DE1923253A1 - Semiconductor device with metallic contacts and / or metallic layers - Google Patents

Description

DII-IN *. D. * U.-IK «.M..C.

OIFL .- ^ Y «. UH,

HÖGER - LEGAL RIGHTS - SRtESSBACH - HAECKER

PATENT LAWYERS IN STUTTGART

A 37 309 b ■ April 28, 1969 b-35

Texas Instruments Incorporated 13500 North Central Expressway Dallas, Texas, USA

Semiconductor device with metallic contacts and / or metallic layers.

r.Ui invention relates to a semiconductor device with under Use of metallic contacts made from molybdenum and / or metallic layers on the semiconductor surface and / or on an insulating layer over the semiconductor

909850/1284 ORIGINAL BATHROOM

A 37 509 b - 2 -

April 28, 1969
b-35

surface. In particular, the invention does not relate to fully hermetically sealed integrated circuits.

In the manufacture of semiconductor devices, there has always been a desire to find better and cheaper ways to accomplish them Encapsulate semiconductor devices. Until recently, it was common practice to mount semiconductor devices on a header made of metal and glass and then to cover them with a metal cylinder closed on one side. In recent times, plastic housings have prevailed, since they are less expensive than the previous packages, their cost often exceeds the cost of the semiconductor device exceeded itself.

Currently, transistors, diodes and integrated Circuits are often encapsulated in plastic. Preferred types of plastic are epoxy or silicone polymers, with which the semiconductor devices are encapsulated. It is somewhat more expensive to use a firmly adhering organizational

Ur

see an adhesive such as an epoxy resin to attach metallic cap on a ceramic carrier. Regardless of the special type of encapsulation however, it has been found that the closure is not completely hermetically sealed as it is in use of metal and glass to encapsulate transistors is the case where the leak rate is in the Order of 10 "cm / see or even less in the case of helium. Plastics are relatively permeable to a wide variety of gases, and it has to be It should be noted that the penetration of the atmospheric gases surrounding the semiconductor device along the interfaces between metallic lines and the plastic

9 O ^f ) η h υ / i 2 f]

BATH ORIGINAL

A 37 309 b - 3 -

in the direction of the active component is even more serious.

In today's stabilized planar semiconductor devices, the penetration of gases from the surrounding Atmosphere in the cladding is less of a serious problem in terms of surface changes of the semiconductor device itself, however, the thin metal layers corrode extremely easily as a result of the ingress of gases Establish contacts with certain semiconductor areas and with the help of which interconnections can be established; particularly Water vapor that has penetrated the envelope is dangerous. The corrosion of the thin metal layers plays a role a relatively minor role in the case of individual devices, since these only have relatively small metal film layers have, but the corrosion is extremely detrimental at the transition points between feed lines and connection points, if different metals were used, which then practically form a galvanic element form.

However, a monolithic integrated circuit has one larger number of active and passive components, such as transistors and resistors, created by diffusion in the surface area a semiconductor die have been formed; on the surface of the latter is usually located an insulating layer on which layers of metal are in turn attached to the various components to connect to one another in the desired manner through openings in the insulating layer. Often there is a large number of individual elements of the semiconductor die must be connected to one another, the length of the thin Metal layer on the surface of a single semiconductor die considerable and much larger than at

9Q9850 / 1284 BATH ORIGINAL

A 57 309 b - 4 -

an individually encapsulated semiconductor element. Eb is natural that the risk of destruction of the device as a result of corrosion is greater, the greater the attack Surface of the metallic connections is. The risk of destruction is particularly great with "Large Scale" Integration "semiconductor devices, typically having interconnections arranged in multiple levels, between which electrically insulating layers are provided; the uppermost metallic connection B layer is then always exposed to the penetrated gases and vapors, while the underlying metallic connecting layers are protected by the covering insulating layers.

Om estimate the quality of the encapsulation using plastics BU becomes the semiconductor device in a usual test brought to a room with an elevated temperature and high humidity, all in one typically over a period of about 1000 hours. In addition, there is a voltage during the test period applied to the semiconductor device. Destruction in the form of circuit interruptions is then common the consequence of corrosion of the connecting lines. Just the integrated circuits are particularly susceptible to this, because relatively large areas of a material made, for example, of silicon Semiconductor wafer are covered by electrochemically active metal layers. Though there is considerable doubt whether such a test is actually a measure of the susceptibility of the semiconductor devices in real use represents, this type of test is widespread and represents a serious hurdle for all manufacturers of ■ Plastic encapsulated semiconductor devices.

