Unexpected Influences on Ultra-Sensitive Bubble Levels
K. P. Trout (1) and Charles A. Gaston (2)
1 Pennsylvania State University York Campus 1031 Edgecomb Avenue York, PA 17403 Telephone: 717 771 4136, e mail: [email protected]
2 Pennsylvania State University York Campus 1031 Edgecomb Avenue York, PA 17403 Telephone: 717 771 4155, e mail: [email protected]
SME numbers: A.6.5; A.8.1; B.1.5; B.3.1; B.6.3; C.3; D.3.3; D.7.10; D.10
Key words: Bubble Level, Spirit Level, Leveling, Wafer Fabrication, Surveying, Machine Installation, Tilt Sensing, Mechanical Effects of Light
ABSTRACT
It is common knowledge that light can produce chemical and electronic changes. However, observable mechanical effects of light are rare. It came to our attention that the bubble in an ultra sensitive level would move toward a flashlight beamed at the level from one end. We hypothesized a number of possible mechanisms and investigated which ones could explain the phenomenon. The investigations described here conclude with a surprising explanation of the phenomenon. The phenomenon is caused by a thermally generated pressure differential between the two ends of the bubble. The bubble shift is large enough that people using these levels in highly sensitive processes, such as the installation of wafer fabrication machines, must be careful in order to avoid inaccurate measurements.
INTRODUCTION
Electronic inclinometers and levels are commonly used in heavy construction grading, ship and barge leveling, deviation surveys, continuous casting, and weapons platform leveling (Jewell Instruments
Website 2007), as well as laser leveling, oil platform leveling, vehicle tilt sensing, headlight leveling, survey instrumentation, crane leveling, monitoring long term movement of buildings, engineering leveling of pipes and beams, and gun sight leveling (Level Developments Limited Website 2007).
Ultra sensitive bubble levels are useful in applications like these, as well as in the installation of sensitive machinery, such as wafer fabrication machinery.
An acquaintance of one of the authors installs wafer fabrication machinery which must be very precisely leveled. He was using a bubble level far more sensitive than an ordinary carpenter’s level.
In a poorly lighted area, he used a flashlight to see the level better, but was amazed to see the bubble move as he watched. When this news reached us, we bought one of the levels so that we could investigate. Simple observations (i.e. getting the bubble tube nearly level and shining a light on it) easily confirmed that the reported phenomenon was real. (See Figure 1.) The mystery: what physical mechanism produces that movement?
Figure 1 - Shining a flashlight on one end of the level shifts the bubble about one division.
INITIAL EXPERIMENTATION
Several different mechanisms were hypothesized to explain the bubble movement:
Light activated changes in surface tension;
Photoelectric effects producing unbalanced charges;
Thermal warping of the metal tray holding the glass tube;
Thermal warping of the glass tube;
Differential expansion between the metal and glass;
Thermal expansion of the glass tube;
Density changes from thermal expansion of the liquid;
Evaporation on one side of the bubble and condensation on the other;
Some mechanism other than any of those listed.
Early efforts sought to analyze which direction the bubble should move, and to quantify, even roughly, how big an effect might be produced by candidate mechanisms. Most of these efforts were frustrated by lack of knowledge about the liquid, the bubble and the interior surface of the glass tube.
If there is any direct effect from light, it isn’t essential. After a series of experiments with various optical filters, a strobe light and a laser, it was discovered that breathing on one end of the level caused the bubble to shift toward that end. Heat obviously was the cause, but there remained at least six possible mechanisms. One thing was clear; the bubble always moved toward the source of energy (if it moved at all; a 5 mw laser pointer produced no visible reaction). Warm breath blown on one end of the level produced dramatic, rapid movement of the bubble and relatively rapid return toward the original position.
Two more of the hypothesized mechanisms were eliminated by separating the glass tube from its metal support tray (after soaking the assembly in acetone for a week). With the metal tray removed, the light sensitivity remained. The effect does not involve the metal tray.
Thermal warping of the glass tube does not seem to be a factor because, with the metal tray removed, light or heat from the side or underneath produced the same effect.
These additional observations now had our hypotheses narrowed to three:
Thermal expansion of the glass tube;
Density changes from thermal expansion of the liquid;
Evaporation on one side of the bubble and condensation on the other; plus, of course, the ones we never thought of:
Some mechanism other than any of those listed.
