单晶片清洁和干燥

The cleaning technique of interest to this work is based on megasonics energy distributed over a large area of the wafer  surface. Applied in a single wafer tool design, the distribution  of the megasonic energy via a quartz rod is very efficient at particle removal. The basic design of the megasonic quartz  rod and movement of the wafer under the rod is depicted in figure one [1, 2]. The top and bottom chemical dispense  nozzles allow for identical or independent chemistries to be directed at the wafer during the cleaning cycle. The power  density applied to the wafer is approximately 5 W/cm2 . This level of power density insures efficient particle removal with no wafer or feature damage [16]. For comparison, water  megasonically excited and dispensed through a spray nozzle onto the wafer surface provides a more concentrated power  source. A typical spray nozzle power distribution is 1300 W/cm2 .

The spin dry, while easy to implement is not very effective at preventing recontamination of the wafer surface or eliminating water marks [5]. A drying method that maintains  the cleanliness level achieved with the clean and in many cases can improve it is based on the Marangoni principle. Combining rotational forces with the surface tension forces of the Marangoni principle results in the “Rotagoni” drying method (figure 2) [6, 8, 17]. The basic operation of the  Rotagoni dry is as follows. The wet wafer is horizontally rotated and a moveable arm containing two dispense nozzles  is brought to the center of the wafer. The forward nozzle  contains a DI water dispense and the trailing nozzle contains a  N2/tensioactive vapor dispense. The arm is then moved from the center of the wafer to the edge drying the wafer.

The main idea behind the Rotagoni dry is removal of the liquid on the wafer surface by a surface tension gradient.  Therefore, any material suspended in the liquid will be removed as well. In contrast, a spin dry removes the final amount of liquid by evaporation. Material suspended in the  evaporated liquid will then deposit onto the wafer surface.

Fig1

All megasonic cleaning and drying data was collected on a Verteq Goldfinger 200 mm single wafer platform. Typical Rotagoni drying conditions were 300 to 500 rpm, DI-UPW flow 200 ml/min, and a N2 flow of 2 SLM, resulting in a dry time of less than 25 seconds. The spin dry at 1800 rpm  required 25 seconds. All processing was done with room temperature DI (19 ºC). Particle analysis was performed by measuring LPDs with a KLA-Tencor SP1-TBI.

To directly compare a spin and a Rotagoni dry, particle  addition studies were performed. The blanket silicon wafers  were first processed through a wet bench with an IMEC clean resulting in a hydrophilic wafer surface [18] then transferred  dry to the Verteq Goldfinger. The wafers were then cleaned  with a megasonic deionized water cycle followed by a dry. By carrying out the process in this manner, two conditions were studied. The first is the ability of the dry to remove particles  suspended in the liquid above the wafer as a result of the megasonic cycle. Second, particles in the ambient can provide  an additional source of contamination and this effect was magnified by carrying out the analysis in a class 1000 clean  room. Turbulence created by the higher spin speeds of the spin dry draws particles from the ambient towards the wafer  surface. The results from this analysis are shown in figure three. The net result is that a spin dry added particles to the wafer surface and the Rotagoni dry slightly reduced the  particle count on the wafer surface.