P-80 PC Flaming-Brown Micropipette Puller

P-80 PC Flaming-Brown Micropipette Puller
P.O. BOX 3592
(415) 459-7327
The Model P-80/PC Brown-Flaming Micropipette Puller combines a proven pulling technology with programmability to produce a very versatile instrument. The pulling mechanism is
derived from the P-77/P-80 series of pullers, which have demonstrated the ability to pull a
complete range of pipette profiles. Added to this mechanism is the ability to program different pulling sequences; thus, allowing ease of use for pulling a multiplicity of pipettes on one
The P-80/PC is a 'velocity sensing' puUer. This feature allows the puller to indirectly sense the
viscosity ofthe glass, giving the P-80/PC the ability to pull pipettes from all glasses save quartz.
Even difficult to pull formulations, such as aluminasilicate glasses, are handled with relative
The P-80/PC can fabricate pipettes for use in such researches as intracellular recording, patchclamping, microinjection and microperfusion. However, realizing the full potential of this instrument is dependent on a complete understanding of the way it implements the puUing
process. To this end we urge that this manual be read in its entirety. To aid in understanding
the function of the instrument, sample programs are already loaded in memory (as discussed
in subsequent material).
For that setting in which a number of users must work with one device, or for that single user
whose investigations require a variety of pipettes; the P-80/PC is the answer.
The Model P-80/PC is shipped to you wrapped in a plastic bubble wrap and surrounded by approximately 6 inches of loose fill on all sides. Please take note of this method of packaging.
Should it ever be necessary to ship the puller to another location, the same method of packaging should be employed.
IMPORTANT: Improper packa^nf! is a form of abuse, and as such can he responsible for voiding the warranty WKPTP shipping damage is .sustained at a remit of such packing.
All material shipped with the instrument is contained within or is attached to the plastic wrapping.
After removing the puller from its shipping carton, cut the tape holding the plastic wraps. Be
careful not to cut through the power cord at the back of the device. Open the plastic cover by
lifting the hinged front and remove any filament containers. These containers are usually located on the left side in front of the main valve on the nitrogen tank. Those pullers ordered
without a nitrogen tank will have their filaments shipped in a small container attached to the
front of the plastic wrap. While the puller cover is still raised, open the valve atop the nitrogen
cylinder by turning it counter-clockwise (as viewed from the valve end of the cylinder).
If you have ordered your puller without a nitrogen cylinder, regulator and cradles it will be
necessary to secure a separate cylinder and regulator to deliver nitrogen at 50 psi, via the attached length of tubing, to the gas solenoid. This nitrogen source need not be dry or medical
grade; however it should not be a source that is subject to changes in moisture content (as is
the case with most piped in sources). Should it be more readily obtainable, compressed air is
also suitable, subject to the same moisture considerations.
Place the puller in a location where there is a free flow of fresh air on all sides. The fan draws
air in through the vents on the sides and exhausts out both ends of the heat sink. NEVER
Since the P-80/PC is a microprocessor controlled device, it should be accorded the same system wiring precautions as any 'computer type' system. If microprocessor based systems in the
lab require line surge protection for proper operation, then the same protection should be
provided for the P-80/PC.
Plug the instrument into a power receptacle ofthe correct voltage and frequency. NOTE: A
power source that is essentially electrically noise free is desirable. The control circuitry of the
puller uses digital logic that may be susceptible to transient spikes that can be caused by faulty wiring or noise producing machinery, such as centrifuges or other equipment utilizing SCR
control circuitry on the same power lines.
Lift the front of the plastic cover; open the main tank valve located at the left end of the
nitrogen tank. Open the valve fully (counter-clockwise). It should not be necessary to tum off
this valve between electrode pulling sessions. The high pressure meter should read above 1500
psi. Adjust the regulator to a reading of 50 pounds on the low pressure meter. If this reading
is at a value higher than 50 pounds, tum the regulator knob counter-clockwise one tum. No
change will be observed in this pressure reading until an elecfrode pipette is pulled. The pressure will then drop. After several electrodes have been pulled, the pressure will stabilize at a
lower value. Now readjust the pressure to 50 pounds.
Pressures greater than 50 pounds are not recommended.
Programs 0 and 1 are preprogrammed into the puller by the manufacturer. Both programs
were developed using 1mm O.D. borosilicate glass with an I.D. of .5mm. Program 0 will pull
a micropipette of less than .1 micron and program 1 will pull a patch-pipette of about 1 micron.
To changefromone program to another press the reset button on thefrontpanel, and then
select the program on the keypad. DO NOT USE OTHER PROGRAMS OR CHANGE THE
To execute program 0 or 1, place a piece of glass in the carriers and press the pull key on the
This section presents a basic mechanical description of the P-80/PC, with particular emphasis
on terminology. Knowing the names of the various parts greatly facilitates communication between the investigators and the factory when discussing adjustments or service problems. In
addition, various controls and adjustments on the top of the instrument are located and
described. Those adjustments which are considered part of maintenance procedures are dealt
with in Section 6.
GAS SUPPLY TANK AND REGULATOR: The Model P-80 gas supply consists of a tank
filled with nitrogen of approximately 2 1/2 pounds at a pressure of 1800 psi, a regulator with
high pressure and lowpressure gages, an electronically controlled solenoid valve, a micrometer
flow valve, and a nozzle.
The gas should last several years with normal use. Continuous lowering of high pressure might
indicate a leaky system, depending on usage. The tank may be refilled by any supplier of compressed gases. Refill only with nitrogen.
When installing a refilled tank, make certain all the fittings are tight.
CAUTION: Be certain to shut off the main tank valve before removing the tank or the
regulator from the tank. The tank is at high pressure and should be handled with care.
CYLINDER MAIN VALVE: (Fig.l, a) This valve is closed at the time of shipment. Upon
receipt of the instmment, this valve should be opened all the way. This is a double seat valve
and should not be left partially opened; small leaks may occur. If the instmment is to be left
unused or stored for any length of time (two weeks or more) this valve should be closed.
NITROGEN REGULATOR: This regulator isfittedwith an industry standard CGA 580 connection. The high pressure and low pressure gages read, respectively, cylinder pressure and
pressure of gas delivered to the micrometer flow valve. The large central knob adjusts the
deUvery pressure and is normally set at 50psi.
GAS SOLENOID: (Fig. 1, c) This solenoid is activated during the hard pull phase of the pulling cycle. It admits nitrogen to the micrometerflowvalve.
MICROMETER FLOW VALVE: (Fig. 1, d) This micrometer valve (needle valve) controls
theflowof gas to the filament. Despite the indicated range of values, this valve has a working
range of approximately 100 units from no flow to maximum flow. The micrometer markings
refer to thousandths of an inch and are referred to as numerical indication x 1000 (i.e. a reading of .1 inches is called 100).
NOZZLE: (Fig. 2, e) The nozzle conducts gas from the flow valve to the filament area. It is
a press fit into an o-ring located at the front of theflowvalve. The nozzle tip is usually located
from 1 to 2 millimeters below the filament and centered on it.
NOZZLE HOLDER: (Fig. 2, f) Maintains the nozzle in position below the filament. The
holder contains two adjustments. The screw on the top allows the nozzle to be rotated and
moved in and out of the front of theflowvalve. The screw that secures the holder to the filament block can be loosened allowing the nozzle to move up and down.
FILAMENT BLOCK ASSEMBLY: (Fig. 1, g) The filament block assembly is made up of
several pieces of hard black nylon. Wires supplying current to the filament are attached to
threaded 'posts'. This current is carried to the filament via the upper and lower heater jaws.
Note that these jaws are slotted and may be moved up and down by loosening the screws that
secure them to the front of thefilamentblock assembly. If the jaws are moved, make sure that
the securing screws have been tightened; otherwise poor currentflowcan result. It is desirable
to keep the gas solenoid/flow valve assembly lined up with the nozzle to prevent leaks at the
o-ring. Note the two screws in slots at the back of the filament block assembly. Loosening
these screws allows the gas solenoid/flow valve assembly to slide left or right.
ANGLE PLATE: (fig. 2, h) The angle plate secures thefilamentblock assembly to the cover
plate; it contains two important adjustments. Note the chrome plated screws in slots at points
A and A' and the locking screws in slots at points B and B'. The chrome plated screws are
'eccentrics'; by rotating them with a screwdriver thefilamentblock assembly can be moved up
and down (A) or forward and back (A') to adjust the position of thefilament.Loosen the locking screw associated with each 'eccentric screw' before turning, and tighten after completing
the adjustment.
COVER PLATE: (Fig. 2, j) The cover plate conceals the entry of the pulling cables in to the
base of the instmment. It is attached to the top by two screws, in slots, at points E and E'.
Loosening these screws allows thefilamentblock/angle plate assembly to move forward and
back over large distances.
NOTE: The movements of the cover plate and the jaws constitute the 'coarse adjustments' of
filament position, while the eccentric screws are the 'fine movements'.
PANELS, LEFT and RIGHT: (Fig. 3, k) The panels are the angled surfaces tiiat provide
mountings for the puller bars and their bearings, the spring stops, the bumpers, and the upper
cable pulley assemblies. Except for minor differences in shape, the left andrightpanels are
identical. Note the three socket-head cap screws that attach each panel to the top. These
screws are used to align the pullers bars. TTieir use, if necessary, is covered in the maintenance
UPPER CABLE PULLEY ASSEMBLY: (Fig. 3, m) This assembly conducts the pulling cables
from the puller bars to the centrally located (and concealed) lower pulley assembly. Note that
this assembly is attached to its panel by two screws, in slots, and contains a large eccentric adjustment screw (G). This eccentric screw is used to adjust cable 'tension' its use is covered in
the maintenance section.
BUMPERS: (Fig. 3, n) The bumper stops the motion of its associated puller bar.
SPRING STOP: (Fig. 3, r) This assembly consists of the puller bar, glass clamp, clamp (wing)
nut and cable retaming screw. The puller bar is made of mild steel and coated with a controlled thickness of hard chrome. Glass is loaded into the groove near the tip of the puller bar
and is held in position by tightening down the clamp nut. The cable retaining screw holds the
cable in a shallow groove at the end of the puller bar, and forms the 'resistance' against which
the cable ends pull.
V-BEARINGS: (Fig. 3, s) These bearings are the guides for puller bar motion. They are made
of stainless steel and should never be oiled (see maintenance section). Note that these bearings are mounted on stainless steel bushings one of which is round with the other two being
hexagonal. The hexagonal bushings are used to adjust position and ease offravelofthe puller
bars (see maintenance section).
PULL CABLE: This cable conducts the pulling force of the solenoid to the puller bars via the
upper and lower pulley assemblies. It is made offlexiblemetal with a nylon coating. NEVER
PINCH OR DISTORT IHE CABLE. The cable is terminated witii crimped-on clamps or
The IQH is the metal plate on which is mounted the panels, cover plate, etc. Two other items
are the cradles which support and hold the gas cylinder and the cover stops to which the plastic cover is secured. The base contains the transformers, circuit board and pull solenoid assembly.
The aim of this section is to provide the user with the information necessary to operate the P80/PC. It begins with some important definitions and descriptions of the front panel controls.
Program: A program consists of one or more cycles, which when executed in sequence will
'pull' the capillary glass inserted in the instmment, A program can be up to 16 cycles in length.
Cycle: A cycle consists of a HEAT VALUE (range 000 TO 999), a PULL value (range 000 to
255), a VEL.(VELOCITY) value (range 000 to 255) and TIME value (range 000 to 255).
Loop: A loop consists of one or more cycles that are repeated by the instmment until glass
separation is achieved.
Ramp Test: A program, resident in the ROM, designed to facilitate program alterations when
it is necessary to change filaments. This test is discussed in subsequent text.
Power: When pushed to the up position, this switch applies power to the instmment.
Reset: When momentarily pushed, this switch resets the microprocessor to an initialized condition. It is used to change from one program to another, and in other circumstances as
described below.
Keyboard: There are three 'groupings' of keys on the keypad: numerical/decision, editing and
control. They function as follows:
Numerical/Decision (0-9): These keys are used to enter the number of the program being
chosen, the various values for HEAT, PULL etc. and to make yes/no (1/0) decisions in certain
Editing (CLR, ENT, NEXT, LAST): These keys are used for entering, deleting and editing
programs. They allow one to move forward and back through a program, enter new values,
and clear out unwanted values. In addition, the CLR key is the access key to the Ramp Test.
The function of the various keys will be explained in the context of instrument operation in
subsequent text. Control (PULL, STOP): These keys control the initiation and cessation of
program execution and control of the Ramp Test.
Apply power to the iostrument. After an automatic 'power on' reset, the display will appear
as follows:
he P-80/PC is shipped with two programs already stored in memory. Program 0 is for a
micropipette in borosilicate or hard glass. Program 1 is a patch-type pipette in borosilicate or
hard ^ass. Programs 0 and 1 were written for 1mm OD/.5mm ID borosilicate glass (and a gas
flow micrometer value of 100). Assuming that 1mm OD/.5mm ID glass is available, proceed
in the following manner. Raise the cover and load a piece of glass into position. This is best
done by loosening the glass clamp on therightor left puller bar and sliding the capillary through
the clamp until it projects about 1 centimeter beyond the clamp. Release both puller bars and
pull them to the center until they stop. Now sUde the glass through the filament and into the
glass clamp on the puller bar and tighten the two wing nuts on the glass clamps. The wing nuts
can be tightened quite a bit without breaking the glass, but a tremendous amount of force is
unnecessary. In this particular instance the user may wish to leave the top in the up position
in order to watch the pulling process; however, in normal use the cover should be down whenever a pipette is being pulled. Now press the number 1 on the keypad; cycles 1 and 2 of program
1 will appear. Press the PULL key and the puller will execute program 1. Whether one obtains a patch-type pipette or not depends on several factors. The programs contained in
memory were written for a particular environment (ambient temperature and humidity) and
type of glass. Remove the pulled pipettes from the glass clamps, close the cover and toggle
Once again the display shows the sign on message:
At this point there are three options: 1) choose a program, load glass and execute the pull; 2)
create a new program or edit an old one; 3) mn the Ramp Test. The user should now create
a program, (please note that the programs that will be written in the following text are not
meant to pull pipettes, but are intended as an exercise to help develop an understanding of
First press a key other than 0 or 1. The display should come up with no values for HEAT,
PULL, etc. For example, if 3 were pressed, the display would look like this:
3 01HEAT=
3 02HEAT=
with the cursor blinking in the leftmost position of the heat value on the first line.
If there are already numbers or symbols entered as program values, make sure that this
program was not entered by another user of the puller. Unused program areas are usually
cleared before a puller is shipped, but occasionally random values or test program values are
inadvertentiy left in memory. Since the program values for heat may be sufficient to damage
the heater filament, we recommend that unused programs be cleared completely before
proceeding. The clear function is outlined below under the heading 'CLEAR'.