Eventually we bought a second level to be used for destructive testing. The major goal was to find out what liquid was inside, but by opening it in the right way we also could learn something about the pressure in the bubble.
We clamped the level (with metal tray still attached) in a vertical position and videotaped it as we ground off the bottom tip. (A funnel beneath the level was positioned to direct the draining liquid into a vial.) As the tip was penetrated, the bubble expanded from 6.5 mm to 8.5 mm vertical length. The weight of the liquid would produce an expansion of just 0.4%, so the bubble must have been at a gage pressure of nearly a third of an atmosphere.
A 30% pressure increase could not occur accidentally. Our lab is at an elevation of 540 meters. To get that pressure differential by elevation change, the level assembly operation would need to have been done a kilometer below sea level.
There must be some advantage to manufacturing the levels under pressure. Maybe it just minimizes the evaporation of the liquid while the fill tube is being sealed.
We did not attempt to determine the composition of the bubble, but the liquid smelled like methanol.
It was sent out for analysis; unfortunately, it was handled improperly and allowed to evaporate before the analysis could be done.
The big surprise for us was that after the liquid drained, the tube became cloudy. A small portion of the tube was cut away and examined under optical microscopes. The inner surface had been sandblasted or etched or simply not polished after grinding to shape; it was uniformly rough at a scale near the limits of optical microscopy. Apparently, this etched surface is completely wetted inside the intact level, and therefore the bubble moves smoothly in response to tiny forces. (When we attempted to make a level with smooth glass tubing and water, the bubble would “stick” to dust or imperfections, and would resist moving until the tube was tilted several degrees.)
PRIOR STUDIES AND INDUSTRY CONTACTS
We found no references regarding light or other energy displacing the bubble in a level, nor about the construction of such levels. The French company (Ducourret SA) that manufactures the levels seemed very reticent to share any information. There was a bit of a language barrier, but they implied that the information we sought was proprietary.
There are other companies producing similar levels (even seven times as sensitive!), and we found a contact at a company in England (Level Developments Limited) who was much more receptive – at first. In response to our first inquiry he claimed that they could explain the phenomenon, but the information was rather specialized and not normally given out. He even volunteered that if our efforts could benefit them, they could “help us out”. We wrote back, explaining what we knew and what we hypothesized. We asked a few specific questions, including whether they were aware of any published papers. The response to our second note was rather terse: “I don’t think I can add any more information than you already know about this phenomenon.”
We can speculate that this second company also was concerned about “proprietary” information (and that the initial encouraging response was unauthorized) … or that they really don’t know any more than we know.
EXPERIMENTATION, DATA AND CALCULATIONS
To obtain quantitative data, equipment was designed to hold the level and tilt it by measured amounts
(see Figure 2). The main feature was a lever arm supported at one end by a flexure and at the other end by a micrometer shaft. Knowing the length of the lever arm, we could calculate the angle of tilt produced by any given change in the micrometer support. This apparatus allowed us to initialize the level before any given test, and to calibrate the amount of change indicated as the bubble moved relative to the level‘s inscribed division marks.
Figure 2 - Calibration Apparatus: With a flexure hinge at the right of the long grey lever arm and a micrometer adjustment at the left, this apparatus permits precise centering and cali- bration of the bubble level mounted on it. Also visible in the photo are a camera for recording bubble position, a watch-holder to include time and date in the data pictures, a light meter to measure light source intensity and a cardboard shield to block light from parts of the level.
The bubble tube itself is about 60 mm long and 10 mm in diameter. The walls of the tube are about one mm thick. The bubble inside the tube is about 20 mm long. The glass bubble tube comes glued into a steel tray 79 mm long, with a cross section like the bottom half of an octagon.
One marked division (2 mm of bubble movement) corresponds to 0.006 degrees or about 0.0001 radians of tilt. (Tests with the level on its side or upside down indicate that the tube is not curved; the inside surface is barrel shaped.)
A small incandescent flashlight directed at one end of the level from several centimeters away causes the bubble to shift roughly one division toward the light.
Considering the liquid wrapped around the bubble to represent a U tube manometer (see Figure 3), the pressure and force differentials for a one division shift can be calculated as follows: