Sharp tips are critical for obtaining high resolution images in scanning probe microscopy (SPM), particularly in samples with large variations in topography. For tuning-fork-based SPM, such tips are commonly obtained by electrochemical etching of metallic wires (e.g., tungsten). Electrochemical etching of metallic wires is the preferred means of preparing tips for scanning tunneling microscopy (STM), and techniques for obtaining sharp tips have been investigated extensively. However, the requirements for STM and tuning-fork-based SPM are different. In particular, the wires used in STM are typically 250500μm in diameter, while the wires used for tuning-fork-based SPM are usually an order of magnitude narrower in order to minimize loading of the tuning fork: 2550μm and sometimes down to a few micrometers in diameter. Consequently, many of the recipes developed for etching thicker metallic wires for STM tips do not give optimal results for smaller diameter wires. The authors describe here a modification of the etching circuit of Ibe et al. that significantly improves the reproducibility and reliability of the etching process for thin wires, and discuss the parameters that affect the aspect ratio of produced tips.

Tips for scanning tunneling microscopy (STM) and tuning-fork-based scanning probe microscopy (SPM) are obtained by electrochemically etching a metallic wire in a solution of KOH or NaOH.1–11 While different metals can be used, tungsten (W) is frequently used for both STM and SPM, and we focus here on the etching of W wires in KOH solution. Figure 1 shows a schematic of a typical setup of the electrochemical cell used for etching: a tungsten wire is dipped into a glass beaker filled with a few molar solution of KOH. A positive bias is applied to the tungsten wire, which serves as the anode; a loop of stainless steel wire serves as the cathode.

Water is hydrolyzed at the cathode, producing hydrogen gas and [OH] ions that travel to the anode and react with the tungsten anode, producing soluble tungstate ions,1 

cathode:6H2O+6e3H2+6[OH],anode:W+8[OH]WO42+4H2O+6e.

While one might expect the latter reaction to occur over the entire length of the tungsten wire below the surface of the KOH solution, the literature suggests that etching occurs preferentially at the surface where the meniscus forms, as shown in Fig. 1. The explanation given for this preferential etching is that etching is slower at the top of the meniscus as the density of [OH] is reduced, while the surface of the tungsten wire below the meniscus is protected by a downward flow of heavier WO42 ions produced by the electrochemical reaction.1 As etching progresses the neck narrows. Eventually, the weight of tungsten wire below the surface fractures the wire, resulting in a sharp tip. However, if the bias is not turned off immediately after the wire breaks, the small section of the tungsten wire below the surface of the KOH solution that is still connected to the circuit continues to be etched, dulling the tip. Consequently, it is important that the bias applied to the anode be switched off as soon as possible after the wire breaks in order to obtain a sharp tip. While this can be done manually, typical human reaction times are on the order of tens of milliseconds, which is not fast enough or reliable enough to obtain reproducibly sharp tips. A number of authors have developed circuits that automatically switch off the bias within less than a microsecond once the wire breaks. In particular, the circuit developed by Ibe et al.1 appears to be widely used. We shall discuss this circuit in more detail below.

The molarity of the solution and the voltage bias applied across the cell affect the etch rate of the wire. A number of other factors have also been reported to affect the sharpness of the tips as well as their aspect ratio, including the shape of the meniscus,1 the evolution of gas bubbles from the cathode that disturb the meniscus (see, for example, the setup in Ref. 3), and the length of the wire below the surface of the electrolyte solution, which determines the weight of the lower part of the wire and the point at which the wire fractures. Most of these studies focused on making tips for STM, with metallic wires 250500μm in diameter. Our interest is in etching wires of far smaller diameters for tuning fork SPM: down to 5μm, but more typically 2550μm. This reduction of a factor of more than ten in diameter results in a completely different regime in terms of the mass of the wire, etch current, and etch time so that many of the recipes used for etching wires for STM do not work well. To illustrate this, we focus on etching 50μm diameter tungsten wires, although the techniques we describe can be adapted in a straightforward manner to smaller diameter wires. We will discuss an improved etching circuit, the design of the electrochemical cell, and solutions to a set of technical problems associated with etching small diameter wire.