Press a series of three numbers such as '333'. Notice that these numbers are loaded in the
HEAT value, and thattiiecursor has moved on to the PULL value. IF THREE NUMBERS
entered for TIME, the cursor moves to the next HEAT value and the display 'scrolls' to the
next cycle. For PULL now enter two digits such as 10. In order to complete the entry, press
the ENT key. The two digit entry isrightjustified and the cursor moves to the next position.
Enter a value for TIME (IE. 80) and press ENT. Enter another set of values, such as 320 for
HEAT, 60 for PULL, 10 for VELO, and 80 for TIME. Currently the cursor is placed in tiie
HEAT value for cycle 03. Press LAST; this will move the display back to the 'last' cycle. Press
LAST again, and the display will look as follows;
3 01HEAT= 333 PULL= 10 VEL.= 30 TIME= 80
3 02HEAT= 320 PULL= 60 VEL.= 10 TIME= 80
Now press NEXT; the display scrolls up one line, and the cursor is on the 'next' line of the
USE THE ENT, NEXT AND LAST KEYS. Remember, the cursor only moves totiieright.
If the cursor were in the PULL value position on line 01 above, and one wished to change the
HEAT value; press ENT three times to arrive at the HEAT value on line 02, and then press
LAST. The cursor will be in the correct position for entering the new value. PRESSING ENT
It is appropriate at this time to define the units attached to HEAT, PULL etc.
HEAT: A change of one unit (ie 333 to 334) represents a change of 50 milliamps in the current through the filament. Generally changes will be made in steps of about 5 units since in
most cases smaller changes will have no effect.
PULL: A change of one unit represents a change of 4 milliamps in the current through the pull
solenoid. In the case of pull strength useful changes are 10 units or more to see an eft^ect.
VELO: One unit represents a change of one or more millivolts of fransducer output depending on the transducer being used. Useful values for velocity range from 10 to 100 with the
lower values being used for patch and injection pipettes and higher values for micropipettes.
TIME: One unit represents .5 milliseconds if the velocity value is 1 or greater. One unit represents 10 milliseconds in the VEL.(VELOCrrY) = 0 mode. See subsequent text for further
Assume for the moment that the two line program entered above, if executed, caused a glass
capillary to sfretch but did not cause the glass to separate. What happens next? The puller is
'aware' of the fact that the glass has not separated, and will go back to line 01 of the program
and tiy again; in effect it begins 'looping'. It will continue to do so until the glass separates.
This looping capability is very useful. For example, consider the following two-line program:
3 01 HEAT= 350 PULL= 10 VEL.= 25 TIME= 80
3 02 HEAT= 320 PULL=
VEL.= 10 TIME =
Assume that, after loading a piece of glass into the puller and executing the program above,
that the filament came onfivetimes before the glass separated. This indicates that the puller
was into the third time through the program (looping) when the glass separated. Further more,
assume that the result of the pull wasn't quite the pipette profile being sought; possibly because there was too much heat on the last pull. Then one might constmct a new program that
read like so:
3 01 HEAT= 350 PULL= 10 VEL.= 25 TIME= 80
3 02 HEAT= 320 PULL=
VEL. = 10 TIME =
303 HEAT= 350 PULL= 10 VEL.= 25 TIME= 80
3 04 HEAT = 320 PULL =
VEL. = 10 TIME =
3 05 HEAT= 300 PULL= 10 VEL.= 25 TIME= 80
Note the reduction in the heat value in cycle 05. This illustrates how the looping capability
can be used to create a multi-step program designed to pull in one program execution. At the
completion of the pull the puller will report on the number of times it looped and which line
it was on at the time the pull took place. In the two line example above it would report the
a 2 o.
The display then shows thefirstline of the program, and is ready for another pull. NOTE: IF
Another user decides to use the program above, but the glass he/she is using is slightiy different (thinner wall, different composition). Three possibilities exist: 1) glass pulls in same
number of cycles; 2) glass pulls in less cycles, or; 3) glass pulls in more cycles. IF A PULL OCCURS AT A CYCLE OTHER THAN THE LAST ONE IN THE PROGRAM, THE CURSOR WILL BE LOCATED ON THAT LINE AT WHICH THE PULL TOOK PLACE. In
the case of our program above, if the pull took place during the cycle on line 03, then the cursor would be on line 03. Likewise, if the pull takes place on a cycle as the puller begins to loop
(because pull did not occur by line 05) the cursor will be sitting at that location.
A value of '0' entered for either TIME or VEL.(VELOCITY) has special meaning:
IfVEL.= 0; PULL is disabled (= 0); HEAT is on at programmed value for duration of TIME
programmed (10 milliseconds/unit) and cycle executes only once, no looping. This allows the
one to use the puller as a 'polisher' for patch-type pipettes and the like.
If TIME = 0 and VEL. is not equal to 0; the gas solenoid is disabled (no active or gas cooling).
This allows the pulling of special pipette shapes. Most often used to pull long tube-like shapes
such as are used for microinjection or micro-perfusion.
If both TIME and VEL.(VELOCITY) are nQl equal to zero; then the value of TIME is the
delayfromthe simultaneous turning off of thefilamentand turning on of the gas solenoid, la
the turning on of the hard or solenoid pull. Increasing the value of TIME will increase the effectiveness of the active cooling prior to the hard pull. It is, to a degree, the equivalent of
changing the micrometerflowvalve.
There is always the possibility that the puller will be given a set of values which 'stall' its operation. An example might be where the HEAT value has not been set high enough to melt the
glass, thus the glass can not be pulled and no velocity can be achieved. If it appears that a situation of this type has arisen, press the STOP key. This action aborts program execution and allows editing to take place. One could press RESET, but this requires that the program number
be reentered.
Finally, it should be mentioned that all programs entered into memory (to a maximum of ten)
remain there even after the power is turned off or the RESET switch is toggled. A special
memory 'chip' that carries its own battery back-up will retain stored information for as long
as ten years without power being applied to the instmment. Miracle that this is, it is strongly
suggested that one keep a written record of programs in case of unexpected difficulties.
CLEAR: When a new program is being entered into memory in an area occupied by another
program it is helpful to be able to 'clean out' the old program. Also, it may be desirable to
remove all the valuesfromthe last cycles of a long multi-cycle program to allow for fine tuning
of these final cycles. This clearing of program values is accomplished by the CLR key.
The CLR key sets all values to 0 from the line on which the cursor is located to the end of the
program. Thus, if the cursor is on line 05 of a seven line program and your response to the
isa'l'. Only the values up through line 04 will remain intact. Clearing out a whole program
simply requires that the cursor be on Ime 01 before the 'yes' response.
The 'no' or '0' response to the above question provides access to the most unique feature of
the P-80/PC, the Ramp Test. If one answers 'no' to the question above, the following display
NO=0,RAMP= 1
A '0' response returns the user to the current program. A T response enters the Ramp Test.
The next display after entering '1' is:
A length of capillary glass is loaded, the cover lowered and the PULL key depressed. On the
display a number will be seen to be incrementing at the rate of 10 units per second. Events
take place as follows: 1) the puller increments the heat at the rate of ten units per second; 2)
when the heat output begins to soften the glass, the puller bars begin to move apart. When a
certain velocity (the value of which is stored in ROM) is achieved the heat is turned off and
the ramp test value is shown on the display. In order to mn the Ramp Test several times, it is
necessary to press RESET, choose a program number (any number will suffice), answer 'no'
to clearing values, and 'yes' to Ramp Test. Glass may then be loaded and the test mn again.
Recall that one has been programming the P-80/PC based on the characteristics of the filament that is installed. Since no twofilamentsare exactly alike, there must be some way to adjust programs when a filament wears out or is damaged and must be replaced. The answer is
the Ramp Test. One of the first actions that should be taken is to mn the Ramp Test with the
glass that will be used for pipette fabrication. An average of several tests gives a number that
relates that particular filament to
that particular batch of glass. If thefilamentmust be changed, or a new batch of glass is obtained; the Ramp Test can be used to establish a new Ramp Test value to act as a guideline
for adjusting program values. It is necessary for the user to keep track of Ramp Test values.
CAUnONl Because of the large power reserve of the regulated heater power supply, it is very
easy to bura out the filament if the heater value is set too high. The recommended starting
heater value is the ramp test value.
At a heat setting of ramp value plus 15, a Imm O.D., 0.50mm I.D. glass capiQary tube should
pull in 4 to 6 seconds after the start button is pressed. K the pull takes longer than eight seconds,
and you are trying to pull a fine micropipette, increase the heat value by about five. Then try
pulling elecfrodes until the pull takes place in less than eight seconds after the start button is
If the pull occurs in less than three seconds after you start, decrease the heat value by five. For
2mm O.D. tubing, the pull should occur between 15 and 25 seconds after the start. Make corrections as outlined for the smaller tubing.
The position ofthe glass within thefilamentwill also affect the time it takes to pull an electrode.
When using a trough filament the glass should be about .5mm above the bottom of the filament and centered front to back. The position of the glass with respect to the filament may
easily be adjusted with the two eccentrics (A and A' in figure 2). The two locking screws B
and B' should be loosened before adjusting the two eccentrics. In the case of a box filament
the glass should be in the center of the filament.
The heat setting can also affect the length and size of the tip. Higher heat settings will give
longer and finer tips. A heat value of the Ramp Test value plus 15 will generally give a very
fine tip.
If should be noted that at high heat settings (filament white hot) the filament life is greatly
reduced. It is suggested that a setting of ramp value plus 15 be used initially and elecfrode
length be controlled by gas valve adjustment.
For patch-pipettes and injection pipettes a good starting point is the ramp test value. The time
for the first pull will be in the 10 to 20 second range.
Low values of pull strength settings in the range of 40-75 will give larger tips, while settings
between 150-250 give the smallest tips. The pull strength can be set to any value desired with
no danger of damaging the instmment. The hard pull is turned after the velocity reaches the
trip value programmed for the cycle and after the TIME value has elapsed. If the velocity
value was greater than 0 the TIME value determines a delay from
when the filament is tumed o^nitrogen is turned on to when the hard pull begins. This time
is in .5 millisecond units and a setting of 80 (40 milliseconds) is recommended as an optimum
value for fine tips.
The velocity value is generally between 60 and 120 for micropipettes and 10 to 30 for patchpipettes. TTiis value is related to the speed at which the two electrode carriers are moving.
The lower the velocity value the slower the speed of the carriers at which the trip point will
occur. At that time the heat will be turned off and the nitrogen will be tumed on. After a delay
determined by the time value the hard pull is turned on.
The gasflowcontrol valve is located above and behind the filament. It is the primary means
of adjusting the length of the electrode tip. It is a precision unit and will give reproducible
For a first try, set the control valve to a reading of 100. Increasing the gasflowwill produce
shorter tips, and, conversely, decreasing theflowwill produce longer tips.
If the gas flow is decreased too much, the electrode will not form a tip. At air flow settings
about five units below the value needed to form a tip, the glass will break and form tips of
about one micron. At still lower settings, the glass will form a wispy fiber. The very finest tips
for a given pull and heat will be formed at an air setting 10 units higher then where the 1 micron
tip was formed. Long tips can be formed by using wider filaments or by using higher heat settings, and conversely still shorter tips can be formed by using narrowerfilamentsor lower heats.
Filaments narrower than 2nim can not form as fine a tip as the wider filaments.
Electrodes will not be formed if the gasflowis set too high.
We suggest the following method of inserting the glass capillary tubing into the carrier clamps,
to prevent damage to the filament.
Use either the left or right carrier; move the carrier away from the filament until the carrier
is latched by the spring clip. Open the capillary clamp; hold the glass tubing about two inches
from one end, and with the two-inch end facing the filament, lower the glass into the clamp
and tighten the clamp. The glass should now be in the groove with one end about 1/2-inch
from the filament.
Release the spring clip latch and move the carrier toward the filament. If the filament is correctly positioned, the glass will pass through it. Hold the two carriers toward the center by
placing two fingers of one hand on the finger bars. Loosen the clamp holding the glass and
slide the glass in its groove toward and into the other clamp groove. Center the glass and
tighten both clamps.
The horizontal filament is quite easy to work with. This filament should be centered between
the two clamps, and the air jet should be centered under the filament about 2mm below the
When using the standard 3mm troughfilament,the glass tubing should be positioned just above
the filament and centered between the two sides. This position can be adjusted by using the
two eccentric cams, located on the aluminum angle piece which holds the filament assembly.
Slightiy loosening the two screws, which lock the filament assembly in place, the filament can
be moved in relation to the glass tubing by turning the appropriate cam.
he heaterfilamentsare easily replaced by loosening the two clamp screws holding the filament
in place. Slide out the old filament, slip in a new one, and position it over the afr jet. Then
tighten the two screws.
The length of the electrodes pulled can be varied, as previously stated, by decreasing the heat
or by decreasing the air flow. It can be changed by usingfilamentsof different widths. Widths
of 1.5nim to 6mm trough filaments can be used. Electrodes pulled using a 1.5mm filament
will be very short and will have large tips.
Tips of l-2u can be formed using a l.Smmfilament,with low filament temperatures and weak
pull sfrengths.
The tip size will decrease with increasing filament width until a width of 3mm is reached. Increasing the filament width beyond 3mm will produce longer tips with a more gradual taper
(which may penetrate better in some cases). However, the tip will not be any smaller.
Another type offilamentwhich can be used is the box type heater filament. The box configuration is particularly useful with thick wall or double-barreled glass, since the box filament
deUvers more heat to the glass. This results in faster heating without the necessity of increasing the temperature of the filament. (Note that the HEAT value must be increased in order
to reach an operating temperature.) The boxfilamentalso heats the glass in a more symmetrical fashion than trough filaments, so that the pipettes produced tend to be more sfraight and
more concentric than those pulled with a trough filament.
The boxfilamenthas two primary limitations. First, it requires more current to heat to a given
temperature than the same size troughfilament.Thus it is possible to use wider trough filaments without exceeding the maximum current capacity of the puller. Second, the box configuration reduces the cooling effect of the gas jet. For this reason the boxfilamentis not
recommended when very short pipettes are to be formed.
The air jet should be centered directly under the box filament. The glass capillary tubing
should be centered within the box filament
The optimal size of the boxfilamentappears to be 3mm wide, 3mm high and 3nim deep. To
produce short, large tips, a boxfilamentof 1.5mm width, forming a box 2mm on each side, may
produce more straighttipsthan a trough filament.
