.0 to 1.6 μm (long wavelength region of optical communication), base 1
The material of flO is Ge or rn-Ga-A1・As-
Compound semiconductor that combines some of P (described above)
etc. Furthermore, in the ultraviolet region of 0.3 μm or less and the far infrared region of 1.6 μm or more, we devised a combination of a photodetector substrate that is sensitive at that wavelength, an etching method for that substrate, and a substrate that is transparent at that wavelength. By doing so, the present invention can be applied.

The operation of this device will be described below.

In the regions 50 and 51 of the opposing reflective films 11 and 21, respectively, Fabry-Bello interference occurs depending on the reflective film spacing d, and as a result, significant wavelength dependence occurs in the transmission spectrum and reflection spectrum of this region. In particular, when the reflectance of the reflective film is high, only light with wavelengths near the wavelength at which the spectral transmittance is maximum is transmitted, and most of the light with wavelengths in that range is blocked. There are multiple maximum wavelengths of this spectral transmittance (referred to as selected wavelengths), which are given as follows. Here, m is the resonance order (m=1°2.
3,4,,,), n is the refractive index of the medium between the reflective films, φ
is the phase shift that occurs during reflection on a reflective film. Ignoring the wavelength dependence of φ, 1 m can be considered to be proportional to d. In this device, a wavelength-selective light-receiving device is realized by making use of this principle and making the cavity spacing of the interference region variable.

In order to improve the wavelength resolution, it is sufficient to increase the reflectance of the reflective film and set the resonance order m high. In this example, a Tie film and a SiO film were alternately used as the reflective film.
A laminated material is used, and its reflectance is 95% or more over the entire wavelength range used, and the phase shift φ is approximately 180'. The resonance order is determined based on the condition that a unique transmission peak exists in the wavelength range used. For example, if the wavelength range to be used is set to 780 to 90 nm, which is the short wavelength region for optical communication, and the resonance order m is 4 or less, the selected wavelength can be determined at any wavelength within the near-infrared region described in Section 2. Even if the wavelength is scanned, higher or lower order wavelengths are not included in this region. Therefore, in this example, m=4. Further, since vacuum is used as a medium, n=1. Under the above conditions, in order to scan the selection wavelength λ4 in the range of 780 to g90nm, d should be calculated to be approximately 15
What is necessary is just to make it variable in the range of 60-1780 nm.

Generally speaking, if you try to use it in a wider wavelength range, m
needs to be set small. For example, visible light region (40
The device that can be used in the wavelength range (0 to 750 nm) needs to be m-1, and the reflective film needs to be made of Ag, AQ, etc., which has a substantially constant reflective film over this wide range of wavelengths.

In this embodiment, the electrostatic attraction generated when a voltage is applied between the ii poles 17 and 27 is used as a means for making the cavity interval variable. Since this method does not require a driving element such as a piezoelectric element outside the interference filter, the overall structure of the element can be simplified and miniaturized, and the vibration resistance of the element can be improved.

The electrostatic force generated between the electrodes 17 and 27 is given by the following equation.

Here, ■ is the voltage applied between the electrodes, and ε. is the permittivity of vacuum, and ε is the relative permittivity of the interelectrode medium.

Note that since d decreases as F increases, the relationship between V and d becomes quite complicated.

Furthermore, in this embodiment, in order to adjust the selected wavelength to a predetermined value,
The capacitance C between the electrodes 17 and 27 is detected. This capacitance C is inversely proportional to the reflective film spacing d, so C and λ4
There is a one-to-one relationship.

Therefore, capacitance! ! By controlling the electrostatic drive voltage so that ic is constant, the selected wavelength can be stabilized.

In this embodiment, in order to use the electrodes 17 and 27 for both capacitance monitoring and electrostatic drive, an alternating current with a frequency higher than the highest frequency of the electrostatic drive signal is used as the capacitance detection signal, and the coupling capacitor is is used to separate the two signals.

Note that an electrode dedicated to capacitance detection and an electrode dedicated to electrostatic drive may be independently provided inside the element so that these signals may be mixed on the electrode.

Next, the characteristics of the variable wavelength light receiving element of this example will be described.

The greatest feature of this device is that silicon, which is easy to etch, is used as the substrate 10, and by precisely anisotropically etching it, the controllability of the gap distance is dramatically improved while maintaining the surface precision of the reflective surface. It is something that has been raised.

FIG. 5 shows a cross-sectional view of the element according to this example when it is driven.

