Nowadays when people say 'integrated circuit' they usually mean a monolithic IC, where the entire circuit is constructed in a single piece of silicon. The fabrication process can be extremely complex, involving many different techniques such as etching, ion implantation, oxidation, deposition of new material, passivation, doping etc. The end result can be a 'chip' containing tens of millions of interconnected transistors.
What were the first such monolithic ICs and who made them? This turns out to be quite a complex piece of engineering history. Nowadays it is well-documented that Jack Kilby of Texas Instruments and Robert Noyce of Fairchild Semiconductor independently conceived of the idea at roughly the same time in 1958/59 and both went on to demonstrate working devices, albeit using different materials and technology. The two companies filed patents and fought a lengthy legal battle over the rights. They eventually settled and cross-licensed their technologies.
However, the commercial development of the integrated circuit was not easy. The first devices were very costly, and only the military and NASA needed the benefits of miniaturisation badly enough to pay the high price. Throughout the 1960's, new manufacturers entered the field, different IC technologies were invented, and complexity slowly increased from small-scale integration, through medium-scale, to large-scale. This was a period of enormous change, and I have tried to document the important milestones below. There is quite a lot of supporting material on the Web, but some of it is anecdotal and I have found such accounts to be unreliable. The key references that I have used are:
- The Chip Collection at the Smithsonian Museum
- an article from Electronic News of January 3, 1972, unfortunately now only available in the Internet Archive. This article suffers from the oddity of not mentioning the manufacturer Motorola at all. I am perplexed by this omission, as Motorola were involved from the start of the IC industry, and created several historic chip families.
- a 1963 book, "Introduction to integrated semiconductor circuits" by Adi J. Khambata, published by John Wiley and Sons, Inc., New York, which contains a considerable amount of information on the commercial types available then.
I do not possess all the IC's below, in fact I am seeking many of them, as well as original data for many of them. It is noted in the text where I am seeking examples of any particular type: if you have some for sale or exchange, please Conversely, I am happy to help anyone looking for information on these devices.
Finally a brief explanation of date codes on ICs. Most chips bear a date code. More modern ones use four digits yyww, where yy is the year and ww is the week in the year. For example, 8324 is week 24 of 1983. Early examples sometimes use three digits, yww, where the year has a single digit. Most often this implies that the decade is 1960, for example, the SG42A below has code 649, week 49 of 1966. However, there are cases where the decade is 1970, such as the RCA 4000 series CMOS chips shown below.
| 1960 | Since TI and Fairchild were the co-inventors of the IC, you might expect that they would release the first commercial devices, and in fact this was so. In some places on the Web the Fairchild 900 series is credited with being the first to market, in 1961, but the documented evidence does not support that: the Chip Collection gives a very specific date of March 1960 and price for the first announced TI chip, the SN502, and Khambata states unequivocally that "In 1960, Texas Instruments announced the introduction of the earliest product line of integrated logic circuits. TI's trade name is 'Solid Circuits' for this line. This family, called the series 51, utilized the modified DCTL circuit...". Finally, "Electronic Design" magazine announced the Texas devices in March 1960,and Fairchild prototype chips in November 1960. These back issues used to be freely available on the Web, but seem to have gone now. | ||
| 1960 |
Texas Instruments first widely advertised integrated circuits were the SN51x series, as stated by Khambata, and they are described in TI's "Semiconductor / Components Catalog". The SN502 seems to have only been available briefly, although the Smithsonian page states clearly that it was "commercially available". Apparently early TI TTL data books state that the SN502 used mesa technology rather than the planar technique that Fairchild developed and which became ubiquitous. SN stands for 'semiconductor network', a TI name for the IC that seems to have replaced 'solid circuits' and a prefix that they still use today. TI's early production was entirely used by NASA or in US military projects, and consequently these chips are not well-known and are difficult to obtain. Some of them exist in the Smithsonian
Chip Collection.
My image shows one in a 10-lead flat-pack encapsulation, although I believe that 14-lead was also used. The circuit technology used transistors with additional base resistors compared to the DCTL (direct-coupled-transistor-logic) technology to which Khambata refers. Each resistor had a parallel capacitor to speed up switching, and this technology was therefore known as RCTL, for resistor-capacitor-transistor logic, a term (and technology) that was used only briefly.