909850/12

A 37 309 b - 5 -

Common metals used to make contacts on semiconductor devices are aluminum, molybdenum and tungsten. Aluminum is preferred in semiconductor devices used with a single element, while tungsten and molybdenum in conjunction with gold are excellent metal systems for integrated circuits supply. In semiconductor devices with a single element and with pads Made of aluminum, gold wires are usually used to connect the aluminum connection points to the metal leads to connect which lead to the outside world. Will aluminum in a non-hermetically sealed device used, ion currents occur between the various metals aluminum and gold as soon as water vapor has penetrated into the interior of the encapsulation and then forms an electrolyte, which is usually sufficient Has conductivity. The combination of aluminum and gold represents a particularly active galvanic element with a voltage of about 3 volts.

At the beginning of the reaction, the aluminum forming the anode oxidizes to Al, while hydrogen at the cathode development takes place. The aluminum ions thus released react immediately with water according to the following Equation:

2Al + 3H _o 0-> Al _o 0 _x + 6H ⁺

whereby insoluble and insulating Al ₂ O * is formed .. The creation of this insulating protective skin naturally reduces the reaction speed and thus protects the anode from further dissolution to a certain extent. However, this anodically formed oxide is still relatively permeable, and irregularities in the

909850/1284

A 37 309 b - 6 - 1 QO Q O C ο

28 ·. April 1969 I 3 ^ J Z D J

Oxide layers also lead to the oxidation process continuing. Usually an attack then takes place at strictly localized points near the cathode, so that pores arise, with the aluminum being broken down and transported away _{in the form of AlO 2 ions.}

However, the aluminum can also corrode in other ways. Since the solution becomes basic near the cathode, nearby areas, in which no tension arises, dissolve, correspond to the following reaction to:

Al + 20H ~ -> A10 ₂ + 2H ⁺ + 3e '

In this case, no protective skin is formed. This reaction consequently continues until the line is interrupted. The dissolved aluminum does not settle as Metal at the location of a cathode, since the hydrogen evolution. treatment is preferred. If you put an external tension on the System, this accelerates the evolution of hydrogen in the cathode which forms a cathode as a result of the bias Areas while the negatively biased metal areas will not corrode. The reactions to the Anodes are accelerated, and there is now also an evolution of oxygen, with part of the oxygen migrates to the cathode, where it is again reduced to water. If aluminum wires are used instead of gold wires, bo one prevents the formation of a galvanic element in this way, but such an aluminum line also corrode when an external voltage is applied.

The molybdenum-gold system behaves somewhat differently. Since some oxides of molybdenum are water-soluble, so

- 7 -

909850/1284

A 37 309 b - 7 -

April 28, 1969

For example, the water-containing form of Mo ₂ Oc, molybdenum is not passivated as quickly as is the case with aluminum. As a result, this system itself forms a galvanic element again and quickly shows signs of corrosion, with the molybdenum dissolving on the anode until an electrical interruption occurs. If a voltage is also applied, the reactions at both electrodes run faster. As long as one does not apply relatively high bias voltages in the order of magnitude of more than 5 volts to the electrodes, the evolution of oxygen plays practically no role, because the dissolution of the molybdenum is electrochemically more favorable. This shows the following listing of the potentials of different electrodes:

Mo- * Mo ⁵⁺ + 3e ~ 0.2 V

Mo + 3H ₂ 0-> Mo0 ₅ + 6H ⁺ 6e ~ -0.1 V

40H ~ - * 2H ₂ 0 ₂ + 4e ~ -0.4 V

2H ₂ 0- * 4H ⁺ +0 ₂ + 4e ~ -1.2 V

In the case of high external voltages, the gold also dissolves at anodic points, but since the oxidation potential of gold is rather strongly negative, it predominates the evolution of oxygen. Nevertheless, there will be some gold on the anode according to the following equation dissolve:

Au + XH ₂ O- »Au ⁺ (aqJ + e ~ -1.68 V

Gold naturally forms stable oxides, so that the dissolution of the gold at anodic points leads to an over the entire anodic surfaces are evenly removed

909850/1284

A 37 309 b - 8 -

April 28, 1969

tion leads and as a result no pores occur. The gold ions become the closest cathodic in the electrolyte Where metallic gold is deposited.

