Five Short Stories on The History of Microfluidics

1. ONCE UPON A TIME IN THE HISTORY OF MICROFLUIDICS

Microfluidics timeline. Figures caption ( chronological order): (a) PDMS chemical structure; (b) Dow Corning is founded in Midland (1943) to work on silicones (image source –IHS media); (c) the first working germanium IC created by Jack Kilby (1958) (image from Texas Instruments); (d) SEM image of a patterned PDMS microstructure obtained by replica molding [8]. (e) a PDMS microfluidic chip. PDMS was introduced as material for microfluidics by Whitesides in 1998; (f) digital microfluidics for manipulation and control of discrete droplets and bubbles; (g) an example of organ-on-chip, a device that allows cells to be seeded in order to recreate the microenvironment and microarchitecture of a specific human tissue or organ; (h) a paper-based microfluidic device can be used as low-cost, portable diagnostic tool, resulting to be particularly useful in emergency conditions and in resource-limited areas.

Fathers of integrated circuit technology

Integrated Circuits (IC) technology and photolithography development go hand in hand. The first developments of ICs were carried out by Werner Jacobi, a German engineer, in 1949, and by the British radar scientist Geoffrey Dummer a few years later. However, Jack Kilby and Robert Noyce are considered co-inventors of the IC technology, despite working separately (Figure 1 and Figure 2).

 Figure 1. Robert Noyce and his silicon ICs.

In 1958, Jack Kilby started working at the Micro-Module program at Texas Instruments, a program sponsored by the US Army Signal Corps whose aim was to create components of uniform size and shape with integrated wires that could be easily assembled to make circuits. That summer, Kilby had no vacations and started working alone in the lab, developing the idea that both passive (resistors and capacitors) and active (transistors) components could be made on the same piece of semiconductor material. By September of the same year, he showed the first working integrated circuit made in germanium. Six months after Kilby had patented his invention, Robert Noyce, a physicist and co-founder of Fairchild Semiconductor, understood the limitations of germanium as material for integrated circuits and started working on silicon ICs, which is why ICs are said to have two fathers. This discovery radically changed the society, being used to create computers and electronic devices for applications in medical, transportation, entertainment and communication fields, and allow achievements such as the Apollo moon program.

Figure 2.  Jack Kilby shows an integrated circuit, 1982.

First came photolithography

Photolithography was invented in the early 1950s, when the U.S. National Bureau of Standards (NBS, then US Army Diamond Ordinance Fuze Laboratory, DOFL) promoted a program to develop new methods obtaining small electronic circuits that could be easily integrated into military proximity fuzes [1]. From 1952, Jay Lathrop and James R. Nall started using photoresists to pattern germanium. They were able to project light through a specially modified trinocular microscope. Thus, they created a germanium transistor that could be easily incorporated into miniaturized hybrid transistor-to-ceramic circuits.  Lathrop and Nall wrote a paper and patented their discovery in 1958-1959, coining the word photolithography (Figure 3). Lathrop himself declared: “The operation actually involved etching, not lithography. However, photolithography sounded higher tech to us than photoetching … and this misnomer has been used ever since”. Lathrop and Nall were not the only researchers who worked on this topic. From 1955, Andrus Jules and Walter L. Bond at Bell Laboratories also developed etching (photoengraving) techniques for producing patterns on silicon semiconductors by using oxide layers, an invention patented in 1964 (US Grant US3122817A).

Figure 3. The patent written by Lathrop and Nall about semiconductors construction, 1959

2. PARENTS OF MICROFLUIDICS

As described by Whitesides [2][3][4], microfluidics has four parents: molecular biology, molecular analysis, national security and microelectronics.

The oldest one is considered to be molecular analysis, which includes methods such as gas-phase chromatography (GPC) or capillary electrophoresis (CE)[5][6]. These techniques, developed from the 1950s and 60s, allow the separation of chemical compounds or biomolecules by flowing small amounts of a sample in narrow tubes or capillaries, reaching high sensitivity and resolution (Figure 4).