TI solid circuits, namely the SN510 and SN514, were the first integrated circuits to orbit the Earth, aboard the IMP satellite, launched by the US on November 27, 1963, as documented in a NASA technical report, again this seems to be no longer available on the Web. There was also a variant for 'severe environment applications', the SNR51x series. These were selected devices that were centrifuged, burnt in, and subjected to radiographic inspection. I am keen to buy some of these. If you have any for sale or exchange, or have data sheets or books about them, please |
||
| 1961 |
Fairchild's first devices were packaged in what was essentially a transistor can, firstly in metal, then a later ceramic option, with either 8 or 10 short leads (later named TO-99 and TO-100 respectively). The circuit technology was essentially DCTL, but with small base resistors, and so it came to be known as RTL, for
'resistor-transistor logic',
although the name
may not be entirely accurate.
Fairchild themselves used the word 'micrologic' (uL) to refer to many of their early families of ICs: these uL90329 resistor-transistor logic devices, dated 1964, are RTuL 3-input NOR gates. Confusingly, Fairchild documentation uses three-digit, five-digit, six-digit, and sometimes nine-character part numbers to designate what is apparently the same device, in documents of different ages. There's a good source of information in the
World Power Systems archives.
Fairchild RTuL was used in the Apollo Guidance Computer (AGC), which landed on the moon in 1969. This is claimed to be a bold early use of integrated circuits, and indeed the decision to use ICs was made at a time when the technology was very new, but by the time of the moon landing, RTL had been obsolete for years. Images of the chips could formerly be seen on the History of Recent Science & Technology site, now deceased, and on the NASA Office of Logic Design site. Both show 10-lead flatpacks, and the NASA site states that these were the dual 3-input NOR gate used in the second version of the AGC, called 'Block 2'. They state that the original 'Block 1' version, which was used in earlier unmanned Apollo missions, employed 6-pin TO-47 packages containing a single 3-input NOR gate, a poor image of which can be seen on this site which even calls the package TO-5, a transistor encapsulation. These single and dual gate devices might correspond to the commercial Fairchild uL903 (shown in the image) and uL915 types. Interestingly, the NASA site has a document showing that they purchased functionally identical chips from Fairchild, Texas Instruments, Motorola, Transitron and Westinghouse. The last two companies did not become major IC manufacturers. The very first Fairchild devices were simply denoted by a one-letter identifier: B, C, F, G, H or S. I would like to obtain some of those, if you can help, please |
||
| 1961 |
Whilst researching material for this site, two lamentable facts have become clear to me. Firstly, the UK used to have many world-leading electrical/electronic research and manufacturing organisations that are now all gone, and secondly, much of the history of these companies, and the important developments that they made, is documented poorly, if at all. Consequently, the Web gives the impression that all advances in semiconductors were made in the USA, which is quite incorrect.
Several UK companies were active in early IC development, including STC, Plessey (before they became GEC-Plessey), Ferranti and GEC (Marconi-Elliot). Little documentation remains of these early efforts: if you have any, or were involved in the work of any of these companies, please as I am keen to understand the true history of this period in the UK, for which a number of contradictory references can be found on the Web. One source of information is the book "A History of the World Semiconductor Industry" by P.R.Morris, which is viewable on Google books. This states that Plessey were the first UK company to become involved with IC development, in 1957, although commercial production was delayed until 1965, a lost opportunity of monumental proportions. Alex Cranswick, a former Plessey employee, tells me that Plessey had a very high speed silicon bipolar process in 1968 and there were two significant ranges of devices that Plessey were producing with this, a range of ECL logic (SPnnn) and a logarithmic amplifier (SL521). The latter went into a number of military projects and the ECL was certainly evaluated and possibly used by ICL for their top of the range computers. I would be very interested to know details of these first Plessey ICs. Another book on Google, Corporate Vision and Rapid Technological Change by Peter Swann, states that Ferranti produced their first ICs, the series called 'Micronor I', in 1964. Correspondence with ex-Ferranti employees seems to confirm this date. Another apparently reputable reference, Technology Transfer and the British Microelectronics Industry, 1950-75: Confused Signals by John F Wilson, states that Micronor I development for the UK Navy started in 1961, but the first sales were made in 1963 (probably late in the year). Other Web articles give the 1961 date but are probably derivatives of the Wilson original. I have included Micronor lower in this table, at the dates that I believe are correct. However, I have located what I believe are some Ferranti RTL devices that possibly could have preceded Micronor. See my dedicated page for details. As far as I know, STC did not take any monolithic ICs to commercial production, although they did make hybrid devices. GEC changed into Marconi-Elliott and did make commercial devices, although I have little data on them. If you have information about the early development of ICs at any of these UK companies, please |
||
| 1962 |
It seems reasonable to expect that Motorola would follow TI and Fairchild and announce an RTL family at about this time, however an article based on the 1976 bicentennial edition of Electronic Design shows that Motorola released the first
emitter-coupled-logic
(ECL) chips in the MC300 series in 1962. This was an inspired move, as ECL was destined to become a significant technology that ultimately attained the highest switching speeds of any IC, whereas RTL was limited in performance and was rapidly made obsolete.