Mainly because of the considerably lower solubility of tungsten oxides in aqueous media - compared to the Molybdenum oxides - the tungsten-gold system is considerably more corrosion-resistant than the molybdenum-gold system. So is for example the Oxyd WPv only in hot solutions with extraordinary high pH-value, while the corresponding molybdenum oxide MoO, already in water at room temperature in Lö-r sung goes. As a result, the corrosion rates set in motion by a self-generated voltage are proportionate low, since the tungsten passivates itself. The application of an external voltage mainly leads to an electrolysis of the water and a slight dissolution of gold as a negligible anodic reaction; tungsten only slowly dissolves in anodic areas with a high pH value.

From the previous discussion of corrosion processes it can be seen that electrical contacts forming transitions from the metal-gold system as well as corresponding interconnections for semiconductor devices as they can be used in hermetically sealed enclosures, not suitable for use with calibrated completely tightly encapsulated Devices are suitable. It is necessary that the materials used have good adhesion aa silicon and Show silicon oxide and no intermetallic bonds occur between the metal of the transition and the The metal forming the transition must not corrode, or at most only the slightest signs of corrosion can occur.

9 -

909850/1284
BATH ORIGINAL

A 37 309 b 9 -

genes are tolerated in not fully encapsulated semiconductor devices, even if the environment is a has high humidity.

The invention is based on the problem of these corrosion phenomena in the case of not completely hermetically encapsulated To avoid semiconductor devices, and starting from a device of the type mentioned in the introduction, this object is achieved in accordance with the invention achieved in that the contacts or the metallic layers in addition to molybdenum a modifier metal or contain metalloid, which is more resistant to corrosion than molybdenum. This system is particularly suitable for those semiconductor devices that are exposed to high levels of moisture, because it will then also join not hermetically sealed encapsulation does not show any signs of corrosion. The specified combination of materials is liable also good on silicon and silicon oxide. Such a barrier-forming metal for the improved electrical contacts and electrical interconnections do not form intermetallic compounds with gold.

Further features and details of the invention emerge from the attached claims and / or from the following Description, which explains the invention with reference to some embodiments shown in the drawing serves; show it:

Fig. 1 is a plan view of a semiconductor plate, in which a planar transistor has been formed and that has an insulating layer on its surface has, the openings for attaching contacts hatj

FIG. 2 shows a section along the line 2-2 in FIG. 1;

- 10 -

208850 / 12ΐ

A 37 309 b - 10 - _q?

April 28, 1969 I 3 <£ O Z 0 O

Fig. 3 is a schematic section through a

Device for high-frequency sputtering, such as those used for attaching the contacts on the Semiconductor wafers can be used;

Pig. Figure 4 is a plan view of the Pig die. 1 after attaching contacts and connection points;

Pig.5a a Schnxtt and a plan view of the semiconductor wafer shown in and ο Pig.4 after attachment of the leads and being embedded in plastic, whereby the cutting direction of the Pig.5a is indicated by the line 5a-5a in 5 b. However, the upper part of the plastic capsule has been cut off in FIG. 5b in order to be able to show the device more clearly;

Pig. 6 shows a section through an integrated circuit with only a single plane lying connections;

Pig. 7 is a plan view showing the arrangement of FIG

Shows circuit elements in one of the functional units of a semiconductor die, that is shown in Figure 8 and requires interconnections in several levels;

8 shows a plan view of this semiconductor wafer which contains a plurality of puncture units;

9 shows a circuit diagram of one of the puncture units of Fig. 7 and

Fig. 10 several sections to explain the production the integrated circuit shown in Figure 8 corresponding to the line 10-1 in Pig. 8.

9 oy iuu ^; / "2 aa

A 37 309 b - 11 -

The invention provides contacts and interconnections for Semiconductor devices that are corrosion resistant and either non-encapsulated or non-hermetic encapsulated semiconductor devices can be used to advantage. A homogeneous mixture is specified Amounts of molybdenum and a modifier metal are used, which has greater corrosion resistance than molybdenum has to increase the corrosion resistance of the latter, The advantages achieved by the use of molybdenum are described in U.S. Patent 3,341,753. It is worth noting that molybdenum is extraordinary adheres well to silicon and silicon oxide so that it does not form intermetallic compounds with gold and acts as a barrier between gold and the semiconductor material represents what is important when considering gold as a coated Metal used where it is both a protective layer and a low resistivity conductor.

at
One of the requirements that must be made of a modifier metal is that it must be more corrosion resistant than molybdenum with which it is to be used. This means that the modifier metal may only be slightly soluble in aqueous acids - with the exception of elutic acid; furthermore, the oxides of the modifier must be more stable than the oxides of molybdenum, and they must be only slightly soluble in a wide variety of acids. The modifier metal must also be able to form passivating oxide layers at a wide variety of oxidation potentials; finally, the modifier must be highly resistant to corrosion. But it should also have a relatively high melting point and low self-diffusion,

- 12 -

9098 50/1284

April 28, 1969 ^{| 3} " ^Z0J

However, none of the modifier metals that meet the above requirements are metallurgical with gold sufficiently stable to be used as a barrier layer in a metal-gold contact.