In FIG. 5, it can be seen that the thinned portion 77 of the substrate 10 due to etching is greatly deformed, and as a result, the gap distance is uniformly reduced throughout the central light-receiving portion 100. As shown in FIG. 7, this device has a rectangular shape with length L of 31 m and groove R of 4+u, and the thinnest part 77 where the silicon substrate IO is etched has a thickness H of 1 μm and a length. It is a square with a length F of 160 μm and a width of K = 2 μm square, and the area of the electrostatic drive electrode is 3.7511. Further, the opening length J of i*80 is 600 gm.

The relationship between the driving voltage and the gap distance at this time is shown in FIG.

It can be seen that the entire required wavelength range can be scanned with a driving voltage of 16 to 2 QV. Note that the electrode spacing before voltage application is 1.7
If it is set to 8 μm, it is possible to drive at an even lower voltage, but considering manufacturing variations in the electrode spacing M, the design value is changed to 2.
It was set as μm. In FIG. 7, only M has the unit of "μm", and the rest have the unit of r zyt j.

For comparison, silicon substrate 1 without any particular etching process
8 is a cross-sectional view of the wavelength variable light receiving element when driven using 10.
As shown in the figure. The size of this element is that the vertical Q is 2rtrx-'
f11r is elongated with a 20π shoulder, and spacer 13
0 is formed only at both ends in the lateral direction. The inner diameter n between the spacers is +6xm. During driving, the entire portion 400 of the substrate 110 sandwiched between the spacers 130 is curved, so that the area where the gap 401 is considered constant is the central portion 40.
It becomes a small part of 1a. Furthermore, since the substrate is thick and requires a large force for bending, the spacer spacing is widened to avoid a significant increase in the electrostatic drive applied voltage. However, this increases the size of the entire device and increases the cost of the package (preferably hermetic seal). FIG. 6 shows the relationship between drive voltage and gap distance. At this time, scanning can be performed within the applied voltage range of 70 to 88V. However, due to the high voltage, many elements suffer from dielectric breakdown due to discharge.

(G) Effects of the Invention As explained in detail above, according to the wavelength-selective light-receiving element of the present invention, the element can be significantly miniaturized, resulting in cost reduction, and the electrostatic drive voltage can be reduced. This can solve the dielectric breakdown problem.

Therefore, the present invention is expected to be widely used in small-sized portable spectrometers, color discriminators, color time-division detection type image sensors, wavelength variable light receiving elements for optical communication, and the like.

[Brief explanation of drawings]

FIG. 1 is an exploded configuration diagram of a main part of a wavelength-selective light-receiving element according to an embodiment of the present invention, and FIGS. 2(a) and (b) respectively show the wavelength-selective light-receiving element during manufacture according to the above embodiment. FIGS. 3(a) and 3(b) are explanatory diagrams of the main part configuration for explaining the dicing method of the wavelength-selective light-receiving element according to the above embodiment, respectively, and FIG. A configuration explanatory diagram showing a situation in which a wavelength-selective light-receiving element is attached to a stem, FIG. 5 is an explanatory diagram of the configuration when the wavelength-selective light-receiving element according to the above embodiment is driven, and FIG. 6 is a diagram showing wavelength selection according to the above embodiment. FIG. 7 (aXb) is a characteristic diagram showing the applied voltage-selective wavelength characteristics of a photodetector and its comparative example, and FIG.
Xc) is an explanatory diagram of a main part configuration showing a conventional example. 10... Silicon substrate, 20... Glass substrate,
l!・21... Reflective film, 16... Light-receiving layer (p-type diffusion layer), 17... Conductive layer for electrostatic drive/capacitance monitoring (p-type diffusion layer), 18... Anisotropy Etching stop layer (p-type diffusion layer)
, 27... Electrostatic drive/capacitance monitoring electrode, 30...
・Spacer, 40...electrode for electrostatic bonding. Tube Diagram 2 (a) Chassis 4 Diagram of the drawer (a) Lever diagram

Claims (1)

[Claims] 1. A semiconductor substrate having a light-receiving portion on one surface, and a light-transmitting substrate that is bonded to the semiconductor substrate to face the one surface via a spacer, thereby forming a cavity that causes Fabry-Perot interference. and a cavity interval variable means capable of varying the cavity interval between both substrates, the semiconductor substrate being formed by etching in a region between the light-receiving layer forming region and the spacer disposing region on the other surface. A wavelength-selective light-receiving element characterized by having a groove.