The image shows a 358 AC-coupled JK flip-flop with DC set and reset inputs and buffered outputs. Despite the very early date code of 1964, this is quite a complex SSI device, incorporating sixteen transistors. It had a switching time of about 16 nanoseconds. The 'G' suffix indicates that this example is packaged in a metal can (in this case a 10-lead TO-100), an alternative 'F' suffix denotes a 10-pin flatpack. Other manufacturers used different suffixes, as described in some other entries below. |
||
| 1962 |
Motorola did in fact also announce their first RTL series at about the same time, but I have no confirmation of the exact date. If you can help me with that, please
Motorola made RTL devices with numbers MC7xx, MC8xx and MC9xx. Although the series do not fully match, equivalent devices (eg MC715, MC815 and MC915) provided the same functionality but with different temperature range specification, intended for different markets. Motorola used a number of names for these series, including MRTL, 'new MRTL', 'mW MRTL', and 'plastic MRTL'. To confuse matters further, they also issued DTL with MC8xx and MC9xx numbers, see later for more details.
The MC700 on the left dated 1967, with the 'G' suffix for the metal can (in this case a TO-99 with only 6 leads), is a buffer.
The MC888 on the right, also dated 1967, with a 'P' suffix denoting the plastic
dual-inline package,
is a dual 3-input NOR gate with dual buffered outputs and a switching time between 20 and 65 nanoseconds depending upon the output used.
|
||
| 1962 |
The next major event in the development of the integrated circuit was the announcement of the first
diode-transistor-logic
(DTL) by the new company Signetics. There is
anecdotal evidence
on the Web that these were an immediate commercial success, so much so that they were a threat to TI and Fairchild's RTL lines, and subsequently all the major manufacturers changed over to DTL. However, information about Signetics actual chips is very scarce indeed. The Motorola 'Microelectronic Data Book' for 1969 has a section on interchangeable components that lists Signetics DTL types in four series, SE1xx, CS7xx, NE1xx and NE7xx. It is then possible to find out their functions on the
DataSheetArchive
site, where they are shown as Philips devices, since Philips Semiconductors bought Signetics in 1975. This SE113 on the left is a dual 3-input NOR gate, the 'K' suffix denoting the TO-100 can. On the right the NE157K is a dual 3-input NAND gate. Other Signetics' part number suffixes were 'A' signifying the dual inline package and 'G' and 'J' which I am not sure about but which probably denote two flatpack types.
If you know more about Signetics DTL, or have any examples of them, please |
||
| ~1963 |
|
||
| 1963 |
In this year Sylvania announced the first chips in the Sylvania Universal High-level Logic (SUHL) family. This was the first commercially successful
transistor-transistor-logic
(TTL) family to be created, although the first chips using this technology were made in 1961 by Pacific Semiconductors (a US early-stage company that became TRW - no relation to the modern asian company of the same name).
