IBM microscope sees reactions in action

Researchers at IBM have devised a way to observe liquid chemical reactions as they occur with a transmission electron microscope. The technique has already been used to capture images showing how copper atoms bond to each other and to electrodes.

Such information could eventually lead to smaller circuits or more efficient chip manufacturing processes because chip designers will have exact information on what happens when molecules meet.

"They (chip designers) know the right conditions for growing copper films, but they don't know the basic physics," said Frances Ross, manager of IBM's nanoscale materials analysis. On Tuesday, Ross received the Burton Medal, awarded each year by the Microscopy Society of America to one scientist under the age of 40.

The technique could also be used to study rust in action, especially the nuances of underwater corrosion.

"We understand the basic idea, and know a lot about rust in the air, but it is pretty hard to study it underwater," she said.

IBM's method--detailed in an upcoming article in the August issue of Nature Material--essentially combines attributes of different types of scientific microscopy.

Atomic force microscopes can be used to observe reactions at the atomic level in liquids. Unfortunately, they can only capture images at around the rate of one image every 30 seconds. These devices therefore deliver accurate, but somewhat static, information.

Transmission electron microscopes, by contrast, produce up to 30 images a second, the same as a standard video camera.

Unfortunately, transmission microscopes--which shoot electrons through micron-thin samples of materials and then form an image from data about the resulting paths of the electrons--depend on placing the sample in a strong vacuum. That's fine for observing reactions between solids and gases, but it doesn't work for reactions with or inside liquids.

"In ordinary circumstances, the liquid would just boil away," Ross said.

Biological samples viewed under a transmission electron microscope have to be initially dehydrated, which can change their shape.

To get around that problem, IBM devised a cell chamber that captures a layer of liquid and the elements to be studied between two silicon nitride membranes.

"You can think of the cell as an extremely sophisticated (microscope) slide," she said.

Rethinking chip basics
So far, the technique is prompting chip researchers to rethink some of their basic assumptions. Copper is the basic metal in chip interconnects, wires that connect different circuits. Interconnects are formed by bonding copper atoms floating in a liquid to an electrode. Initially, groups of atoms form clusters on the surface of the electrodes. The clusters subsequently merge with other clusters until sheets (or wires) are formed.

Researchers have generally believed that the clusters begin to first form on defects or anomalies on an electrode surface, similar to how icicles form on irregularities on a roof. As a result, some have theorized that it could be possible to control interconnect growth by devising designer electrodes.

That doesn't appear to be the case, Ross said. The IBM group conducted a reaction on an electrode and copper atoms 30 times and the clusters began to form in a different place each time. (The reaction was reversible so the same electrode was used all 30 times).

"All of the sites seem to have an equal chance," Ross said. Instead, cluster formation may have to do with affinities caused by the spacing and size of the clusters themselves.

The technique could also be used in the polishing and etching stages of chipmaking. After the circuits are "drawn" into silicon wafers, excess metals and other materials have to be burned off before the wafer can be cut into chips.

"We should be able to see the initial processes," she said.

Although it's primarily known for computers, IBM remains one of the dominant companies when it comes to microscopes. Researchers won a Nobel Prize for developing the scanning tunneling microscope, which is used in nanotechnology.