The ideal modifier metal is insoluble in molybdenum and does not form compounds with it. However, every metal is that meets the requirements outlined above, either soluble in molybdenum or forms a compound with it. The requirement that the modifier metal have no compounds may form with molybdenum and should not dissolve in it, with regard to the etchability of the Mixture of molybdenum and the modifier metal, because the etchability changes with the percentage of one added element in a mixture. Etching becomes more and more difficult, the greater the percent solubility, and compounds that form interconnections are most difficult to etch once the desired circuit pattern is formed shall be. Examples of modifier metals having all of the desired properties except that they do not form a compound with or are soluble in'Molybdenum, are the following elements: titanium, tantalum, chromium, zircon, Hafnium and silicon.

The preferred modifier metal is titanium, which is soluble in molybdenum in any ratio, but not soluble in molybdenum VeroirJungen educates. The remaining modifier metals mentioned above are also suitable for the purpose according to the invention, but they are made in their order Less preferred for the following reasons: Tantalum and chromium do not form compounds with molybdenum, but they are soluble in molybdenum in any proportion. The higher oxides of chromium are water-soluble and therefore not so corrosion-resistant if you have a high positive preload

909850/1284

A 37 309 b. - 13 -

April 28, 1969

application. Zircon forms compounds with molybdenum, as is the pail with hafnium and silicon.

A homogeneously mixed layer of molybdenum and a modifier metal, which is preferably titanium, can easily be opened by high frequency or triode spray using cathodes produced using the usual powder metallurgy technology.

The lower limit of the proportion of the modifier metal in the mixture with molybdenum is determined by the fact that this proportion of the modifier metal increases the corrosion resistance of the molybdenum to a sufficient extent. Incidental impurities with a small amount of the above-mentioned modifier metals, as they can unintentionally get into the molybdenum, have no effect on the corrosion resistance of the molybdenum from which the contacts are formed, for example. So a noticeable amount of modifier metal has to be intentionally added to the molybdenum. In practice, the lower limit for the proportion of the modifier metals is around 3% by weight. The passivation or corrosion resistance of the contact increases with the proportion of the modifier metal in the mixture with molybdenum. Exceeds dsr proportion of the modifying metal, however, approximately 20 i "at this value shows an excellent corrosion resistance of the contacts - that the modifier metal also brings adverse properties of the mixture with them.

The upper limit for the proportion of the modifier metal in the mixture with molybdenum is determined by two things:

-H-

90985GM284

A 37 309 b - 14 -

April 28, 1969

Firstly, by the proportion of modifier metal that the molybdenum matrix can absorb without the modifier metal reacting with the gold layer and consequently increasing its specific resistance, and secondly by the etchability of the layer, which increases with the proportion of modifier -Metal is decreasing. Furthermore, it becomes difficult to precisely define the metallic contact of molybdenum and modifier metal when the modifier metal reacts with the adjacent gold layer. The mixture of molybdenum and modifier metal is extremely difficult to etch if the proportion of the modifier metal exceeds about 35 i> , and the mixture becomes metallurgically unstable and leads to a reaction of the modifier metal with the gold when the proportion of the Modifier metal is greater than about 60 #. Contacts containing more than about $ 35 or 60 # of modifier metal may still be used, but more care is required and more time is required to manufacture as it becomes more difficult to get the individual contacts and interconnects in the correct shape to manufacture. Most conveniently, about £ 20 modifier metal content. The percentages given are averages and do not exactly apply to every modifier metal, since they differ slightly in their properties, so that the percentages given are shifted.

The Pig. 1 and 2 show a semiconductor die 10 in which a transistor has been formed which has a base 11 and has a bitter 12 and wherein the remaining plate is the Collector 17 forms. It was created using the usual planar technique, namely by successive diffusion of donors and acceptors and the intermediate formation of silica masks. In detail is on

- 15 -

909850/1284

the publication "Integrated Circuit-Design Principles and Fabrication" by Warner and Fardemwalt, McGraw-Hill (1965) referenced; Further details can be found in the publications: "Semiconductor Technology", McGraw-Hill (1965) " \ " Physics and Technology of Semiconductor Devices "by Rover, Wiley & Sons (1967). ■ "

