This article presents an overview of different laboratory automation sample transport technologies, using the LUO concept to examine the taxonomy of these approaches to better understand their similarities and differences, strengths and weaknesses. Accompanying articles will present detailed information about the mechansims and properties of each technology.
Sample transport based on moving fluids through tubing or channels was one of the earliest transport mechanisms used for laboratory automation, and remains an active area of technology development today. Over the years the scale has decreased from ml/minute flows to nanoliter or picoliter/minute flows, and from channel diameters of several millimeters to several micrometers. However, the basic LUO architecture remains the same at a conceptual level.
- Sample Transport: A continuous, moving carrier stream containing liquid samples and reagents are transported and combined in a tubular channel.
- Sample Processing: The samples and reagents pass through the tubular channel from one sample processing device to another with each device performing different LUO's, such as mixing, distillation, dialysis, extraction, ion exchange, and incubation.
- Data Collection & Handling: The completed reaction mixture passes through a detection device and subsequent a signal is recorded.
Basic fluidic principles
The flow of a fluid through a channel can be characterized by the :
Re = ρ u L / μ
Re = Reynolds Number (non-dimensional)
ρ = density (kg/m3, lbm/ft3 )
u = velocity (m/s, ft/s)
μ = dynamic viscosity (Ns/m2, lbm/s ft)
L = characteristic length (m, ft)
Figure 1: a) Laminar Flow b) Turbulent Flow
Due to the small dimensions of microchannels (100 nanometers or less), the Re is usually much less than 100, often less than 1.0. In this Reynolds number regime, flow is completely laminar (Figure 1a) and no turbulence occurs. In microflow the classic mechanisms used to induce turbulent mixing (coils, chambers) in macrochannels become less effective or totally ineffective, and mixing occurs by diffusion only. The transition to turbulent flow (Figure 1b) generally occurs in the range of Reynolds number 2000. Laminar flow provides a means by which molecules can be transported in a relatively predictable manner through microchannels. Note, however, that even at Reynolds numbers below 100, it is possible to have momentum-based phenomena such as as flow separation.
Pressure Driven Flow
Detailed technical article: Small_volume_pipetting
There are two common methods by which fluid actuation through channels can be achieved. In pressure driven flow, in which the fluid is mechanically pushed through the device via positive displacement pumps, such as metering/syringe (Figure 2), peristaltic (Figure 3) pumps; vacuum or air displacement pumps; or centrifugal (Figure 4) force. One of the basic laws of fluid mechanics for pressure driven laminar flow, the so-called no-slip boundary condition, states that the fluid velocity at the walls must be zero. This produces a parabolic velocity profile within the channel (Figure 1a). The parabolic velocity profile has significant implications for the distribution of molecules transported within a channel. Pressure driven flow can be a relative inexpensive and quite reproducible approach to pumping fluids through channels. While most common in the macro-fluidic domain, pressure driven flow is also amenable to miniaturization.
Another common technique, mostly in the microfluidic domain, for pumping fluids is that of electroösmotic pumping. If the walls of a microchannel have an electric charge, as most surfaces do, an electric double layer of counter ions will form at the walls. When an electric field is applied across the channel, the ions in the double layer move towards the electrode of opposite polarity. This creates motion of the fluid near the walls and transfers via viscous HGH Advanced forces into convective motion of the bulk fluid. If the channel is open at the electrodes, as is most often the case, the velocity profile is uniform across the entire width of the channel (Figure 4). However, if the electric field is applied across a closed channel (or a backpressure exists that just counters that produced by the pump), a recirculation pattern forms in which fluid along the center of the channel moves in a direction opposite to that at the walls (Figure 5). In closed channels, the velocity along the centerline of the channel is 50% of the velocity at the walls.
One of the advantages of electrokinetic flow is that the blunt velocity profile avoids many of the diffusion nonuniformities that occur with pressure driven flow. However, sample dispersion in the form of band broadening is still a concern for electroösmotic pumping. Another advantage to electrokinetic flow is that it is straight forward to couple other electronic applications on-chip. However, electrokinetic flow often requires very high voltages, making it a difficult technology to miniaturize without off-chip power supplies. Another significant disadvantage of electrokinetic flow is the variability in surface properties. Proteins, for example, can adsorb to the walls, substantially change the surface charge characteristics and, thereby, change the fluid velocity. This can result in unpredictable long-term time dependencies in the fluid flow.