Figure 4. (A) Professor Stellan Hjertén, who first used capillaries to perform electrophoresis, with an automated version of the capillary free zone electrophoresis apparatus in 1967 (image source). (B) A logical NOR IC component of the computer controlling the Apollo spacecraft (1960s) and (C) the “Baby Blue”, a prototype of PCR machine: the software controller is integrated with the thermal cycling block, 1986 (image from Science Museum Group).

The most famous parent of microfluidics is microelectronics [7]: at the beginning, researchers tried to directly fit fabrication methods and materials from microelectronics to microfluidics: photolithography as well as silicon and glass were the first players on the microfluidics stage. Only later did microfluidics split from microelectronics and semiconductors technology by using new specific microfabrication methods and materials [8].

A lesser known but pivotal forerunner of microfluidics is military detection. Starting from 1994, DARPA (Defense Advanced Research Projects Agency of USA) substantially contributed to the growth of microelectromechanical systems (MEMs) and the development of miniaturized and portable “laboratories on a chip” with the main goal of the detecting chemical and biological weapons.

Molecular biology, especially genomics in the 1980s, strongly contributed to microfluidics birth and evolution as its fourth “parent”. The interest of scientists in studying and sequencing nucleic acids led to the development of sequencing machines capable of working with small samples to ensure a high sensitivity read-out [5]. An example? The famous PCR (polymerase chain reaction) technique used to amplify a DNA sequence by means of heat was developed in the early 1980s by Kary Mullis. The reaction requires small amounts of liquid, usually 10-200 µl, so precision equipment is necessary. At the beginning, due to the lack of automated equipment, the reaction was a time-consuming multi-step process that had to be performed manually. The first commercial machine, a simple thermal cycler, was developed in 1987; it rendered the process reliable and its small dimensions gave the possibility to miniaturize operations as well as to work outside of the lab.

3. FROM THE FIRST MICROFLUIDIC LAB-ON-A-CHIP TO THE µTAS CONCEPT

Starting in the 1950s, there was an increasing interest in designing miniaturized systems and components, which had recently become possible thanks to the development of new technologies to create complex 3D micro-patterns on semiconductors. Research started in miniaturizing sensors, transducers and other components and to then integrate them with microcomputers in order to obtain portable (small dimensions) integrated platforms to use as environmental or medical monitors/ measurements systems.

A pioneering work was carried out by Stephen Terry from Stanford university, who in the mid-1970s produced a miniaturized gas chromatograph (GC) integrated on a silicon wafer [9] (Figure 5). The miniature GC was composed of gas supplier, sample injection system, capillary column and output miniaturized thermal conductivity detector. The injection valve and the capillary were fabricated through a micromachining technique onto a silicon wafer while integrated circuit (IC) processing methods, developed starting from late 1950s, were used for the detector microfabrication. This device was considered one of the firsts examples of “laboratory-on-a-chip”.

Figure 5. The article written by S. C. Terry and published on IEEE Transactions on Electron Devices in 1979 about a miniaturized gas chromatograph [9].

Following in the footsteps of Terry, Andreas Manz, a Swiss researcher and analytical chemist, was one of the first to use microchip technology in the field of chemistry to shrink a laboratory to the size of a chip in the 90s [10][11]. In 1990, he published a paper in which he introduced the concept of miniaturized “total chemical analysis system”, abbreviated to “ µ-TAS”, for chemical sensing, i.e. a microfluidic device capable to perform all steps in an analysis [12] (Figure 6). He demonstrated that the µTAS allows faster and more efficient sample separation (chromatographic or electrophoretic), shorter transport times and reduced consumption of reagents compared to chemical sensors and conventional analysis systems. Moreover, the fabrication of a multi-channel device allowed experiments to be run in parallel.

Figure 6. A scheme representing an ideal chemical sensor, a TAS (total chemical analysis system and a miniaturized µ-TAS, as described by Manz et al. [10].