In the planar process, an oxide layer 13 is formed on the surface of the platelet 10, the layer thickness must be thicker over the collector 17 than over the base, so that the usual stepped training results. Pure high frequencies, the active parts of the transistor should be extremely narrow, so that the elongated emitter 12 for example 0.05 mm wide and less than 1/4 mm long. The base 11 then has an area of, for example about 10-2 mm. Two openings 14 and 15 are provided for the base contacts, while one opening 16 is provided for the Emitter contact has been placed; the latter is the same as that used for diffusing in the emitter became. Because of the extremely small size of the base and emitter contact areas, the contacts themselves must pass over the silicon oxide layer be expanded so that lead wires can be attached. The semiconductor die itself has for example an edge length of 0.75 mm and is 0.1 mm thick.

To create a layer of a mixture of molybdenum and a ' To apply ModifikatopMetall on the plate 10 - this layer forms an emitter contact 18 and a base contact 19 (see Fig.4) -, the plate 10, which is part of a larger semiconductor wafer, together with a Number of other such disks mounted on a carrier plate 20 in a conventional HF spray device 21 (see Figure 3). For the sake of simplicity, the formation of the molybdenum

- 16 -

909850/1284

April 28, 1969

Titanium contacts described. Contacts from other compilations can be made in the same way. The carrier plate 20 serves as an anode in the high-frequency spray circuit, while an atomizer plate 22 is provided as the cathode, which contains the metal on the semiconductor wafer 10 should be knocked down. It goes without saying that there are several atomizer plates, if different metals are to be knocked down, in the present case, the atomizer plate 22 consists of the desired Mixture of molybdenum and titanium. It can be manufactured in the usual way by powder metallurgical processes; ' but you can also buy them from manufacturers such as Materials Research Corporation. The atomizer plate 22 consists of a homogeneous mixture of molybdenum and titanium powder, so that any composition is possible. It is carried by a holding plate 23, which is electrically via a switch with a not HF generator shown is connected.

To deposit the gold, a gold atomizer plate 24 is attached to a holding plate 25, which in turn is also connected to the HF generator via a switch. Of course, Figure 3 only represents is a schematic drawing so that the arrangement of the various elements is not exactly taken from it werder can. In order to obtain uniform metal layers on all semiconductor dies, each sputter plate must be at least as large as each die to be coated, and the spacing between each atomizer plate and the surface of each semiconductor chip be equal. As a result, each sprayer has a device within its chamber 26, with whose help the selected atomizer plate for the respective

- 17 -

909850/128 /;

A 37 309 b - 17 -

April 28, 1969

ligen crackdown process can be adjusted.

For example, argon is now fed into the sprayer through an inlet 27 'at a pressure of about 5 to 15 yes Hg 21 passed, whereupon the high-frequency energy is applied between carrier plate 20 and holding plate 22; as appropriate a frequency of about 15 MHz has been found and the process of precipitation should last that long be that a molybdenum-titanium layer is formed on the semiconductor wafer 10, which is approximately 2500 S thick. The HP energy is then switched off and applied between the carrier plate 20 and the gold atomizer plate 24. It then remains switched on until there is a gold layer, for example 10 000 Ül, on the molybdenum-titanium layer originated. HF energy is then switched off, the flow of argon is interrupted, and the semiconductor wafer 10 taken from the device 21. Of course, the molybdenum-titanium layer can also be applied in another way, for example by atomization. Gleiolies applies to the gold layer. But you can also use the usual * vapor deposition technique grab if this seems cheaper.

Then the excess parts of the gold and Molybdenum-titanium layers are removed by cutting the semiconductor wafers selectively etches, particularly using photoresist masks. This is particularly suitable for this Photo resistance material EMEH from Eastman Kodak, which in a certain way with ultraviolet light through a mask is exposed through at the points where the gold and the molybdenum-titanium layer are to remain. The unexposed Parts of the photo resistive layer are then through removed a demineralizer. The photo resistance mask is then only above those parts of the gold

- 18th

909850/1284

April 28, 1969 »^ ο ^ j ο

and molybdenum-titanium layer, which are to form the emitter and base contacts as well as the extended lines, such as she lets the Pig.4 recognize.