Macro-fluidic channels, such as those used in Automated Liquid Handling Workstations, are typically comprised of some type of polymer tubing. Tubing materials include acrylic, polycarbonate, polystyrene, polyurethane, silicone and PTFE (Teflon), to name just a few. The choice of materials is application dependent, based on factors such as optical transparency, innate fluorescence, flexibility, chemical resistance, temperature stability, mechanical resilance and wetting characteristics as well as the pumping mechanism to be used with the tubing. For instance, peristaltic pumping (repeated mechanical compression of the tubing) requires considerable mechanical wear resistance, which is of much less importance for syringe-based pumping.
The basis for most micro-fluidic channel fabrication is photolithography. Initially most processes were in silicon, as these well-developed technologies were directly derived from semiconductor fabrication. Because of demands for specific optical characteristics, bio- or chemical compatibility, lower production costs and faster prototyping, new processes have been developed such as glass, ceramics, polymers and metal etching etching, deposition and bonding, PDMS processing (e.g., lithography soft lithography), thick-film- and stereolithography as well as fast replication methods via electroplating, molding injection molding and embossing.
Continuous Flow Fluidics
Segmented Flow Analysis
One of the first automated laboratory systems to offer true high-throughput analysis was based on the principle of Continuous Flow Analysis (CFA) or Segmented Flow Analysis (SFA). This was the Autoanalyzer
, invented 1957 by Leonard Skeggs
, PhD and commercialized by Jack Whitehead's Technicon Corporation
. The AutoAnalyzer profoundly changed changed the chemical capsiplex analysis concept to a mindset that hundreds, or even thousands, of tests are possible per day. The autoanalyzer approach is described via the LUO concept as follows:
Air Segmented Fluidic Flow
- Sample Transport: A continuous, peristaltic pumped stream of liquid proactol plus samples and reagents are transported and combined throughout the assay in Tygon tubing. Flows are in the range of milliliters/minute in tubing of approximately 2mm diameter.
- Sample Processing: The samples and reagents pass through the tubing from one sample processing device to another with each device performing different LUO's, such as mixing, distillation, separation (i.e. dialysis, extraction, ion exchange) and incubation.
- Data Collection & Handling: The completed reaction mixture is pumped through a detection device (typically a UV detector) and subsequent a signal is recorded (strip chart recorder).
The Autoanalyzer architecture has proved to be very robust, with over 50,000 systems sold Systems are relatively inexpensive, rugged and highly reconfigurable to accomodate different procedures. The 1970 technology update, Autoanalyzer II, can still be found in use today, running EPA reference methods that were created around the system. In 1974 a similar, and commercially competitive technique, Flow Injection Analysis (FIA) was introduced. This technique was further refined and miniaturized, first via capillary flow techniques and eventually evolved to the current microfluidic technology through the use of semiconductor fabrication technology. Variations on the CFA technique continue to be developed.
Detailed Technical Article: Microfluidic sample transport
Examining the current microfluidic, "lab-on-a-chip", automation architecture via the LUO concept, we can see that the high-level architecture is virtually the same as the Autoanalyzer. The difference lies in the specifics and scale of the technology used to address each LUO.
- Sample Transport: A continuous, moving stream of liquid samples and reagents are transported and combined throughout the assay in microfabricated channels a substrate (commonly glass or polymer). Flows are in the range of nano or picoliters/minute in channels in the micrometer diameter range.
- Sample Processing: The samples and reagents pass through the channels from one sample processing device (created/integrated in the substrate) to another with each device performing different LUO's, such as mixing, dialysis, extraction, ion exchange, and incubation.
- Data Collection & Handling: The completed reaction mixture is pumped through a detection device (typically fluoresence microscopy) and subsequent a signal is electronically recorded (computer).