Thus, in 1993, he created a µTAS on a glass chip that could perform capillary electrophoresis of amino acids in few seconds [11]. The device was fabricated in Pyrex glass by micromachining and was composed of capillaries with length from 1 to 10 cm, with a cross-section of 10×30 µm. The results demonstrated the possibility of creating a miniaturized laboratory-on-a-chip that could be used for complex analyses.

One year later, the Chemical and Biological Microsystems Society (CBMS) organized the first workshop on µTAS to present its basic concepts and technologies.

4. KEY ACTORS OF MICROFLUIDIC FABRICATION STAGE: SOFT LITOGRAPHY AND PDMS

SOFT LITHOGRAPHY

The development of techniques capable of patterning small structures was fundamental in microelectronics as well as optoelectronics; starting from the 1960s, the most used approach to fabricate integrated circuits and other micro-components was photolithography [8]. The name indicates a range of different techniques that provides for the transfer of a pattern through a light source (UV, X-ray, …) from a photomask to a photoresist on a solid substrate. However, photolithography turned out to be a difficult process when working with non-semiconductors materials, such as glass and polymers, and from the late 1980s new non-lithographic microfabrication processes started being developed. Researchers were looking for a cheap technology, capable of patterning 3D structures (also on nonplanar surfaces), to control the surface chemistry, that could be used on a wider range of materials, and thus able to overcome all the drawbacks of photolithography. And so, soft-lithography was born. The main difference when compared to photolithography was that an elastomeric mold could be used to transfer the pattern instead of a rigid photomask and that a wide range of materials (organic and biological molecules, polymers, etc.) could be directly patterned.

Some of the most famous soft lithography techniques like replica molding (REM), microtransfer molding (µT), microcontact printing (µCT) and micromolding in capillaries (MIMIC) became very successful microfabrication techniques in microfluidics (Figure 7).

Figure 7. Fabricating an elastomeric microfluidic device using rapid prototyping for the silicon master and then a replica molding to obtain the elastomeric replicas [13].

PDMS

The siloxanes, the macromolecules that compose the backbone of PDMS, were firstly characterized by the English chemist Frederick Stanley Kipping in 1927 (Figure 8). Considered as one of the founding fathers of silicon chemistry, he started studying the materials which are now globally known as silicones from 1899 and coined the term silicone in 1904. In 1943 the Dow-Corning Corporation was established as a joint venture between Corning Glass and Dow Chemicals and it became the first silicones manufacturer, following Kipping’s method [13].

Figure 8. Frederick Stanley Kipping, the British chemist considered the father of silicon chemistry.

The first microfluidic devices were usually made of silicon and glass since the fabrication techniques derived from microelectronics were well known [3]. However, these materials had some issues: silicon is expensive and cannot be coupled to optical microscopy because of its opaqueness while both silicon and glass have low gas permeability, which make them inappropriate for microfluidics applied to biology. Other materials were then investigated for potential use in microfluidic platforms. Researchers were looking for alternative compounds that could be optically transparent, easy to process, flexible and cheap compared to previous ones. Organic polymers seemed to be a good option and in the late 1990s the George Whitesides group of Harvard university introduced a new concept of low-cost microfluidics by using poly(dimethylsiloxane), known as PDMS, as new material for microchips rapid prototyping (Figure 9).

Figure 9. Scanning Electron Microscopy (SEM) of the channels of the first PDMS microfluidic device fabricated by replica molding by Whitesides group from Harvard University in 1998 [13].

5. MICROFLUIDICS AND BIOLOGY: TOWARDS HUMAN-ON-A-CHIP

One of the main advantages of PDMS compared to the traditional materials used in microelectronics was its compatibility with cells and the ease to culture simple organisms such as the C. elegans [3][14]. In the late 1990s microfluidic devices were created for cell biology applications, such as sorting or patterning cells and proteins, cells-based biosensors, studies of cocultures [15][16][17]. Albert Folch and his coworkers were able to selectively pattern cells and cells co-cultures on PDMS channels using collagen and fibronectin as templates as well as with no surface modification (1998-1999)[16] (Figure 10).