The semiconductor wafer is then etched in order to cover the parts of the gold layer 27 that are not covered by the mask remove. A suitable etchant for gold is an alcoholic solution of iodine and potassium iodide. The now exposed Parts of the molybdenum-titanium layer 28 are now also etched away, for example by means of a basic solution made of potassium ferricyanide, so that the contacts 18 and 19 arise, which consist of the firmly adhering layers 27 and 28} the molybdenum-titanium layer 28 adheres also extremely good at the silicon oxide layer 13 and the surface of the semiconductor die 10. The latter is now ready to be assembled and packaged can. An example of a non-hermetically sealed encapsulation is a plastic casing 5 like them Figures 5a and 5b show. After the semiconductor die has been attached to a collector lead 9, a gold wire 6 is connected to the base contact 19 as well as to a base lead 8 connected, for example with the help of so-called "ball bounding", and another gold wire connects between the emitter contact 18 and an emitter lead 10. Then the plastic casing 5 is over the Semiconductor wafer 10 is formed.

In order to ensure a good obm's contact low resistance between To create the silicon and the molybdenum-titanium layer, it is necessary that the surface area of the Silicon in which the contact is to be formed, one has a high density of impurities and is either n- or p-type; As is known, this applies to all contacts

- 19 -

909850/1284

A 37 309 b - 19- ■ _1QO ", _O

Difficult-to-melt metals, used as a doping material either boron or phosphorus, the surface concentration should

1Q / 3

tration greater than 2x10 ⁷ atoms / cDr, especially large

21

Ser than 10. Of course, electrical contacts can also be used to silicon surfaces with a lower density of impurities are generated, but the contact resistance increases on contact with the decrease in the concentration of impurities.

In typical transistors, the η-conducting emitter usually has a very high concentration of impurities, especially in the area of its surface, since the emitter zone has been created by a second diffusion. The base zone usually has a lower density of impurities, however, it also has a high degree of doping, at least on the surface, so that contact is low Resistance can be produced. If this is not the case, before the contact material is deposited In a further diffusion process, a flat p-conducting diffusion zone is additionally created. The diffusion. can be made through openings of approximately the same size and approximately in the same place as the openings 14 and 15, the have been formed for the production of the base contacts, and these are preferably the same openings. In the case of integrated circuits, there are special diffusion steps even more necessary to generate high surface concentrations in the area of the contacts. This is because of the Collector contact is created on top of the semiconductor die in an area that is epitaxial Layer of low impurity density or act around the first diffusion zone in a triple diffused area can, which usually has a relatively low concentration of impurities, to the two following diffusion steps to be able to perform. Furthermore is also

- 20 -

909850/1284

the base region of the transistor or an integrated circuit usually simultaneously with the formation by diffusion generated resistances. Since the specific resistance of the material that forms these resistances is relative should be high, the impurity density in the base zone must be quite low. As a result, it should the concentration of dopants in the area of the collector, the base and the resistors are increased where Contacts must be attached.

The Fig.6 shows an integrated circuit in section, the Interconnections are only required in one level. It is formed in a p-type silicon wafer 30 and comprises a transistor at the left end with collector 31 »base 32 and emitter 33» it is produced by diffusion is an n-pn transistor. At the right end there is a resistor in an insulating zone 34, the resistor is itself formed by a p-conducting diffusion zone and is denoted by 35. Before the second diffusion step is carried out for introducing η-conductive impurities, i.e. before the formation of the emitter 33, is carried out in a Insulating layer 36, which may consist of silicon oxide, for example, has an opening introduced at the point at which the collector contact is to be formed; As a result, a diffusion region 37 with a high n + impurity density is formed at the same time as the emitter 33. Diffusion area 38, 39 and 42 with a high p + impurity density are then produced by selective diffusion of boron, wherein the insulating layer 36 can be used as a mask. Then in the insulating layer in the usual way by means of a poto resistor layer and an etching process created openings, namely at those points where contacts to the transistor and the resistor to be produced. Finally, in the manner described, a

- 21 -

909850/1284

A 37 309 b - 21 -

April 28, 1969

Molybdenum-titanium layer 40 is applied and this is covered by a gold layer 41, whereupon these layers selectively are removed, whereby the desired line and contact pattern arises. The collector 31 is connected to one end of the resistor by an interconnection 39, which runs over the insulating layer. This interconnection has contacts and interconnections just like the rest of the connections The integrated circuit has a layer 40 made of a mixture of molybdenum and titanium, which is attached to the insulating layer 36 and the exposed surface parts of the semiconductor wafer 30 firmly adhered, as well as a gold layer over it 41, which in turn adheres firmly to layer 40.

A particular diffusion of p + impurities to produce a low resistance at the contact can then become unnecessary if an extremely thin PiIm made of aluminum or platinum is applied to the silicon surface and im Case of platinum is sintered into this, so that platinumilioid arises before the molybdenum-titanium layer is applied.

Pig.7 shows an integrated circuit, the interconnections needed in more than one level. A semiconductor wafer 70, for example made of silicon, has a number of functional elements formed in it. The figure shows 16 such elements, however, there could be many more. Each of these functional elements 71 to 86 contains a number of transistors, resistors, capacitors and the like, which when connected together produce the desired circuit. So For example, the functional element 73 can be the one shown schematically in Pig * 9 and in the plan view in Pig.8. Circuit include. In the example shown these are p-n-p transistors 90 bis

- 22 -

909850/1284
BATH ORIGINAL

A 37 309 b - 22 -

April 28, 1969

andn-p-n transistors 94 to 100; there are also three input ports A, B and X and an output connection G are provided. These connections as well as a voltage supply connection V correspond to the connections marked with the same letters in Pig. 7

The connections B ₁ D , J and 0 of the functional elements 73, 76, 81 and 86 are connected to one another by means of an intermediate connection 87. In the same way, the connections V, F, L and R are connected to one another by means of an interconnection 88 and the connections X, H, M and Q are connected to one another by means of an interconnection 89. If one considers that there are already a large number of electrical interconnections in a first level connecting the various transistors to one another and to other circuit elements and connections, it can be seen that the interconnections 87 to 89 must necessarily cross some of the interconnections of the first level, such as this can be seen in Pig.8. For this reason and since the interconnections for the functional elements are expediently made separately from those for the individual circuit elements, they are formed in a second level according to Pig.7 and an insulating layer is placed between them and the interconnections of the first level.

The transistors and other switching elements can be formed in or on the semiconductor wafers 70 in a known manner, for example using epitaxial technology or by diffusing in doping substances. FIG. 10 shows

- 23 -

_T

909850/1284

BATH ORIGINAL

A 37 309 b -23-

April 28, 1969

in section a part of the integrated circuit of FIG before applying any metallic interconnections. An n-p-n transistor 94 comprises an η-conducting collector, which is formed by the semiconductor wafer itself, as well as a diffusion-created p-type base 110 and an η-conducting emitter 111. A resistor R .. is given by / P-type and diffusion-created zone 112 is formed, which is formed simultaneously with the base 110 of the transistor 94 generated. An insulating layer 113, which can consist of silicon oxide, for example, lies on the surface of the semiconductor wafer and is graduated, as the drawing shows, what of the successive Diffusion steps originates. Openings are then formed in the insulating layer, followed by those in the first To be able to ohmically connect level metallic interconnections to the semiconductor components.

In a next step, a thin and, for example, approximately 1200 Å thick molybdenum film 116 is formed on the surface the insulating layer 113 is deposited, which is in ohmic contact at the openings of the insulating layer with the semiconductor material is located. The film 116 can also consist of a mixture of molybdenum and titanium, however, the corrosion resistance of this mixture is not yet required for this film, as it is of a further one Insulating layer 119 (see Figure 12) is covered and protected. To deposit the molybdenum film you can the most varied of processes can be used, as has already been explained with reference to the molybdenum-titanium layers has been. A layer of gold 117 is then deposited on the molybdenum film by any of the known methods, which can be for example 7,500 pounds thick. In their place, however, other highly conductive metals, How copper is provided, that metallurgically with molybdenum

- 24 -

909850/1284

A 37 309 b '- 2'ί -

is stable. On top of the gold layer is a second and example about 1200 S thick molybdenum layer 118 formed, whereupon the three layers of molybdenum, gold and molybdenum are successively etched in the manner previously described so that the interconnections of the first level are created; it becomes, for example, the interconnection 104 is formed which ohmically forms the base 114 of the transistor 94 connects to one end 112 of the resistor R ^; one Interconnect 101 is ohmically connected to emitter 111 of transistor 94 while an interconnect 102 ohmically connects the collector 70 of the transistor 94 to the voltage supply connection V (see Pig. 11). Each interconnection thus has a lower layer of molybdenum (116), an intermediate layer of gold (117) and one upper layer made of molybdenum (118).

An insulating layer 119, for example made of silicon oxide, is then deposited on the by any of the known methods Molybdenum layer 118 applied, for example by vapor deposition or spraying and then selectively etched, in order to expose the surface of the molybdenum layer 118 only where the voltage supply terminal V is located (see Fig. 12). The purpose of the insulating layer 119 is that To separate interconnections of the two levels from each other electrically.