There is a size limitation on boxfilamentsthat can be used with the Model P-80PC electrode
puller. Boxfilamentswider than 3 to 4mm may exceed the maximumfilamentheater current
that the P-80PC can deliver, thus limiting thefilamenttemperature.
The position of the two pulleys which carry the cablesfromthe solenoid to the carriers is adjustable. This adjustment should be made only if the two electrodes formedfromone pull are
of quite different lengths. This inequality is generally caused by the jet not being aimed at the
center of thefilamentbut may also be caused by unequal cable tensions. (To avoid unnecessary cable adjustment, be certain that the air jet is correctly positioned before proceeding.)
The adjustment is made by moving one or both of the pulleys to equalize the tension on the
two cables. It should be explained at this point that there are two sets of stops in the system.
There are the stops in the carrier slots against which the carriers rest, and a stop to prevent the
solenoidfrombeing pulled out of its housing. The adjustment of the pulleys must be made so
that the carriers will still come up against their stops while the solenoid is not against its stop.
The two cables should not be under tension when the carriers are against their stops, this is
the position they would be in just before pulling an electrode.
You should be able to press on either cable between the carrier and the pulley and there should
be about a 2nim deflection before the solenoid hits its stop. If the deflection is more or less,
the pulley position should be changed. This is done by loosening the two screws above the pulley andtuming the chrome eccentric cam to move the pulley in smallincrements until the two
cables are of equal tension. If the carrier no longer stops against its stop in the slot, but stops
against the cable, then the cam must be adjusted back until the carrier once more hits its stop.
It is important that the carriers come up against their stops with no tension on the cables. If
there is tension, the initial pull will depend on how tightly you hold thefingerstops when the
glass is clamped in the carriers. If this happens, the electrodes will not be consistentfrompuU
to pull.
If the cable breaks or gets a bad kink, it will have to be replaced. This is done as follows.
First, take off the plastic cover by removing the three screws which hold it in place. Now the
two screws which hold the cover plate down must be removed. The cover plate is the aluminum
plate on top of which thefilamentassembly sits, and below which the two cable wires can be
seen coming out on their way to the outboard pulleys. With these two screws removed, the
completefilamentassembly, including the air solenoid and air micrometer, can be lifted up
and moved back out of the way.
Next, with the power cord unplugged, remove thefivescrews which hold the front panel in
place. Swing the top edge of the panel forward so that the panel is face down on the table.
The solenoid assembly can now be seen.
The next step is to remove the brass slug on the top of the solenoid. It is held on by two hex
screws. The brass slug should now slide up and off of the shaft of the solenoid. The old cable
can now be slipped out of the slot in the brass slug. At this point note the path of the cable as
the new cable will be strung the same way as the old. To remove the old cable, use a wire cutter to cut the cable near the electrode carriers. Now pull out the cable. Slip the swaged end
of the new cable in the brass slug and replace the brass slug. Be sure to get the hex screws
tight. Both wires go through the first guide, and then each wire must be fed through a pulley
and its accompanying guides.
Using a small screwdriver, loosen the screw at the outboard end of the electrode carrier and
remove the short piece of wire and its swage. Feed the wires through the outboard pulleys so
that the wires lie across the electrode carriers.
For the next steps, the electrode carriers must be held in toward the center. This can be done
with a mbber band around the twofingerbars to bring both carriers in to the center. Loosen
the two screws which lock down the two outboard pulleys and center the pulleys in their travel.
Now slip a swage on one of the wires. The wire must now be pulled on in order to lift the
solenoid. With the solenoid against its upper stop, position the swage over the hole at the end
of the electrode carrier and crimp the swage with a swaging tool. Using a wire cutter, cut the
excess wire off and tighten the screw down on the wire. This most be done for each wire.
The cover plate andfilamentassembly can now be replaced. It may be necessary to readjust
thefilamentposition in relation to the glass tubing. The instmctions for adjusting the outboard pulleys should now be followed to get the correct tension on the cables. This is somewhat easier to do with thefrontpanel down, to see the relationship ofthe solenoid and its stop.
Thefinalstep is to replace the front panel. Be sure that no wires are pinched between the
front panel and the cabinet.
PROBLEM: What glass should I use?
The type of glass and the wall ratio to O.D. (outside diameter) are two of the most important
variables in controlling tip size. For example using program 1 which we used to form the
pipette in the SEM micrograph included with this puller when shipped from Sutter Instmments; borosilicate glass with an O.D. of 1mm and an I.D. of .50 will give tips of .06 to .07
miCTon. Using the same settings borosilicate glass 1mm O.D. and .78mm I.D. will form tips of
.1 to .12 micron. Aluminosilicate glass with an O.D. of Imm and an I.D. of .58mm will form
tips of .03 to .04 microns again with the same settings.
In general the thicker the wall in relation to the O.D. of the glass the finer the tip will be, and
the thinner the wall the larger the tip will be. Thin wall glass will give the best results in most
experiments as it will have the largest pore for a given tip size. This means it will have a lower
resistance and will allow for easier injection of solutions. However in many cases with small
cells the thin wall glass will not form tips fine enough to obtain good penetrations. In this case
heavier wall glass must be used.
PROBLEM: The resistance of my pipettes is to low, how do I pull a higher resistance pipette?
Thefirstpoint to note is that there is very little correlation between tip size and electrode resistance. Most of the resistance of a microelectrode is in the shank of the electrode behind the
tip. Electrode tips which are .1 micron in diameter can vary in resistance from 20 Megohms
to 1000 Megohms depending on the length of the electrode and what is used for the filling
solution. If the same solution is used then resistance may give an indication of how well the
electrode will penefrate a cell as the electrode with the higher resistance will probably have a
longer shank and a smaller cone angle at the tip. This combination will aid in the penetration
of cells where the cell is not a surface cell,
PROBLEM: OK but I still want a smaller tip than I am getting.
The first thing to try in most cases is to increase the heat value. This will generally decrease
the tip size but it will also give a longer shank. If the higher resistance is not a problem this is
generally the best solution. Continuing to increase the heat is not the final answer as to high
a heat can lead to larger tips. In general with 1mm O.D. .5mm I.D. borosilicate glass the finest
tips will be formed when the glass pulls in 4 to 5 seconds after starting the pull.
If the electrode is now too long and causes the resistance to go too high to pass the necessary
current for example, then the next step is to increase the pull strength. In general a pull
sfrength of 125 will give tips of less then .1 micron. Increasing the pull to 250 will reduce the
tips size about 5-10%. We recommend a pull of about 150 in most cases.
The last major variable to adjust is the nitrogenflow.If in the case of 1 by .5 borosilicate glass
the pull takes place in 4-8 seconds thetipsize will not change with a change in the cooling flow.
The only change will be in the length of tiie shank. If however the heat is such that the pull
takes place in more then 8 seconds, increasing theflowwill increase thetipsize. An increase
in gasflowwill shorten the tip and a decrease will lengthen the shank.
PROBLEM: How do I increase the size of my patch-pipette?
Thefirstthing to try is to reduce the heat. Try dropping the heat 5 units at a time to see if this
will increase the size of the tips. If this does not work increase the air micrometer in units of
5. The pull should generally be set to 0 when pulling large tipped (1-10 micron) pipettes.
PROBLEM: The tips of my patch-pipettes vary in size from pull to pull.
This can happen when a pipette is formed in two or more loops. If the pipette is formed in
three loops in one case and then on the next pull it forms in four loops thetipswill not be the
same. Adding one unit in the velocity value will in most cases cause the pipette to be formed
in three loops or subfracting 1 unit should cause the pipette to form in 4 loops. It is always
good technique when a program is developed that produces a desired pipette, to tiy increasing and decreasing the velocity value to be sure; that you are in a stable region. The best procedure in developing a very reliable pipette program is to change the velocity value both up
and down imtil the number of cycles to pull the pipette changes. Then pick a value halfway
between for thefinalvelocity value.
PROBLEM: I need to form an injection pipette with a 1 micron and 20 to 50 microns long.
How do I do this?
Try a program in which thefirsttwo lines of the program have a pull value of 0 a velocity value
of 10 to 30 a time setting of 80 and use the ramp value for the heat. The third line should have
the same heat value, a pull value of 150, a velocity of 30 and the time should be 0.
The idea behind this program is to reduce the size of the glass on thefirsttwo cycles and then
on the third cycle we give a hard pull with the air turned off. Normally if the air is turned off
a long wisp will result, but since we have greatly reduced the size of the glass and with a very
hard pull the glass will tend to separate when it is about 1 micron in diameter.
PROBLEM: The elecfrodes are bent. How do I make them pull straight?
This problem occurs most often when using the trough filament. Going to a box type of filament will produce much more straight pipettes. The bend in the pipette has no effect on the
pipettes tip and should cause no problems unless you are penetrating quite deep in the tissue
with the electrode and you are aiming at a certain site. Then the bend may lead the pipette to
the wrong area. The boxfilamentis not a complete improvement on the troughfilamentas
the gasflowis much less effective with the boxfilament,and you give up much of the length
control that the gas gives with the trough filament.
PROBLEM: The filament does not light up when I press pull.
There are a number of possible reasons why this might happen. First look and see if the filament has burned out. In some cases it may be necessary to lcx)sen the screws holding the filament in place as a very fine break may be hard to see. If the filament is OK, try running the
ramp test and see what happens. If you have just changed the filament it is quite possible that
the new filament needs a very different heat value than what you have been usmg. It is always
a good idea to mn the ramp test each time you change the filament. If you run the ramp test
and the heat value reaches 999 without the filament heating up check the screws holding the
filament in place and if they are tight then check the two nuts connecting the filament wires
to the posts in back of the filament block. If these are tight then the problem is probably on
the circuit board.
PROBLEM: One electrode is much longer then the other electrode.
This is caused by one of two things. First check the tension on the two cables as explained in
the maintenance section. If the cables have the same tension then the gas jet must be aimed
more at one side of the filament than the other. This can be corrected by moving the gas jet
until both sides are the same length.
The P-80/PC micropipette puller is controlled by a Z-80 microprocessor. Three digital to
analog converters control the heat, pull and velocity values. The heat power supply is a
precision constant current switching unit which will vary less than 10 millamperes with a plus
or minus 10% change in the ac line current. The pull supply is a constant current DC power
supply. The velocity trip point is set by a D-A converter. The output of the velocity transducer
is compared to the output of the velocity D-A to determine when the trip velocity is reached.
PROBLEM: The shape and resistance of the pipette changes from pull to pull.
a. In most cases this is due to one or both of the cables to the pipette carriers being setup to
tight. If the cable is adjusted so that the carrier can't come against the stop in the slot in the
center of the pipette carrier. In this case the initial pull tension will depend on how hard the
carriers are squeezed together when the glass clamps are tightened. To adjust, see the cable
adjustment section.
b, A second possible cause of this problem is dirt on the carrier bars or bearings. In this case
clean the carriers and bearings with a lintfreetissueor cloth,
c. If the problem persists mn the ramp test several times. If possible use one long piece of
glass and move the glass over after each ramp test (tum the air adjustment up and the glass
will not separate). If the ramp values are + /- 4 units or less the problem may be with the glass.
If the values are greater than +/- 4 units call Sutter Instmments,
PROBLEM: Display blank, fan not on.
a. Check power cord and wall AC outiet
b. If unit is properly plugged-in and still does not work, remove power cord and check 3 AMP
fuse. If thefiisehas blown, suspect problems with the large transistors mounted on the heatsink on the back of the cabinet.
c. If the fuse is still good, suspect the wiring harness.
PROBLEM: Display blank, fan on.
a. Check the 1/2 AMP fuse. If the fuse has blown suspect the main circuit board and transformer T-3, a DMT 6-15.
b. If the fuse is still good, suspect a loose connection between theribboncable and display
unit or the main circuit board.
c. If the connections are good check the various power supplies located on the main circuit
PROBLEM: Display shows a row of blocks.
a. The microprocessor has failed to properly initialize the display.
b. Press reset and the display should show the proper power-up message. Do not turn power
off and then on rapidly, as this may cause improper power-up. Always allow at least 5 seconds
before turning power back on.
c. If this problem recursfrequentlysuspect the reset timing capacitor, C 2. You may wish to
replace C 2 with a shghtly higher value such as 68 microF.
d. If pressing reset fails to produce the proper power-up message check pin 12 of Ul for a
clock signal and check the address and data lines of U5 to see if the microprocessor is functioning.
PROBLEM: Displayed program values are not correct.
a. Make sure that values were not changed by another user,
b. Always write down the program values and the ramp test value and keep them in a secure
c. If values entered are not held when the power is tumed off suspect the zero-power memory,
Unplug unitfrompower. Remove three screws that hold plastic cover in place; one on each
side and one on the rear.
Remove two nuts and washers that retain heater wires to thefilamentholder and remove wires
from posts. (Fig. 1, H).
Remove six screws that hold top to base; two along each side edge and two along back edge.
Raise up top, move it forward sHghtly and rest it atop cabinet supported by solenoid bracket
and back edge (Fig. 1)
Reaching around the solenoid bracket, unplug the molex connectors at the front edge of the
circuit board and the 26 pin cable. The top may now be lifted clear and set aside.
Remove all other molex connectors.
Remove the eight plastic screws that hold down the circuit board (Fig. 2, locations E and E').
Lift the circuit board clear ofthe chassis and set aside.
Installation of the board is the reverse of the above procedure. Use extreme care in handUng
the connectionfromthe velocity transducer. If the instmctions are unclear, please contact us
via phone or telex for clarification.
1-4726 e3S6
o e OUT i ^
D3 RCr
D4 R t r
r i a K TRIM
R27 J 3 - 3
XL ' ^ o i a
c a HT T < v ^ r
e a M7 T > v M r
3 of
r TO B a D, n » c G T O i a K
FOR 24av o p c R n n c N co»*c:CT c T O n s H . F T O D .
S l M p c x i u n w r t t Nu«*>«r^~
M«w 5 7 .
4 of
Sturoxitnct. 1977. Vol. 1 pp. 813-827. Pcrgamon Press. Printed in Great Briuin.
Department of Physiology, University of (California, San Francisco, CA 94143, U.S.A.
Abstract—Newly developed techniques are described for making both single and dual channel ultrafine
electrodes, and for advancing these electrodes into small cells. Together with our previously described
methods of electrode beveling, all of these techniques are found to be highly successful in rod outer
segments of the toad retina, which average only about 6.0 /mi in diameter. Hence the new techniques
of making and advancing electrodes solve the main problems that have limited intracellular work
in small cells. These techniques should also increase the efficiency of intracellular work in cells of
all sizes. They thus offer promising possibilities for greatly expanding the preparations Aid problems
that may be studied in neurophysiology.