Figure 10. The cellular micropatterning method described by Folch et al. [16]: the microscopic image shows two distinct lines of fibroblasts (stained respectively in green and red), patterned by directly injecting the cells suspensions into the microchannels.

Researchers soon started working on microfluidics devices that could be used as tissue and organ models for drug discovery and development, study of pathophysiology and biological processes in order to overcome the in vitro and in vivo limitations [19]. The organ-on-a-chip technology was being developed. Since the 2000s, many on-chip tissue models of gut, liver, brain, heart, eye, skin, lung, muscle, blood vessels and tumor have been proposed. In 2010, Huh et al. [18] developed a biomimetic device for the mimesis of the lung alveolar-capillary interface in its structural, functional and mechanical aspects. The main difference compared to the previous works was the integration on a single chip of different tissues capable of recreating a functional microenvironment. Their device was composed of two PDMS microchannels separated by a porous PDMS membrane that was seeded with human alveolar epithelial cells on one side and human pulmonary endothelial cells on the opposite side, thus creating the epithelial and the endothelial compartments. The alveolar interface was recreated by injecting air and thus mimicking human breath (Figure 11).

Figure 11. The breathing lung-on-a-chip device proposed by Hu et al. [18]. (left) Chip and cells culture view. The side chambers are used to apply vacuum and recreate the alveolar-capillary movement by the PDMS mechanical stretching and (right) the chip layers, with a porous PDMS membrane between the two channels.

VIDEO ON HUMAN ORGANS-ON-CHIP (YOUTUBE)

In the last ten years, the organ-on-chip technology has evolved rapidly thanks in part to the use of processes alternative to soft lithography, especially for rapid prototyping such as 3D printing [4][20], a technology developed in the early 1980s.  The method is now currently used to produce scaffolds for tissue engineering, such as bio-printed organs, electronics, sensors as well for microfluidic systems [21]. With 3D printing it is possible to obtain three-dimensional structure-embedded devices capable of mimicking a complex microfluidic environment, as occurs in vivo, in a fast and low-cost way. Lee et al. developed in 2016 a one-step 3D bioprinting method to produce organ-on-a-chip platforms and demonstrated the performance of the device by reproducing the liver organ main functionalities. They used poly(e-caprolactone) as platform material, ECM-based hydrogels to simulate the microenvironment and collagen and gelatin hydrogels encapsulating cells as inks to build 3D platforms.

At the beginning of the 2000s, many researchers understood that the in vitro platforms attempting to mimic the organ-level functions were not sufficient to investigate the many physiological, inter-organ connections that occur in vivo. The first “human-on-a-chip” cell culture systems were then designed to miniaturize the body and study organs interactions and metabolism [23][24] (Figure 12).

Figure 12. This picture by Williamson et al. [22] clearly represents the concept of human-on-a-chip: some anatomy sketches from Leonardo da Vinci (1490) showing 3D organs and their interactions are used to show the structure and the aim of this device, i.e. mimic the complex human physiology on a single chip.

Some preliminary studies on the integration of different tissues on the same platform were carried out since early 2000s. In 2004, Viravaidya et al. [25] developed a µCCA (micro cell culture analog) to study bioaccumulation and ADMET (absorption, distribution, metabolism, elimination, and toxicity) pathways of new drugs. The microchip was composed of four chambers: a lung and liver compartments containing living cells and fat and other tissue chambers to distribute the fluids. Similarly, in 2009, Zhang and coworkers [26] designed a multi-channel microfluidic device for culturing human cells from liver, lung, kidney and adipose tissue for drug screening, creating compartmentalized microenvironments. Many advances have taken place in last ten years, making possible an effective body-a-chip technology [27][22][28]. Recently, researchers from MIT developed a body-on-a-chip composed of 10 compartments to mimic at the same time liver, lung, gut, endometrium, brain, heart, pancreas, kidney, skin, and skeletal muscle. The platform allows the investigation of new compounds and the evaluation of potential side effects on human body (physiome-on-a-chip) as well as the modeling of tumors metastases.

References