The insulating layer can also be made of silicon nitride, Aluminum oxide or other inorganic or organic insulating materials. In the preferred embodiment, the insulating layer 119 is formed from silicon oxide, which is sprayed on by means of HP atomization in a layer thickness of approximately 20,000 X. To get a better ohmV Seeing to create contact between the interconnections of the two levels is the top layer of molybdenum 118

- 25 -

909850/1284

A 37 309 b - 25 - _{1 Q o} _

• Selective in the area of the voltage supply connection V etched away, especially with the help of a photo-resistive mask, so that the contact to the gold layer 117 can be made directly.

On the insulating layer 119, for example, about 1200 S. thick molybdenum-titanium layer 120 is formed, and for example by high-frequency sputtering, in particular vapor-deposited followed by a gold layer 121, its thickness is preferably about 7500 Å. Then the gold layer and molybdenum-titanium layer 121 and 120, respectively, are selectively etched to create the desired pattern for the interconnects 122 of the second level; a contact between the two levels is created at connection V between the gold layer 117 and the molybdenum-titanium layer 120. As already mentioned, the gold can easily be etched in an alcoholic solution of iodine and potassium iodide, while the molybdenum-titanium layer is etched using another etching solution is removed, namely expediently / potassium ferricyanide, which can also be used for pure molybdenum. The top gold layer 121 adheres well to the molybdenum-titanium layer 120.

Finally, an outwardly leading gold wire (not shown) can be used, for example Connect pressure and heat to the gold layer 121.

One of the many advantages of the system shown in Figure 12 is the extremely good adhesion of molybdenum and the mixture of molybdenum and titanium (layers 118 and 120, respectively) to the Insulating layer 119. Improved adhesion due to the titanium is observed when one Mixtures of molybdenum and titanium with up to 3 pounds of titanium are used. The latter increases the adhesion of the molyb

- 26 -

909850/1284

A 37 309 b - 26 -

b-35.,.

dans on silicon oxide or on glass surfaces as a result of the extremely strong Ti-0 bonds that arise ^ and improves the adhesion to gold through Ti-Au bonds, which occur at the interface between gold and the molybdenum-titanium system develop. If more levels of interconnections and contacts are required, all of them can go to the top ate consist of pure molybdenum while this should be formed from a combination of molybdenum-titanium with gold; the top gold layer has the advantage that you can easily attach lead wires to the outside. If you want a supply line directly to the lower one Bring the molybdenum layer up, it can be useful to create a gold layer only there on the molybdenum, where this is exposed through an opening in the overlying insulating layer, so that the gold wire can be attached to the relatively small gold layer can be connected.

In order to demonstrate the great advantages achieved by the invention with regard to corrosion resistance, some examples are to be listed numerically below:

In a conventional water drop test with a voltage of 6 volts on a number of thin lines made of the metal to be tested and a drop of 10 "-proentigen phosphoric acid on the lines to complete the circuit, a three-layer metal strip made of molybdenum-gold-molybdenum fails within 20 Seconds since the anodic layers of molybdenum had been etched away. A triangular strip with 80 $ molybdenum, 20 & Ti tan-Gold-80 # molybdenum and 20 $> titanium lasted at least 20 minutes before the difficult-to-peel alloy and gold The dissolution of both metals proceeded at about the same rate.

- 27 -

909850/1284

When testing the metallurgical stability, a metal strip made of a mixture of 19 ° titanium and 81 ° molybdenum, which was approximately 10 ram thick, was provided with a gold layer approximately 7 / 100CDmm thick and then heated to 600 ° C. during One hour in a nitrogen atmosphere. The sheet resistance of the strip was 0.044 ohms at the beginning, but it increased to 0.098 ohms, which shows how little the reaction of titanium with gold is. In another test with a similar material the originally increased 0.064 The sheet resistance amounting to 0hm was not at all during a period of 5 minutes at 500 ° C. Both tests therefore show the great metallurgical stability of the system according to the invention.

- 28 -

909850/1284

Claims (4)

A 37 309 b - 28 - April 28, 1969 P «a t entane requirements 1J semiconductor device »it using molybdenum ^ produced metallic contacts and / or metallic layers on the semiconductor surface and / or on an insulating layer over the semiconductor surface, characterized in that the contacts or layers made of a mixture of molybdenum and such a modifier metal or »Metalloid are formed, which has a greater corrosion resistance than molybdenum. 2. Device according to claim 1, characterized in that the modifier metal or * -Metalloid titanium, tantalum, Chromium, zirconium, hafnium and / or silicon is " 3 · Device according to claim 1, characterized by a Metal layer of higher conductivity on at least part of the metallic layer. 4. Apparatus according to claim 3, characterized in that the metal layer of higher conductivity is a gold layer is. 909850/1284 if Blank page