In our initial attempts to etch tungsten wires, we used the circuit described by Ibe et al.1 This circuit applies a bias to the anode from the positive voltage supply through a p-channel MOSFET and small resistor (+12 V through 50Ω in the case of Ibe et al.). The cathode is connected to ground through a variable resistor (500Ω in Ibe et al.), the voltage across which acts as a measure of the current going through the electrochemical cell. As the wire is etched, the current through the cell and the corresponding voltage across the variable resistor decreases; this voltage is one input to a fast comparator. The second input is a variable voltage that sets the threshold below which the reaction should be stopped. The output of the comparator is connected to the gate of the p-channel MOSFET. At the start of the etching process, the voltage across the variable resistor is larger than the threshold voltage, the output of the comparator is low, and the p-channel MOSFET is turned on (conducting). As the etching proceeds, the current decreases, eventually falling below the threshold voltage. This causes the output of the comparator to go positive, switching off the p-channel MOSFET, and stopping the etch. With a fast comparator and MOSFET, the switch occurs in fractions of a microsecond.

The problem with this circuit is that the electrochemical cell is neither voltage biased nor current biased so that controlling the etch parameters of an individual 50μm diameter wire is difficult given the variability of the resistance of the electrochemical cell from wire to wire. The etch rate and the ideal threshold cutoff voltage (i.e., the current at which the wire breaks) vary from wire to wire, and as it is not possible to know the threshold voltage a priori. While it is possible to obtain sharp tips with good aspect ratios with the circuit of Ibe et al. (see Fig. 2, for example), there is a large variation in the characteristics of the resulting tips. To improve the reproducibility, we have developed an etching circuit that applies a specified voltage bias between the anode and the cathode of the electrochemical cell and cuts the bias when the current through the cell falls below a preset threshold.

To understand why applying a fixed voltage bias would enhance the reproducibility of the etching process, consider the voltammogram (i.e., the current vs voltage curve) of the electrochemical cell. Figure 3 shows these data for a number of 50 μm diameter wires dipped 1–2 mm in 1.25M KOH solution, with a 20 mil stainless steel wire acting as the cathode. The voltage was supplied by a Keithley 2400 SourceMeter, which also measured the current through the cell. These data were taken by sweeping the voltage across the cell at either 1 or 3 V/min and measuring the resulting current. As can be seen, essentially no current flows (and hence no etching occurs) below a voltage of about 0.8 V. At higher voltages, a sharp drop-off in the current indicates that the bottom part of the wire has etched off; the total time for this to occur after etching starts is on the order of 30–60 s. There is a considerable amount of variation in the voltage at which the wire drops off and the etching current at a specific voltage, even for wires under nominally identical conditions. However, well above the threshold voltage of 0.8V, there is a significant etching current for all wires. For example, if we were to set the voltage across the cell at 1.5 V, all wires would etch at a good rate. Although these rates might vary between runs, the etch circuit is comparing the current, so the cutoff voltage and quality of the etch are independent of this rate. If we were to use the circuit of Ibe et al., however, the variation in the effective resistance of the cell would result in variations in the potential across the cell, leading to exponential variations in the etching current, since the dependence of the current on bias voltage is almost exponential in the etching regime.

In order to be able to apply a fixed voltage bias across the electrochemical cell, we have developed the simple etching circuit shown schematically in Fig. 4. The voltage bias to be applied to the cell is adjusted by means of a potentiometer connected to the +Vcc of the circuit. The range over which the voltage can be varied can be controlled by adding a resistor in series: for our application, we targeted a voltage range of 5 V with +Vcc=15V, using a 5kΩ pot with a 10kΩ resistor in series. The voltage from the pot is applied to the electrochemical cell through an op-amp in a follower configuration, so that the voltage applied to the anode is exactly the voltage applied to the input of the op-amp. As a typical inexpensive op-amp (a Texas Instruments LF356 in this case) can supply only a few milliamperes of output current, we use an NPN transistor whose collector is tied to +Vcc to boost the output current. For the inexpensive 2N2222 NPN transistor shown in Fig. 4, chosen because it was at hand, this boosts the maximum output current to 800mA, which should be sufficient for etching, even the much thicker wires required for STM. The transistor can be changed if larger currents are required, so long as the power supply can output the required current.

In order for the voltage from the bias potentiometer to set the voltage applied across the electrochemical cell, the cathode of the cell must be at virtual ground potential. To do this, we connect the cathode to the inverting input of a second LF356 op-amp whose noninverting input is connected to ground, with the inverting input connected to the output of the op-amp through a feedback resistor, in a standard current preamp configuration. Thus, the inverting input is at ground potential, and the voltage at the output of this op-amp is directly proportional to the current flowing through the electrochemical cell. For our parameter range, we chose a feedback resistor of 10kΩ so that the (negative) voltage at the output of the op-amp is a direct measure of the current, with 1 V corresponding to 100μA.