The image on the left shows an SG52A 2-2-2-3 AND-OR-INVERT gate (SG = Sylvania gate?) in a 14-pin flat-pack dated 1966. There exist also SF-series flip-flops (SF = Sylvania flip-flop?) and SM-series devices (possibly 'Sylvania module' or 'Sylvania microcircuit', as they contain more complex circuits such as adders and shift registers). SUHL was second-sourced by Raytheon (RF and RG series) and by Texas Instruments (F and G series). The image on the right shows an SG220 in a gold-plated dual-inline package dated 1969 - I believe this is a quad 2-input NAND gates device, but I do not have definitive data.
I have very little data on these devices, if you have any please |
||
| 1963 |
 As mentioned above, the UK firm Ferranti made its first sales of the DTL family Micronor I in this year, in the series ZSS, ZST and ZSP with single-digit numbers (two-digit numbers were used by Micronor II described below). These are quite rare, but I do have some data on the devices, and one example of them: the ZST2 shown on the right.
They seem to have been unusual circuits, different from the DTL families that the major IC vendors released in 1964. The Ferranti devices are relatively simple, with only one or two transistors, and often have multiple outputs isolated by diodes, rather than inputs. They use 8-pin TO-5-type cans. It's a bit hard to know how these could be used as logic gates, and in fact an ex-Ferranti employee told me that his team developed Micronor II in response to the inadequacies of Micronor I.
I'm still looking for specimens of these so if you know how I can obtain any, please
All the Micronor I devices received "Services Qualification Approval" and each is assigned an equivalent CN number. My second image on the left shows the schematic of one of these, designated CN7, equivalent to ZSS1A, which shows the unusual circuit used. I have been unable to find information on the CN series, such as is available for the CV series of valves and semiconductors. If you know how I can obtain such data, please As well as the logic family Micronor, Ferranti made linear devices in the family Microlin, also in 8-pin TO-5 cans, with prefixes ZBA, ZLA, ZLD, ZQT and ZDC. In addition, they sold a number of what they called "custom-built circuits", both linear types and digital logic. These were devices that had been "conceived ... outside the Electronics Department", presumably for one customer (probably military), but were considered to be of more general use. There was a ZGS series, plus several with a type number EXPnn, and some use an unusual package, either 4-lead or 6-lead flat-pack. I have no examples of either Microlin or the custom-built types, so if you know of a source of any, please |
||
| 1963 |
Also in this year Texas started production of linear integrated circuits, in the SN52x series. My image shows an SN726 from 1967, this is the same device as the SN526 but with a different temperature range.
I am seeking the SN52x types, if you have any, please
|
||
| 1964 |
About this time Motorola started making DTL chips. The fact that TTL was already available from Sylvania shows the uncertainty in the market at the time and the co-existence of competing technologies. Motorola again developed several families: the MC200/250 series, MC830 series, MC850 series and MC930 series. The image shows an MC207G dual 3-input NAND gate and an MC209G single RST flip-flop, both date coded 1964, and therefore very early production units.
|
||
| 1964 |
In this year Texas Instruments issued DTL in their SN530 series, an example can be found in the
Smithsonian Chip Collection.
Apparently this series was long-lived, but examples are hard to find, suggesting that TI supplied most of their production to the military. My image shows an SN534 dual AND/OR gate and an SN530 single JK flip-flop. Unfortunately their date codes are on the underside, but other chips on this board are all dated 1964, making these very early units.
Interestingly, the Smithsonian states that in 1962 TI received an order to supply chips for the Minuteman II program, and I believe that these were used in particular for the missile guidance computer. In 1962 TI only had RCTL devices, but the Smithsonian article states that TI had to 'develop and deliver' nineteen specific circuit types. The TI "Semiconductor / Components Catalog" from 1965 shows a separate family called "Minuteman Series DTL" containing only seven types; SN337A, SN341A, SN343A, SN344A, SN346A, SN347A, and SN359A. These are not shown in the same catalog from 1963. If anyone can clarify the various Minuteman devices used, or has examples of them, please Other oddities exist in the TI DTL families. The SN730 series is described in some places as being the same functional devices as SN530 but with an extended temperature range, however the "Semiconductor / Components Catalog" from 1965 shows this series as a distinct family of 'low power RTL'. I am also puzzled by some SN63xx devices that I possess, which could be designed for yet another temperature range, but for wich I can find no documentation. If you can tell me anything about them, please |
||
| 1964 |
Of course, not all integrated circuits are switching logic, there exist many types of linear devices.
