MULA, an affordable framework for multifunctional liquid automation in natural- and life sciences with a focus on hardware design, setup, modularity and validation

Abstract

The implementation of automation has already had a considerable impact on chemical and pharmaceutical industrial laboratories. However, academic laboratories have often been more reluctant to adopt such technology due to the high cost of commercial liquid handling systems, although, in many instances, there would be a huge potential to automate repetitive tasks, resulting in elevated productivity. We present here a detailed description of the setup, validation, and utilization of a multifunctional liquid automation (MULA) system that can be used to automate various chemical and biological tasks. Considering that such a setup must be highly customizable, we also designed MULA with respect to modularity, providing detailed insight as far as possible. Including all 3D-printed parts and the used Hamilton gastight micro syringe, the total construction cost is approximately 700 €. This allows us to achieve a highly reliable and accurate system that exceeds the precision of a classical air displacement pipette while still retaining the ability to use closed vial (septa) setups. To encourage other groups to adopt this setup, detailed instructions and tips for every step of the process are provided, along with the complete CAD design of MULA and control code, which are freely available for download under the CC BY NC 3.0 license.

Keywords: 3D printing; Biology; Chemistry; Liquid handling; Micro-syringe; Pipetting.

Graphical abstract

Fig. 1

CAD model of MULA: a) assembled frame with 3D-printed parts (orange) and 2040 V-slot profiles (white); b) Sampling head of MULA. For visibility reasons, some parts, like the electronics board case, are not depicted in this illustration. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Scheme 1

Command-chain overview of MULA: The user inputs parameters to the PC, which are transformed into G-code commands, which the software then sends to the control board. The firmware on the board uses these instructions to control the stepper motors, which make MULA move accordingly and conduct the experiment.

Scheme 2

Decision tree for the steps to flash the firmware to the control board.

Fig. 2

CAD model of MULA in different orientations: a) Diagonal view b) side-view c) top-view.

Fig. 3

Assembled structural frame of our implementation of MULA: A double cuboid made from 2040 Nut 6 aluminium profiles with dimensions of 800 x 600 x 600 mm. 2040 V-slot profiles are illustrated in white.

Fig. 4

CAD model of the assembled X-axis. Note that several parts were omitted, including screws, pulleys and belts.

Fig. 5

CAD model of the assembled Y-axis. Note that several parts were omitted, including screws, pulleys and belts.

Fig. 6

CAD model of the assembled I-, Z-axis. Note that several parts were omitted, including screws, pulleys and belts.

Fig. 7

Moveable rack assembly.

Fig. 8

Drag chain setup of MULA.

Fig. 9

Octopus control board; with a) showing the fan connector and fan power jumper (12 V). b) Show the driver's settings for UART mode. c) showing the sensorless DIAG Pings and the connected cable for the Z-endstop. The connected stepper motors to the X₁-,Y-, Z-, I- and X₂-axes are shown on the top left (more information at https://3dwork.io/en/btt-octopus/).

Scheme 3

Overview of the correlation of steps, mm and µL.

Scheme 4

Pronterface tutorial: a) When opening Pronterface, the first step is always to establish a serial connection. In our case, COM-Port 3 is assigned to MULA. However, this can differ between setups. b) To conduct a first-vial calibration, start by homing all axes and then manually move the X- and Y-axis to the right position above the vial. Lower the Z-axis to verify and then request the position of the head by sending the M114 command via the terminal. c) To execute a G-code file, simply load the file and click the button labeled “Print” to start the experiment.

Fig. 10

The Syringe Volume calibration.exe file. The software, as depicted, gives the possibility, depending on which syringe is installed, to take 10%, 50%, or 100% of the maximum volume. In addition, the user can choose between different pulling modes to calibrate the syringe. Furthermore, the backlash correction can be calibrated as well by changing the config file.

Fig. 11

GUI of the program for general liquid handling.

Fig. 12

Visual explanation of several parameters from the Rack section in the configuration file. For the depicted 30Vial rack, the following values are given: vials_per_row = 10; columns = 3; dx_s, dy_s = 15 [mm], solvent_y_increment = 35 [mm], number_of_solvents = 4 (when the waste is located somewhere else; otherwise set to 3 and calibrate the position of the most right solvent as waste coordinates.

Fig. 13

Solvent layering experiments with MULA: a) optimization of the best dispensing velocity. Interestingly, going very slow did not yield the best results because the liquid then simply dropped down from the needle, disturbing the lower layer. We have found that a feed rate of 240 works best with a type 5 needle tip to ensure the liquid is dispensed via the glass vial wall. b) Single crystals of D-glucose after several layering attempts of a saturated aqueous D-glucose solution with acetone.

Fig. 14

Validation results of MULA in comparison to an air displacement pipette with water and acetonitrile (ACN) as solvents. a) Both MULA and the tested pipette show comparable accuracies within the ISO 8655 limits when water is used as a solvent. However, when an organic solvent like ACN is of interest, MULA delivers significantly better accuracy. b) For small volumes (100  µL), MULA delivers slightly better repeatability between measurements. For larger volumes (500, 1000 µL), interestingly, under the tested conditions, the pipette fails to fulfill the ISO 8655 limits, while MULA again yields significantly lower repeatability errors.