This current output provides the positive input to a fast comparator (a Texas Instruments LM306). The negative input to the comparator is provided from a potentiometer that sets a (negative) voltage proportional to the threshold current below which the etching process should be stopped. When the current through the cell is more than this threshold current, the output of the comparator is low. Following the design of Ibe et al., the output of the comparator is connected to the gates of two MOSFETS. One (p-channel) MOSFET is in series with the voltage bias circuit and is in the “on” (conducting) state when the output of the comparator is low. The second (n-channel) MOSFET connects the top of the bias potentiometer to ground and is in the “off” (nonconducting) state when the output of the comparator is low. Thus, in this state, the desired bias voltage is applied to the electrochemical cell.

As the etching progresses, the current through the electrochemical cell progressively decreases, and at some point drops below the preset threshold current. When this happens, the output of the comparator goes high (aided by the pull-up resistor connected to +Vcc), switching off the p-channel MOSFET and turning on the n-channel MOSFET, with the result that the voltage at the top of the bias potentiometer is set to zero, and no potential is applied to the electrochemical cell, stopping the etching process. A light emitting diode (LED) connected to the output of the comparator indicates when the etching process has stopped. The switching times of the LM306 comparator as well as the MOSFETS are nominally better than 30 ns so that the turn-off happens in fractions of a microsecond.

A two-pole, three-position switch, one pole of which is connected to the positive input of the comparator provides control of the etching process. When the switch is in the “Start” position, the positive input of the comparator is tied to 5V, or any voltage larger in magnitude than the maximum (negative) voltage that is provided by the current threshold setpoint pot. With the anode disconnected from the cathode, this allows one to set the desired potential at which the etch occurs by adjusting the Vbias pot. With electrolyte in the cell, setting the switch to the “Start” position will start the etching process. With the switch in the “Stop” position, the positive input of the comparator is grounded, sending the comparator’s output high, thus stopping the etching process. During the etch, the switch is kept in the middle position so that the positive input of the comparator is connected only to the output of the current preamp.

With a specific voltage bias, this circuit proves very reliable in switching off the etching when the current is below the threshold setpoint. Ideally, this bias should be chosen by measuring a few voltammograms for a specific electrochemical cell configuration and electrolyte molarity as in Fig. 3, but may also be chosen by simply observing what voltage gives a good etch rate under an optical microscope. Another consideration is that for very small currents at the tail end of etching thin wires, the voltage offsets of the op-amps and comparators should be taken into account. Unfortunately, just establishing these parameters is not sufficient to reproducibly fabricate sharp tips with the desired short aspect ratio, as the design of the electrochemical cell and the etching rate all strongly affect the process, particularly for small diameter wires. We discuss these issues in more detail below.

In order to determine the best technique to reliably obtain sharp tips, we have investigated a number of different electrochemical cell configurations. For this study, the tungsten wire to be etched was connected to a small circuit board using pressed indium contacts, with the voltage bias being applied to the circuit board contact. We have also used the final setup to etch wires mounted on tuning forks. The cathode was a 20 mil diameter soft stainless steel wire that could be molded into the desired shape. The simplest cell is similar to the one shown schematically in Fig. 1. For the “beaker,” we used a 2 ml glass vial.

Preserving the shape of the meniscus is critical to obtaining good aspect ratio tips.1 During the etching process, hydrogen gas is produced at the cathode; this production can be rather violent at higher etching rates, and hydrogen bubbles that are generated easily disturb the electrolyte solution around the wire.1,3,8 A higher molarity solution means faster etching rates. While some researchers have reported using up to 7M solution, we found that with 5M solution, the etching reaction was too vigorous, producing many bubbles, even for small applied voltage bias. Consequently, we restricted ourselves to KOH solutions of molarity less than 1.25M. For the same reason, we also formed the cathode in a loop in contact with solution only at the surface of the electrolyte, as we found that bubbles formed from a loop placed deep in the electrolyte solution emerged at all points at the liquid-air interface, including near the wire. We also tried encasing the cathode in a glass sheath by placing the stainless steel wire in the broken end of a small pipet. This did indeed confine the hydrogen bubbles to within the glass pipet, with very few near the tungsten wire, but also significantly cut down on the wire etch rate for otherwise identical conditions of electrolyte molarity and applied voltage bias, presumably due to the reduced diffusion of [OH] from the cathode to the anode.