The Fairchild 702 is generally regarded as the first commercially successful
operational amplifier
chip, although other manufacturers produced earlier IC amplifiers, in particular TI with their SN520 and SN521, and also the manufacturers Westinghouse and Sperry Semiconductor, who never became major players in the IC market. The image on the left shows a version from 1968 in a flatpack. The image on the right shows a later ceramic dual-inline version dated 1973.
|
||
| 1964 |
930 series DTL, introduced by Fairchild in response to Signetics, and copied pin-for-pin by several other manufacturers, was the most commercially successful DTL series. The images show a 9945 gated flip-flop in a plastic flat-pack dated 1966, the trailing 59 signifying the operating temperature range of 0-75 degrees Celsius, and a 9930 dual 4-input NAND gate in a 14-pin ceramic dual-inline package dated 1968.
|
||
| 1964 |
In this busy year the Texas Instruments SN5400 series of TTL chips first appeared. These were aimed at the high-specification military/space market where wide temperature tolerance is needed. The 5430 shown is a single 8-input NAND gate, a slightly unusual function, but an early example as it is dated 1965. As well as flatpacks, the SN5400 series is more commonly found in ceramic DIL packaging (denoted by a 'J' suffix on the part number).
|
||
| 1965? |
At about this time, the UK firm Ferranti released a second family of DTL ICs called MicroNor II. In fact, they produced eight series of devices, utilising all three different IC package types (can, flatpack, and dual-inline) plus two switching speeds. The 200 and 300 series with part numbers ZN2xx and ZN3xx utilised proprietary circuit designs and were available in flatpacks or in dual-inline (suffix E). The series 50, series 80, series 110 and series 130 were all in TO-5 cans (part numbers ZSx5x, ZSx8x, ZSx11x and ZSx13x where the third letter is either D for diode arrays, F for flipflops, S for gates, or T for 'power gates') and were compatible with the Fairchild 930 series DTL. My composite image shows a ZSF131B single JK flip-flop dated 1968 and a ZN244E dual 4-input gate dated 1971.
I have some data on these devices but would welcome more, if you have any, please |
||
| 1966 |
The Motorola MC1000/1200 series from this year was a faster version of MECL and became known as MECL2. The MC1001 shown is a six-input OR gate, this example made in 1969. It had a switching time of about 5 nanoseconds.
|
||
| 1966 |
The Texas Instruments SN7400 series is possibly the most successful series of ICs of all time. It is a version of their older SN5400 TTL series, but with a reduced temperature range specification. It really took off when TI started making low-cost plastic dual in-line chips (part numbers with an 'N' suffix). Faster Schottky versions (74S) and low-power versions (74L and 74LS) soon followed, and even CMOS equivalents. The image shows the SN7401 quad 2-input NAND gates with open collector outputs, in a tiny flat-pack (the scale is mm) in its chip carrier. It is dated 6752, which makes it a very early production example, probably intended for military or NASA use. Incidentally, date codes with week 52 are frequent: I suspect that they are incorrect, and the machine has somehow defaulted to the last week of the year.
If anyone can explain this, please
|
||
| 1966 |
Texas Instruments also issued a few specialised DTL series for use in calculators, before those devices switched over to using MOS LSI chips. There were several series, although each had only a few types in it. My image shows an SN15862N triple 3-input NAND gate. I am unsure of the date at which this series was available, but my earliest examples are dated 1966. In addition to this SN158xx series, There were also an SN39xx series and an SN45xx series as documented by
Brent Hilpert. Old calculators also used the TI SN75xxx series of interface chips, primarily for display drivers.
If anyone has examples of the SN39xx and SN45xx types, please
|
||
| 1967 |
In this year, as far as I can tell,
Philips of the Netherlands
issued its 'FC family' of DTL ICs, which it sold directly and through its UK arm Mullard Ltd, and possibly its German subsidiary Valvo GmbH. There were several series: FCHnnn were gates, FCJnnn were flip-flops, FCKnnn were monostables, FCLnnn were level detectors (Schmitt triggers) and FCYnnn were diode expanders. Some of the devices were compatible with the Fairchild 930 series, showing Europe lagging the USA by about 3 years.