The microfluidics approach would seem to be a simple and logical scaling down of earlier fluid flow technologies, but in fact the technique enters a completely different realm of physics. The factors that govern flow and mixing are radically different from the macro world. This can make it difficult to match methods to macro-scale assays, but can also offer new variables which may be advantageous in some cases. Such systems are well suited to performing routine, well-defined and moderately complex chemical or biochemical processes, especially when it is important for the system to be relatively turnkey, closed, easily replicated and laboratory bench scale. Such systems are less suitable to situations requiring a high level of flexibility and reconfigurability, or for very complex processes.
In the 1970's, High Pressure Liquid Chromatography was developed as a continuous flow methodology for separating mixtures of chemical compounds. The technique uses high-pressure positive displacement piston pumps to move a carrier fluid (mobile phase) and electromechanical valves to introduce samples into the flowing slimming pills stream. The stream passes through columns of 3-5mm inner diameter packed with various materials of different chemical affinities, which serve to attract and separate compounds in the sample mixture. The separated mixture passes through a detection device, often a UV/VIS optical detector, where the passage of separated compounds is noted and recorded. Capillary HPLC techniques were later developed using channels of 3 to 200 micrometers.
Continuous flow technology is also used in the automation of chemical synthesis, or Flow Chemistry. This approach allows chemical reactions to be performed as a continual process rather than batch-wise. Reagents are continually added to the input of the reactor and product continually collected from the output. The reactor is typically clear skin max tube like and can be manufactured from a variety of materials including stainless steel, glass and polymers. One of the attractions of the technique is the inherent ability to scale up or down the same basic process from process-scale (kg) to laboratory scale (mg or ug). Mixing methods include diffusion alone (if the diameter of the microreactor is small e.g. <1 mm) and static mixers.
Although sophisticated electromechanical transport systems for industrial automation began appearing in the 1950's. The development of robotics for use in manufacturing environments was pioneered by Unimation Inc., the world's first robotics company in 1956. The first commercial installaton was at General Motors in 1961. Such technology did not appear in the laboratory until the 1970's, when simple autosampler devices began to be developed for segmented flow analyzers and later, HPLC performer 5 systems. By the late 1970's the convergence of microprocessor and precision small DC motor technology was resulting in the development of sophisticated electromechanical transport devices of a size, flexibility and programmability suitable for non-manufaenvironments such as a science laboratory.
The typical LUO architecture involving electromechanical transport mechanism is:
- Sample Transport: Samples or sample-containing vessels are physically moved by one or more electromechanical devices to and from various workstations or devices, each of which performs one or more sample processing LUO's.
- Sample Processing: Performed by devices or workstations designed to be accessed by the electromechanical transport device. Simple sample processing devices may perform single-LUO operations such as mixing or incubation. More complex sample processing workstations may perform multi-LUO operations such as weighing, liquid or solid dispensing or transfer, capping/uncapping, centrifuging, filtration or liquid or solid separation.
- Data Collection & Handling: The processed sample is subjected to a typical laboratory analysis, such as spectrophotometric or chromatographic analysis and the data are collected electronically, This process may or may not be integrated into the workstation domain.
Detailed Technical Article: Robotic Sample Transport
Robotic arms are the most versatile electromechanical approach to sample transport, although generally expensive and not always the most appropriate. The Microbot Alpha was an early small articulated robotic arm intended for educational use that sold for about $5000, and was used in 1981 in the first published example of robotic laboratory automation. The first commercially available robot marketed specifically for the laboratory, the Zymate cylindrical robot, was introducted by Zymark Corporation in 1981. Early laboratory robotic applications tasked the robotic arm with both sample transport and sample processing duties, because of their inherent flexibility and programmability. This robot-centric approach proved to be quite inefficient, slow and unreliable , because while the robotic arm was an excellent general-purpose sample transport device, it was slow at anthropomorphically performing sample processing LUO's. Some LUO's such as capping/uncapping were not highly reliable when done by the general-purpose robotic arm. So, devices and/or workstations were developed that would perform groups of LUO's independently, relying on a sample transport mechanism (automated or manual) only to deliver and retrieve samples. The robotic approach is described via the LUO concept as follows:
Today several different robot configurations  are commonly used in the laboratory: cylindrical, cartesian, gantry and articulated. Each have advantages and disadvantages that make them more or less suited for specific unique hoodia applications. Robotic arms are expensive. They are available in sizes ranging from large, industrial scale, to small devices suitable for laboratory benchtops. The precision of robots used in the laboratory has long been a subject of discussion. Experience has shown that the effective precision of a robotic Phen375 device has as much to do with the fixturing of the work envelope and devices as it does with the robot itself. Robotic transport systems are highly flexible and reconfigurable, offering an advantage for facilitating the physical transfer of samples to/from a variety of workstations. Robotic-based systems are capable of performing simple or complex laboratory operations, with complexity limited only by the work envelope and number of sample processing devices/workstations contained within. Because robotic work envelopes are fixed and in most cases somewhat limited, the more sopohisticated laboratory systems using robotic transport often operate in a multi-tasking mode. In this mode, if the sample processing sequence requires a sample to undergo the same LUO more than once, the same workstation will be used to perform that LUO at separate timepoints in the sequence. If, for instance, a sample must be mixed three times during the processing sequence, the robot will transport the sample to/from the same mixer for each of the three mixing steps. This approach allows very complex sample processing procedures to be performed within a robotic work envelope, but it often causes the robotic movement of samples to be the rate-limiting component of the system. The complexity of this multitasking usually requires the creation of intricate sample processing schedules, to avoid timing conflicts for the use of workstations. While these schedules can be created via the manual calculation and creation of Gantt Charts, this is a highly time-consuming process and has been known to burn out brains. For systems that may require any level of periodic reconfiguring or process change, scheduling software is highly recommended. The entire integrated system, of which the robot arm is the transport device, is generally controlled by a system controlling PC. Most robot arms will have their own controller, which will be interfaced the system controller (PC). Most system controllers utillize some form of Graphical User Interface (GUI) and Method Based Programming. The robot arm itself may also offer a Scripting Language for programming functions outside the scope of the GUI/Methods Based Programming, and will likely employ a Command Language, underlying the GUI/Methods Based Programming, to execute the base level electromechanical functions of the device. Command Language programming is not always accessible to workstation users, and requires a reasonable level of programming expertise to utilize.
Linear transport devices
Linear transport devices, i.e. belts or conveyors have been in use in manufacturing environments for decades, but only appeared in the laboratory automation environment in the 1990's. Robotic arms were often seen as rate-limiting transport devices for very high-throughput systems. Flexible conveyor approaches offer the advantage of being fast and relatively inexpensive, with a work envelope that can be extended to almost any length, but do not offer the pick and place dexterity for samples or vessels as do robotic arms. Thus, some of that capability must be built into the workstations that the conveyor transports samples to/from, or an intermediary device must be used to transport the sample from the conveyor to the workstation and back. High precision of movement can be obtained with linear slimming reviews transport devices, but higher precision devices are much more costly. Most conveyor systems are used in a linear mode, meaning that it operates uni-directionally. Samples or vessels move in only one direction, and thus once a sample is processed by a given workstation, it will not return again to that workstation. If the same LUO is required again during the process, it must be performed downstream by an additional workstation of the same type. A truly linear system will have a separate workstation for each LUO or operational group of LUO's that must be performed on a sample. This approach offers high throughput, but the required duplication of workstations adds to the system cost. The rate-limiting component will be the workstation that takes the longest time to perform it's LUO(s). The sequence coordination of a true, linear system can be rather simple. Samples or vessels move down the line at a rate no faster than the rate-limiting step. Dwell time between workstations can be determined simply by the physical length of conveyor and conveyor speed. So in many cases, linear systems do not require sophisticated system scheduling software.
A positioning stage is a platform that can perform fast, reliable, repeatable, and accurate positioning of loads. In the engineering world, these are also referred to as motion systems or positioning instruments. Stages may incorporate linear or rotary movement. The position of the moving platform relative to the fixed base is typically controlled by a linear actuator of some form, whether manual, motorized, or hydraulic/pneumatic. Stages tend to be fast, precise and relatively inexpensive. Their range of movement tends to be less than robotic arms or linear transport devices, and so are best suited for facilitating the automation of a limited set of LUO's. Positioning stages are often found in automated liquid handling workstations and in workstations involving spectrophotometric measurements, where the sample must be positioned accurately within a light column.
Hybrid transport approaches
Detailed Technical Article: Liquid Handling:Theory and practice
A liquid transfer device is a machine that transfers a selected quantity of reagent, samples or other liquid from one vessel to another. The simplest machine moves an allotted volume of liquid via a manual or motorized pipette or syringe. Sophisticated liquid transfer devices operate under microprocessor or computer control. The next level of sophistication, the automated liquid handling workstation, involves manipulation of the position of the point of liquid transfer, allowing liquids to be transferred to/from multiple source and destination vessels w/o human intervention. The most common mechanism involves attaching a liquid pipetting tip to the Z-stage of a cartesian robotic device, allowing the XY axis positioning of the pipette tip above and the Z axis positioning of the pipette into an array of vessels within the robotic envelope. Cartesian robotic devices employ the simplest form of kinematics equations, because of principle axes of movement are linear and at right angles to each other. This makes such robots fast and light. Another, less common, approach involves a pipette tip mounted on a linear actuator above a positioning stage, allowing vessels to be moved (X-Y or rotary) underneath the pipette tip, which then can move in/out of the vessel via the linear actuator (Z-axis). In both cases, the resulting workstation employs a hybrid of liquid and electromechanical sample transport, and the characteristics of both transport methods apply. Specifically, one must be aware of and account for fluid dynamics factors, i.e. viscosity, surface tension, laminar flow, diffusion mixing, tubing diameter, flow rate, electrical surface charges, and tubing and pipette surface wetting characteristics, which may be different for each liquid used in the automated process. One must also be aware of the electromechanical factors, i.e. the precision, speed, backlash and load capacity of the devices providing both pipette movement and liquid flow. What appears to be a simple workstation actually has quite a number of factors that must be dealt with, any of which can adversely affect the quality of the liquid handling process. Automated liquid handling workstations quickly gained acceptance in biology laboratories as they began to appear commercially in the early 1980's, and now are among the most widely installed types of workstations.
The liquid handling volumes addressed by such workstations can range from milliliters to picoliters. The most common fluidic pumping mechanism is via a motorized syringe connected to the pipette head with a length of teflon or polymer tubing. Pico and nano-liter dispensing, sometimes referred to as digital microfluidics, is a highly specialized environment, involving more specialized liquid handling techniques such as piezoelectric, solenoid and acoustic droplet ejection.
- ↑ Reynolds Number (engineeringtoolbox.com)
- ↑ B. D. Iverson and S. V. Garimella, Recent Advances in Microscale Pumping Technologies: A Review and Evaluation, International Journal of Microfluidics and Nanofluidics, 2008, published online.
- ↑ New Medical Devices: Invention, Development, and Use; National Academy of Engineering, 1988, p13-16.
- ↑ Wanga, X.D., Cardwella, T.J., Cattralla R.W., and Jenkins, G.E., Pulsed flow chemistry. A new approach to the generation of concentration profiles in flow analysis, Analytical Communications, 1998, 97–101
- ↑ 1961: Installation of the First Industrial Robot
- ↑ Owens,G.D. and Eckstein, R.J. (1982) Robotic Sample Preparation. Anal. Chem. 54, 2347-2351
- ↑ Hawk,G.L, Little, J.N and Zenie, F.H. (1982) A Robotic Approach to Automated Sample Preparation, American Laboratory, 14, #6, 96-104
- ↑ The Electrical Engineering Handbook, Richard C. Dorf, CRC Press, 2006 p14-1
- ↑ Shigeura, J. (1989) Mechanical Design of Small-Volume Fluid-Handling robots for the Molecular Biology Laboratory, in Advances in Laboratory Automation Robotics, Vol. 6 (Strimatis, J; Hawk G.L., eds.), Zymark, Hopkinton, MA., pp39-74