A micropipette puller for forming ultrafine tips with short tapers
Requirements and background
Description of micropipette puller
(Characteristics of single barrel electrodes
Summarized advantages of the new micropipette puller
An ultrafine dual channel micropip<-tte electrode
Dual channel electrodes formed from thick septum theta tubing
Characteristics of dual channel electrodes
A high-speed stepping hydraulic microdrive
Features desired
High-speed modifications of the Kopf hydraulic microdrive
Characteristics of high-speed steps
Automatic stopping of electrode upon cell penetration
Testing of intracellular recording techniques in small cells
DEVELOPMENT of the micropipette electrode by LING
& GERARD (1949) provided the technical basis for
major advances in neurophysiology. Since then, however, continuing limitations of microelectrode techniques have largely determined which preparations
and problems can be studied and some of the main
trends in the field. In particular, high quality and systematic intracellular work has been confined almost
entirely to large cells, because of difficulties in penetrating small ceUs and maintaining normal activity
after penetration. These limitations have been especially severe in the vertebrate brain, spinal cord and
retina, where the vast majority of cells is smaller than
20/xm in the critical dimension. Thus intracellular
work in the vertebrate central nervous system has
been devoted almost exclusively to selected large cells,
and the search for especially large cells in lower vertebrates and invertebrates has become a well-cstab—^
Abbreviation: SEM, scanning electron microscopy.
NSC. 2/6—A
lished trend. It seemed evident that the purposes of
neurophysiology would be greatly advanced if the
limitations of microelectrode techniques could be
markedly reduced or abolished, so that cell size would
no longer be a critical consideration. It was also evident that the technical problems could not be solved
on a casual basis, so we undertook a systematic
attack on these problems. This was done partly to
expedite our own research in vertebrate photoreceptors, but mainly with the broader goal of removing
technical barriers that have prevented intracellular
neurophysiology from being applied systematically to
small cells.
Requirements and bacligromd
We have previously reported methods for the rapid
and precise beveling of Pyrex micropipettes with tip
diameters as small as about 0.1 /an (BROWN & FLAMING. 1974; 1975). During that work the need became
evident for an electrode puller that was more adequate than available pullers for the requirements of
small cells. These requirements include the reliable
making of very fine tips, while minimizing both the
flexibility of the tip and its electrical resistance. Even
the optimal application of electrode beveling to small
cells requires the dependable formation of very fine
tips, combined with the requisite stiffness for accurate
and reliable beveling.
Most electrode pullers follow the two-stage plan
of .ALKXANDER & NASTUK (1953), whose design was
followed closely in the puller-; made by Industrial
Science Associates. As illustrated by ALE3CANDER &
NASTLK (1953). the electrode tips are typically about
12 mm long: this length is from the beginning of the
taf)cr to the tip itself, and it cannot be reduced appreciably. With stringent attempts to pull small tips, we
obtained unbevcled tips about 0.15/im in diameter.
as demonstrated by scanning electron microscopy
(SEMI. These tips were reasonably stiff and beveled
well (BROWN & FLAMING. 1974), but were obtained
only with considerable difficulty and low reliability.
.Also, in the outer segments of toad rods we find that
electrode tips of about 0.15 nm diameter, even when
minimally beveled, are too large for satisfactory
results. This preparation requires ultrafine tips, which
we define as tips having diameters of O.lO^m or less.
For such small tips the puller described by LIVINGSTON & Dl GGAR (1934) is widely used. We have found
this puller to provide tips as small as 0.05 /an. but
the tapered portion is 25 mm or more in length.
Hence the dc. resistances are quite high, typically
.•too 400 MQ when filled with 5 M K acetate. Also.
the extreme flexibility of the tip makes these electrodes ditlicult to bevel and unsuitable for reaching
target cells that are deeply buried in the brain or
spinal cord.
ft was first shown by Lux (1960) that cooling a
glass capillary by an air stream while the tip is being
formed can greatly shorten the tip. This principle was
carried further by CHOWDHLRY (1969), who directed
two symmetrically opposed air jets to the forming
electrode tip; this shortened the tip and yielded tip
diameters down to about 0.05 /im. We initially built
a puller following the description of Chowdhury
(1969). and confirmed the validity of the principle,
but results were disappointing because of the great
variability between electrodes at specified settings. We
have now develop)ed a puller incorporating features
found desirable to take advantage of the air jet principle, while also providing versatility and highly reliable results. In achieving these goals, many details
proved important
Description of micropipette puller
Figure 1 is an overall view of our puller, the major
features of which will now be described. It is a twostage horizontal puller, in which both electrode car-
riers are connected by stainless steel cables to a vertically oriented solenoid that is not visible in Fig. 1.
The correct strength of slow pull is provided by the
weight of the solenoid plunger and its attachments.
It proved necessary to use a d.c. solenoid, to avoid
variations of pull strength that occur when the strong
pull can be activated at various phases of an altemating current Smooth and consistent pulling motions
are provided by each electrode carrier being mounted
in V-groove ball bearing wheels. The outboard cable
pulleys are adjustable laterally to give equal tension
on both cables, and after this adjustment the two electrodes resulting from each pull are almost identical.
The heating element is a flat ribbon of 90% platinum and 10% iridium. This ribbon is 0.002 in.
thick, about 2 mm wide, and formed into a loop
about 3.5 mm in diameter. The heating element is
powered through a Variac, preceded by a regulating
transformer and heavy duty transformer that reduces
the maximum voltage to 3 V. Provision is made for
monitoring both voltage and current in the heater
circuit and we usually use values of about 1.8 V and
12 A for single barrel electrodes. Whenever the voltage required for a given current shows a significant
upward trend, the heating element is replaced; this
has proved helpful for maintaining reliable results
over a long period
In the air jet system a bottle of compressed nitrogen
is used as the source of our "air'. This minimizes oxidation of the heating filament and assures clean gas.
with consistent low moisture content for control of
its cooling effect. The gas pulse is formed by an electrical gas valve. The pressure from the nitrogen bottle
has been kept at 50 lb/in*, and gas flow during the
pulse is controlled by a needle valve that is set by
a micrometer. Figure 2 is a closeup view showing
the heating filament and gas jets. Note that the
nitrogen is delivered symmetrically through orifices
positioned just above and below the heating element
These orifices are shaped like flame spreaders, the
long axes of which coincide with the axis of the ribbon filament. This assures efficient delivery of gas to
cool the heating filament rapidly, while the filament
itself protects the forming electrode tip from being
deformed by the gas currents. We found it both undesirable and unnecessary for the nitrogen to cool the
electrode tip directly. Instead, the gas need only cool
the filament rapidly to stop it from continuing to
deliver a relatively large amount of heat while the
tip is being formed. The heating element and gas jets
are mounted on a single block that may be adjusted
to center the heating element on various sizes of glass
tubing. For prevention of tip bending, it also proved
necessary to provide one-way catches that stop the
electrode carriers from rebounding after striking their
rubber dampers. The electrode tip is thus protected
against re-entering the vicinity of the heating element
while it is still delivering sufficient heat to deform
the fine tip.
For pulling electrodes a switch tums on the heat.
FiC. I. Overall view of electrode puller
:^ ::::^:fe,a;ii:i;
Fic. 2. Closeup view of heating coil. ^
jets and electrode clamps.
FIG. 4. Sample electrode tip photographed by scanning electron microscopy. Airflow setting of 75.
Tip diameter measured 0.075/<m; after correcting for the gold coating, the true diameter was about
0.045/jm. The electrical resistance of the other member of this pair of electrodes was I15Mn.
Scale = 0.20 itm.
FIG. 7. Scanning electron micrographs of dual channel electrode tips formed from thick septum theta
tubing. A: a larger tip broken to a diameter of about 0.4 nm. to illustrate that the relative cross-sectional
dimensions of the theta tubing are maintained down to the tip. B: an ultrafine tip with a measured
outside diameter of 0.085 /im; after subtracting the gold coating, the corrected diameter was 0.055 /im.
Scale = 0.40 /im.
FIG. 10. Receptor layer of the retina from the toad. Bufo marinus. The retina was fixed and cracked,
and the exposed receptors were photographed by scanning electron microscopy, as described by STFINBKRG (1973). The predominant receptors are red rods, with long columnar outer segments that are
separated from the inner segments by a narrow gap crossed by many fine calycal processes. A few
green rods may also be seen, which have shorter columnar outer segments, and inner segments that
taper rapidly to a thin process. A few very slender cones may also be seen, interspersed among the
bases of the rods. Scale = 10/im.
FIG. 8. Photograph of modifications to the Kopf hydraulic unit to make a high-s(>eed stepping
Microelectrode techniques for small cells
and the remainder of the pulling cycle is controlled
automatically. When the capillary has been softened
and drawn over a preset distance by the weak pull,
a precision type of optical switch is activated. This
optical switch is mounted on a micrometer with a
non-rotating spindle, which is visible on the right side
of Fig. 1. The length of the slow pull prior to activating the switch may thus be accurately controlled. The
optical switch tums off the heat initiates the gas
pulse, and trips a delay circuit, at the end of which
the pull solenoid is activated. It proved necessary to
initiate the gas pulse about 40 ms before the strong
pull, apparently because of the delay required for the
gas pulse to cool the heating filament. During the
strong pull the tip is formed in 5-8 ms, though the
solenoid is activated for about I s. The duration of
the gas pulse is arbitrarily set at about 200 ms. Hence
the gas pulse normally lasts from about 40 ms before
the strong pull to well after the electrode tip has been
Ctiaracteristics of single barrel electrodes
Single barrel electrodes were formed from Pyrex
'Omega Dot' tubing of outside diameter 1.0 mm and
inside diameter 0.5 mm, a solid glass fiber of 100/mi
diameter having been fused to the entire length of
the inner wall. To illustrate the effect of the gas pulse,
electrodes were pulled with all values constant except
the gas flow. The gas flow micrometer was set respectively at 55, 75, 100 and 150 thousandths, which gave
increasing gas flows. The low power micrographs of
Fig. 3 show the length and overall configuration of the
resulting electrodes. At the lowest gas flow the tip
length from the beginning of the taper to the very
tip was 13.0 mm. These electrodes were very gmilar
in length to the electrodes formed by an Industrial
Science Associates puller. With increasing gas flows
the rapidity of taper increased and the tip length was
reduced to about 6.0 mm. Within the illustrated range
of gas flows, electrodes formed at any given gas flow
were consistently similar in tip length and configuration. The largest gas flow of Fig. 3 cannot be
exceeded appreciably, however, without obtaining
considerable variability between electrodes.
Prior to measuring their d.c. resistance, electrodes
were filled with 5 M K acetate by injection (rpm the
back, and allowed to.stand o\emight. after which any
bubbles in the shaft were removed by reinjection. Tips
examined by SEM were not filled because the
electrolyte was difficult to remove and often contaminated the tip. These electrode tips were mounted in
either epoxy or solder glass (Coming No. 7570). then
carefully cleaned. The detailed procedures in preparing electrode tips for high resolution SEM proved
quite critical and are available upon request. To prevent electron charging during SEM observations, a
conductive gold coating was deposited with a minimum thickness of 150 A. All measurements of tip diameter must thus be corrected by subtracting at least
300 A, or 0.03 imi.
Gas flow
FIG. 3. The length and configuration of electrodes, formed as described in the text, with four gas
flow settings that delivered progressively greater amoimts of gas during the nitrogen pulse.
Using the same puller settings as in Fig. 3, including the same four gas flows, we measured both tip
diameter and electrode resistance. In some cases a
single pair of electrodes yielded a measurement of tip
size on one member of the pair, while the other could
be used to measure electrode resistance. In such cases
the electrode resistance associated with a given tip
size and configuration should be determined with particular accuracy. Figure 4 shows an SEM photograph
of a sample electrode pulled at a gas flow setting of
75. This electrode tip had a measured diameter of
0.075 /im. so the true diameter was 0.045 /mi. and the
electrical resistance of the other electrode of this pair
was 115 M i l •
Table I summarizes our results on tip length, tip
diameter, and electrode resistance. Tip length was
taken from Fig. 3. Note that measurements of tip size
have been corrected for the gold coating. For both
tip size and electrode resistance, the values given are
averages, with the number of measurements entering
each average given just afterwards in brackets. Surprisingly. Table 1 shows no clear change of tip diameter over the range of gas flows used. The average
corrected tip diameter for all twenty-five electrodes
in this series was 0.052 //m, and none of the tip diameters exceeded 0.07 /im. iTie smallest occasional tips
had diameters of only 0.02 /im or 200 A. It seems a
meaningful comparison that this is only about three
times the thickness of a cell membrane.
By contrast with tip diameter, electrode resistance
decreased steadily and markedly with increasing gas
flow (Table 1). Between the smallest and largest gas
flow, the average electrode resistance decreased from
135 to 44 MQ. a factor of about 3. Since this decreased electrode resistance was not related to tip size,
it must have resulted from the more rapid taper of
the shorter electrodes. At the highest gas flow setting
in Table I, the corrected average tip diameter was
about 0.05 fan. while electrode resistance averaged
44 MQ. This electrode resistance is very low for such
fine tips.
If an electrode tip tapers too rapidly, in the region
just behind the tip. the value of an ultrafine tip may
be compromised by the electrode being too large
where it finally lodges in the cell membrane. We have
not encountered significant problems of this typcj
partly because the taper immediately behind the tip
was typically quite gradual, the cone angle of the tip
in Fig. 4 measuring about 6". In spite of this gradual
initial taper, these ultrafine tips have demonstrated
the requisite overall stiffness for beveling readily by
our previously described techniques, as indicated by
the gradual lowering of electrical resistance during
While the described puller forms ultrafine tips reliably, it has thus far not proved capable of forming
reliable tips with diameters much in excess of 0.15 /im.
This does not seem a significant disadvantage, partly
because the ultrafine short tips should be preferable
for many applications where larger tips have had to
suffice. When larger tips are required, whether for
greater strength, lower electrical resistance, or to facilitate injections, it seems desirable that these larger
tips be obtained by beveling the ultrafine short tips.
The advantages of the short taper may thus be
retained, and the larger tip may be formed in a manner assuring its optimal sharpness.
As tip size increases, the difficulty of penetrating
a cell with an unbeveled electrode probably increases
approximately as the area of the tip. thus rising
exponentially as the square of the tip diameter. Beveling forms an almost straight cutting edge, with which
the difficulty of cell penetration should increase approximately linearly with tip diameter. Thus the
advantage of beveling for penetrating cells should increase with tip size, and this seems confirmed by our
observations. With ultrafine tips the advantage of
beveling, from the standpoint of cell penetration, has
proved significant but not critically important. Any
desired tip diameter in the range of 0.1-0.5/im may
be obtained reliably by beveling to the requisite electrode resistance with 0.05 fjm alumina, and electrodes
with tip diameters mainly in the range of 0.2-0.3 /im
showed dramaticaUy improved intracellular performance after beveling (BROWN & FLAMING. 1974). Tips
larger than 0.5 /im may be beveled reliably with diamond dust.
Summarised advantages of the new micropipette puller
This puller combines the advantages of ultrafine
tips with the capability of forming quite short tips.
Conveniently, gas flow may be used to vary tip length
with little or no change of tip diameter. From the
standpoint of absolute values, significantly smaller
Electrode tip
* Tip diameters have been corrected for the thickness of the gold coating, as described in the text
Microelectrode techniques for small cells
tips can be obtained than demonstrated with previous
pullers. These tips can also be only half as long as
the shortest tips provided by conventional two-stage
pullers. With given settings and simple precautions,
electrodes of given characteristics are formed with
high reliability. In addition to single barrel electrodes,
the heavier glass for our dual channel electrodes may
also be used with little change in the required settings.
The short tips provided by this puller confer a variety of advantages. Compared with the Livingston
type of puller, the electrode resistance of ultrafine tips
may thus be reduced by at least a factor of 7. In
addition to decreasing the noise level, this should improve the ease with which dyes and ions may be injected into cells, and our experience bears this out.
The resulting stiffness of the shorter tips is very helpful in beveling, which improves the performance of
even ultrafine tips, especially when injecting substances such as Procion yellow. The stiffness of these
short tips should also be advantageous for penetrating cells, and for reaching target cells that require
penetrating deeply or through tough tissue. The
shorter tips likewise reduce cap)acitative coupling
between the two sides of a double barrelled electrode,
which has proved advantageous for our ultrafine dual
channel micropipettes.
Finally, the combination of this puller with precision beveling techniques can reliably provide electrodes of any desired size ranging from the ultrafine
tips formed by the puller to indefinitely large tips.
Al) such electrodes should exhibit the advantages conferred by the rapidly tapering tip. combined with the
extremely sharp tip formed by beveling. It appears
that a versatile technique is thus provided for making
high performance micropipette electrodes.
For many critical types of intracellular work, such
as voltage clamping, two micropipette channels are
required into the same cell. For cells that cannot be
visualized, dual channel micropipettes are generally
required. If the cells are also small, as are most cells
of the retina and brain, these dual channel micropipettes present some difficult and conflicting requirements. Ideally these electrodes should be comparable
to single channel micropipettes in penetrating small
cells readily and providing stable normal responses.
From this standpoint the dual channel tip should be
provided within the ultrafine range of extemal tip diameter. With such small tips the electrical independence of the two channels tends to be strongly compromised, and this must be prevented. Also, with such
fine channels the electrical resistance of each channel
tends to be quite high, but must be held to a satisfactory upper limit. A dual channel micropipette electrode will now be described that is readily made, and
that meets all the above requirements, while also
offering some additional advantages.
Dual channel electrodes formed from thick septum theta
We first experimented with conventional dual channel electrodes formed by fusing two capillary tubes.
When compared with single channel electrodes that
were similarly pulled and beveled, and tested in the
same types of photoreceptors, the single channel electrodes were highly satisfactory but the dual channel
electrodes usually yielded smaller responses that
deteriorated more rapidly. It seemed apparent that
this resulted partly from the non-circular tip asymmetrically stretching the cell membrane, with the long
axis of this stretching being undesirably great for
small cells. Also, at the fusion of the two capillary
tips, there are p)Otential leakage channels between the
inside and outside of the cell.
To obtain round tips we used "theta' tubing, which
contains a flat septum fused into a circular tube.
Techniques for making such tubing have been desdribed by KLMP & DEHN (1975). The theta tubing
used initially was the standard type then supplied by
W. R. Dehn (^ & D Optical Systems. Inc.. P.O. Box K
198. Spencerville. MD 20868). The septum of this tubing proved only about two-thirds as thick as the outer
wall. Tips were pulled with this tubing and tested
by injecting only one channel. Microscopic observation a few minutes later showed that not only was
the injected channel filled to the tip. but the other
channel was filling backward from the tip. Since fluid
crossed between the two channels at the tip. of course
the channels were not independent in either ionic
composition or electrical characteristics. This undoubtedly resulted from the thin septum attenuating
and disappearing slightly before the tip was formed
by the thicker wall of the outer tubing.
It seemed likely that if the relative dimensions were
reversed by making the septum thicker than the outer
wall, considerable advantages would result. Theoretically the septum should then extend somewhat
beyond the outer wall, forming a "spear" in the center
of the electrode that should improve cell penetration.
It also seemed likely that this extension of the septum
would effectively separate the ion pools utilized by
the two electrode tips, thus decreasing the electrical
coupling between the two channels.
Thick septum theta tubing proved more difficult
to obtain, because of problems in fusing a thick septum with a flat edge into a round tube, but this was
satisfactorily achieved by W. R. Dehn. Figure 5 shows
a high contrast photograph of the cross section we
now use and find satisfactory. Note that the septum
is about twice as thick as the outer wall. The outside
diameter of our tubing averages about 1.6 mm. Since
this tubing is made by hand, the outside diameter
varies from about 1.3-1.9 mm. but by far the greater
portion of a given batch falls within 1.5-1.7 mm. All
but a few of the smallest pieces had large enough
FIG. 5. High contrast photograph of cross-section of thicK
septum theta tubing used for dual channel electrodes.
channels to be filled by injection with 31 gauge stainless steel tubing, and even the largest theta tubing
could be drawn by our puller.
Of course theta tubing requires a procedure to electrically separate the back ends of the two channels
so that an independent contact can be made with
the fluid in each channel. As shown schematically in
Fig 6. a high-speed Dremel tool was clamped and
used to rotate a wheel that had diamond abrasive
embedded in both edge and outer surface (Horico
Diaflex No. 86 x SO. obtainable from suppliers of
dental instruments). The theta tubing was held and
advanced along the wheel so that the glass on one
side of the septum was cut down about 12 mm. The
septum was then surfaced with the diamond wheel,
to obliterate any capillary channel that could remain
if a portion of the outer wall were allowed to project
out from the septum's surface. Fluid can also creep
along the abraded surface of the septum, so the proDiamond
A « i s of
FIG. 6. Schematic illustration of method for separating the
two channels at the back end of a dual channel electrode
formed from theta tubing.
jecting portion of the longer channel was coated with
a nonconductor material (Dow Coming No. 734
RTV)L This silicone rubber compound provides a permanently nonconductive coating; being hydrophobic,
it even resists being temporarily compromised by surface condensation. A coating of this material was
placed all the way around the projecting glass of the
longer channel; if applied only part of the way
around, a capillary channel can occur where the
material terminates, which destroys the value of the
The requisite puller settings to provide tips of given
size and length were little changed when shifting from
I.Omm single channel tubing to the theta tubing. The
heat setting increased, however, to Soften the greater
mass of glass in the theta tubing. Also, the gas flow
had to be increased to cool this hotter filament fast
enough so that the length of the tip was not thereby
Characteristics of dual channel electrodes
Ultrafine tips were readily obtained with the thick
septum theta glass. After correcting for the gold coating, the tip diameters usually measured about
0.07/irtL One of the smallest tips, photographed by
SEM, is shown in Fig. if. Though not as fine as 3
the smallest tips obtained from single channel tubing,
the corrected diameter of the illustrated dual channel
tip was only 0.055 /ira
In spite of strong efforts, it has not yet been p>ossible to visualize the theoretical spear protruding from
the tips of these electrodes. With ultrafine tips the
very small apertures are filled by the gold coating,
so end-on views give little information. In side views
like Fig. 7^, the tip sometimes narrows distinctly n e a r S
the end, as if a spear were being seen in edge view;
if so, the gold coating seems to smooth out and
obscure the details. Figure 7^1 shows a tip broken /^
to a diameter of about 0.4 /im, which is sufficient to
reveal the dual apertures. Note that the relative
dimensions of the tubing are retained down to the
tip, the septum remaining twice as thick as the outer
For most test purposes each channel was injected
with 5 M K acetate, which rapidly and reliably filled
the tip. This rapid fiUing seems intrinsic to theta tubing, probably resulting from the capillary action at
the sharp angles where the. septum joins the outer
wall. When only one channel of these electrodes was
filled, no backward filling of the other tip could be
seen under the microscope. The coupling resistance
was measured conventionally as /?, = £//, where R,
is the coupling resistance when a voltage, £, is
measured through one channel as the result of passing
a current /, through the other channel. These
measurements were made in the physiological saline
bathing the retina, since coupling resistance rises after
penetrating a cell. Control electrodes formed from
thin septum theta tubing showed high coupling resistances, ranging from 5 to 35 MQ. With thick septum
Microelectrode techniques for small cells
theta tubing, coupling resistances in the range of be visualized, or if their locations are well known.
l(X)-200Kfl were readily obtained. Our coupling re- It is thus desirable that the electrode be advanced
sistances were thus as low as reported by WERBLIN by a train of steps, each of which has a good chance
(1975). who used similar measuring conditions, but of penetrating any cell that the electrode has closely
who fused two capillary tubes together and then used approached; this feature is especially helpful in
beveling to separate the two tips by a distance of masses of tissue, such as the central nervous system.
0.5-1.5/im. Since thick septum theta tubing provides Finally, it is desirable that the stepping rate can be
similarly low coupling resistances with such fine tips. relatively fast, with automatic stopping of the elecand without beveling, this result strongly supports the trode up)on cell pyenetration. in order to find cells
presence of the theoretical spear separating the tips; quickly while also minimizing the damage resulting
it likewise indicates the efficacy of this spear for low- from cell pyenetration. An electrode advancer will now
be described that offers all of these desired features.
ering the coupling resistance.
The use of our puller to provide ultrafine tips comHigh-speed modifications of the KopJ hydraulic microbined with short tapers has sp>ecial advantages with
the dual channel theta glass. Since the ultrafine tip
contains two channels, and also a thick septum, each
We began with the Kopf hydraulic microdrive
channel must have a considerably smaller cross sec- (made by David Kopf Instruments. 7324 Elmo St.
tion than that of a single channel electrode. Hence Tujunga. CA 91042). The hydraulic unit of this
it is particularly impwrtant to minimize tip length for device, which may be used separately from the drive
reducing the resistance of each channel. The shortest unit, is useful for preventing vibrations at the drive
tips tested measured about 6 mm from the beginning end from being transferred to the electrode. When
of the taper to the very tip. The resistance of each this hydraulic unit is used with the Kopf stepping
channel then averaged about 150 Mfl, which is satis- drive, a train of 1.0/im steps is also provided, \\hcn
factorily low for most purposes. This compares with tested for intracellular recording in the outer segthe average value of 44 Mfi that proved possible with ments of toad rods, this stepping hydraulic drive
single channel electrodes of similar tip length and yielded few cell penetrations. This was not surprising
slightly smaller outside tip diameter.
because the Kopf advancer was not designed for rapid
When compared with two capillaries fused steps, and the main limitations appeared to he in the
together, the capacitative coupling between the chan- drive unit. The stepping motor can or>erate at a maxinels of theta tubing must be greater per unit length mum of only 200 steps s. so the motor itself requires
of tubing because of the flat septum separating the about 5 ms for a 1.0/im advance. Also, if coupled
two channels. Capacitative coupling was reduced, directly to a standard fine-thread micrometer, this
however, by the shortness of the tips. Compared with motor would give a single step of 2.5 /im. Hence the
a thin septum, the thick septum is also helpful in Kopf drive interposes gears to provide 1.0/jm
this respect. As a consequence of these two factors, advances. For rapid steps the inertia of these gears
we have not encountered any significant problems is undesirable.
resulting from capacitative coupling.
The design of stepping motors has advanced greatly
in recent years. The new motor chosen was a Model
HDM-150-500-4-HS. made by Responsyn Products
Corp.. Bldg. 3G, Sixth Road. Woburn IndusHYDRAULIC MICRODRIVE
trial Park. Wobum. MA 08101). When coupled diFeatures desired
rectly to a standard fine-thread micrometer, this
While intracellular recording in small cells may be motor provides single steps of 1.0/im. so ilie interimproved markedly by techniques of electrode mak- p)Osed gears may be eliminated. This motor also has
ing, techniques of electrode advancement are also a sp)ecially designed low inertia rotor and is relatively
crucial. Certain features of an electrode advancer are strong. These features permit stepping rates up to
1500/s. so each step is made in only about 2 3 ms.
especially desirable. The electrode should move
through the tissue by rapid steps, to improve cell {pen- Though this motor required a new power supply, it
etration, and these steps should be short to minimize was otherwise controlled by the Kopf electronic conthe distance the cell is pienetrated. Lateral electrode trol unit. The method of coupling this motor to the
vibrations, which are usually associated with rapid driving micrometer is critical, because the axial
electrode advances, must also be minimized because motion of the micrometer must be taken up b> the
of their damaging effects up>on the cell membrane. coupling. Our best solution to date is a metal bellows.
With rapid single step advancers, such as those using designed to transmit torque without any significant
electromagnetic or piezoelectric principles, the step rotary twisting of the bellows, while the bellows can
must be taken back and the electrode advanced by change in length to accommodate the axial motion
other means before the rapid step can be repeated of the micrometer spindle (custom design No.
SK-7160. obtained from Servometer Corporation. 501
Little Falls Road, Cedar Grove. NJ 07009). When
Such advancers are most useful if the target cells may attached to the motor and micrometer with light
magnesium adapters, this coupler minimized rotary
inertia and introduced no detectable backlash. Figure
8 shows our drive system coupled to the Kopf hydraulic unit. Just forward of the coupling bellows is
a disc that activates a microswitch at either extreme
of the 25 mm of travel These switches stop the motor
and thus protect the instrument against overtravel.
In the hydraulic unit itself, the standard tube
proved too small to reliably transfer a rapid fluid
pulse. Tube length is also critical, since a longer tube
decreases axial vibrations but reduces the velocity of
an advancing step. These problems were solved by
exchanging the standard tube for an 8 ft Teflon tube
with respective inside and outside diameters of 0.060
and 0.120 in.
When a hydraulic unit was new.' or when new rolling diaphragms had just been installed, it proved
necessary to exercise the unit to obtain high-speed
steps. This was apparently required to seat the diaphragms and to attain their maximum flexibility, and
it was done by using a computer program to drive
the unit back and forth for about 250 cycles. After
this initial exercising, high-speed steps were obtained
reliably. If the instrument was unused for a period
of time, however, it proved necessary to renew the
Characteristics of high-speed steps
Little information has been provided on the movement characteristics of most electrode advancers,
partly because convenient and inexpensive devices for
measuring rapid movements have not been available.
Such information is necessary to evaluate the critical
details of the movement and the requisite instrumentation is now available. We measured axial movement
at the electrode end of the advancer by using a
Unimeasure/80, a device utilizing the Hall effect
(obtained from Unimeasure, 180 S. Lake Avenue,
Pasadena, CA 91101). In this application movements
of 0.1 fan could be resolved readily, and the response
time for recording movements was not greater than
0.25 ms.
In Fig. 9 the ciontrol record was obtained with an
unmodified Kopf stepping hydraulic microdrive. The
maximum velocity attained during this 1.0/im control
step was about 0.27 /an/ms, and about 7 ms were
required from onset to completion of the step. The
same hydraulic unit was tested with the described
high-s[>eed modifications. Figure 9 shows that a single
1.0/im step then began after a considerably shorter
latency; such a step was always completed in two
phases, each about 0.5 fan in amplitude, with an intervening notch that seemed to result from the inertial
and resonant characteristics of the systeriL Movement
velocities were measured accurately from records
made with higher gain and expanded time scales. For
a 1.0/im step, the first and second phases had respective velocities of 0.94 and 0.69 /im/ms. Since the stepping motor proved not to reach full velocity on the
first, step, higher movement velocities were attained
by giving preset bursts of steps at about ISOO/s.
Figure 9 shows that these high-frequency bursts
became essentially fused into a single larger step, with
a small pause always separating the total motion into
two phases. With a 2.0 fan advance, the respective
velocities of the first and second phases were 1.56
and 0.85 /im/ms, while a 3.0 fan advance gave respective velocities of 2.0 and 1.4/im/ms. With highfrequency bursts of 4 or 5 steps, there was little
further increase of movement velocity.
From the standpoint of [)enetrating cells, while also
minimizing cell damage, three aspects of a small but
rapid electrode advance are probably important
First, acceleration should be great, so that full velocity wiU be available over as much of the step as
possible. Second, the velocity of advance should be
as high as possible, to assist in cell penetration. Third,
vibrations following the step should be minimal, since
after-vibrations can damage the cell and reduce both
the quality and duration of intracellular recordings.
Figure 9 shows very high accelerations, full velocity
being reached almost instantaneously for advancing
stepw of 1-3 fan. For the case of a l.O/an step, velocities of about 1.0/im/ms have been reported for both
electromagnetic and piezoelectric advancers (FISH et
1971; TUPPER &
RIKMENSPOEL, 1969)L With a
1.0/im step, the velocity of our first phase of movement was in that same range, while the velocity of
the second phase was slightly less. With a 3.0/im step,
our velocities during both phases of movement were
about double those attained during a 1.0/im step.
Figure 9 shows vibrations following a 1-3/im
advance to be quite small on both a relative and
absolute basis, and these vibrations appear much less
significant than those reported with electromagnetic
devices (Fiai et al., 1971; TOMITA, 1965). In summary,
for all three critical asp>ects.of the advancing step,
our results are similar to or better than the characteristics reported for previous devices that have been
designed for high-speed electrode advances over short
FIG. 9. Records of advancing movements by a standard
Kopf stepping microdrive (control) and by the same hydraulic unit following high-speed modifications. All records
made with a Unimeasure/80. using a bandpass of
0-10.000 Hz. For further description, see text
Microelectrode techniques for small cells
Automatic stopping of electrode upon cell penetration
lular work in small cells have thus been tested in
the isolated, inverted, and p>erfused retina of the toad
Bufo marinus. Intracellular recording was concentrated in rod outer segments, which are shown by
SEM in Fig. 10. The densely represented red rods
have long cylindrical outer segments: the scarcer
green rods have cylindrical outer segments of similar
diameter but are shorter. Because of their size. Fig.
10 shows that the outer segments of the red and green
rods are the only reasonable targets that are available
for intracellular recording in this preparation until
reaching the inner segments of the red rods. A total
of 49 rod outer segments, obtained from toads of
three different sizes, were measured in the fresh state
immediately after being shaken off into toad Ringer
solution on a microscope slide. The diameters ranged
from 5.0 to 7.5 fan and averaged 5.9 fan.
As a preparation for testing microelectrode
methods, the outer segments of toad rods have two
major advantages. First, they provide a dense papulation of structures of uniform size. When penetrating
this layer of the retina, there is no doubt about the
sizes of cells p>enetrated. unlike the case of a tissue
containing cells of various sizes. Second, they provide
an esp)ecially difficult preparation for intracellular
recording. In addition to their small size, they are
also rather free lo move, particularly at their tips,
when contacted by an electrode. Hence microelecWhile the described controls are quite convenient
trode methods that are effective in this prepiaration
the major advantages are to find cells quickly, while
should be equally or more effective in most other
minimizing the damage resulting from penetrating a
cell. More high-quality recordings may thus be
Testing was done mainly with ultrafine single barrel
obtained during a given experiment With each step
electrodes formed from Omega Dot tubgiving an axial advance of only I /im. and the
spjecial puller, then filled with 5 M K
advancer always stopping immediately after the step
that p)enetrates a cell, the distance of membrane pen- acetate, after which their resistances averaged about
etration is limited to about 1 fan and in some cases 70 MQ. The retina was placed on the convex surface
may be considerably less. The imp>ortance of control- of a glass contact lens, which was mounted in a perling the depth of cell jjenetration is readily appre- fusion chamber on a microsco{)e stage. The receptor
ciated by examining the electrode taper. Though our surface was visualized with infrared illumination and
puller provides electrodes that taper only slightly in a 40 X water immersion objective, combined with a
the reglbn of the tip. Fig. 4 shows that the diameter high-resolution television camera and monitor. Conincreases about 0.1 fan for each 1.0/im of axial dis- tact ofthe electrode tip with the receptors was usually
tance behind the very tip. For small cells this is a seen accurately by movement of the tip of a single
significant increase of diameter, so the advantages of outer segment. At our angle of penetration, an
ultrafine tips will be compromised unless the distance advance of about 155 fan was then required along
the electrode track before reaching the inner segments
of penetration is severely limited.
of the red rods.
In obtaining the described control of electrode adDuring such penetrations intracellular recordings
vancement, a computer is convenient but not necesswere
readily obtained. Several successive recordings
ary. The main control functions could also be
obtained with a relatively simple electronic unit fea- along a single penetration were not unusual, and as
many as 7 outer segments have been recorded along
turing a voltage comp)arator.
a single electrode track. It has thus proved easy to
make 30-40 intracellular recordings during a single
experiment. More important, we obtained membrane
potentials up to 46 mV and hyperpolarizing light reRECORDING TECHNIQUES IN
sp>onses up to 32 mV. These values are similar to the
largest values that have been reported in a variety
The capabilities of intracellular techniques can be of vertebrate photoreceptors, though most previous
assessed accurately only by testing them in a well- work has been in larger and more firmly held structures, such as inner segments of Necturus and turtle.
defined preparation. All of our techniques for intracelThe electrode advance was programmed from a
Nova 2/10 mini-computer, using the extenial control
terminal of the Kopf control unit The computer program specified the stepping rate and the total distance
the electrode was to be advanced. If the stepping rate
is too high, interactions between steps can result in
undesirable vibrations. We used a rather low rate of
10 steps/s, but Fig. 9 suggests that rates up to about
40 step>s/s could probably be used, and these higher
rates might be useful for deep p)enetrations in the central nervous system. An A-D converter monitored the
d.c. potential recorded by the advancing electrode,
and whenever a negative d.c. shift occurred that
exceeded a preset trigger level, the computer stopp>ed
the advance. Thus electrode advancement was initiated by pressing a button, and the electrode
stopped aiUomatically when a cell was p>enetrated
whose membrane potential exceeded a selected trigger
level. After recording from a given cell, the advancer
could be restarted by an override signal. Upon reaching the preset total depth, the advancer was stopped
by an intemal signal. At that p)oint another depth
run could be initiated if desired, but the electrode
was usually withdrawn. A high-sp)eed retum was provided to withdraw the electrode a preset distance at
about 1000 steps/s.
Of the numerous possibilities offered by dual chanThe largest responses were usually the most stable,
probably because both of these results depend upon nel electrodes, some sample applications have been
penetrating the cell with minimal damage, and most tested in the outer segments of toad rods. Experimencells with large responses have given stable recordings tal voltage clamping has been done, and ions have
that were not lost sp)ontaneously. In some cases been injected intracellularly through one channel
neither the membrane potential nor light response has while recording their effects on light responses with
shown any detectable deterioration for 2-3 h. Hence the other channel. Also, these electrodes have been
we typically spend only a short time obtaining a high used in both outer and inner segments of toad rods
quality recording and perform an. entire experiment to make Procion yellow injections with one channel
on only one or two cells. This extreme stability per- while 5 M K acetate was used in the other channel
mits experiments not previously possible, such as for low resistance recording. Finally, these electrodes
studies of dark adaptation in rod outer segments. It have proved satisfactory for voltage clamping in rod
also permits a variety of observations to be made inner segments in the retina of the snapping turtle
on the same cell, so that the results can be compared (COPENHAGEN. OWEN & BROWN. 1976). Hence diese
electrodes may be used for a variety of dual channel
As expected, there was a tendency for penetrations experiments.
to be made more readily at deeper levels of outer
segments, where these stmctures can move less freely
when contacted by the electrode. Recordings from the
outer tip)s, however, at the very pxjint of contact or
Since our test results were obtained in a quite diffionly a few /im beyond, sometimes yielded large and cult preparation for intracellular recording, it appears
stable responses. For the tips of outer segments, it that the new techniques solve the main problems that
proved useful to advance by 3.0/im steps, as illus- have pertained to intracellular work in small cells,
trated in Fig 9. Appiarently the higher velocity and while also offering significant advantages in large
greater distance of the 3.0/tm advance was helpful cells. In the current context of neurophysiology, these
for penetrating these small structures when they were techniques promise two especially broad advantages,
especially free to move.
the limits of which can be established only by future
First it seems highly advantageous that cell
The repxjrted results pertain to both beveled and
unbeveled electrodes, but the beveled electrodes size should no longer be a crucial consideration in
generally performed better. Apparently our ultrafine choosing an experimental preparation, so that this
tips are sufficiently small that beveling was not critical choice can be based more exclusively on considerfor cell penetration but was still helpful. For injecting ations directly related to experimental purp>oses.
Procion yellow, however, beveling dramatically Second, it is desirable for cells of all sizes that
reduced problems of electrode plugging. This permit- methods be available for conducting intracellular
ted current of at least 2.0 nA to be injected for 5 min work with greater efficiency. This should improve the
or more without any significant spontaneous fluc- data base in many cases, save experimental time and
animals, reduce the funds required for each project
The dual channel electrodes performed at least as and speed overall experimental progress.
well as the single channel electrodes, and perhapjs
It seems likely that these techniques will find pareven better, for readily obtaining high quality intra- ticularly extensive applications in the vertebrate cencellular recordings. This was in spite of the dual chan- tral nervous system, where the amount of intracellular
nel electrodes having slightly larger tips. Hence the work that will ultimately be required makes the need
ease of cell penetration further indicates the presence for experimental efficiency especially compelling. For
of the theoretical spear on the dual channel tips. This example, it has been demonstrated in the retina that
theoretical spear has shown no indication of being much signal processing is accomplished by relatively
especially fragile, since the low coupling resistance small cells generating only slow px)tentials, which can
was sometimes maintained through several pen- be studied in detail only by intracellular methods. If
etrations. Though all of our electrode tips are stif- this principle holds generally in the central nervous
fened by having relatively short tapers, the thick sep- system, as may be expected, intracellular recording
tum must further stiffen the dual channel tips. This is urgently required in small cells of the brain and
feature may likewise aid in penetrating cells, and it spinal cord. Many other problems in the central nershould be especially helpful in penetrating deeply, or vous system, such as the detailed connectivity
through tough tissue, en route to target cells. Because between cells, likewise require intracellular techof its advantages for cell and tissue penetration, the niques.
thick septum theta tubing also has promising appliThe use of intracellular recording has thus far been
cations as a single channel electrode. For this case severely limited in the vertebrate central nervous sysno special treatment is required at the back end, con- tem, piartly because most cells have been too small,
tact being desired with both channels, which may thus and partly because of troublesome circulatory and
be used in pjarallel as a single channel having half respiratory pulsations. Methods of reducing the pulthe resistance of an individual channel
sations have long been available, and further reduc-
Microelectrode techniques for small cells
tion seems pxjssible. Also, most cells.in the central
nervous system are significantly larger than the outer
segments of toad rods, and more movement can probably be tolerated in those larger cells. Though seldom
mentioned, a potentially very helpful factor is an
apparent sealing process that occurs after a cell has
been penetrated. In rod outer segments this has been
revealed by a gradual increase in both membrane
px>tential and light responses during the period just
after penetrating a cell. Similar effects have been seen
by other investigators and suggest a sealing of the
membrane around the electrode tip. perhaps a molecular type of wound healing. T h e time required for
such healing may be expected lo decrease with
smaller tips that cause less damage, and with our
that if ultrafine tips can be inserted into brain cells
for relatively short periods, the sealing process may
occur rapidly and thereafter prevent the recording
from being lost by small pulsations. Also. E. MAVFRI
(personal communication) has noted in large .4ply.^ia
cells that hyperpolarizing current increases the rate
at which a cell's input resistance increases to a stable
value just after penetration. This indicates that hyp)crpolarizing current may further speed the sealing process. In view of the experimental possibilities and the
available techniques, it thus appears that the intensive
apphcation of intracellular recording to the vertebrate
central nervous system is now a goal that is not only
desirable, but also realistic,
ultrafine tips the sealing process was usually complete
i_ . -irv
r- II •
m about 30 s. Following this sealing process the
° ,
attachment between electrode tip and membrane was
astonishingly strong. Someumes the electrode could
then be moved forward or backward by 20/im, or
even more, without losing the intracellular recording
from an outer segment. These observations suggest
, ,
. . , , • , , . „
Acknowledgements—we thank Mr S. WINSTON for elec.
, ^
tronic work and for suggestmg automatic electrode stopp.„g ^ „ j ^^ ^^^ j ^ ^ ^ ^ ^ . ^ ^ ,„ ^ ^ j ^ ^ ^ ^ ^ ^^^ ^ ^ ^ ^
WISEMAN for construction of the mechanical equipment,
w e also thank Ms M. T. MAGLIO for assistance with the
scanning electron microscopy. This work was supported
by grant No. EY-00468 from the National Eye Institute.
ALEXANDER J. T. & NASTUK W . L. (1953) An instrument for the production of microelectrodes used in electrophysiological
studies. Rei. Scieni. lustrum. 24, 528-531.
BROWN K. T. & FLAMING D . G . (1974) Beveling of fine micropipette electrodes by a rapid precision method. Science.