In 1972 Philips issued a further family, the FJ family of TTL ICs, compatible with the TI SN74 series. |
||
| 1967 |
At about this time, a number of IC manufacturers produced various medium-powered AF amplifiers in idiosyncratic packages that were based on the standard dual-inline variety. Some of them had odd pin arrangements and protruding metal lugs, designed to be bolted or soldered to a heatsink, such as this General Electric PA237, and also the
Sinclair IC-10.
Others had heatsinks clamped onto the plastic DIL package, such as the slightly later
Sinclair Super IC-12.
None of these designs has stood the test of time, and nowadays most 'power' IC's use a single-inline package that is either free-standing or bolted to a heatsink using a protruding metal flange or just a simple hole through a package with metal back plate.
The PA237 from 1967 was able to deliver 2 Watts into a 15-ohm load. It was one of a number of GE amplifier chips. GE ICs in general are relatively rare. |
||
| 1967 |
The Fairchild uA709 op-amp. was even more successful than the rather quirky uA702 and paved the way for the classic 741 (not shown here). The image on the left shows an original Fairchild 1967 example in an 8-lead TO-99 can. That on the right shows a 1970 device made by Microsystems International, a Canadian company that was a wholly-owned subsidiary of Bell Canada and Northern
Electric, in a dual-inline package. Both top and bottom of the chip are covered by gold, only from the side is it possible to see white ceramic inside. The use of Roman numerals for the date code is quite extraordinary.
|
||
| 1968 |
The Motorola MC1600 series was their third generation of ECL, known as MECL3. This was the ultimate in high-speed monolithic technology, with switching times of 1 nanosecond, at least until microprocessors reached GHz clock speeds more than 30 years later. However, it was not commercially popular, probably because it required very careful circuit design, and the high speed resulted in high power dissipation. Motorola therefore created a more successful fourth generation, called MECL 10K because the part numbers were MC10xxx, with a 2 nanosecond speed and less than half the power requirement of MECL3.
My image shows a 1662 quad 2-input NOR gate with suffix 'L' denoting a ceramic dual-inline package, dated 1973. |
||
| 1968 |
The arrival of the RCA CD4000 series of CMOS chips, originally known as COS/MOS, is particularly well-covered by Wikipedia. This was the last major IC technology to be developed, and is still used for all ultra-scale circuits such as microprocessors. Nevertheless it started like the others with simple logic circuits as shown in the image, a CD4012 dual 4-Input NAND gate and a CD4009 hex invertor. These examples have the suffix A indicating unbuffered outputs.
|
||
| 1969 |
This year saw the arrival of Intel's first commercial chip, the 3101, a 64-bit static RAM using Schottky TTL technology. Intel had started an internal competition between the 3101 using tried technology and the 1101 using a PMOS technology. Although the 3101 narrowly won, in the long term it was MOS that dominated.
|
There were other IC pioneers apart from Intellux that have been largely forgotten. This image shows a 301CJ chip by the company Amelco, which became part of Teledyne. This company, formed by ex-Fairchild employees, was a leader in early IC manufacture. It made RTL and then DTL.
Westinghouse were one of the pioneers of IC manufacture, which they called "molecular electronics", a term that had a broader meaning than just monolithic silicon ICs, and has come back into use in recent years. Their initial work was funded by the US Air Force, and they also supplied chips to NASA for the Apollo Guidance Computer. As far as I know from consulting old data books, they did not market any RTL but went straight to DTL such as the example I show: a WM201T dual 3-input NAND gate with tri-state outputs. It is coded MED24 - if anyone knows what this signifies, a date code perhaps, please
Now we have come to the end of the 1960's, and in that decade we have seen the development of RCTL, RTL, ECL, DTL, TTL and MOS. Intriguingly, there has been no new IC technology since then (or has there? Are there really gallium-arsenide or silicon-on-sapphire or even silicon-germanium IC families? I don't think so, at least not in general use.).
There are some other devices not shown above which I am keen to find. These include:
- any ICs from the 1960's by General Electric, Sperry Semiconductor, Pacific Semiconductors, Transitron or AMELCO.
- the Siliconix AO2A and other early Siliconix types
- early UK or European ICs by Mullard/Philips, Ferranti, Plessey, STC, Siemens etc.