The main problem that we encountered in etching tungsten wires by dipping them in a beaker with KOH solution was that the etching of the tungsten wire was not confined to the region near the meniscus, but occurred throughout the length of the wire that was immersed in the solution. An example of this is shown in Fig. 5(a), where the etch was stopped early in order to demonstrate clearly the etching along the entire immersed length of the wire. If the etching current is small so that the etch is gentle and takes a relatively long time of a few minutes, the immersed length of the wire will be almost completely etched before it finally floats away. For more vigorous etches with larger etch currents, the immersed section of the wire is still etched substantially but very nonuniformly so that the wire may break before the immersed section is completely etched, but it may break at a number of different points below the surface. An example of such a wire is shown in Fig. 5(b), where the long aspect ratio and the nonuniformity of the wire taper clearly show this effect.

The substantial etching of the immersed section of the wire is clearly different from what has been reported by other researchers for thicker wires for STM. We speculate that this may be due to the smaller amount of tungstate ions produced in etching thinner wires. In any case, the uniform etching along the entire immersed length of the wire has detrimental consequences in terms of the sharpness, shape, and aspect ratio of the tip. By the end of the etch, the section of the wire below the surface of the electrolyte has been almost completely etched, so one cannot depend on its weight to result in a clean break. Instead, during the final stages of the etch, the wire is more likely to bend at the weakest point just before fracture, resulting in a bent tip as shown in Fig. 5(b).

To avoid this problem, we use a small loop of the same 20 mil stainless steel wire with a drop of electrolyte on it, instead of a beaker. The circuit board holding the tungsten wire to be etched is mounted on a precision xyz stage that allows one to insert the wire through the center of the loop without touching the stainless steel wire. A drop of electrolyte solution is then placed on the loop and held there by surface tension, forming a meniscus both above and below the loop. As the wire is etched throughout the length that is in contact with the electrolyte, it is important to minimize the size of the drop on the loop. This can be accomplished by carefully “wicking” away the excess solution with the small fibers at the end of a torn laboratory paper wipe. Using this technique, one can get the thickness of the electrolyte layer in contact with wire to be of order 100μm or so. Special care must be paid to the stability of the drop at this size, as evaporation and sputtering can cause rapid mass loss. In a brief amount of time, a matter of seconds, the amount of solution is insufficient to maintain the meniscus, and the etching stops. By trial and error, we have found that the ideal amount of solution is such that the initial thickness of the drop is about the thickness of the stainless steel wire (20mils500μm). An example of the setup with the drop is shown in Fig. 6.

Our first attempts using the loop setup used long tungsten wires, under the assumption that the weight of the wire below the loop would be sufficient to form a clean break with a sharp tip, since the portion of the wire below the loop was no longer being etched. However, even with 5mm of wire below the loop, this was not the case. The reason is that the surface tension is strong enough to keep the bottom section of the wire attached to the droplet even after the top and bottom sections of the wire separated. As the etched section of the wire reaches a thickness of a few micrometers, the lower section of the wire bends up and attaches itself to the lower surface of the droplet, resulting in a bent tip, very similar to the tip in Fig. 5(b).

To combat this, we attempted to attach a weight to the bottom section of the wire. As has been noted before, the mass of the attached weight is critical. With a larger mass, one is likely to prematurely fracture the wire, resulting in a wider, jagged tip. An example of such a tip is shown in Fig. 7, where the wire was weighted with a partially hanging self-closing tweezers. To obtain a smaller mass, we attached a small (3mm diameter) ball formed from plastic clay (“PlayDoh”12) to the ends of the wire. When the mass of this PlayDoh ball was of the correct magnitude, reproducibly sharp tips could only be obtained if the tungsten wire to be etched had no bends or kinks and was perfectly vertical, which is extremely difficult to achieve in practice. More likely, after being mounted on the holder or tuning fork, the tungsten wire had kinks and bends (e.g., the wire in the loop in Fig. 6) or was not vertical (the left-most wire in Fig. 6). Due to the tensility of the tungsten, the mass of the PlayDoh ball was not sufficient to straighten the wire before the etch so that the ball at the tail end of the wire was at an angle. If the wire was not straight or had kinks, the PlayDoh mass would move just prior to wire separation, resulting in a bent or fractured tip.