.M.Y. 185, 693-695.
BROWN K. T. & FLAMING D . G . (1975) Instrumentation and technique for beveling fine micropipette electrodes. Brain
Res. 86, 172-180.
CHOWDHURY T . K. (1969) Fabrication of extremely fine glass micropipette electrodes. J. Scieni. Insirum. 2. 1087 1090.
COPENHAGEN D . R.. OWEN W . G . & BROWN K. T . (1976) Electrical properties of snapping turtle rods: Evidence for
electronic coupling. Proc. Assoc. Res. \ ision and Ophlhal. Spring Meeting. Vol. 28.
FISH R. M . . BRYAN J. S.. MCREYNOLDS J. S. & RIES J. J. (1971) A mechanical microelectrode pulsing device to facilitate
the penetration of small cells. IEEE Trans. Bio-Med. Eng. 18. 240-241.
KUMP W . R. & DEHN W . R. (1975) Fabrication techniques for multichannel microelectrodes. Fusion 22. Book II.
LASSEN U . V. & STEN-KNUDSEN O . (1968) Direct measurements of membrane potential and membrane resistance of
human red cells J. Physiol.. Land. 195. 681-696.
LING G . & GERARD R. W . (1949) The normal membrane potential of frog sartorius fibers. J. cell. comp. Physiol.
34. 383-396.
LIVINGSTON I- G. & DUGGAR B. M . (1934) Experimental procedures in a study of the location and concentration
within the host cell of the virus of tobacco mosaic. Biol. Bull. mar. hiol. Lab.. Woods Hole 67, 504-512.
Lux D. (1960) Microelectrodes of higher stability. Electroenceph. din. Neurophysiol. 12, 928-929.
STEINBERG R. H . (1973) Scanning electron microscopy of the bullfrog's retina and pigment epithehum. Z. Zelltorsch.
mikrosk. Anat. 143, 451-463.
TOMITA T. (1965) Electrophysiological study of the mechanisms subserving color coding in Ihe fish retina. Cold Sprin^i
Harh. Symp. quant. Biol. 30, 559-566.
TUPPER J. T. & RIKMENSPOEL R. (1969) Piezoelectric driving device for glass capillary microelectrodes. Rev. Scieni.
Instrum. 40. 851-852.
WERBLIN F . S. (1975) Regenerative hyperpolarization in rods. J. Physiol.. Land. 244. 53-81.
{.Accepted 7 September \911]
Journal of Neuroscience Methods. 6 (1982) 91-102
Elsevier Biomedical Press
Micropipette puller design: form of the heating
filament and effects of filament width on tip length
and diameter
D.G. Flaming and K.T. Brown
Department of Physiology. University of California, San Francisco, CA 94143 (U.S.A.)
(Received November 9th, 1981)
(Revised version received January 1 Ith, 1982)
(Accepted January ]2tli, 1982)
Key words: micropipettes—micropipette pullers—heating
microelectrodes—micropipette lip length
Though the ultrafine short tips provided by our micropipette puller are helpful for many preparations,
longer tips are still needed for some cases where target cells for intracellular work are deeply embedded in
tissue. Because the original 'loop' type of heating filament proved unsuitable for widening to provide the
longer tips, we studied simpler forms of heating fllaments. By comparison with the common loop design, a
rectangular trough fllament proved to have many advantages without any significant disadvantage. In
particular, this type of fllament may be used at varying widths to provide long tips as well as short ones.
Thus our micropipette puller can now provide tip lengths that extend upward by continuous gradations
from about 6 to 27 mm, while maintaining tip diameter in the ultrafine range of 0.1 /im or less. These long
tips have proved more reliable and free from bending than long tips traditionally provided by the
Livingston puller. By using narrow filaments in conjunction with other parameters that influence tip size,
tips have also been formed with diameters up to 1.5 /tm and lengths of only 2-3 mm.
It has usually been assumed that the heating filament of a micropipette puller
would ideally be cylindrical to soften a glass capillary imiformly around its central
axis. In the original design of our airjet micropipette puller, that assumption was
followed by forming a flat ribbon of platinum-indium into a loop (Brown and
Flaming, 1977). With that heating filament the puller was shown to provide very
reliable tips in the ultrafine range, as defined by outer tip diameters of 0.1 fim or
less. These tips could also be much shorter than previously attained for such fine
tips, as little as 6.0 mm when measured from the earliest discernible taper to the
ultimate tip. In obtaining these short tips it was necessary to use a relatively narrow
heating filament (2.0 mm). It also proved critical to cool the filament rapidly as the
tip was formed.
0165-0270/82/0000-0000/$02.75 O 1982 Elsevier Biomedical Press
Though the ultrafine short tips have proved advantageous in many preparations,
long tips are still required in some cases where a considerable depth of nervous tissue
must be penetrated before reaching the target cell. This is because the adhesion of
nerve tissue to glass, combined with the rapid taper of a short tip, cause significant
resistance to penetration. To date the most practical solution of that problem is a
relatively long slender tip, similar to those provided by the well known Liviiigston
puller (Livingston and Duggar, 1934). Hence one goal of this work was to increase
the versatility of our puller by making it readily adaptable to form longer ultrafine
tips. In so doing, it was anticipated that the advantage of high reliability could be
extended to the longer tips.
An effective simple method for producing longer ultrafine tips is a widening of
the heating filament. This requires that filaments of various widths be easily
interchangeable, which proved inconvenient with loop filaments. More important,
when the rather long band that forms the loop filament was widened appreciably, its
power requirements exceeded the design of the power supply; also, the airjet system
was insufficient to cool such filaments rapidly. As a result, the tips formed were
inconsistent and unsatisfactory. In experimenting with other filament shapes we
learned that the form of the heating filament can be simplified without any
significant disadvantage and with a variety of advantages. These include the use of
interchangeable heating filaments that can be widened for increasing the length of
ultrafine tips or narrowed for increasing tip diameter while further reducing the tip
The experimental work of this report was conducted on our own design of
micropipette puller, in part because research on micropipettes is more difficult and
of less assured significance when conducted with pullers of low reliability. Since the
heating element is a critical component of all micropipette pullers, it is believed that
the general principles revealed in this work should apply similarly to other puller
The loop filament
The 3 filament forms used in this work are shown in Fig. 1. All filaments were
made from a sheet of 90% platinum and 10% iridium that was 0.002 in. thick.
In Fig. 1 the top drawing shows the 'loop' type of filament in our original puller
design. Our 'standard' filament of this type has remained 2.0 mm in width, as was
used in forming the micropipettes described with the puller design (Brown and
Flaming, 1977). The width of this filament can be reduced to I.S mm to obtain even
shorter tips, while still adequately shielding the tips from the airjet so that tip
bending does not become a problem. The filament width can also be increased to
about 2.5 mm without encountering the problems that occur with still wider
filaments. At that filament width, tip lengths of about IS mm may be obtained, but
still longer tips are required to cover the desired range of applications. Because of its
horizontal orientation the loop filament can sag at high filament temperatures if the
horizontal 'legs' of the loop are too long. That problem was solved by shortening the
filament legs to 4.3 mm, as shown in Fig. 1.
4.0 mm
Fig. I. Scale drawings of the 3 kinds of heating filaments used in this work.
While theory suggests that tip bending would be minimized by orienting the glass
capillary tubing in the center of the loop filament, for uniform heating of the glass,
this has proved not to be the case. Instead, tip bending is empirically minimized by
placing the glass capillary as close as possible to the loop near its lower leg, as shown
in Fig. 1. This suggests that uniform heating of the glass capillary is not as important
as generally believed for forming straight tips. It also suggests that tip bending is
strongly influenced by other factors that are minimized by the illustrated configuration.
In the original description of our puller, the air tube leading from a small
micrometer-controlled orifice was divided at a Y-junction to form two airjets. which
were mounted symmetrically above and below the loop of the heating filament.
When we subsequently removed the heating filament and held a strip of paper
between the two airjets, puffs of air were noted to deflect the paper randomly in one
direction or the other. So the air in any given puff was passing mainly through one
channel or the other, rather than being divided about equally between the two
channels. Upon consulting with airflow engineers, it was learned that this result may
be expected when an orifice limiting the airflow precedes a Y-junction. Though the
random variation of effectiveness between the two airjets had no obvious consequences for the micropipette tips, it may slightly reduce the reliability of tip
characteristics. In any event, since each puff of air was passing primarily through
only one airjet, a second one appeared unnecessary. We thus kept only the lower
airjet and this has proved satisfactory. As shown in Fig. 1, this airjet is located near
the part of the loop that delivers most of the heat to the glass capillary. Other
portions of the loop deliver less heat to the glass; though they are also farther from
the airjet, they appear to be cooled adequately by heat conduction through the
filament to the site cooled most directly by the airjet.
The horizontal band filament
In studying the feasibility of simpler filament forms, we looked first at the
ultimate simplicity of a straight horizontal band, as shown in the middle drawing of
Fig. 1. This band was placed about O.S mm below the glass capillary, rather than
over it, to take advantage of convective heat rise and thus improve the efficiency of
heat delivery to the glass capillary. With this type of filament the location of the
airjet proved less critical, and it was mounted about 2.0 mm below the center of the
horizontal band, where it could efficiently cool that portion of the filament delivering the most heat to the glass capillary. There is some advantage in having a
reasonably large separation of the airjet from the filament, since this allows the puff
of air to cover better the portion of the filament to be cooled; also, the lesser velocity
of resulting air currents in the vicinity of the forming micropipette may reduce
problems of tip bending.
Micropipettes formed from this type of heating filament showed no significant tip
bending, even when the filament was widened to S.O mm to form tips as long as
about 23 mm. In fact, this form of heating filament gave less difficulty with tip
bending than the loop filament. It appears that several factors contribute to
understanding this result. First, asymmetric heating of the glass capillary is probably
significant only while the capillary diameter remains fairly large. When the capillary
diameter has become quite small, as it has following the rapid initial taper, the
temperature gradient between the lower and upper sides of the tubing apparently
becomes too small to contribute to tip bending. Second, the positioning of the glass
capillary to prevent tip bending is much more critical with the loop filament than
with the horizontal bend. This suggests that complex patterns of convective airflow,
which must occur within the loop, may contribute to tip bending. By contrast, with
the horizontal band filament the convective airflow is only upward and thus less
complex. Third, if the glass capillary is centered in a loop filament, tip bending is a
consistent problem when tips are formed normally by use of the airjet. If the airjet is
inactivated, the normal length of the tip is quite straight, but an extremely long
slender filament extends well beyond the normal tip length and makes the micropipette unusable. These observations indicate: (I) that a minimum amount of air
cooling is required to form usable tips; and (2) that under certain conditions the
airjet contributes'significantly to tip bending. This aspect of tip bending may also be
less severe with the horizontal band filament because of its simpler form and the less
complex resulting air currents. Fourth, it is often considered that a horizontally
oriented micropipette puller will give gravity-bent tips, especially if they are relatively long. In our puller the fast pull is completed within S-8 ms (Brown and
Flaming, 1977). This is probably too little time for the tip to be bent significantly
downward by gravity. Altematively, this effect may be present but almost exactly
cancelled by other factors, such as upward air currents resulting from convection
and the airjet.
The trough filament
Compared with the loop filament, the horizontal band filament delivers considerably less heat to the glass capillary. Thus the horizontal band is not satisfactory for
borosilicate (Pyrex) tubing that is thick-walled or large in diameter, nor is it
satisfactory for aluminosilicate tubing. The horizontal band filament also bends
either upward or downward as the filament lengthens during heating; this alters the
distance between the filament and the glass capillary, which is a critical variable for
pulling micropipettes. Both problems are overcome by the rectangular 'trough'
filament shown in the lower portion of Fig. 1. The bottom of the trough functions
much like the horizontal band filament, but the sides of the trough filament
contribute to the total heat delivered. Also, the sides of the trough filament can tilt
slightly as the filament lengthens during heating, thus taking up most of this effect.
While a slight downward movement of the trough filament may still occur during
heating, this should be consistent in both direction and amount and should not
reduce the reliability of the micropipettes that are formed. In adapting our micropipette puller for horizontal band and trough filaments, it was only necessary to
redesign the filament mount; this was done with special attention to convenience in
changing heating filaments. When using double-barrelled micropipettes based upon
two capillary tubes fused together, the two tubes should be equidistant from the
bottom of the trough to avoid tip bending. Provision has also been made for this
special case by modifying the tubing clamps on the pipette carriers.
After some experimentation, the dimensions chosen for the trough filament were
4.0 mm for the length of the bottom segment, 3.0 mm for the height of the sides, and
3.0 mm for the filament width. These values suffice for most purposes and hence
provide a satisfactory standard, from which alterations may be made for special
purposes. In evaluating this type of filament we have used primarily borosilicate
tubing with outer and inner diameters of 1.0 and 0.5 mm, respectively, a solid fiber
0.1 mm in diameter being fused along the entire length of the inner bore for
convenient filling of micropipette tips. We refer to this as standard tubing, since it is
the type most commonly used in neurophysiology. One reason for choosing the 3.0
mm width for the trough filament was that it proved to form very similar tips from
standard tubing as were provided by the loop filament at a width of 2.0 mm. Also,
the heat delivery from this trough filament has proved adequate for all types of glass
that are now commonly used in neurophysiology. This includes borosilicate tubing
up to at least 2.0 mm in outer diameter and 0.5 mm in wall thickness. It also
includes aluminosilicate tubing with outer and inner diameters of 1.0 and 0.75 mm,
respectively, the only dimensions in which this type of tubing is thus far readily
available. With such heavy borosilicate tubing, or aluminosilicate tubing, trough
filaments appreciably narrower than 3.0 mm deliver too little heat to form tips. But
with these tubings wider trough filaments, up to 6.0 mm, are useful if longer tips are
desired. With standard tubing the width of the trough filament may be varied from
1.5 to 6.0 mm, for the control of tip length, without encountering any special
problems. Even at the width of 6.0 mm, the power supply has proved adequate to
heat this filament to the burnout point, and tip length may be modulated adequately
by the airjet system. These features result from the trough filament having a total
length of only 12 mm between the binding posts, compared to 18 mm for the loop
filament. With the trough filament we have not noted any significant problem of tip
bending, even at the maximum filament width of 6.0 mm. It thus appears that the
trough filament combines the advantages of the horizontal band and loop filaments.
Evaluation of trough filament
Comparison with loop filament
When compared with the loop filament, the trough filament offers the following
(1) The trough filaments are easier to form.
(2) The trough filaments are easier to mount in the puller and are then more
easily interchanged to vary filament width.
(3) The trough filament is less vulnerable to being struck when positioning the
glass capillary because there is no need to thread the glass capillary through a loop.