Our efforts to address the stated technical challenges have yielded the following procedure to obtain reproducible results with the 50μm and narrower wires. After mounting to the tip holder or tuning fork, the length of the tungsten wire is cut to approximately the correct length. While this can be done with a pair of scissors, we prefer to simply etch the wire at the desired length using a fast etch rate in order to avoid undue stresses at the wire mounting point. The fast etch results in a tip that may be bent, jagged, or blunt depending on the circumstances. We then use one or a few finishing etches to obtain the desired tip aspect ratio and sharpness. If the end of the wire is blunt at the start of this final etching process, one can produce a smooth but relatively blunt tip by dipping only the tip of the wire in the electrolyte for etching. Such tips would be suitable for imaging on relatively flat samples, and an example is shown in Fig. 8(a), achieved using 2.0V bias in 1.25M KOH solution. If a sharper tip is desired, one can first etch the wire in the normal way to form a taper and then form a well defined tip by immersing only the last few micrometers in the solution. A tip etched with the latter technique is shown in Fig. 8(b). While this tip is not as sharp as the one shown in Fig. 2, the process is far more reproducible. Further examples of this reproducibility can be seen in Fig. 9, etched by different operators at different times.

In summary, we have presented a new circuit design for etching wires for scanning probe microscopy wires that greatly enhances the reproducibility of the etching process. While the focus of this paper has been on techniques for etching thin tungsten wires for scanning probe microscopy, the circuit and general techniques can be applied to different metal wires, as well as the thicker wires usually used for scanning tunneling microscopy.

This research was conducted with support from the U.S. Department of Energy (DOE), Basic Energy Sciences, under Grant No. DE-FG02-06ER46346.

1.
J. P.
Ibe
,
P. P.
Bey
,
S. L.
Brandow
,
R. A.
Brizzalora
,
N. A.
Burnham
,
D. P.
DiLella
,
K. P.
Lee
,
C. R. K.
Marrian
, and
J. P.
Colton
,
J. Vac. Sci. Technol. A
8
,
3570
(
1990
).
2.
A. H.
Sørensen
,
U.
Hvid
,
M. W.
Mortensen
, and
K. A.
Mørch
,
Rev. Sci. Instrum.
70
,
3059
(
1999
).
3.
Inger
Ekvall
,
Erik
Wahlström
,
Dan
Claesson
,
Häkan
Olin
, and
Eva
Olsson
,
Meas. Sci. Technol.
10
,
11
(
1999
).
4.
E.
Abelev
,
N.
Sezin
, and
Y.
Ein-Eli
,
Rev. Sci. Instrum.
76
,
106105
(
2005
).
5.
C.
Albonetti
,
M.
Cavallini
,
M.
Massi
,
J.-F.
Moulin
, and
F.
Biscarini
,
J. Vac. Sci. Technol. B
23
,
2564
(
2006
).
6.
Till
Hagedorn
,
Mehdi
El Ouali
,
William
Paul
,
David
Oliver
,
Yoichi
Miyahara
, and
Peter
Grütter
,
Rev. Sci. Instrum.
82
,
113903
(
2011
).
7.
Matthew M.
Jobbins
,
Annette F.
Raigoza
, and
S.
Alex Kandel
,
Rev. Sci. Instrum.
83
,
036105
(
2012
).
8.
Yasser
Khan
,
Hisham
Al-Falih
,
Yaping
Zhang
,
Tien
Khee Ng
, and
Boon S.
Ooi
,
Rev. Sci. Instrum.
83
,
063708
(
2012
).
9.
G.
Basnet
,
J. K.
Schoelz
,
P.
Xu
,
S. D.
Barber
,
M. L.
Ackerman
, and
P.
Thibado
,
J. Vac. Sci. Technol. B
31
,
043201
(
2013
).
10.
H.
Chen
,
W.
Xiao
,
X.
Wu
,
K.
Yang
, and
H.-J.
Gao
,
J. Vac. Sci. Technol. B
32
,
081801
(
2014
).
11.
J. A.
Morán Meza
,
C.
Lubin
,
F.
Thoyer
,
K. A.
Villegas Rosales
,
A. A.
Gutarra Espinoza
,
F.
Martin
, and
J.
Cousty
,
Carbon
86
,
363
(
2015
).