(4) The trough filament requires less power because it is shorter than the loop
filament, thus offering less electrical resistance and requiring less voltage to attain a
given filament temperature.
(5) The trough filament requires a lesser airflow to provide the minimum cooling
needed to form usable tips. This is partly because the trough filament is shorter:
also, upper portions of the loop filament that contribute to heating the glass, but are
rather inefficiently cooled, have been eliminated. Because of the lower minimum
airflow that is required by the trough filament, the range of airflow available to
control tip length is increased.
(6) As a consequence of items 2, 4 and 5 above, the trough filament may be
greatly widened to form longer tips, as required for certain preparations.
By contrast, we have not identified any significant disadvantage of the trough
filament by comparison with the loop. Though the total heat delivered to a glass
capillary by the trough filament is somewhat less than that delivered by a loop
filament of the same width, this does not appear significant since all tubings
commonly used in neurophysiology are readily handled by either filament. In any
special cases requiring even more heat delivery, the 'standard' trough could readily
be modified by lengthening its sides or bringing them closer together.
Measurement of tip diameters and lengths
Fig. 2 shows the effects of varying filament width upon both tip length and tip
diameter. In obtaining these results, the width of the trough filament was varied
while keeping all other pulling variables as constant as possible. For each filament
width, a filament was burned out and the reading on the digital scale of relative
filament current at the burnout point was noted. These burnouts did not occur at
bends in the filament but were consistently at the center of the filament just above
the airjet. Though the heating current must be quite uniform along the filament,
between the two binding posts, the color temperature of the filament is always
highest in the center where the burnout occurs. This is undoubtedly because the
heavy metal binding posts act as heat sinks that draw off heat from both ends of the
filament. For filaments 3.0-6.0 mm in width the filament current was then reduced
slightly below the burnout point, a reduction of 10 units on the digital readout, and
that value was used to form all micropipettes with a given filament width. The
burnout temperature may be assumed quite similar for filaments of different width,
and when filament widths of 3.0-6.0 mm were used at a constant airflow setting,
filament current could be reduced over a considerable range below the burnout point
without significantly affecting either tip size or tip length. For filaments narrower
than 3.0 mm, however, tip size increased sharply as filament current was reduced.
Hence for these narrower filaments the filament current was reduced only very
slightly below the burnout point, an attempt being made to match the color
temperature at the center of the filament with the color temperature of the wider
filaments. These procedures provided near-maximum filament temperatures that
remained approximately constant with varying filament width and had little or no
effect upon tip measurements.
The length of the slow pull was always 0.50 in. The strength of the slow pull on
each micropipette carrier was determined by an 80 g weight suspended from a
flexible cable, which passed over a pulley and was thence attached to the carrier. The
digital readout for strength of the fast pull was always set at 1250, which provided
SOV across the pulling solenoid. For the airjet system the pressure was SO psi. Of
course the minimum airflow needed to produce usable tips increases with filament
width. The micrometer control of the needle valve was thus set to provide the
minimum airflow required for usable tips with the 6.0 mm filament, and that airflow
setting was used for all fllament widths. The actual reading of the airflow micrometer was 0.075 in. But the reading to obtain the same airflow might be somewhat
different on another micropipette puller of our design. This is because the zero point
on the airflow micrometer does not provide an identical airflow on all instruments.
Standard glass tubing was used in this work. As with the horizontal band
fllament, the glass tubing was located about 0.5 mm above the bottom of the trough
filament, and the airjet aperture was about 2.0 mm below the filament, as shown in
Fig. 1.
We have now developed a convenient and reliable method for mounting micropipette tips for high resolution scanning electron microscopy (SEM), as described in
a separate paper in preparation, and that mounting method was used in this work.
Prior to SEM observations the tips were coated with 10 nm of gold. This has proved
very close to the minimum thickness of gold that sufflces for these puqxises. and it is
somewhat less than the 15 nm used previously (Brown and Flaming, 1977). Photographs illustrating the resolution that we attain with ultraflne tips at a magniflcation
of about 40,000 X have already been provided (Brown and Flaming, 1977). Tip
diameters were measured from such photographs, and twice the thickness of the gold
coating was subtracted to obtain accurate diameters for the glass tips. Following our
previous practice, tip length was measured from the very beginning of the taper to
the ultimate tip.
Filaments were used at 6 different widths, which cover the usable range of
fllament widths under the conditions described. On any given day of SEM observations, all micropipettes were formed at a given fllament width and then measured
that same day. The number of tips successfully measured in these daily fllament-width
groups varied from 12 to 15 with an average of 14.4. The order of filament widths
examined was as follows: 1.5, 3.0, 5.2, 2.25, 4.0. 6.0 and 3.0 mm. The final retest at
the 3.0 mm fllament width was to determine whether any change had occurred
between the initial test and the retest of that filament. The respective test and retest
values were 12.6 and 12.0 mm for tip length and 0.079 and 0.073 /im for tip
diameter. These differences between test and retest values are too small to require
consideration in the interpretation of our results.
Fig. 2 shows that when the 6.0 mm fllament was used with minimal air cooling.
the tip lengths averaged 26.8 mm. In our experience ultrafine tips cannot be made
shorter than about 25 mm on the Livingston puller. Hence our range of tip length
now extends to somewhat above the shortest ultraflne tips provided by the Livingston puller. As filament width was reduced, tip length decreased progressively and
markedly. As this occurred, tip diameter remained quite constant at about 0.08 fim
down to a filament width of 3.0 mm. which gave a tip length of 13.0 mm. This
flnding supplements earlier evidence that over a considerable range of tip length, our
puller can provide ultraflne tips of quite constant size (Brown and Flaming. 1977).
Below a fllament width of 3.0 mm. Fig. 2 shows that tip length continued to decrease
rapidly, while tip diameter began to increase. A smooth curve fitted by eye to the
d a u on tip diameter suggests that tip diameter did not increase to any statistically
signiflcant extent until tip length had fallen to about 10.0 mm. In any event, at a
filament width of 2.25 mm the tip length had fallen to 7.9 mm and the tip diameter
had risen only to 0.1 fim, which is still within the ultraflne range. Our deflnition of
the ultraflne range of tip sizes is operationally useful because we find that tips in this
range function extremely well for intracellular work in the outer and inner segments
of toad rods, which have average diameters of only about 6.0 /im (Brown and
Flaming, 1977). At a filament width of 1.5 mm. Fig. 2 shows that tip length had
fallen further to 5.4 mm but tip diameter had risen sharply to 0.18 fim. That rise of
tip diameter is quite signiflcant, at least for certain preparations of small cells. In the
outer and inner segments of toad rods, for example, we have found that the
frequency of cell penetration with 0.18 fim tips drops to about one-third of that
obtained with ultrafine tips.
The results shown in Fig. 2 are not inconsistent with our earlier report that
Tip DiometBr
Width of Heating Fllomtnt (mm)
Fig. 2. Tip length and tip diameter as a function of width of the trough type of heating filament. Vertical
error bars indicate^ 1 S.E. of the mean. For further details, see text.
ultraflne tips could be obtained down to tip lengths of 6.0 mm (Brown and Flaming.
1977). This is because tip length was reduced in Fig. 2 by decreasing fllament width
while holding airflow constant. When the fllament becomes quite narrow, the total
heat delivery becomes inadequate to heat the glass sufficiently to form ultraflne tips.
Hence when tip length is reduced by decreasing filament width, a point is inevitably
reached beyond which a further reduction of tip length is accompanied by an
increase of tip diameter, as Fig. 2 illustrates. By contrast, our earlier results were
obtained with a loop fllament of flxed width, the tips being shortened by increasing
airflow. In that case the glass was always thoroughly heated at the beginning of tip
formation, only the rate of cooling of the tip being modulated by the airflow, and
that procedure appears preferable for producing ultrafine tips of minimum length.
In certain applications such as patch clamping, it is desired to have outer tip
diameters up to about 1.0 /im to provide inner tip diameters of about O.S /im
(Sigworth and Neher, 1980). Fig. 2 indicates that narrow filaments can be useful
toward that goal, while having the additional advantage of producing particularly
short tips. We have also determined that thin-walled tubing increases tip diameter,
as theoretically predicted (Brown and Flaming, in preparation). By using narrow
filaments and thin-walled tubing, with respective outer and inner diameters of 1.0
and 0.75 mm, we have recently produced tips only 2-3 mm long with tip diameters
up to about 1.5 fim. Under given pulling conditions the tip sizes are consistent, a
result that appears not to have been achieved previously for such large tips. This
may solve the problem of consistently forming tips of appropriate sizes for patch
clamping. Since these tips are formed in the puller, rather than being pulled to a
small diameter and then broken, it is also possible that the additional step of Are
polishing may be eliminated. Hence these tips must now be evaluated in patch
clamping applications. Experiments are likewise being conducted to study the
interacting effects of all the critical variables in micropipette tip formation with
respect to both tip length and tip diameter.
Applications of short and long ultrafine tips
The importance of varying tip length for different requirements is well illustrated
by the intracellular retinal work of our research group. The ultraflne short tips have
proved valuable in the isolated and inverted toad retina, where intracellular recordings have been made from the outer and inner segments of rod photoreceptors
(Brown and Flaming. 1977). In that work the approach has been directly to the
exposed tips of the outer segments. Because high quality and extremely stable
intracellular recordings have been readily attained, some flndings have been obtained that would otherwise have been very difflcult and time consuming or not
feasible at all (Brown and Flaming. 1978: Flaming and Brown. 1979). Similar
advantages of the ultrafine short tips have been reported to us from many other
investigators. In another project of our research group, intracellular work has been
conducted in the isolated and perfused eyecup preparation of the Eastern grey
squirrel (Charlton et al.. 1981). The retinal target cells in this work have been
primarily horizontal cells and photoreceptors, which in this preparation must be
approached from the vitreous humor. Though the retina itself is only about 200 /um
in thickness, there is always an overlying layer of vitreous humor that is quite viscous
and variable in depth. In order to reach the target cells through the overlying tissues.
it has proved necessary to use longer tips than provided by the original design of our
Comparison of our long tips with those formed by Livingston puller
For intracellular work in the squirrel retina, relatively long micropipette electrode
tips have been formed both by wide trough fllaments in our puller and by the
Livingston puller. The former group of microelectrodes offered several advantages.
First, tip length could be varied gradually up to about 27 mm. and tips considerably
shorter than the maximum could often be used in this preparation, but we have not
been able to obtain ultrafine Livingston tips shorter than about 25. mm. Second,
these electrodes performed consistently well in this preparation, indicating uniformity of tip size and tip conflguration. By contrast, the Livingston tips were quite
variable in performance, and valuable experimental time was often required to
identify a useful microelectrode. Since most biological preparations can be used for
only quite limited periods of time, the reliability of micropipette performance is
often critically important. Third, there was no significant tip bending. By comparison, the Livingston tips showed tip bending, which was sometimes considerable and
troublesome; though this may be avoided at low pull tensions, it occurred consistently at the high pull tensions required to form ultraflne tips.
The requirement of rapid tip cooling in all micropipette pullers
As already mentioned, experience with our puller indicates that it is not possible
to form a usable micropipette tip close to a heating fllament unless the filament has
been deliberately cooled. In our instrument two pipette carriers are pulled apart
symmetrically to form two essentially identical micropipettes, the tips being formed
as they separate at about the midpoint of the fllament width. Though the tips are
then pulled away from the fllament very rapidly, inactivation of the airjet system
results in tips that are so long, slender, and lacking in stiffness that they cannot be
used. Thus, a minimum amount of air cooling is required to form usable tips, while
still greater amounts of air cooling can be used to shorten the tips. The Livingston
puller also makes two symmetrical micropipettes and solves the minimum cooling
requirement by lifting the capillary tubing up and out of the heating fllament as the
tip is formed. So this upward motion of the glass capillary is an intrinsic aspect of
the design, rather than an incidental effect of the pulling motion, and it is the most
obvious probable cause of tip bending in that puller. In single-sided micropipette
pullers, which have only one movable carrier, the tips are cooled by being formed
well outside the heating filament in the direction of the movable carrier. A minor
limitation is that the two tips are quite different in length, so for most work only one
micropipette can be used from each pull. A greater limitation of single-sided pullers,
which also applies to the Livingston case, is that cooling effects cannot be modulated
to vary tip length.
Apphcation of findings to other micropipette pullers
It is noteworthy that the Livingston puller has always employed a trough
fllament, but it is so deep that the sides provide relatively more heat than in the
design we employ. Thus the Livingston puller must provide more uniform heating of
the glass tubing, and in that respect it is probably intermediate between our trough
fllament and a loop filament. It is also noteworthy that several investigators found in
the early 1960s that widening the filament of the Livingston puller produced
sufflciently fine tips for intracellular work in vertebrate photoreceptors, and this
flnding was critical to initiating intracellular work in those small cells. Presumably
that observation used the principle illustrated in Fig. 2 of this paper, which shows
that if one starts from a sufficiently narrow filament, widening the fllament can
reduce the tip size while also increasing the tip length. In any event, the general
principles found in this work may be expected to apply similarly to other puller
designs. Thus, some other horizontal pullers could probably be improved signiflcantly by the rectangular trough fllament. And the general effects of fllament width
upon tip size and tip length may be expected to apply in all micropipette pullers,
whether trough or loop fllaments are used in either horizontal or vertical orientations, though the quantitative details of the effects undoubtedly vary between puller
This work was supported by Grant EY 00468 from the National Eye Institute.
We are also indebted to H.F. Leeper and J.S. Charlton for assistance in evaluating
some of the micropipette electrodes of this work.
Brown. K.T. and Flaming. D.G. (1977) New microelectrode techniques for intracellular work in small
cells. Neuroscience. 2: 813-827.
Brown. K.T. and Flaming. D.G. (1978) Opposing effects of calcium and barium in vertebrate rod
photoreceptors. Proc. Nat. Acad. Sci. (U.S.A.). 75: 1587-1590.
Charlton. J.S.. Leeper. H.F. and Brown. K.T. (1981) Intracellular recording and horseradish peroxidase
morphology of tree squirrel horizontal cells. Neurosci. Abstr.. 7: 619.
Flaming. D.G. and Brown. K.T. (1979) Effects of calcium on the intensity-response curve of load rods.
Nature (Und.). 278: 852-853.
Livingston. L.G. and Duggar, B.M. (1934) Experimental procedures in a study of the location and
concentration within the host cell of the virus of tobacco mosaic. Biol. Bull.. Mar. Biol. Lab.. Woods
Hole. 67: 504-512.
Sigworth. F.i. and Neher. E. (1980) Single Na* channel currents observed in cultured rat muscle cells.
Nature (Lond). 287: 447-449.
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF